technologies

Technologies

Polar Science at the Edge: ERICON Aurora Bore-alis, the European Icebreaker

A. Bergamasco1, L. De Santis2, R. Azzolini3, F. Falcieri11, Institute of Marine Sciences, CNR, Venezia, Italy2, Department of Biological Oceanography, National Institute of Oceanography and Ex-perimental Geophysics, Trieste, Italy3, Polar Research Coordination Unit, CNR Research Area of Roma “Tor Vergata”, Roma,[email protected]

Abstract

The ERICON-AURORA BOREALIS Project represents a comprehensive assem-blage of strategic and coordinating activities to prepare the basis for decision makingby interested European and international authorities and Funding Agencies in rela-tion to developing a world class Research Infrastructure to be operated in the PolarRegions (Arctic and Antarctic). The ERICON-AURORA BOREALIS project willgenerate the Strategic, Legal, Financial and Management Frameworks required forinterested National Governments to commit financial resources to the constructionand running of this proposed unique polar environmental facility. Scientific manage-ment frameworks will be assessed including mechanisms to handle dedicated large-scale multi-year or special mission specific research programs will be discussed inparallel with the operational portfolio and requirements. The paper will focus onmoving the project from the preparatory phase to the construction phase, showingthe ship characteristics and a first portfolio for the italian involvement based on aworkshop discussion between several Italian Institutions and the Aurora Borealisteam.

1 Introduction

Polar regions play an important role in theEarth System. In fact, polar ocean basinsare characterised by large areas, perma-nently or seasonally covered by sea ice,very low temperatures, pronounced sea-sonal changes, bordering prominent con-tinental ice sheets, and in the SouthernHemisphere also huge ice shelves. Theseareas control global climate on a broadrange of time scales. Global ocean circula-tion, sea level change, atmospheric forcingand teleconnections are only some of the

topics involved. Again complex interac-tions between biosphere and hydrosphere-atmosphere-cryosphere and sea ice deter-mine the nature of these unique regions.In every place of the Earth, long repeatedtime series observations are critical for un-derstanding the functioning of the climatesystem, but in the polar regions this impor-tance is critical because both Polar Regionswill remain a challange to operate in, due tosevere ice and weather conditions.The polar oceans are potentially most vul-nerable to present and future global en-vironmental change, because small shifts

mailto:[email protected]

Technologies

may cause irreversible consequences. To-day, even the most sophisticated modellingforecasts, as seen in the IPCC AssessmentReport, are limited by insufficient data inspace and time coverage especially in highlatitudes, and a complete understanding ofthe small scale processes interacting withglobal ones. There is a lack of informa-tion about natural physical and biologicalvariability of the oceans, long term shifts inthe cryosphere and ecosystems due to theextreme technical and logistical efforts in-volved to operate in these extreme environ-ments.Aurora Borealis with its unique year roundoperational capability will allow crucialnew process-oriented studies in polar re-gions. Expeditions can be planned inde-pendently of the weather conditions, drift-ing pack ice even in completely ice coveredwaters. The advanced scientific drilling ca-pability turns Aurora Borealis icebreakerinto an extremely useful platform for sci-entific deep-sea drilling in regions inacces-sible by other conventional drilling plat-forms.

2 The Consortium

European nations have express a substan-tial interest in studying polar environments,their potential, processes and changes. TheEuropean Research Icebreaker ConsortiumAurora Borealis will offer an unique oppor-tunity for European polar and marine sci-entist to attain a leading position for com-ing decades. The vessel will facilitate ex-tended expeditions into remote regions. Inaddition the vessel will be the first platformtechnically able to operate continuously inthese most sensitive parts of the polar re-gions.The european research icebreaker Consor-

tium ERICON [1] is managed by the Euro-pean Science Foundation comprises fifteenpartners from ten European nations and as-sociated countries.The overall aims are to establish strategic,legal, financial and organisational frame-works for this multi-country research facil-ity. The vessel will be operated as a largescale research infrastructure by EuropeanNations and other partner nations. Actu-ally ERICON AB within the EU FP7 willdevelop the framework for joint owner-ship and operation of the vessel (www.eri-aurora-borealis.eu). A scientific manage-ment frameworks will be established tohandle large scale, multi year, mission spe-cific research programs and science & tech-nology co-operations. Figure 1 give the in-terrelations of the workpackages that buildthe ERICON AB, EU FP7 project, a 48months project with end in 2012.

3 Technical Details

The research icebreaker Aurora Borealiswill be the most advanced polar researchvessel in the world, will be a IACS polarclass 1, year-round operations, for all po-lar waters and multi-year ice. In Figure 2(side view) and Figure 3 (deck view) wehave a first rendering from the technical de-sign. It will be a multi-disciplinary ves-sel for all polar and marine research, with120 berthing capacity (science and crew)and 90 days operational endurance. Thenew technological features will include dy-namic positioning in closed sea-ice cover,deployment and operation of remotely op-erated vehicles (ROV) and autonomous un-derwater vehicles (AUV) from the twinmoon pools. Figure 4 give the bottom viewof the vessel, showing the two moon poolsand the propellers. In Figure 5 we can see

2348

Marine research at CNR

Figure 1: ERICON AB EU FP7 Project , workpackages scheme and interrelation.

a vessel section showing the glass domeroof, retractable with the natural light in-side, the direct access to staging hangarand the watertight lower moon pool cov-ers. The vessel will be a powerful ice-breaker with approx. 65,000 tons displace-ment, about 200 m length, 50 m wide andwith 90 Megawatt diesel-electric propul-sion power. It will have high ice perfor-mance to penetrate autonomously, as singleship operation, into the Arctic and Antarc-tic Ocean with 2.5 m of ice cover, duringall seasons of the year. A complete double-hull design, fully redundant engine roomsand safety equipment, full weather protec-tion, combine high ice breaking capabilitywith stable open water performance.A working deck, see Figure 3, is avail-able for geophysics: an easy use for de-ployment of 1-4 seismic arrays via built-

in ramps for access close to waterline.Themost unique feature of the vessel will bethe deep drilling rig, which will enablesampling of the ocean floor and sub-seafrom 50 m depth up to 5000 m, and 1000m penetration. Aurora Borealis will bethe only vessel worldwide that could un-dertake the drilling capability on the longrun in both Polar regions. Figure 6 give theoverview of vessel features.

4 Targets for Polar Re-search

All of the countries involved in the ERI-CON Aurora Borealis project have signedthe Antarctic Treaty and are members, orobservers, of the Arctic Council. Rep-resentatives of these countries are active

2349

Technologies

Figure 2: Aurora Borealis rendering, Side view.

Figure 3: Aurora Borealis rendering, Deck view.

into the SCAR and IASC programmes, thatprovide independent advice in a variety offields, particularly on environmental andconservation matters. All countries in-volved into the ERICON were also mem-bers of EUROPOLAR ERA-NET Consor-tium, participate to many workshops madeat national level with the aim to carry outa deep survey on the European polar as-sets and programmes. These scientific re-search comprise three major topics: a) cli-mate variability studies: scales and indica-tors of polar climate change; b) state andstability of the cryosphere: biodiversityand ecosystems in polar environments; c)

integrated real-time ice-ocean atmosphere-hydrosphere observations and modeling.In particular a first italian nation-wideworkshop identified a science cluster pro-ducing a first stage italian portfolio of pos-sible activities that can be carried out onboard of the future AB icebreaker. A clus-ter of seven main topics was individuated:

1. Geophysics & Geology (Paleoclimate,Bottom current deposits, Slope stability,Sedimentary processes, Fluids as Gashydrates and mud volcanoes).

2. Polar Oceanography & Climate re-lated processes, (air-sea-ice interac-tions, brine rejection, turbulent mixing,

2350


Figure 4: Aurora Borealis rendering, Bottom view.

Figure 5: Moon pool section view.

Deep Ocean Ventilation, Benthic LayerDynamics, Water formation and spread-ing processes, Abyssal Circulation &Ridge constraints, Ice Cavity dynam-ics).

3. Ecology, Biogeochemistry & Physics ofsea ice (Sea-ice formation and structure,nutrient dynamics inside sea-ice, sea-ice ecosystems, biodiversity associatedto the sea ice annual cycle, & relatedphysical processes, role of platelet-ice).

4. Polar Biology & Ecology (Ecology andBiodiversity of deep polar regions, or-ganic matter fluxes and input pathways,

quantification and qualification of or-ganic matter in sediment).

5. PaleoOceanography & PaleoEnviron-ment (carbon burial-sink in the Arc-tic zone, Retrieve records with theEocene/Oligocene transition to checkfor bipolar glaciation, Bipolarity of Cli-mate Change, Ice sheet evolution inAntarctica, Biostratigraphy).

6. Atmospheric physical & Chemistrycluster ( Aerosols, Sea-ice-atmosphereinteraction processes, Radiation andcloud coverage, Satellite validation ac-

2351

Technologies

Figure 6: Aurora Borealis scheme.

tivities, Direct measurements of tracegases in remote areas (CO2, CO, O3,etc).

7. Polar Chemistry & Biogeochemistry ofmicronutrients (Trace elements and Or-ganic micro-pollutants dynamics at thesea-ice system and sea-sediment inter-face, Processes contributing at transport

and dispersion of micro-pollutants inpolar regions, Effects of climate changeon the transport of micro pollutants atglobal scale).

5 AcknoledgementsThis work was realised through the EU FP7grant.

References[1] European Polar Research Icebreaker Consortium. European Polar Research Ice-

breaker Consortium Annex 1° ERICON AB 7th November 2008, Grant Agreementn. 211796. 2008.

2352

Development of a Sediment Washing Techniquefor the Decontamination of Heavy Metals fromDredged Sediments in the Harbour Area ofTaranto

N. Cardellicchio, C. Annicchiarico, M. Buonocore, A. Di Leo, S. Gian-domenico, L. SpadaInstitute for Coastal Marine Environment, CNR, Taranto, [email protected]

Abstract

Among the research activities of the Institute for Coastal Marine Environment -”UOS” Taranto, the study the anthropic impact on marine ecosystems and the de-velopment of innovative technologies for ”environmental remediation” have takenparticular significance in recent years.Attention has been paid to the ecological risk analysis for contaminated sites andsustainable management of dredged sediment from coastal areas. The latter issue isof great relevance for the development of harbour areas, especially in Italy.Concerning dredged and contaminated sediments, the easiest option is often the land-fill disposal, even if the problem is considerably complicated in relation to the vol-umes to be managed and/or treated and the simultaneous presence of different typesof pollutants. In order to reuse the dredged sediments both civil and industrial sector,in recent years the development of treatment technologies textitex situ has assumedconsiderable importance. Among these, the ”sediment washing” appears very inter-esting for the removal of toxic heavy metals from dredged sediments.For this purpose research has been started aiming to developing a new method ofwashing with chelating solutions on samples of sediments in the harbour area ofTaranto, it’s a site remediation of national interest.Different chelating agents (ethylenediaminetetraacetic acid (EDTA), Ethylenediamine-N,N’-disuccinic acid (EDDS), citric acid) were used at different extraction times.The optimum removal time was found to be 48 h, both with EDTA 0.2 M and EDDS0.2M. Citric acid was not very efficient for the contamination process. The resultssuggest that bioavailability and speciation of metals in sediments were the main fac-tors affecting sediment washing. In particular, metal speciation, using sequentialextraction technique, were determined before and after washing: this approach wasuseful to evaluate heavy metals mobility in treated and untreated sediments. Themethodology developed could have important implications on industry, also in viewof reusing treated material for different purposes (filling, cement production, etc.).


Technologies

1 Introduction

Heavy metal contamination of sediments ishazardous to benthic organisms and needsmore attention in order to prevent accumu-lation of these metals into the food chain.The interaction of contaminants with sedi-ments is a very complex phenomenon andmeans are required to understand this mat-ter more fully. In sediments, most of metalcompounds are persistent and toxic: so,sediment removal by dredging or excava-tion is the most frequent cleanup methodin marine environment. After dredging,sediment washing is a relatively low costtreatment option when compared to othertechnologies commonly considered for theremediation. Different chelating solutionshave shown the capability to remove heavymetals from sediments. The advantagesof using chelating agents in washing pro-cess is the high efficiency of metal ex-traction, high thermodynamic stability ofmetal complexes and normally low adsorp-tion of the chelating agents and their met-als complexes on sediment [1, 2]. Theparameters that influence the washing in-clude the type and chelant concentration,pH, Liquid-to-Solid ratio (L/S) and extrac-tion time. For sediment remediation, thevery important parameters are pH, parti-cle size distribution, type of metals to beextracted and their concentration, distribu-tion and physicochemical forms in the sed-iment. The kinetics of metal extraction isalso a crucial parameter as it can affectthe treatment time and cost. The objec-tive of this research has been to evaluatethe performance of particular chelating so-lutions for removal of heavy metals fromsediments taken in a contaminated marinearea as Mar Piccolo in Taranto. Experi-ment for heavy metals removal are showedand discussed. Batch tests of washing

with an aqueous solution of three chelat-ing agent were performed at selected liq-uid/solid (L/S) ratios and different chelantconcentrations. The objective of the re-searches was to investigate metal mobi-lization and sediment decontamination bywashing treatment. In order to determinemetal mobilization, speciation techniquesof metals using sequential extraction tech-nique have been used. Finally, for eachmetal removal efficiencies have been eval-uated. In addition sediments toxicity testswere used for evaluating the ecotoxicolog-ical hazards of contaminated sediments.

2 Materials and methods

The sediment was sampled in the first in-let of the Mar Piccolo in Taranto (Figure1). Sediment was collected with a VanVeen grab sampler in a station mainly con-taminated by heavy metals. The Mar Pic-colo is located in the Northern area of theTaranto gulf; it is a semi-enclosed basin(surface area of 20.72 km2) with lagoonfeatures. The scarce hydrodynamism andthe low water exchange with the nearbyMar Grande, determine, mainly in the sum-mer, a high water stratification. The basinis influenced by urbanization, harbour andaquaculture activities. The main problemsof environmental impact are nine pipes dis-charge sewages, the ship-yard of the Ital-ian Navy, the largest mussel farm and smallrivers and freshwater springs, which drainin the basin the surrounding agriculturalsoils.After collection, sediments were dried, ho-mogenized and stored in a polyethylenecontainer. In sediments the grain sizedistribution, water content, pH, total or-ganic carbon (TOC), metals and chemi-cal speciations of metals were determined.

2354


Figure 1. Sampling area in the Mar Piccolo of Taranto

Grande

Mar Grande

Taranto Gulf

APULIATARANTO

ITALY

Ionian Sea

Mar

II Inlet I Inlet

Sampling station

Figure 1: Sampling area in the Mar Piccolo of Taranto.

All the analyses were performed in tripli-cate. Grain size distribution, water con-tent, and TOC were measured accordingto the methods prescribed by the APAT &ICRAM [3].Metal content was determined by atomicabsorption spectrophotometry, after sam-ple digestion in a microwave oven at 170°Cand 130 psi with 9 ml HNO3 (J.T.BAKER69%-70%) and 3 ml HF (J.T.BAKER60%). Metal concentrations (mg/kg dryweight) before sediment washing are showin Table 1. In this table there are shownalso the limit values for sediment remedia-tion referred by ISPRA (Institute for Envi-ronmental Protection and Research) in ac-cord to directive 2000/60/EC. and Legisla-tive Decree 152/2006.Metal speciation in sediments was esti-mated through extraction procedure ac-cording to Tessier et al. [4]. Washingexperiments were carried out in a singleextraction experiment with EDTA in thesodium salt form, with citric acid (Cit) and

with S,S-isomer of the ethylenediaminedis-uccinic acid ([S,S]-EDDS) in the form ofNa3-N,N′ salt. Dried sediment was treatedwith each chelant compound, shaking from0.5 to 48 h; this time has been selected onthe basis of previous kinetic experiments[5, 6]. Molar concentration of the chelat-ing agents was 0.2 M, corresponding to achelant dosage of 10 mol per mol of to-tal trace metals. Since citrate, EDTA, andEDDS usually form 1:1 complexes withmost heavy metals, including Zn, Cu andPb [7], the theoretical minimum amount ofchelating agent needed to extract the threemetals is equal to the sum of their mo-lar concentrations. In practice, the chelat-ing agent is often applied in several-foldexcess to maximize metal extraction [8].All extractions were carried out at roomtemperature and pH 5. At the end ofthe extraction period, the slush was cen-trifuged at 3000 rpm and supernatant wasfiltered through 0.45-µm filters and thenacidified with HNO3. The residue from

2355

TechnologiesTABLE. 1: Metal concentrations and limit values for remediation in sediments from the Mar Piccolo of Taranto (contaminated site of national interest)

Concentration Parameter Limit Values

mg/Kg d.w. Metal mg/Kg d.w.

2.76 PP Cd 1.0 25.25 PP Hg 0.8

65.03 P Ni 40* 100**

219.8 P Pb 50 172.53 Cu 45

The metals countersigned by Pand PP are respectively the priority pollutants and priority dangerous pollutants (Decision n° 2455/2001/CE European Parliament, November 20th, 2001)* for sediment with pelitic fraction < 20% ** for sediment with pelitic fraction > 20%

Table 1: Metal concentrations and limit values for remediation in sediments from theMar Piccolo of Taranto (contaminated site of national interest).

centrifugation at the end of the extractionexperiments was washed two times with20 ml of distilled water and the wash so-lutions were discarded. The residual sed-iment was dried at 60 °C and thereaftersubjected to the sequential extraction pro-cedure to evaluate metal speciation. Eco-toxicological evaluations were performedwith textitGammarus aequicauda and tex-titCorophium insidiosum. The general de-sign of toxicity tests was based on thestandard guides for acute sediment toxicitytests with marine-estuarine amphipods andisopods, with some modification [9, 10].

3 Results and discussion

The physico-chemical characteristics of thesediment analyzed and used for testing ofdecontamination process were: % water =60.7%, pH = 7.85, TOC= 8.5% dry weight.For grain size distribution according to theAASHO classification system, the sedi-ment was found to be composed of 3.98%gravel, 4.77% sand, 91.25% silt and clay,

thus belonging to the pelitic fraction.Ecotoxicological results showed that mor-tality of the organisms used was veryhigh compared with the control sediment(68.3% for textitG. aequicauda, 34.6% and40% for textitC. insidiosum) (p < 0.05)However, it should be emphasized that themortality of organisms used in biologicaltests could be caused by other types of pol-lutants, not just metals.Analysis of sediment washing kinetics pro-duced useful information on the relation-ship between metal extraction yield andcontact time. Figure 2 shows the efficiencyof washing treatment in terms of percent-age of extracted metals as a function ofcontact time (2, 24 and 48 h). Wide vari-ations in metal removal efficiency were ob-served depending on both the metal underconcern and the chelating agent used.Among the investigated metals, Ni and Vwere the most difficult to extract from thesediment matrix after 2 h of washing treat-ment, as confirmed by kinetics curves (Fig-ures 3,4). According to the Figure 2a theremediation efficiency was 32.9% for Hg

2356


Figure 2. Percentage of metals solubilised by chemical extraction after a) 2 hours, b) 24 hours and

c) 48 hours

Figure 2: Percentage of metals solubilised by chemical extraction after a) 2 hours, b) 24hours and c) 48 hours.

(EDTA), 10.2% for Ni (EDTA) and 6.1%for V (Cit). After 24 h of washing (Fig-ure 2b) this percentage increases: 58.4%for Hg (EDDS), 71.8% for Ni (EDDS) and61.4% for V (EDDS). After 48 h washing,Pb was removed with maximum efficien-cies (up the about 95.0 % with EDDS and89.3% with EDTA). Concerning Cu it wasremoved until 87.1% with EDDS and until95.5% with EDTA. After 48 h washing, theremoval efficiency increased for all metals.For Pb, the good complexation efficiencywith EDDS and EDTA was demonstratedin various studies [11]. In this work, Cudisplayed by far a slowest extraction ki-

netics. It has been proposed [12] that theslower extraction rate may be due to thefact that Cu is usually associated to organicmatter or other oxidizable species in soiland sediment. Therefore, the slower ex-traction kinetics of the copper may be ex-plained by the formation of a low percent-age of Cu complexes during washing treat-ment. Figures 5 and 6 show the differentextraction kinetic diagrams for Cu and Pbrespectively while Figures 7 and 8 show thespeciation analysis of Cu and Pb in the un-treated sediment (U.S.) and after extractionwith different chelating solutions. In theuntreated sediment, Pb and Cu are associ-

2357

Technologies

K inetic Ni

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

u.s . 0.5 2 4 6 24 48

contac t time

mg/l

E DDS

E DTA

C it

Figure 3. Nickel concentration with time from 0.5 to 48 h in chelating solution (U.S. = untreated sample)

Figure 3: Nickel concentration with time from 0.5 to 48 h in chelating solution (U.S.=untreated sample).

ated to the Fe and Mn oxides fraction (59-55 %).Concerning Cd, removal was similar fortwo chelating agents, EDTA, EDDS with60-75 % removal efficiency after 48 h; lowdecontamination was obtained after 48 husing citric acid. For Cd a reduction in re-moval percentage was observed after a timegreater than 48 h: in fact, after 24 h of treat-ment with EDTA, the removal efficiencywas 95 %, but after 48 h of treatment the re-moval efficiency was 63 % (Figures 2b and2c). The reduction in extraction efficiencyafter a long contact time with the chelat-ing agents, appeared important and maybe attributed either to sorption mechanismsof the newly formed metal-chelant com-plexes onto the active sites of sediment par-ticles, or metal exchange reactions betweenmetal-chelant complexes. Metal speciationanalysis allowed to make some importantconsiderations: the result displays that thechelating agents were capable of extract-ing, in addition to the easily mobile frac-tion (exchangeable and carbonate bound),part of other less mobile forms (i.e., Fe ox-

ides and sulphides). In some cases, an in-crease in the amount of Cu and Pb associ-ated to fraction A (exchangeable) was ob-served (Figures 7 and 8), most probably asan effect of the presence of residual metalcomplexes [13]. The amount of Cu and Pbassociated to fraction D (bound to organicmatter) was in many cases found to in-crease as a result of treatment. The reasonsmay be that, in addition to the inability ofchelating solutions to solubilize such frac-tions, re-adsorption of chelant metal com-plexes or metal ions onto organic matterfraction of the material, may have played arole in the bioavailability and remediation.As regards the methods of decontamina-tion with chelating agents, metal concen-trations in sediments were in some casesbelow the limits of DM 367/2003, mak-ing the treated sediment theoretically beused as beach nourishment. However, theextraction capacity towards various metalsdepends on the form in which the metal ispresent and its relationship with the miner-als that make up the sediment, ie, the geo-chemistry of the sediment itself. As a good

2358


K inetic V

0.00

0.20

0.40

0.60

0.80

1.00

u.s . 0.5 2 4 6 24 48

contact time

mg/l

E DDS

E DTA

C it

Figure 4. Vanadium concentration with time from 0.5 to 48 h in chelating solution (U.S. = untreated sample)

Figure 4: Vanadium concentration with time from 0.5 to 48 h in chelating solution (U.S.=untreated sample).

extraction efficiency was obtained for con-centrations of 0.2 M chelating, correspond-ing to a chelating dosage of 10 mol permol of total trace metals, it follows that theamount by weight of extractant are consid-erable, thus presenting problems of cost.

4 ConclusionsThe batch washing treatment of 48 h,EDTA and EDDS were effective chelatingsolutions for removing heavy metals fromsediment analysed. The concentration ofchelants (0.2 M) and the experimental con-ditions (pH 5) were found to be useful forachieving high extraction yields. The timeof treatment with washing solution of 48hours gives the best results in the metal ex-traction. Ni, Cu and V are extracted withhigh efficiency by EDDS and EDTA. Thebest extractant for Hg and Cd was found

to be the EDDS. For Pb, all three chelantsshow a high capacity for decontamination.Citric acid did not provide satisfactory re-sults so it is not recommended for remedi-ation of sediments. The possibility of ap-plication of the method of sediment wash-ing with chelating agents, such as remedia-tion process of sediment being dredged, ishighly dependent on specific properties ofthe matrices in terms of particle size char-acteristics, mineralogical composition, na-ture, and distribution of contamination lev-els between different fractions. The influ-ence of macro-elements (Ca, Mg, Fe) affectalso the efficiency of extraction. In con-clusion, the method appear good to reme-diation of ex situ contaminated sediment inharbour areas. The studies are in progressfor optimize a mixture of chelating agentsin order to improve the efficiency of extrac-tion.

2359

Technologies

K inetic C u

0.00

1.00

2.00

3.00

4.00

5.00

u.s . 0.5 2 4 6 24 48

C ontact time

mg/l

E DDS

E DTA

C it

Figure 5. Copper concentration with time from 0.5 to 48 h in chelating solution (U.S. = untreated sample)

Figure 5: Copper concentration with time from 0.5 to 48 h in chelating solution (U.S.=untreated sample).

K inetic Pb

012345678

u.s . 0.5 2 4 6 24 48

contac t time

mg/l

E DDS

E DTA

C it

Figure 6. Lead concentration with time from 0.5 to 48 h in chelating solution (U.S. = untreated sample)

Figure 6: Lead concentration with time from 0.5 to 48 h in chelating solution (U.S.=untreated sample).

2360


C u s pec iation

0

20

40

60

80

100

U.S . E DDS E DTA C it

spec

ies distibution Res idual

Org matter+s ulphides

F e and Mn ox

C arbonate

E xchangeable

Figure 7. Copper speciation before and after washing (U.S. = untreated sample)

Figure 7: Copper speciation before and after washing (U.S.= untreated sample).

Pb s pec iation

0

20

40

60

80

100

U.S . E DDS E DTA C it

spec

ies distribution Res idual

Org matter+s ulphides

F e and Mn ox

C arbonate

E xchangeable

Figure 8. Lead speciation before and after washing (U.S. = untreated sample)

Figure 8: Lead speciation before and after washing (U.S. = untreated sample).

2361

Technologies

References[1] R.J. Abumaizar and E.H. Smith. Heavy metal contaminants removal by soil wash-

ing. J. Haz. Mater., 66:151–21, 1999.

[2] G.W. Gee and Y.W. Bauder. Particle-Size Analysis In: Methods of soil Analysis.pages 383–411., 1986.

[3] APAT & ICRAM. Manuale per la movimentazione dei sedimenti marini, Ministerodella Ambiente e della Tutela del Territorio e del Mare. 2007.

[4] A. Tessier, P.G.L. Cambell, and M. Bisson. Sequential Procedure for the Speciationof Particulate Trace Metals. Anal. Chem., 51:844–850, 1979.

[5] D. Cermigna, A. Polettini, R. Pomi, E. Rolle, L. De Propris, M. Gabellini, andA. Tornato. Comparing sediment washing yelds using traditional and innovativebiodegradable chelating agents. pages 24–27, 2005.

[6] A. Polettini, R. Pomi, E. Rolle, D. Cermigna, L. De Propris, M. Gabellini, andA. Tornato. A Kinetic study of chelant-assisted remediation of contaminateddredged sediment. J. Hazard Mater, 137:1458–1465, 2006.

[7] A.J. Francis, C. J. Dodge, and J. B. Gillow. Biodegradation of metal citrate com-plexes and implications for toxic-metal mobility. Nature, 356:140–142, 1992.

[8] J.H. Linn and H.A. Elliott. Mobilisation of Cu and Zn in contaminated soil by NTA.Water, Air, Soil Pollut., 37:449–458, 1988.

[9] ASTM. Standard guide for conducting 10-d static sediment toxicity tests with ma-rine and estuarine amphipods. 1993.

[10] SETAC Europe. Guidance document on sediment toxicity assessment for freshwa-ter and marine environments. 1993.

[11] C.N. Neale, R.M., Bricka, and A.C. Chao. Evaluating acids and chelating agentsfor removing heavy metals from contaminated soils. Environ. Prog, 16:274–280,1997.

[12] P. Vandevivere, F. Hammes, W. Verstraete, T. Feijtel, and D. Schowanek. Metaldecontamination of soil, sediment, and sewage sludge by means of transition metalchelant [S,S]-EDDS. J. Environ. Eng, 127:802–811, 2001.

[13] B. Sun, F. Zhao, E. Lombi, and S.P. McGrath. Leaching of heavy metals fromcontaminated soil using EDTA. Environ. Pollut, 113:111–120, 2001.

2362

Use of Hydroacoustic Methods for the Identifica-tion of Potential Seabed Habitats for Small PelagicFish Schools in the Strait of Sicily

B. Patti, M. D’Elia1, A. Sulli2, G. Tranchida1, A. Bonanno1, G. Basilone1,G. Giacalone1, I. Fontana1, S. Aronica1, G. Buscaino1, L. Caruana1

1, Institute for Coastal Marine Environment, CNR, Capo Granitola (TP), Italy2, Department of Geology and Geodesy, University of Palermo, Palermo, [email protected]

Abstract

Contemporary data on both fish distribution and seabed characteristics can beused in ecological studies to investigate the distribution and the behaviour of bothdemersal and pelagic fish schools in relation to the nature of the sea bottom. In thisstudy we describe a method to classify the sea bottom by a single-beam echosounderused during an echosurvey carried out on the north-western part of Southern Sicil-ian continental shelf. The study area was divided in two contiguous regions (la-belled ZONE 1 and ZONE 2), characterized by different dominant texture, ‘sand’for ZONE 1 and ‘clayey-silt’ for ZONE 2. The acoustic classification evidenced dif-ferences in terms of bottom types between the two investigated zones. Though theaverage bottom depth of the investigated transects is higher in ZONE 1, this area isdominated by substrates having greater backscatter, identifying relatively “harder”bottom types. On the contrary, in ZONE 2 “hard” substrate is confined to the shal-lower inter-transects regions, with the bulk of seabed deeper than 50 m classified as“soft” bottom. The acoustic classification of the seabed was already used in the con-text of a recent study, indicating a preference of small pelagic fish schools for softbottom in ZONE 2, where their occurrence was higher and the bulk of total biomasswas concentrated.

1 Introduction

A variety of remote hydroacoustic tech-nologies, which employs a single or amulti-echo, are used to analyze and mapseabed characters in term of texture andgrain size [1, 2]. The most widespreadcommercial acoustic bottom classificationsystems are RoxAnn and QTC-View. Theyuse normal incident echosounders and arebased on the measurement, analysis and in-terpretation of the shape and energy fea-

tures contained in the bottom acoustic sig-nal [3, 4]. The knowledge of the bottommaterial and topography is very importantin different scientific fields such as geol-ogy, offshore industry and fishery, mainlywhen the activity is focused on demersalfish. Over the last few decades there wasa rapid increase in the use of hydroacoustictechnologies also for schooling pelagic fishand for the biomass evaluation of pelagicfish populations [5, 6]. Echosounders com-monly used to this aim work at fixed fre-


Technologies

quencies and are able to acquire informa-tion from both the fish distribution alongthe water column and the bottom seabed.Despite this, information from the sea bot-tom is usually discarded because the acous-tic signal is used exclusively to obtain dataabout pelagic fish biomass and the struc-ture of the fish schools, to the aims ofthe sustainable management of fisheriesresources. However, the consensus thatsingle species stock assessment alone isnot sufficient to manage fisheries sustain-ably has been constantly increasing overthe last two decades. Specifically, it hasbeen argued the importance of the so-calledecosystem approach for the effective andsustainable management of fish popula-tions [7, 8, 9, 10, 11]. For instance, thequality of scientific advice on exploited fishpopulations can be significantly improvedby the knowledge of the effects of environ-mental conditions on the recruitment or fishpopulations. In this context, the contempo-rary collection of information on fish dis-tribution and seabed characteristics couldbe greatly useful in multidisciplinary stud-ies. To this aim, the use of a single beamechosounder was firstly investigated byD’Elia et al. [12], who also demonstratedthe preference of small pelagic fish schoolsfor softer (and finer) subtrates. In thisstudy we describe the rationale of methodtherein used to classify the sea bottom bya single-beam echosounder (SIMRAD EK500), based on the comparison between hy-droacoustic bottom data and granulometricanalysis of sediment samples.


2.1 Study area and acoustic sur-vey

Bottom acoustic information were ob-tained from hydro-acoustic survey con-ducted over the continental shelf of theSicily Strait aboard the N/O Salvatore LoBianco in June 1998. The surveyed areawas divided in two sectors, labeled ‘ZONE1’ (north-western sector, over the Adven-ture Bank) and ‘ZONE 2’ (south-easternsector, continental shelf off the central partof the Southern Sicilian coast) (Figure 1).These two sectors are characterized by dif-ferent dominant seabed conditions, harder(and coarser) in ZONE 1, softer (and finer)in ZONE 2.

2.2 Backscattering signal by theseabed and bottom samplescollection

The split-beam echosounder SIMRAD EK500, originally designed for biological re-sources, was used to measure the backscat-tering signal from the seabed, using a trans-ducer operating at the frequency of 38 kHz.The choice of this frequency is linked toits more frequent availability due to itscommon use in acoustic surveys aimed atbiomass evaluations of small pelagic fishspecies. Raw data from the bottom (Sv)were analyzed using the post –processingsoftware Echoview v.3 by Sonar Data andMatlab scripts. A GPS system, interfacedto the echosounder, was used to recordthe position of the bottom signal. Themethod of extracting bottom type informa-tion is based on the following considera-tions. First, there is a direct relationshipbetween the mean diameter of the seabedmaterial and bottom surface (and volume)

2364


Figure 1: Study area, with Zone 1 (western sector) and Zone 2 (eastern sector) evidenced.Along the transects the acoustic classification of seabed based on bottom Sv is shown.“Hard” seabeds (Sv > median value) is displayed in dark grey, “soft” seabeds (Sv≤me-dian value) in light grey. The sediment sample sites are also shown with white. Shepard’sand Folk’s sediment classification labels are also given under the symbols.

backscattering value for normal incidentechosounder [13]. It means that to a largergrain size corresponds a greater backscat-tering surface or volume coefficient. Thereason for this is related to the bulk den-sity of the sediment, which mainly de-pends on porosity. Clayey-silt and Silty-clay sediments have a higher porosity val-ues than sand [14], so they bind morewater and have lower density than sandysediments. The bulk density is in rela-tion to the acoustic impedance of a sed-iment: the higher sediment density, thehigher acoustic impedance and the greaterbackscattering coefficient [15, 16]. Theintensity of the energy scattered back tothe instrument is also a function of the

bottom roughness, due to the grain sizeof the particles. Gravel is rougher thansand which in turn is rougher than silt andclay [17]. Starting from this considera-tion, for each acoustic transect of Figure1 the Sv volume backscattering values ofthe seabed line were extracted by meansof Echoview software, using the maximumSv algorithm. Then, a moving average to50 terms (= acoustic pings), correspond-ing to about 200–250 m, was calculatedto reduce variability in the Sv bottom val-ues. The median value of these averagedSv bottom data was used to classify the bot-tom type (0 for Sv bottom values ≤ me-dian value, identifying ‘soft’ seabed areas,and 1 for Sv bottom value > median value,

2365

Technologies

Figure 2: Shepard’s (a) and Folk’s (b) diagram, showing the classification into grain sizefor the analyzed samples.

identifying ‘hard’ seabed areas). In or-der to verify the results, physical samplingof the bottom was accomplished with graband box-corer, in proximity of the acousticprospected transects and a grain size anal-ysis was conducted on the collected sam-ples. The analysis of the particle size ofsediment samples required a phase of sam-ple pre-treatment, which included diges-tion with hydrogen peroxide, washing, sep-aration of particles with a diameter lowerthan 500 µm and drying and subsequentanalysis by a laser diffraction instrument(Fritsch model Analysette 22). The coarsesand-gravel fraction (> 500 µm) was driedand divided in 2 fractions using a sieve witha 2 mm mesh size. Finally, the sieved parti-cles were weighted by a simple balance. Afinal normalization process of the fraction,analyzed by the laser instrument, was nec-essary in order to obtain the correct com-positional data. The analysis of particlesize distribution were conducted using the

Wentworth scale, and the classification intograin size class was based on the Shep-ard’s method [18] for samples containingsilt, clay and sand. The Folk’s method [19]was used for the classification into grainsize of sediment samples containing a frac-tion coarser than sand (diameter>2mm, lit-tle pebble or bioclastic granules).

3 Results and discussionThe results of the acoustic-based bottomtype classification are displayed in Fig-ure 1, together with information obtainedfrom the analysis of sediment samples col-lected over all the investigated area. Theacoustic classification shows the differ-ences in terms of bottom types between thetwo zones along the investigated transects.ZONE 1 seabeds, at depth less than 100m (the western sector, over the AdventureBank) are dominated by substrates withgreater scattering strength, indicating rela-

2366


Figure 3: Box plots of bottom Sv for the sediment facies named “Silt-clay” (includingsediment samples classified as clayey-silt) and “Sand” (including sediment samples clas-sified as sandy-silt, silty-sand and sand).

tively ‘harder’ (and coarser) bottom types.In contrast, in ZONE 2 the ‘hard’ substrateis confined to the shallower inter-transectregions, with the bulk of the seabed deeperthan 50 m classified as ‘soft’ bottom. Thegranulometric analysis of sediment sam-ples is consistent with the results of theacoustic classification of the seabed (Fig-ure 2 and Table 1). Although the sedimentsample sites were available only in proxim-ity of the acoustically prospected transects,it is worth noting that the use of the me-dian value (-17.65 dB re m-1) of bottombackscattering volume strength permittedus to approximately separate finer (andsofter) substrate locations (classified as

clayey-silt and silty-clay), correspondingat bottom Sv values lower than the globalmedian value, from coarser (and harder)sediment sites (facies: silty-sand, sandy-silt, and sand) matching with higher bot-tom Sv values (Mann–Whitney U-test, p =0.0055; Figure 3). This paper highlightssome important aspect in the application ofechosounder data for seabed characteriza-tion. If the fisheries oriented instrumentsuch as SIMRAD EK 500 echosounder iscontemporarily used both for fisheries dataand for bottom classification, its use in eco-logical studies could be very efficient andcost saving.

2367

Technologies

Table 1: Compositional data from the granulometric analysis of the collected samples.

References[1] R. Freitas, S. Silva, V. Quintino, A.M. Rodrigues, et al. Acoustic seabed classifica-

tion of marine habitats: studies in the western coastal-shelf area of Portugal. ICESJournal of Marine Science, 60:599–608, 2003.

[2] J.T. Anderson, D.V. Holliday, R. Kloser, D.G. Reid, and Y. Simard. Acoustic seabedclassification: current practice and future directions. ICES Journal of Marine Sci-ence, pages 1004–1011, 2008.

[3] B.T. Prager, D.A. Caughey, and R.H. Poeckert. Bottom classification: operationalresults from QTC view. OCEANS ’95 - Challenges of our changing global envi-ronment conference, San Diego, California, 1995.

[4] S.P.R. Greenstreet, I.D. Tuck, G.N. Grewar, and E. Armstrong. An assessmentof the acoustic survey technique, RoxAnn, as a means of mapping seabed habitat.ICES J. Mar. Sci., (54):939–959, 1997.

[5] R. Muino, P. Carrera, and M. Iglesias. The characterization of sardine (Sardinapilchardus Walbaum) schools off the Spanish-Atlantic coast. ICES Journal of Ma-rine Science, (60):1361–1372, 2003.

[6] B. Patti, A. Bonanno, G. Basilone, Goncharov S., S. Mazzola, et al. Interannualfluctuations in acoustic biomass estimates and in landings of small pelagic fish pop-ulations in relation to hydrology in the Strait of Sicily. Chem. Ecol., (20):365–375,2004.

[7] P.A Larkin. Concepts and issues in marine ecosystem management. Reviews inFish Biology and Fisheries, (6):139–164, 1996.

2368


[8] S. Jennings, M.J. Kaiser, and J.D. Reynolds. Marine Fisheries Ecology. BlackwellScience, 2001.

[9] S.M. Garcia, A. Zerbi, C. Aliaume, T. Do Chi, and G. Lasserre. The ecosystemapproach to fisheries. Issues, terminology, principles, institutional foundations, im-plementation and outlook. FAO Fisheries Technical Paper, (443), 2003.

[10] E. Pikitch, C. Santora, E.A. Babcock, A. Bakun, et al. Ecosystem-based fisherymanagement. Science, (305):346–347, 2004.

[11] G.H. Kruse, Y. Ishida, and C.I. Zhang. Rebuilding of depleted fish stocks throughan ecosystem approach to fisheries. Fisheries Research, (100):1–5, 2009.

[12] M. D’Elia, Patti B., A. Sulli, G. Tranchida, A. Bonanno, et al. Distribution andspatial structure of pelagic fish schools in relation to the nature of the seabed in theSicily Channel (Central Mediterranean). Marine Ecology, (30):151–160, 2009.

[13] M. Manik, H.M.and Furusawa and Amakasu K. Measurement of the sea bottomsurface backscattering strength by quantitative echo sounder. Fisheries Science,(72):503–512, 2006.

[14] A.W. Nolle, W.A. Hoyer, J.F. Mifsud, W.R. Runyan, and M.B. Ward. Acousti-cal properties of water filled sands. Journal of the Acoustical Society of America,(35):1394–1408, 1963.

[15] E.L. Hamilton and T.B. Richard. Sound velocity and related properties of marinesediments. Journal of the Acoustical Society of America, (72):1891–1904, 1982.

[16] M.D. Richardson. In-situ, shallow water sediment geoacoustic properties. Proceed-ings International Conference on Shallow-Water Acoustics, Beijing, China, 21–25April. China Ocean Press, page 163–170, 1997.

[17] C. Mazel. Side Scan Sonar Record Interpretation. Training Manual. Klein Assoc.,pages 14–16, 1985.

[18] F.P. Shepard. Nomenclature based on sand-silt-clay ratios. Journal of SedimentaryPetrology, (24):151–158, 1954.

[19] R.L. Folk. The distinction between grain size and mineral composition in sedimen-tary rocks nomenclature. Jour. Geol., 62:344–359, 1954.

2369

Technologies

2370

Benthic Chamber for Metabolic Measurements ofUnderwater Flora: a New Realization

G. Fasano, A. Materassi, F. BenincasaInstitute for Biometeorology, CNR, Sassari, [email protected]

Abstract

Benthic chambers are widely used in studies concerning both marine sedimentsand flow of solutes through them up to high depth. Benthic chambers were usedalso for studies on metabolism of coral reefs and vegetable biomasses; these studiescan be carried out for use up to a few meters of depth because the data acquisitioninstruments can be placed above the water surface. Difficulties arise when thesemeasurements have to be made to depths of several tens of meters. In these cases,the instruments should be waterproof for pressures up to several bar. For a study, incollaboration with the Department of Biology - University of Pisa, a benthic chamberwas designed and built. The chamber is able to carry out metabolic measurementsof underwater flora at depths between 0 and 50 m. The chamber is large enough tocontain several plants and provides a reasonable compromise with chamber handling.Parameters which can be measured are underwater solar radiation (global, red, green,blue), depth, temperature, pH, conductivity, dissolved oxygen. In the Chamber thereare two operation phases: measurement of water physical-chemical parameters, al-tered by biological activity of plants and a water exchange with external water. Apump system avoids the water stratification and allows the water renewal. Resultsof several tests confirm full compliance of the benthic chamber with respect to therequirements of marine biologists and to the project data.

1 Introduction

Benthic chambers are widely used instudies concerning both marine sedimentsand the flow of solutes through them(in presence or absence of subterraneanmegafauna) up to a depth of several thou-sands of meters [1, 2, 3, 4]. So, ben-thic chamber technique was used to es-tablish the chemical reactions that occurin the bottom water and the top layer ofthe sediments [5]. Benthic chambers wereused also for studies on metabolism (pri-mary production, respiration, etc.) of coralreefs and vegetable biomasses [6, 7]. Thesestudies concerned the air-water interface

or, at most, a few meters below the sea sur-face [8, 9]. Structures carried out for thefirst type of studies are elementary, usuallysmall, and use a few simple instruments ofdetection (individual sensors, or assembledin multiparameter probes, samplers, etc.)that do not require support and manage-ment hardware and software [10]. Struc-tures to conduct studies of the second kindare, usually, more complex. These, how-ever, can be easily carried out for use up toa few meters of depth because the instru-ments (for data acquisition, managementand power supplies) can be placed abovethe water surface. Technological difficul-


Technologies

ties arise when these measurements have tobe made to depths of several tens of meters(10 to 50 m). In these cases, the instru-ments should be waterproof for pressuresup to several bar (1 ÷ 5).

2 Realization

For a study, funded by the Ministry of Ed-ucation, University and Research (VEC-TOR Project Line 7: VulnErability of theItalian coastal area and marine Ecosys-tems to Climatic changes and Their rOlein the Mediterranean caRbon cycles), incollaboration with the Department of Bi-ology of the University of Pisa, a plex-iglas benthic chamber was designed andbuilt. The chamber is able to carryout metabolic measurements of underwaterflora at depths between zero and 50 meters.The chamber is large enough (diameter 900mm, height 1000 mm, volume 0.5m3) tocontain several plants and provides a rea-sonable compromise with chamber han-dling. The instrument, by monitoring phys-ical parameters related to metabolic activ-ity of plants, makes it possible to deducethe “health” of marine waters in the eu-photic zone. The parameters which can bemeasured are:• solar radiation that reaches the chamber

in the ranges: global (400 ÷ 1100nm),red (500 ÷ 700nm), green (480 ÷600nm), blue (450 ÷ 540nm). The in-strument, a SuMaRad (SubMarine Ra-diometer) [11], indicates the quantities asa percentage of the corresponding solarradiation at the sea surface;

• and, by means of a multiparameter probe[12]:

- depth: 0 ÷ 61m ± 0.12m, resolution0.001m;

- temperature: -5 ÷ 45°C ± 0.15°C,resolution 0.01°C;

- pH: 0 ÷ 14 ± 0.2, resolution 0.01;- conductivity: 0 ÷ 100mS·cm−1 ±

0.5 % of reading, resolution 0.1mS·cm−1;

- dissolved oxygen (in two ranges): 0÷ 20mg·l−1 ± 0.1mg·l−1 or 1 %, ofreading, (whichever is greater), 20 ÷50mg·l−1 ± 15 % of reading resolu-tion (in both ranges) 0.01mg·l−1.

Figure 1 shows the benthic chamber, onits trolley, with its measuring instruments:A) SuMaRad, B) multiparameter probe, C)control electronics with battery pack, D)internal circulating pump. In the picture,the charging pump which replaces the wa-ter inside the chamber is not visible. Figure2 shows the front panel of the control elec-tronics within the waterproof case (markedC in Figure 1), which also contains the bat-tery pack. In the picture, A is the magnetickey that allows the programming of thecontrol system, B is the keyboard for en-tering commands, C is the display, D is theconnector for charging batteries. The con-trol electronics is based on an Atmel mi-crocontroller ATMega168 and was devel-oped using Arduino [13], an open-sourceelectronics prototyping platform. Figure3 shows the block diagram of the controlunit.The control unit allows two modes of pumpoperation, selectable by the user (Figure 4).Alternating mode. The two pumps can-not operate simultaneously and the time inwhich a pump is working corresponds tothe interval time when the other is off. Theworking phase of the pump can alternateperiods of activity and inactivity, withoutaffecting the off pump. The user can setthe ratio (in %) between periods of activityand inactivity (duty cycle) during the work-

2372


Figure 1: Benthic chamber on its trolley with annexed measuring instruments: A)SuMaRad (SubMarine Radiometer), B) Multiparametric probe, C) Control electronicswith battery pack, D) Internal circulating pump. The charging pump is not visible in thispicture.

ing time.Independent mode. The pumps operatein a completely independent mode and foreach one the user can set: on time, off timeand the delay to their starting after powerup of the control unit. Each pump alsocan always be set on or off. The controlunit generates a signal whose amplitude de-pends on the working state of the pumps.This signal can be sent to a separate inputof the multiparameter probe, thus allow-ing monitoring of the pumps. These pumpsare specific for oceanographic applicationsand are centrifugal type with magneticallycoupled impeller and adjustable flow rate[14]. Flow rate was set at 7.5 l·min−1

to optimize electrical power consumption.

In alternating mode, there are two opera-tion phases in the Chamber: measurementof water physical-chemical parameters, al-tered by biological activity of plants (cir-culating pump on); water exchange: waterinside the chamber is replaced by externalwater to set new initial conditions (charg-ing pump on). The power supply is pro-vided by a pack of sealed lead acid batter-ies (36 V, 8 Ah). The battery life is morethan 30 h (in alternating mode: circulatingpump 4 h, 50 % duty cycle, charging pump1 h, duty cycle 100 %). Diving and sur-facing operations are facilitated by a valvesystem that allows, in the first case, thecomplete expulsion of air and, in the sec-ond case, water discharge.

2373

Technologies

Figure 2: Front panel of the electronics control within the waterproof case (C in Figure1), which also contains the battery pack. A) magnetic key for programming the con-trol system, B) keyboard for entering commands, C) display, D) connector for chargingbatteries.

Figure 3: Block diagram of the management system of the benthic chamber. The Blockdiagram of the control unit is within the dashed rectangle.

2374


Figure 4: Scheme for the two modes of operation: on the left, Alternating mode: theon and off sequence of the circulation pump, programmed with a duty cycle of 50 %,is visible. The charging pump is running continuously during its operation phase (dutycycle of 100 %). On the right, Independent mode: the two pumps can be both on or off atthe same time, and for each one the on time, off time and the delay to their starting afterpower up of the control unit can be set. Each pump also can be set always on or alwaysoff.

Figure 5: A test to verify the functionality of the Chamber was conducted in the CalaVillamarina (Santo Stefano Island, Archipelago of Maddalena, northern Sardinia) on aPosidonia Oceanica meadow at a depth of 9 m in July 2009.

2375

Technologies

Figure 6: A typical July day: percentage transmittance of the water, the ratio betweenunderwater radiation and outside water radiation in the indicated spectral bands (left or-dinate). Dashed line (right ordinate) shows the plot vs. time of the water temperature.

Figure 7: A typical July day: the plot vs. time of dissolved oxygen (upper line, left ordi-nate) and pH (lower line, right ordinate). Rectangles indicate the time intervals when thewater inside the benthic chamber was replaced with external water.

2376


3 Results and ConclusionsIn order to verify Chamber functionality,a test was conducted in Cala Villama-rina (Santo Stefano Island, Archipelago ofMaddalena, northern Sardinia) on a Posi-donia Oceanica meadow at a depth of 9 min July 2009 (Figure 5). Figure 6 showsas an example the plot vs. time of thepercent transmittance of the water, the ra-tio between underwater radiation and out-side water radiation in the indicated spec-tral bands (left ordinate). The dashed line(right ordinate) shows the plot vs. time ofthe water temperature. Figure 7 illustratesthe plot vs. time of two other measuredparameters: dissolved oxygen (upper line,left ordinate) and pH (lower line, right or-dinate). Rectangles indicate the time in-tervals (adjustable from 10 minutes to 24hours) in which the water inside the benthicchamber was replaced with external water.When the charging pump is off the circu-

lating pump is activated to prevent strati-fication of the water in the chamber. It ispossible to observe how, at night, plant res-piration reduces dissolved oxygen in seawater and gives off carbon dioxide, low-ering the pH (in Figure 7, approximatelyfrom 8:00 pm to 5:30 am); vice versa withthe return of sunlight, plants absorb car-bon dioxide (pH increases) and emit oxy-gen. The measurement cycle is repeatedwith each water renewal. The acquireddata are stored in internal memory of themultiparameter probe and of the SuMaRad.The results obtained confirm full compli-ance of the benthic chamber with respect tothe requirements of marine biologists andto the project data. This benthic chamberis one of the most versatile for measuringthe metabolic activity of marine flora in theeuphotic zone due to its size, instrumentalequipment, automatic operation, operativeautonomy, and low construction cost.

References[1] W. Dickinson and F.L. Sayles. A benthic chamber with electric stirrer mixing.

Technical Report of WHOI, Massachusetts USA, 1992.

[2] D.J. Hughes, R.J.A. Atkinson, and A.D. Ansell. A field test of the effects ofmegafaunal burrows on benthic chamber measurements of sediment-water solutefluxes. Marine Ecol. Prog. Ser. - Wiley-Blackwell, 195:189 – 199, 2000.

[3] A.P. Webb and B.D. Eyre. The effects of two benthic chamber stirring systems onthe diffusive boundary layer, oxygen flux, and passive flow through model macro-fauna burrows. Eustuaries, 27(2):353–362, 2004.

[4] S. Sommer, M. Turk, S. Kriwanek, and O. Pfannkuche. Gas exchange systemfor extended in situ benthic chamber fluxe measurements under controlled oxy-gen conditions: first application - sea bed methane emission measurements at Cap-tain Arutyunov mud volcano. American Society of Limnology and Oceanographic:methods, 6:26 – 33, 2008.

[5] D. Dyrssen, P. Hall, and S. Westerlund. Benthic chamber chemistry. Anal. Chem.,317:380–382, 1984.

2377

Technologies

[6] L.B. Cahoon and J.E. Cooke. Benthic microalgal production in Onslow Bay, NorthCarolina, USA. Mar. Ecol. Prog. Ser., 84:185–196, 1992.

[7] L.B. Cahoon, G.R. Beretich, C.J. Thomas, and A.M. McDonald. Benthic microal-gal production at Stellwagen Bank, Massachusetts Bay, USA. Mar. Ecol. Prog. Ser.,102:179–185, 1993.

[8] J.P. Gattuso, M. Pichon, B. Delesalle, and M. Frankignoulle. Communitymetabolism and air-sea CO2 fluxes in a coral reef ecosystem. Mar. ecol. Prog.Ser.- Wiley-Blackwell, 96:259 – 267, 1993.

[9] J.P. Gattuso, C.E. Payri, M. Pichon, B. Delesalle, and M. Frankignoulle. Primaryproduction, calcification, and air - sea CO2 fluxes of a macroalgal-dominated coralreef community. Journal of Phicology- John Wiley & Sons Ltd UK, 33:259 – 267,1997.

[10] L.B. Cahoon. An instrumented lander for measurement of benthic respiration andproduction. The Diving for Science archive.rubicon-foundation.org, pages 45–51,1996.

[11] G. Fasano and A. Materassi. Progetto e realizzazione di uno strumento per la misuradell’assorbimento radiativo dell’acqua marina: SuMaRad. Rivista di IngegneriaAgraria - Ed. ETS Pisa, 4:1–7, 2003.

[12] YSI. Multiparametric Sonde YSI 6600 V2. https://www.ysi.com, 2009.

[13] Anonymous. Arduino: an open-source electronics prototyping platform.www.arduino.cc, 2009.

[14] Sea Birds Electronics. Submersible Pump SBE 5P. http://www.seabird.com, 2009.

2378

https://www.ysi.com

http://www.seabird.com

A Multidisciplinary Approach for Paleoenviro-mental Reconstruction in Ultra-Shallow Waterthrough Acoustics and Core Sampling

F. Madricardo1, S. Donnici1, A. Lezziero2, F. De Carli2, J.A. De Souza3,B. Bertani41, Institute of Marine Sciences, CNR, Venezia, Italy2, PHAROS s.a.s. di Alberto Lezziero e C., Venezia, Italy3, Center Of Studies in Coastal and Oceanic Geology, UFRGS, Porto Alegre, Brazil4, Information Service of the Water Authority of Venice, Venezia, [email protected]

Abstract

The lagoon of Venice represents a very peculiar environment resulting from com-bined natural and anthropic action. It began to form during the last marine trans-gression about 5500 years ago. Since then, its morphological aspect has drasticallychanged. The changes are due, on one hand, to natural causes like tributary riversediment supply, eustatic variations and natural subsidence. On the other hand, theanthropic action had a strong impact on the lagoon evolution. The first anthropic set-tlements go back in time to the roman age, but big engineering and dredging worksstarted about the 14th century AD and are still ongoing. All these interventions con-tributed to radical changes in the lagoon hydrodynamics. The consequent changein the lagoon morphology can be established by studying the content of the lagoonalsediments. For this purpose, a multidisciplinary research project has been carried outover a large area (about 50 km2) of the lagoon. An extensive high space resolutionacoustic survey in ultra-shallow water (up to a depth 0.5 m) together with geologicalanalysis from 60 cores extracted in the explored area revealed the Holocene sedi-ment architecture. In particular a complex network of buried palaeochannels wasreconstructed thanks to the acoustic techniques that allowed a detailed mapping ofthe lagoonal sedimentary feature.

1 Introduction

The Lagoon of Venice is the result of acomplex interplay of natural and anthro-pogenic factors. Many studies were car-ried out to investigate and to reconstructits origin and its development over timeand space up to its current morphologicaland hydrographic state. The lagoon be-gan to form about 5500 years ago during

the late Holocene marine transgression [1].Since then, a sediment succession of la-goon environments with various morpholo-gies and hydrological regimes overlaid theearliest marine deposits [2]. In histori-cal times, the anthropogenic action had astrong impact on the lagoon evolution. Thefirst settlements go back in time to the ro-man age, but big engineering and dredgingworks (major river diversion, construction


Technologies

of outer jetties of port harbours and indus-trial area, reclamation works, digging outof large artificial channels) started aboutthe 14th century AD and are still ongoing[3, 4]. The actual lagoon morphologies,like salt-marshes, tidal meanders and theirproperties and the present tidal network,had been object of very detailed field andlaboratory observations and hydrodynam-ics modelling studies [5, 6, 7, 8, 9]. Theancient lagoon morphologies, on the otherhand, were reconstructed thanks to geolog-ical investigation based on core sampling(see for example [10, 11, 12] and refer-ences therein). In particular, the positionof ancient coastal strips and the presenceduring the Roman age of large emerged ar-eas were discovered [13, 14, 15]. However,the core sampling gives information on dis-crete points, whereas the extreme spatialand temporal variability of these lagoonmorphologies makes it sometimes very dif-ficult to correlate sedimentary facies. Theextremely shallow waters (less than 1 mdepth) of the Lagoon of Venice had pre-vented for long time the use of conven-tional geophysical survey methods. Thefirst field trials with acoustic methods inthe lagoon were restricted to few surveylines [16, 17]. Other recent works are fo-cused more on the stratigraphic sequenceof the Pleistocene and Holocene sedimentsthan on geomorphological reconstructionand they concern to the southern lagoon[18, 19]. In this context, since 1999,a joint research programme between theCNR Institute of Marine Science (Venice)and the CNR Institute of Acoustics (Rome)together with the PHAROS sas was car-ried out: extremely shallow zones in thelagoon were explored with a multidisci-plinary approach [20]. Within this researchproject, an extensive high spatial resolu-tion acoustic survey was carried out [21].

The Holocene sediments had been system-atically investigated using a single beamechosounder in extremely shallow water,with water levels ranging from about 0.5m to about 5 m. More than 50 km2 in thecentral and northern part of the lagoon fora total of 2000 km of survey lines wereexplored. The acoustics gave indicationsof the subbottom layer content over wideareas and provided an extremely usefulguide for selection of sample sites for cor-ing. In particular, this geophysical investi-gation allowed the detailed reconstructionof buried channel paths, internal stratifica-tions and meandering behaviour.

2 Study area and experi-mental method

The investigated area of about 50 km2 islocated in the Lagoon of Venice (North-eastern Italy), which has a total surface of550 km2 and a average depth of 1.05 m forshallow flats and 5.5 for channels (data ofyear 2000, [9]. It is connected to the Adri-atic sea by the inlets of Lido, Malamoccoand Chioggia (Figure 1a). The surveyedarea is subdivided in two parts: the areaA in northern lagoon subbasin (of about 2km2) close to the islands of Burano, Maz-zorbo and S. Erasmo (Figure 1b) and thearea B between the lagoon margin aroundthe city of Venice to the Lido (Figure 1c).The area A is relatively protected from thecurrents and waves coming from the opensea into the lagoon through the Lido inlet.It is a submerged mudflat with water depthbelow 1 m. In this area we conducted avery detailed acoustic survey along paral-lel lines with 2-3 m spacing. This detailwas necessary to find buried archaeologi-cal structures, of which the northern lagoon

2380


Figure 1: Map of the Venice Lagoon (a) and an enlargement of the survey areas (b,c):The dark grey points indicate the position of the 60 cores extracted in the two areas.

is rich (see for example [22]). In this area,30 sediment cores were extracted (Figure1b) to verify the nature of noticeble acous-tic features identified during the survey.The area B develops all around the city ofVenice and interests both an area of semi-enclosed and of open lagoon, less protectedfrom currents and waves, with an evidenttendency to evolve into a bay environment[23]. Main navigation channels are presentbecause of the proximity to the city andto the industrial area. Here, we wantedto identify the main geological structuresaround the city of Venice. Therefore, thesurvey was carried out along parallel lineswith a 50 m spacing. In addiction, also inthis area, 30 truth samples were extracted(Figure 1c). The modified ELAC LAZ72 single beam echosounder equipped witha 30 kHz LSE 131 transducer was usedfor the acoustic survey. The echosounderwas mounted on a small boat along with aco-located DGPS system Trimble DSM12,which allows a positioning accuracy withinone meter. The flat bottom boat was just

7 m long with a 30 cm draft.Thus it waspossible to operate under extreme shallowwater conditions (down to 50 cm depth).All the survey echograms were processedand interpreted (see [21] for more details).The acoustic anomalies were stored in adatabase with indications about their formand their depth for comparison with theother data. The geological data presentedhere for comparison with the results of theacoustic survey were obtained from 5 bore-holes: SA4, SA5, SA8, SA9 drilled in thearea A and SG25 in the area B. Their geo-graphic positions were determined througha DGPS system, while their lengths var-ied between 8 and 9.1 m, for all the struc-tures identified by the acoustics to be in-tersected. They were sampled in Septem-ber 2004, October 2005 and April 2006with the help of a rotation method withwater circulation. The corer diameter was101 mm. The lithology and sedimentarystructures of all extracted materials weredescribed, using as altitude reference themean sea level (m.s.l.) of 1897, 23.5 cm

2381

Technologies

Figure 2: Acoustical anomalies identified as buried palaeochannels: the black lines rep-resent the spatial extension of the palaeochannels along the acoustic survey; in light anddark are shown the actual channel network and the emerged morphologies, respectively.

below the mean present-day level. In par-ticular, the bottom depth with respect to them.s.l. was 0.54 m for SA4, 0.58 m for SA5,0.83 m for SA8, 0.7 m for SA9, and 1.19m for SG25, respectively. The sedimentcores were sampled for micropalaeonto-logical and radiometric investigations. Mi-cropalaeontological analyses were carriedout on sediment samples, where the or-ganic content was composed of crushedmollusk shells mixed with abundant testsof benthic foraminifera. From each sampleat least 150 foraminiferal specimens wereclassified according to the taxonomic or-der of Loeblich and Tappan [24]. Theirpercentage abundance was used for statis-tical data processing. Radiocarbon ageswere calculated on four samples of shells atthe Beta Analytic Inc. and at the CEDADlaboratories. The samples were analyzedusing the Accelerator Mass Spectrometry(AMS) technique to determine the amountof 14C content. The conventional 14C agebefore present (BP), i.e. AD 1950, con-tain the 13C/12C corrections to the mea-

sured age. The 14C ages were calibratedusing the Calib 501 program [25, 26].

3 Results and Discussion

In the surveyed areas a complex networkof buried channels was identified boththrough acoustic and geological data. Thepeculiar acoustic response of palaeochan-nels helped us in their recognition and clas-sification. They generally show dippingacoustic reflectors interpretable as inclineddeposits due to the channel migration. Thisinterpretation was later confirmed by theaccurate inspection of core sedimentaryrecords. As a result of data processing,interpretation and classification, in Figure2 we mapped the position of the acous-tic features corresponding to buried chan-nels. The detected palaeochannels belongto different stages of the lagoon develop-ment, as was shown by the radiocarbondating on our cores. Observing Figure 2,however, one can say that the lagoon has

2382


certainly been the theatre of channel form-ing, migrating and disappearing, as a con-sequence of the complex hydrological andsedimentological regime in the area. Asobserved in many studies, the channel net-work within the tidal basin is strongly re-lated to its hydrodynamics and sedimentexchanges. Whereas the present tidal net-work, the tidal meanders and their prop-erties, are known and studied from dif-ferent perspectives, the network of buried,and therefore disappeared, morphologieswas never mapped before in such a de-tail. This map give a first indication ofhow much the geomorphology of the la-goon have changed since its origin untilpresent. However, the mapped buried mor-phologies need to be deeply understood,since the acoustic echograms show a gen-eral picture referring to different times andhydrodynamic conditions. In the followingsections, we present the study of two mainburied geomorphological features identi-fied in the acoustic profiles. The first liesin the northern lagoon closer to the Lidoinlet (in the A area), while the second oneis located in the central lagoon closer to themainland (in the B area).

3.1 Palaeochannel in the North-ern Lagoon (area A)

Acoustic sections intersect a widepalaeochannel that developed with a largemeander in an area that now is a submergedmudflat with a mean water depth of about0.7 m. The 2-D reconstruction of the buriedpalaeochannel path for a length of 2.7 kmis depicted in Figure 3. Two examplesof the acoustic anomalies relative to thisburied morphology are shown in Figure 4.In Figure 4a, the reflection pattern shows atypical prograded fill of erosional channels

[27]. The reflecting horizons correspond tochanges in lithology (velocity and density)in the sediment. The related layers areinclined, dipping in a northern direction.This configuration is interpreted as the re-sult of the active lateral accretion of thepalaeochannel through point bar migration.In Figure 4b, a survey profile of the samepalaeochannel is depicted. It was recorded550 m to the west of the one shown inFigure 4a. Here, as well, one can see aprograded fill. In this case, the seismic re-flectors dip in the southern direction, beinga section of a consecutive loop of the samemeander pointing in the opposite direction.This is clear also from the full path recon-struction given in Figure 3. The verticallines in Figure 4 represent the ground-truthsamples (cores) extracted to verify the na-ture of the acoustic anomalies. In Figure4a the locations of cores SA4 and SA5 areshown. The core SA4 was drilled in thecentre of the supposed palaeochannel. Thecore SA5 was located immediately outsidethis buried morphology in order to clearlydiscriminate the sedimentary characteris-tics given by the presence of the acousticanomaly. In Figure 4b the locations ofthe cores SA9 and SA8 are shown. Thegeological investigation (SA4, SA8, SA9)confirmed the presence of the buried chan-nel. From the depth of -9 m below m.s.l.,medium to fine silty laminated sand andsandy silt with frequent millimetric sandlayers are present. The emplacement ofthese layers is related to the changing ofenergy levels, linked to the tidal phasesand sediment transport. These sedimentsare representative of channel facies andcontain a typical foraminiferal associationof the outer lagoon environment. This set-tlements can be explained by proximity ofthis area to a lagoon inlet up to 500-400years BP, when the S. Erasmo island (Fig-

2383

Technologies

Figure 3: Reconstruction map of the path of a buried palaeochannel with the locations ofthe cores SA4, SA5, SA8, SA9.

ure 2) was a littoral strip. (see [28]). Theradiometric ages of the shells found at thebottom of the channel deposits in coresSA4 and SA9, respectively 4276 ± 93 and4341 ± 82 calibrated years BP, are in verygood agreement. They indicate that theybelong to the same channel active 4500years BP at least. Finally, core SA5, co-herently with acoustic data, did not revealthe presence of a channel. The undisturbedstratigraphic sequence revealed subtidal la-goon sediments, whose bottom radiometricage is 4531±87 calibrated years BP. Theyare mainly clayey or sandy silts, containingmollusc shells and characterized by a highbioturbation level. Foraminiferal analysessuggest the occurrence of freshwater inputsin the lagoon environment. This indicationcan be explained by the fact that the riverinputs were important into this area beforethe artificial river diversion between the14th and the 17th century.

3.2 Paleochannel in the CentralLagoon (area B)

Acoustic sections, in this case as well, in-tersect a palaeochannel that developed be-tween Fusina and Venice with a meanderburied in a actual submerged mudflat withmean water depth of about 1-1.5 m. The 2-D reconstruction of the buried palaeochan-nel path for a length of 2.2 km is shownin Figure 5. An example of the acousticanomaly relative to this buried morphol-ogy is shown in Figure 6. In this case,two different fill patterns are visible. First,one has a prograded fill up to the greyline. Here, the inclined reflectors, dip-ping into the northern correspond to thepalaeochannel point bar migration. Then,the buried channel presents a divergent fill.Therefore, the grey line seems to separatetwo different phases: an earlier high en-ergetic regime with sand deposition andchannel migration and a later low energeticregime with a finer filling and apparentlyno migration. The palaeochannel was in-

2384


Figure 4: Reflection profiles of a palaeochannel in different points. The straight verticallines indicate the location of the cores sampled. Horizontal and vertical scales are 300and 7.5 m, respectively.

tersected by the core SG25, whose loca-tion is indicated by the black vertical linein fig.6. The stratigraphic record (Figure6 right) presents mainly clayey-silty sedi-ments from -1.2 to -5.2 m and sandy sed-iments from -5.2 to -6.60 m above m.s.l.The radiometric age of the shell found at-5.2 m between the two facies set the ener-getic regime change about 1600 ± 90 calyrs BP (grey line in fig.6). The two phasescan possibly be related to a variation ofthe area hydrodynamics due to the climateworsening between the IV and VI centuryAD (see [29]).

4 ConclusionsThe multidisciplinary approach allows forthe first time a very detailed reconstructionof paleochannel paths and internal struc-tures, their meandering behaviour and theirevolution related to possible change of thelagoon hydrology. In particular a com-plex network of palaeochannels was recon-structed thanks to the acoustic techniques

that allowed a detail mapping of the buriedlagoonal morphology. In this paper, we il-lustrate two examples of two distinct casesof geomorphological reconstruction: onein the northern and the other in the centralLagoon. The paths and the internal struc-tures of two palaeochannels were mappedand studied, respectively, by correlating theacoustic and geological data. In the firstcase, the acoustic data show a dense strat-ification within the palaeochannel due tothe meander bar migration testifying a cer-tain constancy of the hydrological condi-tions until its complete fill. In the secondcase, instead, the acoustics allowed one toidentify a sudden passage from a more en-ergetic environment to one less energetic,possibly related to a rapid change of the hy-drodynamic conditions.

5 AcknowledgementsThe authors would like to thank the SIN forall the Lagoon of Venice background mapsof the figures we presented. Moreover, we

2385

Technologies

SG25

Figure 5: Reconstruction map of the path of the buried palaeochannel intersected by thecore SG25.

Figure 6: To the left: Buried palaeochannel intersected by the core SG25 (black verticalline). The thick black lines show the channel point bar migration and stratification, whilethe dashed lines indicate paleosurfaces. The grey line corresponds to the channel’s shapeabout 1600± 90 cal yrs BP. To the right: SG25 Core Log. The depth scale is with respectto the lagoon bottom which was 1.19 m below m.s.l.

are very grateful to the referee of this pa-per, whose suggestions were very useful toimprove the paper. This research was sup-

ported technically and financially by theMinistero delle Infrastrutture e Trasporti,Magistrato alle Acque di Venezia.

References[1] P. Gatto and L. Carbognin. The Lagoon of Venice: natural environment trend and

man-induced modification. Hydrological Sciences Bulletin, 26:379–391, 1981.

[2] R. Serandrei-Barbero, R. Bertoldi, G. Canali, S. Donnici, and A. Lezziero. Paleo-climatic record of the past 22,000 years in Venice (Northern Italy): Biostratigraphicevidence and chronology. Quaternary International, 140:37–52, 2005.

2386


[3] L. Carbognin. Evoluzione naturale e antropica della Laguna di Venezia. Mem.Descr. Carta Geol. Ital, 42:123–134, 1992.

[4] L. Carbognin, G. Cecconi, and V. Ardone. Interventions to safeguard the envi-ronment of the Venice Lagoon (Italy) against the effects of land elevation loss,volume 1 of Land Subsidence. La Garangola, Ravenna (Italy), 2000.

[5] S. Fagherazzi, A. Bortoluzzi, W.E. Dietrich, A. Adami, S. Lanzoni, M. Marani,and A. Rinaldo. Tidal networks 1. Automatic network extraction and preliminaryscaling features from digital terrain maps. Water Resources Research, 35(12):3891–3904, 1999.

[6] S. Lanzoni and G. Seminara. Long-term evolution and morphodynamic equilibriumof tidal channels. Journal of Geophysical Research-Oceans, 107(C1), 2002.

[7] M. Marani, S. Lanzoni, and D. Zandolin. Tidal meanders. Water Resources Re-search, 38(11):1125, 2002.

[8] L. Carniello, A. Defina, and L. D’Alpaos. Morphological evolution of the Venicelagoon: evidence from the past and trend for the future. Journal of GeophysicalResearch, 114(F04002), 2009.

[9] E. Molinaroli, S. Guerzoni, A. Sarretta, M. Masiol, and M. Pistolato. Thirty-yearchanges (1970 to 2000) in bathymetry and sediment texture recorded in the Lagoonof Venice sub-basins, Italy. Marine Geology, 258(1-4):115–125, 2009.

[10] L. Alberotanza, R. Serandrei-Barbero, and V. Favero. I sedimenti olocenici dellaLaguna di Venezia. Bollettino della Societa Geologica Italiana, 96:243–265, 1977.

[11] A. Bondesan and M. Meneghel. Geomorfologia della Laguna di Venezia. Noteillustrative della Carta geomorfologica della provincia di Venezia. Esedra editrice,Padova, 2004.

[12] R. Serandrei-Barbero, A. Albani, S. Donnici, and F. Rizzetto. Past and recent sed-imentation rates in the Lagoon of Venice (Northern Italy). Estuarine Coastal andShelf Science, 69(1-2):255–269, 2006.

[13] V. Favero and R. Serandrei-Barbero. Evoluzione paleoambientale della Laguna diVenezia nell’ area archeologica tra Burano e il Canale S.Felice. Lavori SocietaVeneziana di Scienze Naturali, 6:119–134, 1981.

[14] R. Serandrei-Barbero, A. D. Albani, and S. Zecchetto. Palaeoenvironmental sig-nificance of a benthic foraminiferal fauna from an archaeological excavation inthe Lagoon of Venice, Italy. Palaeogeography Palaeoclimatology Palaeoecology,136(1-4):41–52, 1997.

[15] A. Lezziero, S. Donnici, and R. Serandrei-Barbero. Evoluzione paleoambien-tale dell’area archeologica sommersa di S. Leonardo in Fossa Mala (Laguna diVenezia). Geografia Fisica e Dinamica Quaternaria Suppl, 7:201–210, 2005.

2387

Technologies

[16] C.E. McClennen, A.J. Ammerman, and S.G. Schock. Framework stratigraphy forthe lagoon of Venice, Italy: Revealed in new seismic-reflection profiles and cores.Journal of Coastal Research, 13(3):745–759, 1997.

[17] C.E. McClennen and R.A. Housley. Late-Holocene channel meander migration andmudflat accumulation rates, lagoon of Venice, Italy. Journal of Coastal Research,22(4):930–945, 2006.

[18] M. Zecchin, L. Baradello, G. Brancolini, F. Donda, F. Rizzetto, and L. Tosi. Se-quence stratigraphy based on high-resolution seismic profiles in the late Pleistoceneand Holocene deposits of the Venice area. Marine Geology, 253(3-4):185–198,2008.

[19] M. Zecchin, G. Brancolini, L. Tosi, F. Rizzetto, M. Caffau, and L. Baradello.Anatomy of the Holocene succession of the southern Venice lagoon revealed byvery high-resolution seismic data. Continental Shelf Research, 29:1343–1359,2009.

[20] S. Buogo, E. Canal, G.B. Cannelli, S. Cavazzoni, S. Donnici, and A. Lezziero.Geoarchaeology in the Lagoon of Venice: palaeoenvironmental changes, ancientsea-level oscillation and geophysical surveys by acoustic techniques. CambridgeUniversity Press, Cambridge, 2005.

[21] F. Madricardo, S. Donnici, A. Lezziero, F. De Carli, S. Buogo, P. Calicchia, andE. Boccardi. Palaeoenvironment reconstruction in the Lagoon of Venice throughwide-area acoustic surveys and core sampling. Estuarine Coastal and Shelf Science,75(1-2):205–213, 2007.

[22] E. Canal. Testimonianze archeologiche nella Laguna di Venezia. Edizioni delVento, Venezia, 1998.

[23] A. Sarretta, S. Pillon, E. Molinaroli, S. Guerzoni, and G. Fontolan. Sediment budgetin the Lagoon of Venice, Italy. Continental Shelf Research, 30-8:934–949, 2010.

[24] A.R. Loeblich and H. Tappan. Foraminiferal Genera and their Classification. 1987.

[25] M. Stuiver, P. J. Reimer, E. Bard, J. W. Beck, G. S. Burr, K. A. Hughen, B. Kromer,G. McCormac, J. Van der Plicht, and M. Spurk. INTCAL98 radiocarbon age cali-bration, 24,000-0 cal BP. Radiocarbon, 40(3):1041–1083, 1998.

[26] M. Stuiver, P. J. Reimer, and R. Reimer. CALIB 5.0. pagehttp://calib.qub.ac.uk/calib/ program and documentation, 2005.

[27] R.M. Mitchum, P.R. Vail, and J.B. Sangree. Seismic stratigraphy and globalchanges of sea level, Part 6: Stratigraphic interpretation of seismic reflection pat-terns in depositional sequences. American Association of Petroleum GeologistsMemoir. Tulsa, 1977.

2388


[28] M. Bonardi, E. Canal, S. Cavazzoni, R. Serandrei-Barbero, L. Tosi, A. Galgaro, andM. Giada. Sedimentological, archaeological and historical evidences of palaeocli-matic changes during the Holocene in the Lagoon of Venice (Italy). World ResourceReview, 9:435–446, 1997.

[29] A. Veggiani. I deterioramenti climatici dell’Eta del Ferro e dell’alto Medioevo.Bollettino della Societa Torricelliana di Scienze e lettere, 45:3–80, 1994.

2389

Technologies

2390

Innovative Approaches to Retrieve Water Opti-cally Active Parameters from Hyperspectral Data:in situ Measurements and Image Processing

F. Santini3, L. Alberotanza2, F. Braga2, R.M. Cavalli1, S. Pignatti31, Institute for Atmospheric Pollution, CNR, Roma, Italy2, Institute of Marine Sciences, CNR, Venezia, Italy3, Institute of Methodologies for Environmental Analysis, CNR, Potenza, [email protected]

Abstract

This paper describes the research activities carried out to enhance the perfor-mances of the application of remote sensing techniques on water bodies. In order toprovide an inexpensive way to gather information linked to the water clarity and qual-ity, several approaches for remotely sensed data interpretation, devoted to producewater optical property parameters, were tested. These approaches range from simpleempirical relations to complex physical models. As regards the latest, a method forthe retrieve of Chl (chlorophyll), CDOM (coloured dissolved organic matter) and aclass taking into account the particulate material (TSM, TSS or Tripton), was as-sessed. At the purpose, a bio-optical model, based on radiative transfer theory, andan inverse procedure were implemented. The method was applied on spectral datacollected during on purpose performed in situ campaigns by means of two spectrora-diometers (OceanOptics and SpectraScan) following the SeaWiFS protocols. Watersamples were collected for validation purposes. Laboratory analysis were performedon the water samples to provide water compound concentrations. The good resultsobtained by applying the model to in situ data encouraged the application to remotesensor data. Various pre-processing steps were carried out on remotely sensed im-ages to make them feasible for the model application.

1 Introduction

Over the past 30 years, many satellite bornesensors have been exploited to gather in-formation on land cover and insight intothe biological activity occurring within thewater bodies. Remote sensing products of-fer many new capabilities in a wide va-riety of applications, including the man-agement of natural resources, inventory as-sessment (e.g. invasive weeds), environ-mental hazard, wildlife habitat characteri-

zation, and water quality monitoring. Thestudy of the water colour (optical charac-terization in the visible spectral range) canbe carried out by using several sensors like:ENVISAT/MERIS sensor (412-1050 nm)with a spatial resolution of 250 m/pixel;OCEANSAT-1/OCM (Ocean Colour Mon-itoring) with a spatial resolution of 350m/pixel and a spectral coverage between402 and 885 nm; TERRA/MODIS (405-14,382 nm) with a spatial resolution of1000 m/pixel, Hyperion with a spatial res-


Technologies

olution of 30 m/pixel and SeaWiFS (402-885 nm) with a spatial resolution of 1000m/pixel. All the above sensors present arevisiting time suitable to study the sea-sonal/annual evolution of weed distributionand water clarity time degradation. Thelaunch of new satellites with an increasedspectral and spatial resolution (PRISMA,EnMap) will further enhance the utility ofremotely sensed information for a waterresource management support system. Inorder to extract water quality information,the optical data can be interpreted usingdifferent methods. Simplest methods arebased on empirically derived relationshipsbetween the measured reflectance ratio andthe chlorophyll concentration. These meth-ods are suitable only for those water bod-ies (like open ocean) where CDOM andparticulate material are of autochthonousorigin and are strictly linked to phyto-plankton abundance. These waters are re-ferred to as Case I waters. Coastal and in-land waters are generally much more com-plex. Because of detritus due to coast ero-sion or brought from the wind or fromrivers, in fact, correlations between com-pounds fall down. To completely describethese water bodies, known as Case II wa-ters, it is necessary to develop more com-plex algorithms. This complexity has beenreached also through the definition of mod-els which link the spectral behaviour ra-diometric field to the optical properties ofwater compounds. This kind of algorithmsis named physically based algorithms andrelate to the radiative transfer (RT) the-ory. Several models have been suggested(an exhaustive reference list can be foundin [1]) with varying degrees of complex-ity. In this paper we present the workcarried out in different study area to en-hances the performances of the applica-tion of remote sensing techniques on water

bodies. In particular we describe the dataand procedure followed to define a physi-cally based model approach. The study ar-eas token into consideration relate to theMontenegrin and Albanian coasts, to theVenice lagoon and to the Kisumu bay (Vic-toria Lake - Kenya), which, although notbeing a case of inland water, represents acase of very optically complex water bodycharacteristic of some coastal areas. In theKisumu bay a field campaign was carriedout on January 2004 onboard a small boat.During the field campaign 17 measure sta-tions were devoted to the acquisition ofspectral signature of the water surface andemerged vegetation. In 9 of these stationswater samples for laboratory analysis werealso collected. The data collected for theAlbanian and the Montenegrin coastal ar-eas were obtained from R/V G. Dallaporta(ISMAR-CNR, Italy) during two sea cam-paigns carried out on May 9-21 and on June24-29, 2008. As regards the Venice lagoonstudy area, four field campaigns were con-ducted over the lagoon and the open sea onJune-July 2001, June 2004 and May 2005.

2 Field data

Remote sensing measurements require agood field sampling strategy to calibrateand validate the data by-products. Wa-ter bodies were characterized, according toOcean Optics Protocols for Satellite OceanColour Sensor Validation [2, 3, 4], by spec-tral measurements and collecting samplesfor laboratory analyses. Spectral signaturesof the water column were derived from theradiance measurements of the sky and thewater.

2392


2.1 Spectral measurementsThe optical measurement above water weredevoted to the retrieve of the Rrs (Remotesensing reflectance), defined as:

Rrs =Lu(0−)

Ed, (1)

where Lu(0−) is the upwelling signal com-ing from just below the water surface (alsoknown as the in-water leaving radiance)and Ed is the irradiance incident (down-welling) on the water surface. The need todefine an ad hoc physical quantity for wa-ter characterization originates from the factthat standard reflectance (defined as the ra-diance coming from the target divided bythe incident irradiance) includes the sur-face reflectance contribution, that dependson illumination geometry. Rrs, on thecontrary, is independent from illuminationcondition and is, therefore, functional tocharacterize water targets. What is directlyacquired is Lu(0+), that is the signal justabove the water surface. This can be ex-pressed as

Lu(0+) = Lu(0−) + aLsky, (2)

where the last term represents the surfacereflectance and is supposed to be propor-tional to the sky radiance (Lsky) a dependson the acquisition geometry. To correctlyassess Rrs three different set of measureswere carried out for Ed, Lu(0+) and Lsky

estimates following the SeaWiFS protocol[3].

2.2 Water samplesDuring all the field campaigns, severalmeasure stations were devoted to thecollection of water samples for labora-tory analysis. During the field cam-paigns carried out in the framework of the

Hypad.COM project along the Albanianand Montenegrin coastal areas (June-July2005) samples of CDOM, TSM (Total Sus-pended Matter) and Chl a at the surfaceand 10 m depth were collected in 30 suit-able stations. As regards the Venice lagoonstudy area, the first two campaigns (Juneand July 2001) supplied only water surfaceradiometric data to validate the remotelysensed data preprocessing chain. The sec-ond two field campaigns (June 2004 andMay 2005) included water body opticalproperty measurements and water samplesfor laboratory analysis. In this case studythe particulate class was represented byTripton, which includes only the inorganicmaterial. Finally, during the field cam-paigns related to the Kisumu bay casestudy (May 2004), water samples were col-lected on nine stations and analyzed in theKMFRI laboratory for the assessment ofChl a, (with fluorescence analysis), CDOMand TSS (Total Suspended Solids). Allmeasure and laboratory analysis were per-formed following the Ocean Optics Proto-cols For Satellite Ocean Color Sensor Val-idation [2, 3, 4].

3 Remote images

During the diverse field campaigns someimages were acquired for model applica-tions. During the Montenegrin and Alba-nian field campaigns a MIVIS sensor, on-board an airborne platform, acquired im-ages along the coastline and along oneof the measure transects. As regards theVenice lagoon, the first campaign wascarried out in concomitance to a Hype-rion overpass, while the second and thefourth campaigns were supported by air-borne platforms equipped respectively witha MIVIS and a CASI sensor. Finally,

2393

Technologies

although the Kisumu bay field campaignwasn’t programmed in concomitance withany satellite or airborne acquisition, fewdays after the field campaign took place, aMERIS Image was cloud free and adequatefor the model application. Remote imagedata were pre-processed, analogously to insitu data, to obtain the Rrs images.Noise reduction. A global de-striping pro-cedure was applied to pushbroom sensordata (Hyperion, CASI), in order to reducethe spectral effects of column-to-columnnoise.Atmospheric correction. The atmo-spheric correction were performed by ap-plying the MODTRAN code. Path ra-diance, transmittance and incoming irra-diance were simulated by using standardurban areas aerosol concentration profilesa visibility matching the acquisition con-dition and illumination geometry corre-sponding to image acquisition time.Surface effects correction. Reflected sig-nal was simulated through Hydrolight soft-ware (Sequoia Scientific, Inc.), given an il-lumination geometry corresponding to im-age acquisition time and a wind speedlower than 5 m/s. The signal over the sur-face was determined excluding path radi-ance from remote sensors signal and divid-ing by the transmittance. Then Rrs, ac-cording to its definition, was obtained sub-tracting surface reflected signal and divid-ing by incoming surface irradiance.Geometric correction. The whole re-flectance dataset was geocoded accordingto navigational data available for each plat-form, in the Gauss-Boaga reference sys-tem.

4 MethodsThe proposed method includes a bio-optical-model (that defines the opticalproperties of the main water compoundsand connects them to the optical propertiesof the water body) and a radiative trans-fer model (that links the optical propertiesof the water to the radiometric field). Thejunction of the two models allows estimat-ing radiometric quantity in relation to theabundance of water compounds and theirproperties of light absorption and scatter-ing. Then, by means of an appropriate in-version procedure, it is possible to retrievecompound concentrations starting from ob-servable radiometric quantities (Rrs).

4.1 Bio-optical modelAt a first step, the set of compounds thatcharacterize the water body has to be es-tablished, and their specific optical proper-ties have to be defined. The elements takeninto account are: pure water, phytoplank-ton (associated to chl a, that represents itsmain pigment), CDOM, and a class of par-ticulate (TSM, TSS or Tripton). The mainassumption concerning the optical charac-terization of the whole water body is thelinear composition of the single constituentabsorption and back-scattering spectra:

a = a0 +n∑

j=1

a∗jCj ; bb = bb0 +n∑

j=1

b∗bjCj ,

(3)where a0 and bb0 refer to the pure water ab-sorption and back-scattering; Cj representsthe concentration of the j − th constituent(chl, CDOM and TSS); a∗j and b∗bj are theoptical properties per unit of concentrationfor the i-th constituent. The optical charac-terization of the constituents was obtainedon the basis of literature data (Kisumu bay)

2394


or by integrating literature data with labo-ratory optical profiles (Venice lagoon). Forthe Venice lagoon case study, chlorophylland CDOM optical profiles were obtainedby laboratory analysis, while in the othercase studies they were token respectivelyfrom Prieur and Sathyendranath [5], andfrom Morel [6]. For the backscatteringof phytoplankton and TSS/TSM/Tripton aprofile based on the scattering phase func-tion of large and small particles, as definedin Kopelevich [7], was utilized. The purewater absorption and back-scattering coef-ficients are those elaborated in Pope andFry [8] and in Smith and Baker [9].

4.2 Radiative transfer modelThe interactions between light and mat-ter within a water body are well repre-sented by the RT equations. A software(Hydrolight v4, Sequoia Scientific, Inc.)which implements a numerical solution ofthese equations was used for this study.This software maintains a high physicalrigour and allows the computations of ra-diance distributions and related quantities(reflectance, irradiance) within the waterbody and upon the surface as a functionof the water compound concentrations, thewater compound optical properties, the wa-ter surface state, the illumination geome-try and other boundary conditions. How-ever, this software allows to retrieve wa-ter compound abundance only through atime consuming iterative procedure, and itis not suitable for processing large amountof data. If we consider that a remote im-age normally involves tens of thousandsof pixels in different spectral bands, it isunderstandable that a simplified analyticaland invertible formulation of the radiativetransfer equations would be precious. Sim-plified analytical formulations can be de-

rived directly from the RT theory under re-strictive conditions [10]. Many treatments[11, 12], in the case of a optically deep andvertically homogeneous water body, agreeon solutions similar to the one reportedhere:

Rrs = F · bb

a + bb, (4)

where a and bb are given by the Eq. 3and F is a parameter related to the illu-mination fields and the optical propertiesof the water surface. The F parameterwas established for each image throughquadratic regressions on the basis of datasimulated with Hydrolight software. In thisway it was possible to define an analyti-cal formulation that is invertible and, at thesame time, coherent with the RT equations.The regression involved several hundredsof simulations, which were carried out us-ing the optical model defined above, con-sidering the illumination geometry relatedto the remote images acquisition time andvarying the abundance of the compoundswithin the in situ measured ranges.

4.3 Inversion procedureUsing Eq. 3 in Eq. 4 and defining the quan-tities it is possible to obtain a linear relationof the following form [13]

Rrs =n∑

j=1

SjCj . (5)

This equation, if applied to the spectralbands of a specified sensor (with m num-ber of bands), corresponds to a linear sys-tem of m equations with n unknowns (n =number of compounds). Since m is gen-erally greater than n, many authors definesome criteria for band exclusion in order tosolve the system. In view of an utilizationof hyperspectral sensors, this approach can

2395

Technologies

result in a loss of information. Therefore, itis proposed to resolve the system through aleast square method that allow to fully de-ploy hyperspectral data [13]. To practicallycarry out the inversion, a suitable softwarewas developed in IDL language. This soft-ware takes as input the bio-optical modeland the remote sensed data and starts a cy-cle on the image pixels. On each step theEq. 4 is computes and the system of Eq.5 is inverted. The outputs of the IDL pro-cedure consist of the water compound con-centration maps [13]. In the Venice lagooncase study a different inversion procedurewas taken into consideration to improve theresults. In fact, the solution found for thelinearized formulation of Eq. 5 doesn’tmatch exactly the optimal solution of theoriginal non-linear problem. They differas much as the modeled Rrs differs fromthe observed Rrs [14]. In the Venice la-goon case study a two-step procedure waspresented for the inversion of the model:the solutions of the linearized formulationwere used as a starting point for an itera-tion procedure based on a first order Taylorseries expansion. For a detailed descriptionof the two-step procedure see [14]. The so-lutions of the linearized formulation havebeen demonstrated to be an optimal start-ing point, as no divergence occurred forthe whole processed dataset. The iterationwas stopped when a refinement of less than0.1% was reached for each parameter.

5 Results and Discussion

Physically based model, differently fromthe empirical ones, has the great advantagethat does not necessitate of large amountsof in situ data for parameter refinement.In fact, the F parameter is assessed onthe base of a model that solves the RT

equations rather than by means of empir-ical data. This doesn’t imply that no fielddata are needed for model validation. Inany case, the quantity of data collected insitu is much less crucial than for empir-ical approaches. A set based on a lim-ited number of stations may be behind thedefinition of rigorous procedure of errorquantifications, but, if supported by his-torical data and by a regional knowledge,it can be adequate for a preliminary as-sessment of models and methods and toquantify their usefulness. The differentsteps of data processing (atmospheric cor-rections and other pre-processing proce-dures, bio-optical model definitions, in situdata analysis, model parameterization etc.)make it difficult to develop a general apri-oristic procedure for final products uncer-tainty estimation. Therefore, a series ofsingle validation steps which involve a pre-processing, a bio-optical and a thematicproduct validation has to be carried out.

5.1 Pre-processing validationAtmospheric correction on remote imagesand the other pre-processing steps werevalidated comparing Rrs images with insitu measured Rrs spectra. For this com-parison, we extracted from the remotesensing data the average value of a squareof diverse pixels centred at the location ofthe in situ measurement to take into ac-count possible local water mixing. The gapbetween field campaign and remote sen-sor overpass can be partially responsiblefor the not perfect alignment between com-pared spectra. Anyway, the fit was ap-preciable presenting an RMS (Root MeanSquares) value generally lower than 15%of the Rrs signal. As an example, in Figure1, it is reported the comparison between theCASI spectra and the in situ ASD spectra

2396


Figure 1: Comparison between CASI and ASD Rrs spectra acquired in two measurestations located in the proximity of a navigable channel of the southern basin (upperspectra) and near the Malamocco inlet (lower spectra) during the 2005 Venice lagoonfield campaign.

as acquired in two measure station duringthe 2005 Venice lagoon campaign.

5.2 Model validationThe model validation was performed bysimulating Rrs with Hydrolight softwareusing as input the water compound op-tical properties and laboratory concentra-tions. Simulated spectra have been com-pared with in situ measurements of Rrs.An analogous indirect validation was car-ried out applying the inverse model to thein situ measured Rrs spectra and compar-ing the retrieved compound concentrationswith the results of the laboratory analysis.Table 1 report the comparison related to theKisumu bay, Table 2 reports the compari-son related to Venice lagoon. The two-stepprocedure performed on the Venice lagoon

case study allowed an improvement of theresults of up to 10 % with respect to thelaboratory water compound concentrations[14].

5.3 Remote data applicationsThe good results obtained by the applica-tion of the inverse physical model to thein situ Rrs spectra encouraged the ap-plication to the pre-processed remote im-ages. The model was thus applied to theMIVIS, CASI and the Hyperion Images ac-quired close to the time the in situ cam-paigns took place. These applications al-lowed the production of synoptic imagesof the water compound concentrations re-lated to the different study areas. As re-gards the Kisumu bay case study, theseproducts were used to analyze the precon-

2397

Technologies

Table 1: Kisumu bay water constituent concentrations obtained by laboratory analysis(column 2-4) and by the model application to the ASD in situ Rrs spectra (column 5-7).

dition for the anomalous proliferation ofwater weed that affects the area [13]. Asthe model doesn’t take into considerationthe bottom contribution, the application tothe Venice lagoon was limited to the opensea and the deeper navigable channel. Asregards the Montenegrin and the Albaniancoasts, no remote images have already beenprocessed. Nevertheless, a large set ofMIVIS images covering the whole coast-line is appropriate for the model applica-tion and will be soon elaborated. Relatedresults and products will be presented in afeature work. In Figure 2 the results ob-tained for the Venice lagoon case study areshowed.

6 ConclusionsThis paper showed the potentiality of in-novative approaches, based on the radia-

tive transfer theory in water body and at-mosphere, to retrieve water optically ac-tive parameters from hyperspectral data.A physically based method for the re-trieve of water quality parameters in opti-cally complex waters was presented. Themodel was validated through comparisonwith in situ collected data and applied toremote images acquired close to field cam-paigns. This allowed to generate synopticimages of the water compound concentra-tions. The comparison of these informationwith the results provided by the analysisof the water samples collected during thecampaign showed a general good agree-ment and demonstrated the usefulness ofthe physical model approach in estimatingwater quality in coastal and internal waterbodies.

2398


Table 2: Venice lagoon water constituent concentrations obtained by laboratory analysis(column 2-4) and by the model application to the ASD in situ Rrs spectra (column 5-7).

Figure 2: Water compound concentrations obtained from Hyperion data on open sea andnavigable channels of the Venice lagoon. From left to right, the maps of Chl, CDOM andTripton.

2399

Technologies

References[1] Z. Lee, K.L. Carder, and R.A. Arnone. Deriving inherent optical properties from

water color: a multiband quasi-analytical algorithm for optically deep waters. Ap-plied Optics, 41:5755–5772, 2002.

[2] B.G Mitchell, M. Kahru, J. Wieland, and M. Stramska. Determination of spec-tral absorption coefficient of particles, dissolved material and phytoplankton fordiscrete water samples. Ocean Optics Protocols for Satellite Ocean Color SensorValidation, pages 39–64, 2003.

[3] J.L. Muller, C. Davis, R. Arnone, R. Frouin, K. Carder, Z.P. Lee, R.G. Steward,S. Hooker, C.D. Mobley, and S. McLean. Above-water radiance and remote sens-ing reflectance measurement and analysis protocols. Ocean Optics Protocols forSatellite Ocean-Color Sensor Validation, pages 21–31, 2003.

[4] C.C. Trees, R.R. Bidigare, D.M. Karl, L.V. Heukelem, and J. Dore. Fluorometricchlorophyll a: sampling, laboratory methods, and data analysis protocols. OceanOptics Protocols for Satellite Ocean-Color Sensor Validation, pages 15–18, 2003.

[5] L. Prieur and S. Sathyendranath. An optical classification of coastal and oceanicwaters based on the specific spectral absorption curves of phytoplankton pig-ments, dissolved organic matter, and other particulate materials. Limnol. Oceanogr,26(4):671–689, 1981.

[6] A. Morel. Light and marine phtosynthesis: a spectral model with geochemical andclimatological implications. Prog. Oceanog., 26(6):263–306, 1991.

[7] O.V. Kopelevich. Small-parameter model of optical properties of sea water. OceanOptics, I: Physical Ocean Optics, 1:208–234, 1983.

[8] R.M. Pope and E.S. Fry. Integrating cavity measurements. Applied Optics,36:8710–8723, 1997.

[9] R.C. Smith and K. Baker. Optical Properties for the clearest natural waters. AppliedOptics, 20(2):177–184, 1981.

[10] C.D. Mobley. Light and Water: Radiative Transfer in Natural Waters. 1994.

[11] C. Giardino, V.E. Brando, A.G. Dekker, N. Strombeck, and C. Candiani. Assess-ment of water quality in Lake Garda (Italy) using Hyperion. Remote Sensing ofEnvironment, 109:183–195, 2007.

[12] V.E. Brando and A.G. Dekker. Satellite hyperspectral remote sensing for estimatingestuarine and coastal water quality. IEEE transactions on geoscience and remotesensing, 41(6):1378–1387, 2003.

2400


[13] R.M. Cavalli, G. Laneve, L. Fusilli, S. Pignatti, and F. Santini. Remote Sensingwater observation for supporting lake Victoria weed management. Journal of Envi-ronmental Management, 90:2199–2211, 2009.

[14] F. Santini, L. Alberotanza, R.M. Cavalli, and S. Pignatti. A two-step optimizationprocedure for assessing water constituent concentrations by hyperspectral remotesensing techniques: An application to the highly turbid Venice lagoon waters. Re-mote Sensing of Environment, 114(4):887–898, 2010.

2401

Technologies

2402

Integration of Earth Observation Data to Improvethe Knowledge of Upwelling Phenomenon in theStrait of Messina

F. Azzaro1, C. Bassani2, R.M. Cavalli2, S. Pignatti3, F. Santini31, Institute for Coastal Marine Environment, CNR, Messina, Italy2, Institute for Atmospheric Pollution, CNR, Roma, Italy3, Institute of Methodologies for Environmental Analysis, CNR, Potenza, [email protected]

Abstract

The work focuses on the data Earth Observation (EO) integration with differ-ent spatial and temporal resolution according to the Group on EO Working Plan2009-2011. In particular, the “Ecosystem observations will be better harmonizedand shared, spatial and topical gaps will be filled, and in situ data will be better in-tegrated with spacebased observations” (GEOSS Societal Benefit Areas) is the mainchallenge applied to the Strait of Messina. The area is a coastal site with a remark-able economic and environmental role addressable to the complex dynamic of thewater body. The strait is located in the centre of the Mediterranean Sea with strongcurrent and vortex due to the morphology of the site. In this area, the upwelling isa hydrological phenomenon that strongly impacts the marine ecosystem. The up-welling systems are characterized by increased ‘biological richness’. The principalindicator of upwelling is low sea surface temperature (SST). Since 1994, SST andphytoplankton were measured by continuous underway surface measurements on-board dedicated research boats. During 1999-2002, these in-situ measurements havebeen integrated with remote sensing image acquired by MIVIS airborne sensor. Theintegration meets the requirements to validate the empirical algorithm applied to re-motely data in order to retrieve the phytoplankton and, the remote sensing provides asynoptic view of upwelling phenomenon by mapping the SST and the phytoplanktoninto the water body of the strait.

1 Introduction

The Earth Observations system constitutesa critical input for advancing the under-standing and monitoring the Earth system,such as climate, oceans, atmosphere, wa-ter, land, natural resources, natural andhuman-induced hazards and for protectingthe global environment, reducing disasterlosses and achieving sustainable develop-ment [1]. In order to ensure comprehen-

sive and sustained Earth observations, acoordination of the efforts, articulated inthe GEOSS 10-Year Implementation Plan,has been built on to add value to existingEarth observation systems, filling criticalgaps and supporting their interoperability.Over the next decade, the global scientificcommunity has been expected to join thepivotal GEOSS Plan. The GEOSS will pro-vide an important scientific basis to clearlyidentify the planning and decision making


Technologies

in the complexity of the everyday life in-cluding public health, agriculture, trans-portation and numerous other areas to keepabreast of the rapid changes occurring inthe society today. In less than two years,the number of participating countries hasnearly doubled, and over 40 internationalorganizations also support the emergingglobal network. The Messina monitoringprojects take place at this strategic moment,when the Earth Observation has becomean essential component of the global ef-fort to deal with global challenges. Thepurpose of these projects is to promote ca-pacity building in Earth observation, onexisting local, regional and internationalinitiatives and sector-specific needs, likeGEOSS planned, in order to achieve com-prehensive, coordinated and sustained insitu, airborne, and space-based observa-tions of the Messina coastal area. The ac-tivities of these projects meet the need fortimely, quality local scientific informationas a basis for sound decision making, andwill enhance delivery of benefits to soci-ety especially in the following initial areas,recognized in the GEOSS 10-Year Imple-mentation Plan: “Improving the manage-ment and protection of terrestrial, coastal,and marine ecosystems” [2]. Hydrodynam-ical processes affect the spatial distributionand temporal development of phytoplank-ton biomass on the world’s oceans andseas. Among the hydrological events, thedivergent current bring nutrients into theupper layer of the water column and mod-ulates the chlorophyll−a distribution [3].Upwelling is a hydrological phenomenonthat strongly impacts the marine ecosys-tem. In fact, upwelling systems belong tothe most productive marine environments,and are characterized by increased ‘biolog-ical richness’ in all levels of the trophicchain. Low water temperature is one of

the indicators of upwelling, and the differ-ence of sea-surface temperatures betweenan upwelling zone and the surrounding wa-ters is a parameter for defining upwellingintensity [4]. In these environments, phy-toplankton growth is primarily regulated bythe availability of allochthon nutrients, pri-marily nitrates, which stimulate the pro-duction of phytoplankton [5].

2 Study area

The Straits of Messina (Figure 1), at thecenter of the Mediterranean Sea, is anarea where strong currents determine fastchanges in the oceanographic conditions.This system, separating the Italian penin-sula from Sicily, is an amphidromic pointfor the tides of the Tyrrhenian in the north-west and the Ionian seas in the southeast.Morphologically, the Strait resembles afunnel-shaped geometry with a northsouthlength of 40 km and a west-east width rang-ing from 3 km near the Tyrrhenian edge,to about 25 km at the Ionian open bound-ary. The narrowest section (Ganzirri–PuntaPezzo), which coincides with the sill re-gion, has a depth of about 80m and dividesthe area into northern and southern sectors.The sea bottom slopes steeply downward toa depth of 1000m at 19 km south of the sill.The northern sector has a gentler slope,and the 400m isobath is located 15 kmnorth of the sill toward the Tyrrhenian Sea.The Straits exhibits strong tidal currentsdriven by both barotropic and baroclinicprocesses, due to strong bathymetric con-straints exerted by the sill and coastal mor-phology [6]. Large gradients of tidal dis-placements are encountered in the Straits ofMessina, because the predominantly semi-diurnal north and south tides are approxi-mately in phase opposition. Due to both

2404


Figure 1: Study area.

phase opposition and topographic constric-tions, the current velocities can attain val-ues as high as 3.0ms−1 in the sill region,and are related to the position of the sun,the phase of the moon, the wind speed anddirection, and the air pressure distribution[7]. Surface water turbulence in the Straitsis mainly influenced by two types of circu-lation, a steady surface current and a turbu-lent mixing, both of which generate discon-tinuity of the thermo–haline distribution inthe surface layer. The steady current hasa maximum speed of 2m·s−1 and a preva-lently north to south direction at the surface(0–30m). Water turbulence is caused bothby internal waves and by tidal currents [8].Internal waves are caused by differencesin water mass densities of the two basinsthat are facing in the Straits. Tidal currentsare caused by opposite tidal oscillations ofthe two basins (max 27 cm) with almostthe same amplitude and period (about 6 1/4h). These conditions lead to the upwellingof deeper water of Levantine Intermediate

Water origin, which is colder, more saltyand more nutrient-rich compared to theTyrrhenian Surface Waters (Atlantic Waterorigin). Harmonic oscillations of the cur-rent flowing from the Tyrrhenian Sea wa-ters into the Ionian Sea (high tide current),and from the Ionian water into the Tyrrhe-nian Sea (low tide current) are encountered,with a brief slack water interval (balancingflow). Because of these particular environ-mental features, the Straits have been stud-ied by many researchers in order to defineforcing factors that determine its currentregimen [9]. Many hydrobiological stud-ies have been conducted over the last fewdecades to describe the influence of com-plex hydrodynamic conditions on biologi-cal parameters [10, 11].

2405

Technologies

Figure 2: MIVIS surveys with superimposed the cruise tracks followed by the R/B“Delfo”.

3 Instruments and meth-ods

Since 1994, the IAMC-CNR (Institutefor coastal marine environment) measurestemperature, salinity, and phytoplanktonby continuous underway surface measure-ments during the dedicated oceanographiccruises. During 1999-2002, remote sensingsurveys were performed with MIVIS air-borne sensor and specific in-situ measure-ments. Specific in-situ measurements meetthe requirements to validate the empiricalalgorithm applied to remotely data in or-der to retrieve the phytoplankton and, the

remote sensing provides a synoptic viewof upwelling phenomenon by mapping theSST and the phytoplankton into the waterbody of the strait.

3.1 Oceanographic cruisesSpatial and temporal distribution patternsof physicochemical (temperature, salinity,and nutrient concentrations) and biolog-ical (chlorophyll−a measured by in situfluorescence emission), data are presentlyavailable for the central-northern side ofthe Straits of Messina. Discrete water sam-ples (nutrients, chlorophyll−a, etc.) basedon conventional methods were collected

2406


during the surveys. The wind, air tem-perature and pressure were measured dur-ing each hydrographic cruise. Automaticreal-time data were obtained by monthlysurface tracking in the area between CapoPeloro (Sicily) to the north and Reggio Cal-abria to the south. This area was coveredon the R/B “Delfo” using zigzag cruisetracks between the Calabrian and Siciliancoasts, adopting the lagrangian method inorder to follow the wave of tide in theStrait. The location of the area and thetrack followed by the boat in each sur-vey is displayed in Figure 2. It iswellknown that the maximum intensity (> 3m·s−1 ) of the tidal currents occur dur-ing spring tides, corresponding to syzygylunar phases. Measurements were car-ried out when the water slackens dur-ing the spring tide period to individuatethe upwelling of deeper waters. Priorto taking measurements, the current fore-cast from “Tide tables of the Istituto Idro-grafico della Marina, Genova” was usedin order to seize the quasi-stationary sit-uations following the dynamic phases ofhigh (high tide current, 3 h/cruise) and low(low tide current, 3 h/cruise) tides. Pa-rameters taken into consideration in thepresent work are temperature and fluo-rescence, continuously (each 30 seconds)measured by means of a multiparameterprobe (Meerestechnich Elekronik) and afluorometer (Haardt-Mod1101LP).

3.2 Remote sensing surveys

3.2.1 Hyperspectral instrument

In situ: The ground-based acquisition havebeen performed by means of the Field-Spec Pro FR spectroradiometer (ASD).The ASD is a portable instrument cover-ing the spectral domain between 350 nm

and 2500 nm, sampled to 1 nm. Remote:The hyperspectral data have been acquiredby the airborne MIVIS (Multispectral In-frared and Visibile Imaging Spectrometer)sensor. The MIVIS is a whiskbroom scan-ner composed of 4 spectrometers coveringthe spectral domain from 420 to 12700 nmby 102 channels. The scanner geometry ischaracterized by a Field of View of 90°.

3.2.2 Methods

In situ: During the remote sensing sur-veys, field measurements were collected toinvestigate spatial variability of physical,chemical and biological parameters, in par-ticular in each station were collected wa-ter samples and performed Optical Mea-surement. The sampling stations werecollocated between Sicily and Calabriacoasts. Samples of water for the retrieveof CDOM, TSM (Total Suspended Matter)and chlorophyll-a at the surface and at 25m depth were collected. At the same timethe radiance upwelling from the sea sur-face and downwelling from the sky wasrecorded. All acquisition and elaborationprocedures followed the Ocean Optics Pro-tocols [12, 13]. The procedure for hyper-spectral sea surface reflectance measure-ment is based on the acquisition of the up-welling and downwelling radiance accord-ing to SeaWifs Protocol [14]. Whereas, thereflectance measurements over land are di-rectly the outputs of the instrument. Thereflectance is obtained from the relative ra-diative measurements achieved by a fastdouble acquisition (target and referencepanel) fixed in a nadir pointing configura-tion. During acquisition time some roleswere respected in order to facilitate Rrsestimation and avoid other undesirable ef-fects. At the same time of the remote sens-ing survey, reflectance spectra of some ma-

2407

Technologies

terials situated in land over Sicily and Cal-abria coastal areas, necessary for the atmo-spheric correction of the MIVIS images,were measured. Remote: The joint MIVIS(CNR LARA) and sailing vessel (CNRIST) campaign were carried out on 25thSeptember 1999 at 2:00 p.m., on 2nd July2000 at 12:00 a.m. and on 15th May 2002at 11:30 a.m. in correspondence of thesyzygy moon phase, when tidal currentsreach maximum intensity. The MIVIS re-mote sensed 3 scenes with a NNE-SSWand SSW-NNE direction at the altitudeof 4000 meters and a scanning speed of16.5 Hz (Figure 2). A further scene wasrecorded over Messina at the altitude of1500m (resolution at ground of about 3m).Preprocessing (sun glint effect): To assessthe suspended and dissolved matter in wa-ter in the visible and near infrared spectralregions it is necessary to estimate with ad-equate accuracy the water leaving radiance[15]. Consequently radiance measured bya remote sensor has to be corrected fromthe atmospheric and the sea surface effectsconsisting in the path radiance and the sunand sky glitter radiance contributions. Thispaper describes the application of the sunglint correction scheme on to airborne hy-perspectral MIVIS measurements acquiredon the area of the Straits of Messina duringthe campaign in July 2000. In the Messinacase study data have been corrected forthe atmospheric effects and for the sun-glitter contribution evaluated following themethod proposed by Cox and Munk [16].Comparison between glitter contaminatedand glitter free data has been made taking

into account the radiance profiles relevantto selected scan lines and the spectra ofdifferent pixels belonging to the same scanline and located outside and inside the sunglitter area (Figure 3). The results showthat spectra after correction have the sameprofile as the contaminated ones, although,at this stage, free glint data have not yetbeen used in water constituent retrieval andconsequently the reliability of such correc-tion cannot be completely evaluated [17].

4 Discussion and conclu-sions

The MIVIS has moreover stressed up-welling areas of the Straits of Messina,previously only hypothesised though neverascertained by the continuous traditionalmonitoring. As a matter of fact, becauseof the great space-time variability of thephysical-chemical and biological parame-ters, the monitoring of the Straits by meansof the vessel sailing was restricted bothfrom time and space points of view. There-fore, monitoring techniques offered by theMIVIS hyperspectral data have providedfor an instantaneous global view of theStraits integrating measures taken contin-uously by the sailing vessel. Then, themonitoring the Straits of Messina by meansof the combined MIVIS and sailing vesselcampaigns during the entire yearly cycle,could allow in the future a better vision ofthe sea circulation and its effects on the bi-ological activity of the system during theseasonal variations.

2408


Figure 3: Comparison between spectra before (in red) and after (in green) the sun glintcorrection, relevant to three different pixels located outside (top and bottom) and inside(central graph) the sun glitter area.

2409

Technologies

References[1] GEOSS. 10-Year Plan Reference Document. 2005.

[2] C. Bassani, C. Cattaneo, R.M. Cavalli, L. Fusilli, S. Pascucci, and S. Pignatti. TheHYPerspectral for Adriatic Coastal Monitoring (HYPAD.COM) project. ISRSE33rd International Symposium on Remote Sensing of Environment. Sustaining theMillenium Development Goals, 2009.

[3] F. Gomez, G. Gorsky, E. Garcia-Gorriz, and M. Picheral. Control of the phytoplank-ton distribution in the Straits of Gibraltar by wind and fortnightly tides. Estuarine,Coastal and Shelf Science, 59:485–497, 2004.

[4] A. Reul, V. Rodriguez, F. Jimenez-Gomez, J.M. Blanco, B. Bautista, T. Sarhan,F. Guerrero, J. Ruiz, and J. Garcia-Lafuente. Variability in the spatio-temporaldistribution and size-structure of phytoplankton across an upwelling area in theNW-Alboran Sea, (W-Mediterranean). page 589–608, 2005.

[5] C.T.A. Chen, L.Y. Hsing, C.L. Liu, and S.L. Wang. Degree of nutrient consumptionof upwelled water in the Taiwan Strait based on dissolved organic phosphorus ornitrogen. Mar. Chem., 87(3-4):73–86, 2004.

[6] T.S. Hopkins, E. Salusti, and D. Settimi. Tidal forcing of the water mass interfacein the Straits of Messina. J. Geophys. Res., 89:2013–2024, 1984.

[7] A. Defant. Physical Oceanography. 2, 1961.

[8] N. Alpers and E. Salusti. Scylla and Charybdis observed from space. J. Geophys.Res., 88(3):1800–1808, 1983.

[9] E. Bohm, G. Magazzu, L. Wald, and M.L. Zoccoletti. Coastal currents on theSicilian shelf south of Messina. Oceanologica Acta, 10:137–142, 1987.

[10] F. Azzaro, F. Decembrini, F. Raffa, and E. Crisafi. Seasonal variability of phyto-plankton fluorescence in relation to the Straits of Messina (Sicily) tidal upwelling.Ocean Science, 3:451–460, 2007.

[11] F. Azzaro, R.M. Cavalli, F. Decembrini, S. Pignatti, and C. Santella. Bio-Chemicaland Dynamical Characteristics of the Messina Straits Water by Means of Hyper-spectral Data. SPIE Proceedings Series, 4154(240-248), 2000.

[12] B.G. Mitchell, M. Kahru, J. Wieland, and M. Stramska. Determination of spectralabsorption coefficients of particles, dissolved material and phytoplankton for dis-crete water samples. In: J.L. Mueller, G.S. Fargion, and C.R. McClain (Ed.). OceanOptics Protocols for Satellite Ocean-Color Sensor Validation. NASA Tech. Memo.NASA Goddard Space Flight Center, Greenbelt, MD., 4(Rev4):39–54, 2003.

2410


[13] J.L. Muller. In-water radiometric profile measurements and data analysis protocols.In: J.L. Mueller, G.S. Fargion, and C.R. McClain (Ed.). Ocean Optics Protocols forSatellite Ocean-Color Sensor Validation. NASA Tech. Memo. NASA Goddard SpaceFlight Center, Greenbelt, MD., 3(Rev4):39–54, 2003.

[14] G.S. Fargion and J.L. Mueller. Ocean Optics Protocols for Satellite Ocean ColorSensor Validation: Revision 2. NASA Tech. Memo, pages 2000–2096, 2000.

[15] F. Santini, L. Alberotanza, R.M. Cavalli, and S. Pignatti. A two-step optimizationprocedure for assessing water constituent concentrations by hyperspectral remotesensing techniques: An application to the highly turbid Venice lagoon waters. Re-mote Sensing of Environment, 2010.

[16] C. Cox and W. Munk. Measurements of the roughness of the sea surface fromphotographs of the sun’s glitter. J. Opt. Soc. Am., 44:838–850, 1954.

[17] R.M. Cavalli, S. Pignatti, and E. Zappitelli. Correction of sun glint effect on MIVISdataof the Sicily campaign in July 2000. Annals of Geophysics, 49(1), 2006.

2411

Technologies

2412

Modular Multiparametric Analyser for AutomaticVessel Monitoring of Nutrients in Sea-Water

F. AzzaroInstitute for Coastal Marine Environment, CNR, Messina, [email protected]

Abstract

This paper describes the characteristics of a prototype of a modular multipara-metric analyser (MicroMAC FAST MP3) for automatic monitoring of sea water andanalytical methods for nutrients. The MicroMAC FAST reactor is an evolution ofthe basic LFA (Loop Flow Analysis) reactor. It has been conceived to assay ammo-nium, nitrate-nitrite and orthophosphate at low concentration in sea water samples.A sample analysis is 3-4 times faster than that obtainable with a standard LFA re-actor. With respect to the previous analyser a temperature control (30-52 °C) onthe measurement cell has been added (NH4, PO4), while the colorimeter and the re-lated links for transporting the sample have been moved beyond the Loop and forma hydraulic-optical set almost completely independent from the main LFA. All thesteps of a wet-chemical colorimetric analysis method are carried out in an analysiscycle sequentially. This new technique allows the preparation of two products of re-action which can be introduced at intervals of 150 seconds in the measurement cell.The statistical test show that the results of automated and manual analyses agree forall the examined parameters (≤4 % RSD). Multiparametric online analyzer: it ispossible to connect the analytical modules to a data logger with analogue and dig-ital signals, in order to have online simultaneous analysis of the sample. A typicalapplication is used during research at sea which vessel does not require an operator.

1 Introduction

In the last thirty years an ever increasingnumber of automatic tools for Colorimet-ric analysis, principally using “ContinuousFlow Analysis” [1, 2] and “Flow Injec-tion Analysis” [3, 4, 5] have been devised.The analysers of this kind, employed forthe determination of phosphorus and nitro-gen salts, generally have detection limita-tions, insufficient to satisfy the analyticaldemands concerning sea-water, generallycharacterized by poor nutrients.The need to have automatic monitoring op-erations during real-time determination of

some chemical parameters (such as nutri-ents) in environments at risk, has made therealization of an automatic chemical anal-yser suitable for onboard installation.The MicroMac Fast MP3 (MultiParamet-ric) is an automatic chemical analyser forrapid measurement of seawater (Figure1), based on analytical technology devel-oped by Systea (an Italian company) andnamed LFA (Loop Flow Analysis). LFAhas permitted levels never reached beforein analytical automation and in analyserminiaturization. All the steps of a wet-chemical colorimetric analytic method arecarried out in an analysis cycle sequen-


Technologies

NH4 NO3 PO4

Fig. 1.

66 cm42 cm

25 cm

NH4 NO3 PO4

Fig. 1.

66 cm42 cm

25 cm

NH4 NO3 PO4NH4 NO3 PO4

Fig. 1.

66 cm42 cm

25 cm

66 cm42 cm

66 cm42 cm

25 cm

Figure 1: Module ammonia, nitrate and orthophosphate.

tially. The LFR reactor components aresequentially connected to form a Loop,that can be opened (sampling position) orclosed (analysis position) by means of 2x3ways valve. The hermetic closed Loop pro-vides full protection against backgroundinterference, a basic requirement for stabletrace analysis. At the beginning of a cyclethe loop is washed and filled with a sam-ple. The sample color is measured for com-pensation. Small amounts of concentratedreagents are added and vigorously mixed.


2.1 MicroMac Fast descriptionThe MicroMac Fast is an evolution of thebasic reactor LFA and has been conceivedfor the fast determination of nutrients inseawater. In fact, the frequency of the

measurements is 4 times faster than thatachieved with previous versions of the in-strument µMAC 1000 [6].With respect to the LFA standard reac-tor, the colorimeter and the circuit hy-draulic connections to bring the sample andthe wash of the flowcell, has been shiftedoutside the loop and make an hydraulic-optical assembly independent from the nor-mal dosing loop (Figure 2). In the Fastsystem, the LFA standard reactor can makenormal and maximum speed, sampling andreagent dosing, similar to the normal LFAreactor. To speed up the reaction, a heat-ing bath (30–52 °C) has been inserted inthe circuit, which is made of a teflon tubecoiled around an aluminium heater (onlyfor modules NH4 and PO4). As in the nor-mal LFA reactor, a SAMPLE/LOOP valvepositioned in sample S and the peristalticpump P1 is activated in direct to draw the

2414


C2 C1

AIR

WASTE

V3

V10Valve

Sample/Loop

V9

V11 WATER

SAMPLE

CALIBRANT

WASTE

COLUMN Cd

AIR Buffer (for NO3)

NO3 R1 480 µl

NO3 R2 480 µl

PO4 R1 200 µl

PO4 R2 200 µl

V 6

V4

Pump 1

Pump 2

WASTE

COLORIMETER

50 mm

V4

Colorimeter circuit

Only for NO3

Loop Flow

Analysis

S

L

NH4 R1 720 µl

NH4 R2 240 µl

NH4 R3 240 µl

Figure 2: Hydraulic schematic diagram for determination of ammonia, nitrate and or-thophosphate.

sample. The sample is drawn up throughthe V9, starts to flow inside the Loop,washing out the liquid present here. At theend of the sampling, the S/L valve is placedin the Loop and the reagents are injectedand mixed as in the normal LFA reactor. Atthe end of the mixing, the S/L valve is po-sitioned in Sample, the valves V10 and V3are activated realising a link between LFAbase and the colorimeter circuit. The pumpP1 is activated in direct for a convenienttime to transfer the product of reaction in-side the colorimeter. The V10 and V3 aredeactivated, disconnecting the colorimetercircuit from the LFA base, which is con-nected to the waste line through of V10valve. An extra prolonged activation of theP1 pump, with V9 connected to the sample,allows for complete disposal of the prod-uct. A part of this has been transferred tothe colorimeter to fill the LFA base circuitwith the next sample. The steps requiredfor sampling are the dosing of reagents andtheir mixing with the sample just aspirated,

the product of reaction of the previous sam-ple which is still left inside the colorimeter.This allows sufficient time to for completecolor development, that reaches maximumefficiency during the mixing and heatingphases in the LFA part of the reactor.Before transferring the product of the re-action into the colorimeter, the raw opticaldensity of the sample is read and stored.The following wash cycle of the flowcelloccurs: activation of V3 valve, activationof V4, activation of P2 (to allow the aspira-tion of a small portion of air to increase thecleaning efficiency of the flowcell), deacti-vation of V4 and activation of P2 (to betterclean the flowcell), deactivation of P2 andV3 and measuring of the sample blank. Aspreviously said, the new product of reac-tion can be transferred to the flowcell bythe modality described above. This newtechnique allows the preparation of twoproducts of reaction which can be intro-duced at 150 second intervals in the mea-surement cell. The analytical time from

2415

Technologies

introducing a sample to having a result is300 seconds. The intensity of the colorof the reaction product is measured on thecolorimeter using a monochromatic lightbeam of specific wavelength.The Fast MP3 is a modular analyser andeach analytical module is independent. Nu-meric results are stored in a system mem-ory together with analytical graphics forfurther evaluation. The results are given inconcentration units.On line monitor: each module can be usedas a stand alone on line monitor or portableanalyzer with IP65 protection, the built-insoftware includes all the functions neededfor on line analysis (16-key keyboard): re-sults on display (2x16 row LCD), internalmemory up to 100 results, remote printoutof stored results, analog output for recorderconnection, RS232 serial port. The PCsoftware, expressly developed under Win95/98 OS, allows a simple and completemanagement of all the analyzer functions,from autocalibration to results presenta-tion.

2.2 Reaction and OD readingThe reaction takes place in all points ofthe reactor and therefore in the colorime-ter flow cell, thus allowing the monitoringof the reaction from time 0 (reagents injec-tion) to the end point.If methods require heating, to speed upthe colour development (i.e. for NH4,PO4) only the colorimeter flow cell will beheated at the requested temperature.At the reaction end point, measured OD(optical density) is stored.

2.3 CalculationsReagent Blank OD: stored and used to cal-culate the calibration factor.

Calibrant OD: stored and used to calculatethe calibration factor.Sample OD: stored together Sample BlankOD and used to calculate Sample concen-tration.Sample concentration = (Sample OD –Sample Blank OD – Reagent Blank OD)x Calibration factor.The wet chemistries used in Fast MP3 arethose recommended by international stan-dards [7].

2.4 Determination of ammoniain seawater

In this automated method, the ammoniaions present in the sample react with phe-nol and hypochlorite in an alkaline mediumaccording to Bethelot’s reaction. Thetrisodium citrate and EDTA are added tothe sample to avoid the precipitation of al-kaline hydroxides, while nitroprusside actsas a catalyst. The indophenol blue is mea-sured at 630 nm.Ammonia reagents. The complexingagent for seawater was trisodium citrate,dihydrate 25 g, EDTA, disodium salt 2g, sodium nitroprusside 0.4 g, deionizedwater (DIW) q.s. 100 ml. The phenolreagent was phenol, solid and colorless 6.4g, sodium hydroxide 2.75 g, DIW q.s. 100ml. The Chlorine reagent was dichloroiso-cianuric acid (D.I.C.), sodium salt 1 g,sodium hydroxide 4 g, DIW q.s. 100 ml.Ammonia standard solutions. The stan-dard stock 1000 mg·l−1 NH3 as N (so-lution A) was ammonium sulphate anhy-drous 4.714 g, chloroform 10 drops, DIWq.s. 1000 ml. One ml of solution A whendiluted to 100 ml of water gives a work-ing standard of 10 mg·l−1 ammonia (solu-tion B). 5 ml of solution B when dilutedto 500 ml of natural low nutrient seawater

2416


(LNSW) gives a working standard of 100ppb NH3 as N. That is actually introducedinto the analyzer for calibration.

2.5 Determination of nitrate-nitrite in seawater

In this automated method, the nitratepresent in the sample is reduced to nitrite ina coppered cadmium column, in a bufferedmedium. The nitrites formed and the onesalready present in the sample, react withsulphanilamide and N-(1-naphtyl) ethylen-diamine in acid medium to give a coloreddiazonium salt, which is measured at 550nm.Nitrate reagents. The washing waterwas deionized water. The sulphanilamide(SAA) reagent was sulphanilamide 2.5g, concentrated hydrochloric acid 25 ml,DIW q.s. 250 ml. The naphtylethylen-diamine (NED) reagent was N-(1-naphtyl)ethylendiamine x 2 HCl 0.375 g, DIW q.s.250 ml. Buffer solution was imidazole 4 g,concentrated hydrochloric acid 1 ml, DIWq.s. 1000 ml.Nitrate standard solutions. The standardstock 1000 mg·l−1 NO3 as N (solutionA) was sodium nitrate anhydrous 6.068 g,chloroform 10 drops, DIW q.s. 1000 ml.One ml of solution A when diluted to 100ml of water gives a working standard of 10mg·l−1 nitrate (solution B). 5 ml of solu-tion B when diluted to 500 ml of LNSWgives a working standard of 100 ppb NO3

as N. That is actually introduced into theanalyzer for calibration.

2.6 Determination of phosphatein seawater

In this automated method, the orthophos-phate present in the sample reacts with

molybdate in an acid medium to formphosphomolybdate, and then with ascorbicacid to form molybdenum blue, whose in-tensity is measured at 880 nm. The anty-monium catalyzes the reaction.Phosphate Reagents. The molybdatereagent was antimony potassium tar-trate 120 mg, ammonium heptamolybdatetetrahydrate 4.3 g, concentrated sulphuricacid 27 ml, DIW q.s. 250 ml. The ascorbicacid reagent was ascorbic acid 10 g, DIWq.s. 100 ml.Phosphate standard solutions. The stan-dard stock 1000 mg·l−1 PO4 as P (solu-tion A) was potassium hydrogen phosphateanhydrous 4.394 g, chloroform 10 drops,DIW q.s. 1000 ml. One ml of solutionA when diluted to 100 ml of water givesa working standard of standard 10 mg·l−1

phosphate (solution B). 5 ml of solutionB when diluted to 500 ml LNSW givesa working standard of 100 ppb PO4 as P.That is actually introduced into the ana-lyzer for calibration.

3 Results and analyses

The colorimetric methods examined havedifferent reaction rates reaching the maxi-mum value of optical density for ammoniaafter 60 minutes, nitrate after 8 minutes andphosphate after 25 minutes [8]. The differ-ence between the value of optical densityread at time T2 and that recorded at timeT1 is proportional to the nutrient concen-tration under examination. The Fast MP3is a sensitive instrument, also able to detectvariations in optical density in the order ofa tenth of mAU; therefore, it is not nec-essary for the colorimetric dose to waitfor the complete development of the color(Figure 3). The value of the time T2 usu-ally results from the compromise between

2417

Technologies

Figure 3: Recorder traces - Orthophosphate product of reaction in 300 seconds, frequencysamples each 150 seconds.

the need for the maximum development ofcolor and the number of analysed samplesper unit of time.An important aspect of the system concernsthe possibility sample blank measurementand colorimeter zeroing, measuring the op-tical density immediately after the additionof reagents and the aspiration in the cell,making the approximation that at T1 thecolorimetric reaction has not yet occurred.With this stratagem it is possible to correctthe value of the optical density, compensat-ing for variations linked to the instability ofthe lamp, salinity or torbidity of the matrix[9], without resorting to dual ray systems.In Table 1 the analytical protocols for thethree considered parameters in sea waterare summarised. The calibration of theanalysers may be done either with a read-ing of a standard (100 ppb) located in thereaction chamber or the factor may be setfrom a series of standard readings at vari-ous concentrations, aspirated as samples.

3.1 Accuracy and precision

The precision of the analysis was evaluatedby analysing a series of standard solutions(15, 10, 7, 5 and 2.5 ppb) 15 times forall the methods and calculating the relativestandard deviation % (RSD) of the mea-surements (Table 2). The analysis preci-sion proves satisfactory with a % RSD av-erage of 2.5 for the ammonia, 4.0 for thenitrates and 2.7 for the phosphates.

3.2 Comparison of manual andautomated chemical analy-ses

Intercalibration of manual and automatedprocedures is necessary to detect system-atic bias in either method, and to ensurethat data from automated analyses are com-patible with data from manual analyses.Stock Standards (15, 10, 7, 5, 2.5, 2 ppbprepared in low-nutrient seawater) wereused to test the agreement between theFast MP3 (lightpath 50 mm) and the spec-trophotometer determinations Varian Mod.Cary 50 (lightpath 50 mm).

2418


Test

1 2 3

Name ammonia nitrate+nitrite phosphate

sample introduction Aspiration Aspiration Aspiration

first reagent Citr. Hypoch. SAA Ammon. Molyb.

second reagent Phen.alk + DIC NED Ascorbic acid

mixing conditions Turbulent

colorimeter 630 nm 550 nm 880 nm

light path 50 mm 50 mm 50 mm

range 0.001 ÷ 1.3 absorbtion units

kind of measurem. end point end point end point

uses blank yes yes yes

sample blank automatic zeroing at the end of sampling

calibration automatic preparation of the working calibrant

units ppb ppb ppb

decimal digits 4 4 4

calib. ratio mode factor factor factor

Table 1: Analytical protocol for the ammonia, nitrate+nitrite and phosphate determina-tions.

3.3 Correlation coefficient

Results indicate good agreement of thetwo methods with a correlation coefficientR2= 0.94 for NH4, 0.98 for NO3 and 0.96for PO4, calculated on n= 86 (Figure 4).The determinable minimal concentrationfor methods nitrate and orthophosphate is2.5 ppb while for ammonium it is 5 ppb,considering that the readings below thatvalue were not perfectly repeatable as theywere for other concentrations. If the val-ues of 2 ppb were removed then the cor-relation coefficient would be R2= 0.99 forNH4, 0.99 for NO3 and 0.99 for PO4, cal-culated on n= 71

3.4 Instrumental drift

A standard solution (15 ppb) for eachmethod in seawater was analyzed, every150 seconds, per two hours. The solutionswere prepared in a 5 liters container and

stirred with a magnetic stirred. The results,shown in Figure 5 do not reveal any instru-mental drift.

3.5 Carryover studies

The effects of carry-over which may bepresent when solutions at highly variedconcentrations are sequentially analysedhave been evaluated. The Broughton car-ryover percentage (K) was calculated ac-cording to the equation: carryover (%) =[(L1 - L3)/(H3 - L3)] x 100 [10]. Threeconsecutive samples with high (H) concen-trations (ex 500 ppb of NH4) were mea-sured, followed by three samples with low(L) concentrations (ex 10 ppb of NH4), andthis sequence was repeated five times. Allthe measurements made as well as for thethree analysed parameters, a carry-over co-efficient superior than 0.3 % has not beenhighlighted.

2419

Technologies

Nutrient Species Known Concentration (ppb)

15 10 7.5 5 2.5

Ammonia Mean Concentrat. 15.01 9.64 7.30 4.68 2.10

Stand. Deviation 0.13 0.16 0.10 0.20 0.10

% RSD 0.84 1.68 1.37 4.13 4.55

Nitrate Mean Concentrat. 15.07 10.01 7.46 5.01 2.47

Stand. Deviation 0.37 0.32 0.18 0.29 0.15

% RSD 2.36 3.18 2.45 5.85 6.23

Phosphate Mean Concentrat. 15.00 10.00 7.52 4.97 2.46

Stand. Deviation 0.11 0.23 0.15 0.18 0.13

% RSD 0.74 2.35 2.01 3.54 5.11

Table 2: Accuracy and precision for each method.

Parameter detect. limit

(ppb)

time analys.

(second)

analysis rate

Samples/h

vol. samp

(ml)

vol. reag

(μl)

ammonia

5 300 24 10 1200

nitrate-nitrite

2.5 240 30 11 960

phosphate

2.5 240 30 10 400

Table 3: Main characteristics for each method.

The three chemical parameters examinedhave different analysis times. The maincharacteristics of the devised automaticmethods are summarised in Table 3.Anomalous values were recorded occa-

sionally during the various experiments,probably due to air bubbles in the circuitthat created errors in reading the sample;these determinations were not included inthe elaborations.

2420


N-NH4

y = 0,9856x + 0,8696

R2 = 0,9449

n=86

0

2

4

6

8

10

12

14

16

0 2,5 5 7,5 10 12,5 15

MicroMac Fast MP3 (ppb)

S

pec

tro

ph

oto

met

er (

ppb

)

a

A

N-NO3

y = 0,9342x + 0,696

R2 = 0,9848

n=86

0

2

4

6

8

10

12

14

16

0 2,5 5 7,5 10 12,5 15

MicroMac Fast (ppb)

Sp

ectr

op

ho

tom

eter

(p

pb)

B

P-PO4

y = 0,8801x + 1,2306

R2 = 0,9649

n=86

0

2

4

6

8

10

12

14

16

0 2,5 5 7,5 10 12,5 15

MicroMac Fast (ppb)

Sp

ectr

op

ho

tom

eter

(p

pb

)

C

Figure 4: Linear regression analysis relationship between automated and manual deter-minations of ammonia (A), nitrate (B) and orthophosphate (C).

2421

Technologies

40

50

60

70

80

90

100

110

0 20 40 60 80 100 120 140

minute

Ab

sorb

ance

x 1

00

0

NH4 15 (ppb) NO3 15 (ppb) PO4 15 (ppb)

at 550 nm

at 880 nm

at 630 nm

Figure 5: Instrument drift - A standard solution (15 ppb) for ammonia, nitrate and or-thophosphate in seawater was analyzed, every 150 seconds.

4 Conclusions

The results obtained suggest that the Mi-croMac Fast MP3 autoanalyzer is reliablefor the automatic performing of chemicalanalyses of ammonium, nitrate-nitrite andphosphates in sea water. The correlationwith traditional methods is optimal, alsoconsidering the other positive experimentssuch as the repeatability of data and theabsence of drift. In fact, the cleaning ofthe loops and the colorimeter is a very im-portant feature for the seawater analysis.Indeed, in flow instrumentation the driftis a limiting factor especially for the or-thophosphate analysis.In the future, with regard to the analysisrate, it would be better that all the moduleshad the same analysis time in such a way to

achieve coastal monitoring with the sameanalysis rate (every 150 seconds).In conclusion, the µMAC-FAST MP3meets the required characteristics for re-peatability, accuracy and reading speed,with universally recognised analyticalmethods. Analysis of nutrients on anequipped boat is the typical application ofMicroMAC Fast MP3 that can run a fullyautomated analysis of nutrients in seawa-ter. Near real-time results, the possibilityof connection with the boat positioningsystem and georeferencing, are the mainfeatures of the application.

5 AcknowledgementsThe study was carried out within MURST(Italian Ministry for University and Scien-

2422


tific Research) Cluster 10-SAM (SistemiAvanzati Monitoraggio) Project. The au-

thor is grateful to A. Marini and M. Gallettafor assistance with the analyses.

References[1] J. Barwell-Clarke and F. Whitney. Institute of Ocean Sciences Nutrient Methods

and Analysis. Can. Tech. Rep. Hydrogr. Ocean Sci., 182:vi – 43, 1996.

[2] X.A. Alvarez-Salgado, F. Fraga, and F.F. Perez. Determination of nutrient saltsby automatic methods both in seawater and brackish water: the phosphate blank.Marine Chemistry, 39:311–319, 1992.

[3] Y. Hiray, N. Yoza, and S. Ohashi. Flow injection analysis of inorganic ortho andpoly-phosphate using ascorbic acid as reductant of molybdophosphate. Chem. Lett.,5:499–502, 1980.

[4] C.B. Ranger. Flow Injection Analysis. Anal. Chem., 53:20A, 1981.

[5] K.S. Johnson and R.L. Petty. Determination of nitrate and nitrite in seawater byflow injection analisis. Oceanogr. Limnol., 28:1260–1266, 1983.

[6] F. Azzaro, E. Crisafi, G. Magazzu, F. Oliva, and A. Puglisi. Un nuovo fotometroautomatico per la determinazione di nutrienti da boa Oceanografica. pages 213–226, 1994.

[7] J.D.H. Strickland and T.R. Parsons. A practical handbook of seawate analysis. Bull.Fish. Res. Ed. Can., 167:1–311, 1972.

[8] N. Cardellicchio and M. Luzzana. Messa a punto di un prototipo di analizzatore col-orimetrico programmabile per la determinazione automatica dei nutrienti in acquadi mare. Atti del Seminario Tecnico Scientifico del Progetto Strategico “Monitor-aggio automatico dell’Inquinamento Marino nel Mezzogiorno” Lecce Italy, pages137–158, 1993.

[9] P.N. Trodelich and M.F.Q. Pilson. Systematic absorbance errors with TechniconAutoanalyzer II colorimeters. Water Reserch, 12:599–603, 1978.

[10] P.M.G. Broughton, A.H. Gowenlock, J.J. McCormack, and D.W. Neill. A revisedscheme for the evaluation of automatic instruments for use in clinical chemistry.Ann. Clin. Biochem., 11:207–218, 1974.

2423

Technologies

2424

R/V Luigi Sanzo: A Fifteen Meters FullyEquipped Boat for Coastal Monitoring

F. Azzaro, G. Zappala, E. CrisafiInstitute for Coastal Marine Environment, CNR, Messina, [email protected]

Abstract

Funded in the frame of the PI-CNR project (Cluster 10 MIUR), a coastal moni-toring boat was designed and built to be used by the Ist. Talassografico of Messina(now CNR-IAMC). Launched in 2002, the boat is named after the first director of theIstitute, Prof. Luigi Sanzo, a world-wide known ichthyologist. The boat contains onewet (8 m2 ) and one dry (10 m2 ) laboratory equipped with refrigerators and freezers.Fresh and salt water lines are available in the lower deck, in the wet laboratory andastern. Deck instruments include one 2000 kg hydraulic crane, one hydraulic winchwith 300 m of 6 mm armoured cable, one hydraulic winch with 200 m of 8 mm steelwire rope. A 1,5 m3 afterpeak is available to hold oceanographic instrumentationunder the 13 m2 wide stern. Instruments for biological, coastal oceanography andmonitoring researches are installed on the boat. A rosette equipped with 12-8 litresbottles and one CTDO probe with altimeter, fluorometer and turbidimeter enable toprofile the water column. Continuous monitoring on surface water is obtained by awater line feeding both a measurement chamber in which a CTDO probe with inte-grated fluorometer-turbidimeter is fitted and a group of automatic colorimetric ana-lyzers measuring NH4, NO3, PO4. Meteorological data are measured by a stationequipped with barometer, thermometer, anemometer, pyranometer. A data acquisi-tion computer integrates data measured by the scientific instruments with navigationdata coming from the GPS.

1 Introduction

Funded in the frame of the PI-CNR project(Cluster 10 MIUR), a coastal monitoringboat was designed and built to be used bythe Ist. Talassografico of Messina (nowCNR-IAMC). Launched in 2002, by theshipyard “NAVALTECNICA” in RomettaMarea (ME). The boat, owned and oper-ated by IAMC, is designed for research inthe coastal zone and instruments testing.It provides quick access to local coastalareas, but can work further offshore andcloser inshore than is customary for a boat

of its size. It also provides a special ca-pability for the Institution’s education pro-gram through support of student projectsand training [1, 2, 3, 4, 5, 6, 7, 8, 9, 10].This coastal monitoring boat rememberswith its name the first director of the Isti-tuto Talassografico in Messina, Prof. LuigiSanzo, a world-wide known ichthyologist.

2 Ship specificationsThe research vessel (R/V) L. Sanzo is ableto operate in the coastal zone, along watercolumn, to perform both basic and applied


Technologies

Figure 1: The boat R/V Luigi Sanzo.

scientific research programmes, survey ac-tivities and monitoring (Figure 1). The R/VL. Sanzo has a cruising range of 350 nau-tical miles with a typical endurance 1 day,occasionally 2-3 days. Its crew includes acaptain and a mate, one officer, and up to11 scientists on one-day trips. The lowerdeck hosts a bow cabin with four berthsand a fully equipped bathroom. Over thatcabin, a flying bridge is available. To ob-tain the maximum ease of use, three steer-ing and engine command sites are avail-able: in the cockpit, on the upper deck, andastern, near the crane (Figure 2a, 2b, 2c).The main features of the boat are as fol-lows:

- overall lenght 15 m;- beam 4,3 m;- max draft 1,2 m;- max speed 30 knots;- cruise speed 25 knots;- propulsion 2 Caterpillar 570 Hp engines;- electrical supply 8 kW - 220 V diesel

generator;- fuel tanks 3000 l;- water reservoirs 2000 l;- electrical bow-prop to maintain bearings

during samplings;- mobile wing system to trim attitude;

- fully air conditioned.

Navigation and safety instruments: Radar,echosounder, VHF, magnetic compass, au-topilot, GPS plotter. Control instruments:Engine tachometer, attitude and helm an-gle indicators. Crane/winches: Hydrauliccrane max load 2000 kg (6 m straddleand 360° rotation), Hydrological winchwith slip ring double-drum with 300 mof 6 mm armoured cable for rosette sys-tem and 200 m cable of 8 mm iron ropefor Side-Scan-Sonar, Sub-Bottom-Profiler,etc. (Figure 2d).

The main deck offers a wide open af-terdeck and a cabin hosting (from sternto bow) the “wet laboratory” (8 sqm), the“common room-dry laboratory” (10 sqm)including a kitchenette, the cockpit (Figure3). Laboratories are equipped with refrig-erators and freezers. Fresh and salt wa-ter lines are available in the lower deck,in the wet laboratory and astern. A 1,5m3 afterpeak is available to hold oceano-graphic instrumentation under the 13 sqmwide stern (Figure 4). The boat is equippedwith sampling and measuring instrumentsfor biological, coastal oceanography andmonitoring researches. Scientific instru-ments: SBE 32C rosette with 12 - 8 liters

2426


ba

c d

Figure 2: The steering commands: in the cockpit (a), on the upper deck (b), astern nearthe crane (c), and particular crane/winches (d).

GO-flow bottles and multiparametric probeSBE 911/plus provided with an indepen-dent data processing and restitution sys-tem for vertical profiles of pH, temper-ature, conductivity/salinity, oxygen, pres-sure, density, redox, altimeter, and SCUFAfluorometer-turbidimeter. Its instrumen-tation includes an acoustic Doppler cur-rent profiler (ADCP), flow-through seawa-ter sampling system, CTD system withconducting wire winch, and meteorologicalstation. Large transducer wells are avail-able for project use, and the transducerscan be changed out from inside the boatwhile in the water. System for contin-uous monitoring on surface waters (Fig-ure 5), comprising a SBE 19 plus CTDOprobe (Temperature, conductivity/salinity,pH, oxygen, density, redox), SCUFA inte-grated fluorometer-turbidimeter (Scufa II)from chlorophyll a (Chla), Systea nutri-ent analyzer Micromac fast MP3 (nitrate-

nitrite, ammonium and orthophosphate)determinations based on Loop Flow Anal-ysis able to detect low nutrient concentra-tions every 150 sec. [8]. The on-board im-plemented measuring system is constitutedby two connected instruments both auto-matically collecting parameters quickly (intime and in space). The output providesa real-time measurements stream and con-sequent data transmission to a PC by amultiplexed serial port. Integrated me-teorological box supplied with: barome-ter, thermometer, anemometer, psycrome-ter (all system is connected to a microcom-puter, true motion data). The scientific in-struments are interfaced together with thepositioning system to the main data acqui-sition computer, the data are available byvideo in the two laboratories.

2427

Technologies

a b c

Figure 3: ”Wet” Laboratory (a left-side, b starboard), and “dry/computer” laboratory (c).

Figure 4: Afterpeak locker of stern with the rosette housing.

3 Ship Activities

The first investigations: that have beenachieved were the Milazzo Gulf monitor-ing. The study was performed over 11months (from February to December 2003)in the framework of Cluster 10 SAM Pro-gramme (Advanced Monitoring Systems)funded by the Italian National Ministry forScientific Research [11, 3]. The adoptedmeasurement approach couples an auto-matic coastal platform and periodic surveysby a research vessel. Meteorological andwater column parameters were hourly col-lected by the platform. The survey strat-egy implemented with the research boat in-cluded a seasonal surface tracking alongthe coast with the on-board automatic sys-tem to measure CT-Fl and nutrients of Nand P and two oceanographic campaigns

in 11 stations along three transects verticalCTD profiles, nutrients of N and P, chloro-phyll a). The study area is the Gulf ofMilazzo, a natural bay of about 25 km2

laying other northern coast of Sicily andopen to the Tyrrhenian Sea. The cen-tral stretch of coast hosts two seasonally-controlled stream outflows. Cruises dataallowed to reconstruct the distribution ofChla integrated in the euphotic layer (0-80 m), which reflects the general hydro-dynamic features of the Gulf. Surface dis-tribution of physico-chemical and biologi-cal parameters obtained along tracking sur-vey are presented in Figure 6. The R/VLuigi Sanzo demonstrated to be a versatile,useful floating laboratory for coastal opera-tions, enabling scientists to obtain sinoptic-ity of observations also in rapidly changingwater masses.

2428


NH4 NO3 PO4

Fig. 1.

66 cm42 cm

25 cm

NH4 NO3 PO4

Fig. 1.

66 cm42 cm

25 cm

NH4 NO3 PO4NH4 NO3 PO4

Fig. 1.

66 cm42 cm

25 cm

66 cm42 cm

66 cm42 cm

25 cm

Pump

PumpCTD

µMAC Fast MP3

sea inlet

outlet

GPS

Acquisition

computer

NH4 NO3 PO4

Figure 5: System for continuous monitoring on surface waters.

Figure 6: The representative surface distribution of physico-chemical and biological pa-rameters (T, S, Chla and nutrients of N and P) obtained in tracking cruises.

4 AcknowledgementsThe authors thank Mr. Franceso Soraci andMr. Gianfranco Pandolfino, crew of the

boat for their help during surveys.

2429

Technologies

References[1] L. Alberotanza, F. Azzaro, F. Braga, R.M. Cavalli, S. Pignatti, S. Salviato, and

F. Santini. Definizione di un modello bio-ottico per le acque dello Stretto diMessina. 1:51–56, 2004.

[2] F. Decembrini, F. Azzaro, M. Galletta, and F. Raffa. Short-term changes of hydro-biological features in the Gulf of Milazzo (Tyrrhenian Sicily). 37:511, 2004.

[3] F. Decembrini, F. Azzaro, A. Bergamasco, and F. Raffa. Real-time tracking andmonitoring of sea-surface water quality: an experimental approach in a gulf ofTyrrhenian Sicily. 7:3763, 2005.

[4] A. Bergamasco, F. Decembrini, F. Azzaro, L. Guglielmo, and E. Crisafi. Caratter-istiche Idrologiche nell’AMP Isole Ciclopi (Costa Ionica Sicilia) e relazione con labiodiversita del comparto planctonico. Biol. Mar. Medit, 12(1):52–62, 2005.

[5] G. Zappala, G. Caruso, F. Azzaro, and E. Crisafi. Marine environment monitoringin coastal Sicilian waters. 95:337–346, 2006.

[6] G. Caruso, C. Garoppo, F. Azzaro, F. Raffa, and F. Decembrini. Comunita micro-bica nello stretto di Messina distribuzione e diversita funzionale. Biol. Mar. Medit.,13:112–113, 2006.

[7] G. Caruso, G. Zappala, G. Maimone, F. Azzaro, F. Raffa, and R. Caruso. Assess-ment of the abundance of actively respiring cells and dead cells within the totalbacterioplankton of the Strait of Messina waters. 99:10, 2008.

[8] F. Azzaro and M. Galletta. Automatic colorimetric analyzer prototype for highfrequency measurement of nutrients in seawater. Marine Chemistry, 99:191–198,2006.

[9] F. Azzaro, F. Raffa, A. Marini, and P. Rinelli. Caratteristiche idrobiologiche delGolfo di Gioia (Tirreno Sud Orientale): estate 2005. Biol. Mar. Medit., 14(2):336–337, 2007.

[10] F. Azzaro, E. Gangemi, A. Marini, and F. Raffa. Monitoraggio idrobiologico inun’area marina protetta: “Isole Ciclopi” (Sicilia orientale). 19:31–41, 2008.

[11] G. Zappala, G. Caruso, and E. Crisafi. The SAM integrated system for coastalmonitoring. pages 341–350, 2002.

2430

Water Quality Remote Sensing of the LargestEuropean Lagoon

C. Giardino1, M. Bresciani1, M. Bartoli21, Institute for Electromagnetic Sensing of the Environment, CNR, Milano, Italy2, Department of Environmental Sciences, University of Parma, Parma, [email protected]

Abstract

Coastal lagoons are transition zones between terrestrial and marine environments,characterized by high rates of primary and secondary productivity and elevated bio-diversity. Worlwide deterioration of these aquatic bodies, resulting from excess nu-trient inputs, has stimulated monitoring actions, addressed in particular to noxiousblooms of microalgae.In this study we use MERIS images to assess water quality in the highly eutrophiedCuronian Lagoon, the largest European coastal bay (1584 km2), bordering the south-eastern part of the Baltic Sea. The algorithm developing was firstly addressed onatmospheric correction and secondly on water quality parameters retrieving. In par-ticular, we focused on chlorophyll-a being an indicator of trophic level and on yellowsubstances because their role in protecting the aquatic biota from ultraviolet solar ra-diation and their influence on overall microbial activity in the water column. TwoMERIS images respectively acquired in spring and summer 2009 were elaborated byusing physically based approaches build using in-situ data collected synchronouslyto the sensor overpass. The results showed the patchy distribution of water qualityin the lagoon with higher average chlorophyll-a concentrations observed in summer(50 mg·m−3) than in spring (27 mg·m−3) and a slightly opposite situation for yellowsubstances (1.6 m−1 in spring and 1.3 m−1 in summer).

1 Introduction - Lagoonecosystems

Coastal lagoons are bodies of shallowsalt or brackish waters separated from thedeeper sea by sand bars, coral reefs, or sim-ilar features [1]. Lagoon ecosystems arecomplex habitats at the interface betweenterrestrial and marine environments, wherelight and nutrient availability and hydrody-namic features promote primary and sec-ondary productivity and elevated biodiver-sity.Highly productive in their ideal state, la-

goons accumulate most anthropogenic andsynthetic pollutants and are liable to be-come eutrophic. The coastal lagoons arehabitats of conservation importance and asin other parts of the world [2] support com-munities which are unique in structure anddiversity. They make up about 13% ofglobal coastal environments [3] and have anumber of physiographic attributes whichincrease habitat heterogeneity not only spa-tially by the presence of a multitude ofecotones providing refugia for diverse ma-rine fauna and flora [4], but also tempo-rally with the seasonal hydrological cy-


Technologies

cle [5, 6]. These environments have alsoa high economic value, as locations forextensive aquaculture, and scientific valueas natural laboratories to investigate short-medium term effects of anthropogenic im-pacts, as well as restoration plans.Unfortunately, in many countries deteriora-tion of lagoon’s habitats has occurred (e.g.,[7, 8, 9]) and to prevent further loss and en-sure the safeguard of existing ecosystems,it is essential to monitor their contiguousuplands, and their water components. Sim-ilarly to any aquatic natural environment,the continuous monitoring of key biolog-ical and environmental parameters mightpresent some major concerns since the phe-nomena to be observed in the lagoon mightbe characterized by a high rate of change,both in time and space. In particular, thecollection of in-situ data with traditionaltechniques may not be sufficient for under-standing the characteristic of such targetswhere the water quality conditions maychange rapidly as a consequence of tidesor meteorological constraints.When deterioration of water quality iscaused by optically active substances, theeffect of these changes can be observedwith optical remote sensing instruments.Remote sensing may therefore present asignificant integrating technique for moni-toring water quality in coastal lagoons. Be-sides to improve the spatial and temporalfrequency of observation, satellite sensorsalso provide useful data to monitor lagoonslocated in remote regions.

2 Backgrounds - Waterquality remote sensing

Remote sensing techniques have been ex-tensively adopted for assessing water qual-

ity in oceans, coastal zones, lakes and tran-sitional ecosystems as lagoons and deltas(e.g., [10, 11]). For their investigations,these aquatic environments have been tra-ditionally split into two main categories,depending by the complexity of their inher-ent optical properties [12, 13]: case-1 wa-ters are those waters whose optical prop-erties are determined primarily by phy-toplankton and related yellow substancesand degradation products (e.g., oceans andseas). Case-2 waters are everything else,namely waters whose optical propertiesare significantly influenced by other con-stituents such as mineral particles, yellowsubstances originated either from degrada-tion of phytoplankton or land runoff andtributaries inputs, whose concentrations donot covary with the phytoplankton con-centration (e.g., coastal zones, estuaries,lagoon and inland waters). In the late1970s and early 1980s the idealized con-cept of case-1 and case-2 water provideduseful guidance for the development of thefirst generation of bio-optical models whilenowadays it might be considered too muchrestrictive [14].In recent years, the developments in waterquality algorithms have mainly been drivenby the advent of the SeaWiFS (Sea-viewingWide Field of View Sensor), MODIS(Moderate Resolution Imaging Spectrora-diometer) and MERIS (Medium Resolu-tion Imaging Spectrometer) satellites andtheir specific capacities to sense the low ra-diances of aquatic ecosystems. The spec-ifications of these sensors, their frequentdata acquisition, high signal-to-noise ra-tio, and number and position of spectralbands, make them better suited than higherresolution sensors with broad bands, suchas Landsat, for regular/real-time monitor-ing applications observing change occur-ring over short time scales. However, the

2432


spatial resolution of ocean colour satellitesis usually only about 1 km, which meansthat the satellite does not show variabilityat a small-local scale and is not adequatefor the monitoring of small to medium-sized targets. Only MERIS, with its spa-tial resolution of about 300 m at Full Res-olution (FR) mode, offers possibilities forimproved coastal remote sensing. Fur-thermore, it also has a better spectral res-olution than SeaWiFS and MODIS. TheMERIS band 9, centered at 708 nm, hasbeen shown to be vital for the detectionof chlorophyll in turbid waters [15]. TheMERIS band 6 around 620 nm was shownto be of high interest in the detection ofcyanobacterial blooms since this band co-incides partially with the absorption peakof the main cyanobacterial pigment phyco-cyanin [16].In general, the parameters that have beenidentified in the literature as detectable inaquatic environments by modern satellitesare: (1) the diffused attenuation as mea-sure for the water transparency in the eu-photic zone (e.g., [17]). The euphotic zonedepth (Ezd) reflects the depth where pho-tosynthetic available radiation is 1% of itssurface value. The value of Ezd is animportant parameter regarding ecosystemssince in the Ezd most of the aquatic lifeoccurs. (2) Green algae pigments mainlyas chlorophyll-a (chl-a) used as a proxy ofphytoplankton biomass and primary pro-duction (e.g., [18]). (3) Total suspendedmatter (TSM) which is placed in suspen-sion by wind-wave stirring of shallow wa-ters and is a tracer for inflowing pollutants(e.g., [19]). (4) Yellow substance (YS, alsoknown as CDOM or coloured dissolved or-ganic matter), which protect the aquaticbiota from ultraviolet solar radiation andinfluence on overall microbial activity inthe water column [20]. (5) The cyanobac-

terial pigment phycocyanin, potentially as-sociated to harmful algal blooms [21].Various methods were developed to esti-mate the water constituents from remotesensing images. Empirically based ap-proaches using experimental data sets andstatistical regression techniques generateempirical algorithms relating the water-leaving reflectances or radiances at the sen-sor in specific spectral bands or band ra-tios/combinations to in-situ water qualityparameter measurements (e.g., [22]). Thesimplicity of the empirical approach meansthat it is easy to implement (especially ininstances where in-situ data is regularlycollected) and that the algorithms are gen-erally robust [23]. However, the algorithmsare limited to the specific constraints of thedata set from which they are derived (andare thus not applicable across seasons orareas), perform accurately only inside theirrange of derivation, and have a limited abil-ity to separate signals from different waterconstituents, especially in instances wherethere is covariance.More complex semi-analytical approachesuse a variety of inversion techniques, eitheras in-water or coupled water atmospherealgorithms, and are potentially more pow-erful in their ability to separate the signal ofdifferent in-water constituents (e.g., [24])but their use of many, if not all of the spec-tral bands available, makes them more sen-sitive to errors in the atmospheric correc-tion [25]. This approach have the advan-tage of being independent of concurrent in-situ measurements which makes them wellsuited to operational water quality monitor-ing systems, however it requires that thebio-optical specific inherent optical prop-erties of the water body are known before-hand.

2433

Technologies

Figure 1: MERIS data acquired on 22 July 2009 with location of in-situ stations sam-pled in March and July 2009. The pseudo true colour MERIS image clearly shows thedifferent optical behaviour of the Curonian Lagoon and the Baltic Sea waters.

3 Case study - The Curo-nian Lagoon

The Curonian Lagoon (Figure 1), belong-ing to Lithuania and to Russian Federation,is located in the south-eastern part of theBaltic Sea, from which it is separated bya narrow sand spit. With a total area of1584 km2, it is the largest lagoon in Eu-rope. This water body receives nutrient-rich freshwater inputs from different trib-utaries, mainly from the Nemunas River,and occasionally wind-driven low-salinitywater from the Baltic sea. Complex watercirculation, variable depth, different sedi-ment features and nutrient availability af-fect primary production and influence theconcentration and distribution of nutrients

and bio-chemical elements, thus leading tocomplex ecosystem.Formed about 9,000 years ago, the Curo-nian Lagoon might be more properly con-sidered as a freshwater body since the saltwater intrusion is limited to the northernarea only, where salinity varies from 0 to8 PSU (Practical Salinity Unit) [26]. Itis a shallow (average depth 3.7 m) transi-tory basin which is highly biodiverse, al-though troubled by water pollution. In par-ticular, cyanobacterial blooms, which wereobserved recently [27] represent a majorconcern for water quality issues and poten-tial risk for human health. The need andimminence for water quality monitoring inthe Curonian Lagoon have been thereforerecognized by the increasing of eutrophica-tion that overall remains the main ecolog-

2434


ical problem in the Baltic coastal lagoonsand estuaries [26].The Curonian Lagoon is also a good site fortesting and developing techniques for re-mote sensing in optically complex waters,dominated by cyanobacteria and, similarlyto both Boreal lakes and Baltic Sea, rich ofyellow substances. The lagoon’s wide size,the spatial heterogeneity of water qualitypatterns, and covariant water constituentspresent a challenging case for remote sens-ing both in terms of algorithm developmentand atmospheric correction.

4 Material and methods

4.1 Field campaigns

Field campaigns represent an importanttask to both develop robust image pro-cessing algorithms and to validate remotesensing-inferred products. In this studythey were in particular performed for: (i)validating the performances of the atmo-spheric correction models, (ii) building asemi-empirical band-ratio based model forchl-a and (iii) assessing the accuracy ofMERIS-derived products. Two field cam-paigns were conducted in March 2009 andJuly 2009 in coincidence to MERIS ac-quisitions. Remote sensing reflectance(Rrs) values were derived from submergedmeasurements of upwelling radiance anddownwelling irradiance subsequently cor-rected for the emersion factor.Water samples were collected at the sur-face and filtered for laboratory analysis ofwater quality parameters. Chl-a concen-trations were determined according to theanalytical method ISO 10260-E [28]. ForYS, the absorbance of the filtered waterwas measured in a 5-cm cuvette, by us-ing another cuvette filled with distilled wa-

ter as a blank. A baseline correction inthe range around 685 nm was applied andabsorbance was converted into absorptioncoefficient of YS at 440 nm as showed inStrombeck and Pierson [29]. In the samestation a multi-wavelength probe was usedin-situ to derive the algal composition ofphytoplankton.

4.2 Image processingMERIS is a medium resolution imaging in-strument carried aboard the ESA’s Envisat-1 satellite and operating since 2002 withina mission covering open oceans, coastalzone and land surfaces. The instrument’s68.5° field of view around nadir coversa swath width of 1150 km. MERIS isdesigned to acquire 15 spectral bands inthe visible near-infrared wavelengths. Ta-ble 1 shows the position of MERIS spec-tral bands determined following a detailedspectral characterization of the instrument.The spectral range is restricted to the visi-ble near-infrared part of the spectrum andthe spectral bandwidth is variable depend-ing on the width of a spectral feature to beobserved and the amount of energy neededin a band to perform an adequate obser-vation. Driven by the need to resolvespectral features of the oxygen absorptionband occurring at 760 nm a minimum spec-tral bandwidth of 2.5 nm was designed.In order to derive water quality parame-ters from MERIS data it is firstly neces-sary to correct image data for the atmo-spheric effects. Compared to other natu-ral surfaces, such as soils and vegetation,the fraction of light reflected from water isvery small [30]. Water-leaving radiancesare commonly less than 10% of the to-tal radiance measured at the sensor whilethe 90% is represented by the signal re-flected towards the sensor by the atmo-

2435

Technologies

Table 1: Position of MERIS spectral bands (band centre and bandwidth) and potentialfield of application.

sphere [30, 31]. Atmospheric correctionmodels are numerous but they are almostall based on the radiative transfer equa-tions that might be resolved in more sim-ple or rigorous way depending by initial as-sumptions. It is also quite common to dis-tinguish the atmospheric correction mod-els into two categories: image-based tech-niques and radiative transfer codes [32].In any case, both the low signals thatneed to be retrieved and the optical com-plexity of coastal, transitional and inlandwaters make quite challenging the atmo-spheric correction of satellite images forwater quality applications. In fact, incor-rect values in obtaining water-leaving ra-diances consequently produce large errorsin retrieving concentrations of water qual-ity parameters [33].In this study two MERIS FR level-1 im-ages, acquired on 26 March and 22 July2009 were corrected for the atmosphericeffects with two radiative transfer codeschosen verifying they performances by us-ing in-situ measured Rrs data. One algo-rithm [34] is implemented in the Basic ERS& Envisat (A)ATSR and MERIS (BEAM)toolbox and it was chosen because appro-

priately designed both for performing theatmospheric correction and for assessingthe water quality parameters in inland wa-ters. In particular the plug-in Case 2-Regional (C2R) was used. The other modelis 6S (Second Simulation of the SatelliteSignal in the Solar Spectrum) [35, 36] thatwas run using appropriate atmospheric andaerosol models. The comparison of atmo-spheric correction outputs with in-situ datarevealed an overall better agreement whenthe BEAM C2R code was used but focus-ing on red and near infrared wavelengths,which are commonly used to assess chl-ain eutrophic waters (e.g., [18, 23], cf. Ta-ble 1), the 6S code performed better [37].

4.3 Water quality algorithms

As seen in the previous paragraph, the C2Ralgorithm appropriately designed to correctMERIS data for the atmospheric effectsand simultaneously providing concentra-tions of water quality in case-2 waters hasbeen recently developed [34]. In this studyit was used to generate the map of YS,since the algorithms used for assessing YSmainly uses the shorter wavelengths [24]

2436


where the BEAM C2R processor providedaccurate values of Rrs. For mapping chl-a concentrations image data atmospheri-cally corrected with 6S code were preferredsince more reliable to capture the relativepeak of Rrs around 700 nm. The 6S-derived Rrs values in bands 9 and 7 werecombined in a band-ratio algorithm devel-oped by using in-situ data [37]. The al-gorithm, which is similar to those devel-oped in eutrophic lakes (e.g., [18, 23], wasenough robust (R2=0.94) to be applied tothe atmospherically corrected MERIS im-ages.For detecting cyanobacterial blooms re-search activities are rather new but in caseof abundant concentrations preliminary re-sults are promising [21]. In this study, Rrsvalues in MERIS bands 6 and 7 were inves-tigated for detecting phycocyanin absorp-tion feature near 630 nm and a small peakin reflectance spectra near 650 nm, charac-teristic of cyanobacteria only.


Widely variable water components condi-tions were measured in the study area:in-situ derived chl-a concentrations var-ied between 12 and 32 mg·m−3 in Marchand between 44 and 73 mg·m−3 in July.The variation range of YS was lower andlimited from 0.7 to 2 m−1 with slightlyhigher values in March than in July (Fig-ure 2). As shown in Figure 2, chl-aand YS were not related (R2 <0.15),suggesting that the Curonian Lagoon wa-ters belong to case-2. The phytoplank-ton composition revealed (Figure 2) that inMarch diatoms and dinoflagellates species(58%) prevailed, while in July the mostpredominant were cyanobacteria species(63%), with concentrations ranging from

17 to 50 mg·m−3. According to Kutser[21], who suggested that MERIS can beused in detecting cyanobacteria if they arepresent in relatively high quantities only,the cyanobacteria map was derived for im-age data acquired on 22 July 2009 only.Figure 4 presents the pseudo true colourMERIS image and related products in theseasons. The pseudo true colour MERISimages qualitatively describe the diversityof waters within the lagoon. The brown-ish yellow colours observed in March aredue to combinations of yellow substancesand phytoplankton, while in July the watersare greenish indicating the phytoplanktonas the major responsible of water colour.Both images show significant colour varia-tions within the lagoon and several patternswith eddies structure. Then, it is also veryevident the differences in water colour ofthe lagoon with respect the nearby BalticSea, caused by the differences in the opticalproperties of their waters. The water qual-ity maps show a quite heterogeneous pat-tern of YS in March while in July the dis-tribution is more homogeneous. The chl-aconcentrations are highly heterogeneous inboth seasons with values definitely greaterthan those observed in the Baltic Sea.The concentrations of both chl-a and YSderived from image data in correspon-dence of the sampling stations are in agree-ment with in-situ data (Table 2), confirm-ing the capabilities of the proposed methodin assessing water quality in this complexecosystem. Figure 5 shows the map de-picting the distribution of the cyanobac-terial bloom observed in July 2009 theevent observed one week later, on 28 July2008. The map shows a degree of cor-relation with the map of chl-a concentra-tion of the same day as revealed by in-situdata too. Nevertheless it is interesting toobserve the patchy distribution of algae in

2437

Technologies

Table 2: Comparison of MERIS-derived products and in-situ data. The statistics, bothfor in-situ and image data, are computed for the 15 and 10 stations measured in Marchand July, respectively.

Figure 2: Chl-a concentrations plotted versus and YS concentrations. The poor correla-tion (R2 <0.15 in both seasons) is typical of case-2 waters.

the lagoon: with higher values in the north-ern part nearby the city of Klaipeda, in thecentral area in correspondence of the maintributary and in the southern part close tocoastlines of the Russian Federation.

6 Conclusions - Limita-tions and challenges

This study showed how remote sensingmight offer an important source of in-formation for monitoring water quality incomplex ecosystems, as the Curonian La-

goon. The results indicate the feasibil-ity of using MERIS for monitoring wa-ter quality in transitional water bodies.The maps produced enabled the temporaland spatial variability and range of wa-ter constituent concentrations to be ob-served in a synoptic manner unequalled byconventional monitoring techniques. TheMERIS-derived products provided insightswhich might be combined with water cir-culations maps, meteorological parameters(wind speed and direction), inventory ofpunctual nutrient inputs and map of sedi-ment features for a better understanding of

2438


Figure 3: Phytoplankton composition measured in-situ (mean value of the stations) inMarch and July 2009.

the Curonian Lagoon ecology.Nevertheless, the algorithm developing inoptically complex waters, both in terms ofatmospheric correction and water qualityretrieval still needs further improvementsto fulfil a real-time monitoring. In par-ticular, efforts are needed to establish ro-bust relationships between remote sensing-inferred data (e.g., Rrs or inherent opti-cal proprieties) and concentrations of waterquality parameters since in optically com-plex waters the relations between those pa-rameters might have a local-regional be-haviour and vary over time. The demon-strated capabilities of MERIS in detect-ing cyanobacterial algal blooms, at leastin case of massive blooms, might alsoreceive increasing attention in the widerscientific community and water managerssince there is evidence to suggest that eu-trophic conditions lead to increasing dom-inance of cyanobacterial species. We be-lieve that further studies on MERIS time-

series will allow accurate, reliable and fre-quent screening of water quality, in particu-lar phytoplankton blooms, in this vast areapermitting to explore the complex dynam-ics and the short-term evolution of bloomsin the Curonian Lagoon. Finally, the find-ings present substantial opportunities forimproving monitoring in other transitionaland coastal waters with similar water qual-ity problems.

7 AcknowledgementsMERIS data were made available throughthe ESA AO-553 MELINOS project. Thisstudy was co-funded by the University ofKlaipeda-Coastal Research and PlanningInstitute. We are grateful to all the peopleinvolved in the fieldwork activities and lab-oratory analyses performed in this study,special thanks are due to Renata Pilka-ityte and Arturas Razinkovas (University ofKlaipeda).

2439

Technologies

Figure 4: The pseudo true colour image and the two water quality products obtainedfrom MERIS data acquired on 26 March (column on left) and on 22 July 2009 (columnon right).

Figure 5: Cyanobacterial bloom in the Curonian Lagoon: on left the map of phycocyaninpigments derived from MERIS data acquired on 22 July 2009, on right the photos of thebloom and of its sampling taken some days later.

2440


References[1] G.K. Reid. Ecology of Inland Waters and Estuaries. 1961.

[2] R.N. Bamber, S.D. Batten, N.D. Bridgwater, and M. Sheader. On the ecology ofbrackish water lagoons in Great Britain. Aquatic Conservation: Marine and Fresh-water Ecosystems, 2:65–94, 1992.

[3] R.S.K. Barnes. The coastal lagoons of Britain. An overview and conservation ap-praisal. Biological Conservation, 49:295–313, 1989.

[4] L.J. Chapman, C.A. Chapman, and M. Chandler. Wetland ecotones as refugia forendangered fishes. Biological Conservation, 78:263–270, 1996.

[5] C. Gordon. Hypersaline lagoons as conservation habitats: macro-invertebrates atMuni Lagoon, Ghana. Biodiversity and Conservation, 9:465–478, 2000.

[6] G. Izzo, G. Massini, G. Migliore, and C. Varrone. Indicatori funzionali della sta-bilita ecologica degli ambienti lagunari. 2004.

[7] J.W. Day, F. Scarton, A. Rismondo, and D. Are. Rapid deterioration of a Salt Marshin Venice Lagoon, Italy. Journal of Coastal Research, 14:583–590, 1998.

[8] A. Newton and S.M. Mudge. Lagoon-sea exchanges, nutrient dynamics and waterquality management of the Ria Formosa (Portugal). Estuarine, Coastal and ShelfScience, 62:405–414, 2005.

[9] W. Gong, J. Shen, and J. Jia. The impact of human activities on the flushing prop-erties of a semi-enclosed lagoon: Xiaohai, Hainan, China. Marine EnvironmentalResearch, 65:62–76, 2008.

[10] T. Lindell, D.C. Pierson, G. Premazzi, and E. Zilioli. Manual for monitoring Euro-pean lakes using remote sensing techniques. page 164 pp, 1999.

[11] V.E. Brando and A.G. Dekker. Satellite hyperspectral remote sensing for estimatingestuarine and coastal water quality. IEEE Transaction on Geoscience and RemoteSensing, 41:1378–1387, 2003.

[12] A. Morel and L. Prieur. Analysis of variations in ocean color. Limnology Oceanog-raphy, 22:709–722, 1977.

[13] A. Morel. Optical modeling of the upper ocean in relation to its biogeneous mattercontent (Case 1 waters). Journal of Geophysical Research, 93:10749–1076, 1988.

[14] C.D. Mobley, D. Stramski, W.P. Bissett, and E. Boss. Optical Modeling of OceanWaters: Is the Case 1 - Case 2 Classification Still Useful? Oceanography, 17:61–67, 2004.

2441

Technologies

[15] R.J. Vos, J.H.M. Hakvoort, R.W.J. Jordans, and B.W. Ibelings. Multiplatform op-tical monitoring of eutrophication in temporally and spatially variable lakes. TheScience of the Total Environment, 312:221–243, 2003.

[16] S.G.H. Simis, S.W.M. Peters, and H.J. Gons. Remote sensing of the cyanobacterialpigment phycocyanin in turbid inland water. Limnology Oceanography, 50:237–245, 2005.

[17] Z. Lee, B. Casey, R. Arnone, A. Weidemann, R. Parsons, M.J. Montes, B.C. Gao,W. Goode, C.O. Davis, and J. Dye. Water and bottom properties of a coastal envi-ronment derived from Hyperion data measured from the EO-1 spacecraft platform.Journal of Applied Remote. Sensing, 1, 2007.

[18] A.A. Gitelson, J.F. Schalles, and C.M. Hladik. Remote chlorophyll-a retrieval inturbid, productive estuaries: Chesapeake Bay case study. Remote Sensing of Envi-ronment, 109:464–472, 2007.

[19] A.G. Dekker, R.J. Vos, and S.W.M. Peters. Comparison of remote sensing data,model results and in-situ data for total suspended matter (TSM) in the southernFrisian lakes. 268:197–214, 2001.

[20] T. Kutser, D.C.Pierson, K. Kallio, A. Reinart, and S. Sobek. Mapping lake CDOMby satellite remote sensing. Remote Sensing of Environment, 94:535–540, 2005.

[21] T. Kutser, L. Metsamaa, N. Strombeck, and E. Vahtmae. Monitoring cyanobacterialblooms by satellite remote sensing. Estuarine, Coastal and Shelf Science, 67:303–312, 2006.

[22] C. Giardino, M. Pepe, P. A. Brivio, P. Ghezzi, and E. Zilioli. Detecting chloro-phyll, Secchi disk depth and surface temperature in a Subalpine lake using Landsatimagery. The Science of the Total Environment, 268:19–29, 2001.

[23] M. Bresciani, C. Giardino, D. Longhi, M. Pinardi, M. Bartoli, and M. Vascellari.Imaging spectrometry of productive inland waters. Application to the lakes of Man-tua. Italian Journal of Remote Sensing, 41:147–156, 2009.

[24] C. Giardino, V. E. Brando, A. G. Dekker, N. Strombeck, and G. Candiani. Assess-ment of water quality in Lake Garda (Italy) using Hyperion. Remote Sensing ofEnvironment, 109:183–195, 2007.

[25] L. Guanter, A. Ruiz-Verdu, D. Odermatt, C. Giardino, S. Simis, V. Estelles,T. Heege, J. A. Domınguez-Gomez, and J. Moreno. Atmospheric correction ofENVISAT/MERIS data over inland waters: Validation for European lakes. RemoteSensing of Environment (in press), 2010.

[26] A. Krevs, J. Koreiviene, R. Paskauskas, and R. Sulijiene. Phytoplankton productionand community respiration in different zones of the Curonian lagoon during themidsummer vegetation period. Transitional Waters Bulletin, 1:17–26, 2007.

2442


[27] A. Paldaviciene, H. Mazur-Marzec, and H. Razinkovas. Toxic cyanobacteriablooms in the Lithuanian part of the Curonian Lagoon. Oceanologia, 51:203–216,2009.

[28] ISO 10260-E. Water quality Measurement of biochemical parameters Spectropho-tometric determination of Chlorophyll-a concentration. 1992.

[29] N. Strombeck and D.C. Pierson. The effects of variability in the inherent opticalproperties on estimations of chlorophyll a by remote sensing in Swedish freshwater.The Science of the Total Environment, 268:123–137, 2001.

[30] G.A. Maul. Introduction to satellite oceanography. Martinus Nijhoff Publisher,Dordrecht, 1985.

[31] P. A. Brivio, C. Giardino, and E. Zilioli. Calibration and validation of satellite datafor quality assurance in lakes monitoring applications. The Science of the TotalEnvironment, 268:3–18, 2001.

[32] Y.J. Kaufman. The atmospheric effect on remote sensing and its correction. In:Theory and applications of optical remote sensing. pages 336–428, 1989.

[33] P.A. Keller. Comparison of two inversion techniques of a semi-analytical model forthe determination of lake water constituents using imaging spectrometry data. TheScience of the Total Environment, 268:189–196, 2001.

[34] Doerffer R. and Schiller H. MERIS regional, coastal and lake case 2 water project- Atmospheric Correction ATBD. Version 1.0. 2008.

[35] E.F. Vermote, D. Tanre, J.L. Deuze, M. Herman, and J.J. Morcette. Second Simu-lation of the Satellite Signal in the Solar Spectrum, 6S: an overview. IEEE Trans-action on Geoscience and Remote Sensing, 35:675–686, 1997.

[36] S.Y. Kotchenova, E.F. Vermote, R. Matarrese, and F.J. Klemm Jr. Validation ofa vector version of the 6S radiative transfer code for atmospheric correction ofsatellite data. Part I: Path radiance. Applied Optics, 45:6762–6774, 2006.

[37] C. Giardino, M. Bresciani, R. Pilkaityte, M. Bartoli, and A. Razinkovas. In situmeasurements and satellite remote sensing of case2 waters: preliminary resultsfrom the Curonian Lagoon. Oceanologia, 52:197–201, 2010.

2443

Technologies

2444

The Erythropoietin and Regenerative Medicine: aLesson from Fish

G. Maricchiolo1, A. Lacquaniti2, D. Bolignano2, A. Favaloro2, A. Buemi2,G. Grasso3, V. Donato2, G. Giorgianni2, L. Genovese1, G. Coppolino2, A.Sfacteria2, M. Buemi21, Institute for Coastal Marine Environment, CNR, Messina, Italy2, University of Messina, Messina, Italy3, Faculty of Mathematical, Physical and Natural Sciences, University of Palermo, Palermo,[email protected]

Abstract

Erythropoietin (EPO), the main haematopoietic growth factor for the prolifera-tion and differentiation of erythroid progenitor cells, is also known for its angiogenicand regenerative properties.In our research we demonstrated the regenerative effects of EPO administration inan experimental model of sea bass (Dicentrarchus labrax) subjected to amputationof the caudal fin.EPO-treated fish (3000 UI of human recombinant EPO-alpha) show an increasedgrowth rate of their fins compared to those untreated. By analyzing fin length atestablished times (15 and 30 days after cut), EPO-treated fish always show an in-creased length compared to untreated ones. Moreover, exogenous EPO adminis-tration induces an enormous increase in EPO-blood levels at each observation timewhereas these levels remained quite unmodified in untreated fishes. Immunochemi-cal analyses performed by confocal laser scanning microscopic observations show anincreased expression of EPO-receptors and PECAM-1 (an endothelial surface markerof vessels sprout) in the regenerating tissue, whereas no signs of inflammation or fi-brosis are recognizable. All these findings confirm EPO as a new factor involved inregenerative processes, also suggesting a potential, future utility for new therapeuti-cal applications in the field of human regenerative medicine.

1 The research context

The regeneration has always been deeplyfascinating and of considerable scientificinterest.The ability to funtional recovery after lossor reduced function of an organ variessignificantly in mammals, amphibians andfish.Among vertebrates, only teleosts fish and

urodel amphibians naturally keep theirability to regenerate injured or amputateparts for life.Regenerative Medicine, a new medical do-main, tries to develop therapeutic pathwaysthrough the stimulation of natural regen-erative processes also in humans. Themechanisms for functional recovery are un-der the observational lens of researchers,whose purpose is to know the natural net-


Technologies

work that regulates the regeneration pro-cess and the intricate connections that leadto the production of hormones and media-tors fundamental to the regeneration of dif-ferent tissues [1].In the field of Regenerative Medicine, Ery-thropoietin (EPO) represents a particularlysignificant subject of research [2]. Erythro-poietin (EPO) is a glicoproteic hormonemainly products in the kidney and to alesser extent in the liver. It is the principalhaematopoietic growth factor for the prolif-eration and differentiation of erythroid pro-genitors cells.EPO acts by binding with EPOr, a tyrosinekinase receptor. Binding EPO-EPOr trig-gers the enzymatic function of JAK2, thereceptors-associated tyrosine kinase, andallows phosphorylation and nuclear traslo-cation of STAT5, leading to progenitor cellproliferation and differentiation [3].The induced-production of EPO has beendemonstrated, not only in the kidney andpartly in the liver, but also in other tissues.This aspect underlines that besides the abil-ity of stimulating red blood cell production,this hormone has many pleiotropic effects;that is to say that it can act on several typesof cells through different mechanisms, car-rying out functions additional to the rolewhich is classically attributed to it.Recents studies suggest that function ofEPO and EPO-receptor (EPOr) are not lim-ited to erythroid linage. EPOr expres-sion has, in fact, been detected in placen-tal endothelial cell lines and EPO is capa-ble of stimulating endothelial cell prolifer-ation in vitro [4]. Moreover, EPO is knownto induce a pro-angiogenic phenotype incultered endothelial cells, stimulated neo-vascularization in the chick chorioallan-toic membrane [5], and the hormone seemsto play an important role in cardiac mor-phogenesis and in myoblast proliferation

[6, 7]. The relationships between EPO andcardiovascular system are demonstrated bythe presence of high levels of EPO andEPO-r expression in the vasculature andthe heart. “Preconditioning” with EPO ac-tivates cell survival pathways in myocar-dial tissue in vivo and adult rabbit car-diac fibroblasts in vitro. These pathways,activated by erythropoietin in both wholehearts and cardiac fibroblasts, are also ac-tivated acutely by ischemia/reperfusion in-jury [8].Moreover, in vivo studies indicate thatEPO treatment either prior to or dur-ing ischemia significantly enhances cardiacfunction and recovery, including left ven-tricular contractility, following myocardialischemia/reperfusion.This amazing link between EPO and hearthis further defined in the Japanese puffer-fish, Fugu rupripes and zebrafish, Daniorerio, where physiological hormone pro-duction occurs in the hearth.This tissue-protective effect of EPO in-volves various natural mechanisms of ac-tion including the angiogenesis, a set offunctional processes appointed to the for-mation of new blood vessels starting fromthe pre-existent ones [9]. Angiogenesis isone of the few natural regenerative activi-ties of humans. The regeneration is a pro-cess by which damaged or lost structuresare perfectly or almost perfectly replaced.It is known that mammals contain sev-eral organ systems capable of regeneration,such as blood and liver, but the major-ity of organs heal by scarring, while nonmammalian vertebrates like urodele am-phibians and teleost fish restore complextissues much more effectively than mam-mals, creating tantalizing examples of suc-cessful organ regeneration [10].Danio rerio is taking an increasingly domi-nant role in the study of the possible mech-

2446


anisms that underlie the processes of tis-sue regeneration. This is a small fresh-water fish that reaches 4-6 cm in lengthin adulthood. In recent years, it has beenincreasingly used as a model for study-ing the development of vertebrates and formany human diseases; in fact small size,big proliferative capacity, short interval be-tween one generation and the next, embryotransparency make this fish an experimen-tal model increasingly used in Regenera-tive Medicine.The final site of hematopoiesis, as for allfish, is the kidney since seventh day fromfertilization. Chu et al., [11] have recentlycloned the gene coding for erythropoi-etin in zebrafish, highlighting in the codi-fied protein a sequence homology equal to90%, 55% and 32% respectively with thecarp, humans and pufferfish EPO. It wasalso shown that the main sites of expres-sion of the gene for EPO are representedby heart and liver.Other authors, on the other hand, believethat in some fish the kidney is not onlythe main erythropoietic organ but also thelargest source of erythropoietin. Wickra-masinghe [12] has in fact shown that super-natants prepared by kidney homogenatesof rainbow trout (Salmo gairdneri) containmore immunoreactive erythropoietin com-pared with autologous plasma, serum andsupernatants obtained by homogenates ofliver or spleen.The mechanisms that underlie the produc-tion of erythropoietin are similar in humansand in fish and are recognized as the mainfactor hypoxia. In fact, in fish, there isa relationship between the increase of re-nal erythropoietin levels, reduced levels ofsplenic erythropoietin and spleen somaticindex. Under conditions of hypoxia, spleencontraction and the subsequent release oferythrocytes justifies the initial increase of

oxygen in the blood, while the erythropoi-etin action is probably responsible for in-creasing levels of haemoglobin which oc-cur at a later stage [13].Even the paper of Rojas et al. [14]highlighted the existence of an impor-tant sequence homology between Hypoxia-Inducible Growth Factors (HIF) 1α and 2αbetween zebrafish and vertebrates on con-firmation of a pathophysiological identity.However, one of the foundations that dif-ferentiate zebrafish, vertebrates and hu-mans is the natural regenerative capacityinvolving fins, skin, heart and brain in thelarval stage of these primordial living be-ings. Regeneration seems to be an ar-chaic tract of the animal world: it waskept in some metazoaria organisms andsecondly lost in others for unknown rea-sons. However, the study of regenerativecapacity of zebrafish is an experimental ba-sis to study and understand why human be-ings have lost this extraordinary ability andwhat routes should be given to ”recall thepast.”The caudal fin of the zebrafish has becomea popular model for studying the molecularmechanisms regulating regeneration due toits accessibility to amputation and its fairlysimple structure, consisting of segmentedbony fin rays [15].Recently, our research group has shownthat, even the sea bass, Dicentrarchuslabrax, as well as the zebrafish, representsanother interesting and simple experimen-tal model.In particular has been shown the regen-erative effects of EPO administration inD. labrax subjected to amputation of thecaudal fin, by analyzing the expression ofspecific EPO-receptors and the angiogenicproperties [2].

2447

Technologies

Figure 1: Schematic representation of the skeleton of a ray in sea-bass fin. Dermal boneforms two segmented, concave and opposing hemirays. The image illustrates the posi-tion of amputation, which was performed below the origin of bifurcation. From Buemi etal.[2]. European Journal of Clinical Investigation. Figure 1. Copyright Wiley-Blackwell.With permission.

2 The research method

The research was conducted at aquacul-ture experimental plant of Institute for theCoastal Marine Environment (TerritorialUnit of Messina, Sicily, Italy) of the Na-tional Research Council (IAMC-CNR).D. labrax were acclimatized for at least 3weeks in 2 m3 indoors tanks (T= 18 ± 0.2°C) provided with a flow-through supply ofaerated seawater and maintained under nat-ural photoperiod.Fish were fed to satiation daily, but werefasted for 24 h before the start of the exper-iments to avoid problems during anaesthe-sia.

2.1 Experiment procedure anderythropoietin administra-tion

Fish were randomly assigned to the ex-perimental group. Before amputation, fishwere anaesthetised with tricaine methane-

sulphonate, MS-222 (0.1g·L−1). In partic-ular fish were placed in the anaesthetic bathand monitored until they reach the plane ofsurgery anaesthesia (Stage 3, plane II ac-cording Stoskopf [5]) characterized by a to-tal loss of reflex and by slowing of hearthand respiratory rate. Fish anaesthetizedwere placed in an operating sling where thegills were irrigated with anaesthetic.The caudal fin was amputated, with a razorblade, one segment proximal to the level ofthe first ray bifurcation. (Figure 1).Following application of the anaesthesia,fish were immediately removed from theanaesthetic bath and placed in a recoverytank containing sea water without anaes-thetic and were monitored until completerecovery of normal swimming.At two established times (immediately af-ter cutting and after 15 days) a dose of3000 UI of Recombinant Human Erythro-poietin alfa (Eprex® Janssen Cilag), wasadministered to 5 fish (experimental group)through a subcutaneous injection at dor-

2448


sal level, while a similar quantity of sol-vent was administered to the others animals(control group).The fish length was measured from the tipof the mouth to the caudal peduncle using aruler. For fin length and segment measure-ments, the skin and muscle surrounding themost proximal ends of the fin rays weredissected away before mounting in 100%glycerol under a coverslip. Measurementswere facilitated with a 10 3 10 reticule inthe eye-piece of the dissecting microscope(stemiscope 2000, Zeiss, Thornwood, NY)at low magnification.Fish were evaluated at established times(baseline, after cutting (T-0), after 15 days(T-15), after 30 days (T-30) by measuringthe length of maximum outgrowth, repre-sented by the distance from the centre ofthe amputation plane to the tip of the re-generating fin.Simultaneously, a dosage of EPO levelswas performed in blood samples obtainedfrom the caudal vein and assayed by com-mercial ELISA kit (Predicta EPO Gen-zyme Diagnostics USA).

2.2 Histological Analysis: Im-munohistochemistry (IHC)and Immunofluorescence(IF)

In order to evaluate the different angio-genic response between the two groups offish (EPO-treated and untreated), histolog-ical analysis was made using immunofluo-rescence and immunohistochemistry.At the end of the study the amputatedfish were euthanized and fixed in 4%paraformaldehyde at 1 day after amputa-tion. The fixed fins were postfixed in 1%osmium tetroxide and embedded in Em-bed 812-Araldite 502 resin. Semithin serial

sections were cut, stained with toluidineblue, and analyzed by light microscopy.For IHC and IF, the samples were fixedin 10% formalin and embedded in paraf-fin. Section of 5µm were deparaffinizedand steamed in 0.01 mol·L−1 sodium cit-rate buffer, pH 6, in a microwave oven for15 min. Only for IHC, endogenous perox-idase activity was quenched by 0,3% hy-drogen peroxide in methanol. Aspecificproteic reactions were blocked by incuba-tion with 5% BSA for 30 min. Slideswere then incubated over night at 4°C withanti-EPOr (rabbit polyclonal clone C-20Santa Cruz) and PECAM-1 (goat poly-clonal clone M-20 Santa Cruz), a 130 Kdtransmembrane glycoprotein belonging tothe immunoglobulin superfamily of celladhesion molecules, mainly found at thecell-cell borders of neighboring endothelialcells primary antibodies followed by an in-cubation at room temperature with a horseanti-goat (Vector Laboratories) and goatanti-rabbit biotynilated IgGs (BioSpa). ForIHC, the reaction was developed by anavidin peroxidase complex (Biospa), re-vealed with Vector Nova Red (Vector Lab-oratories) and counterstained with hema-toxylin.For IF, the reaction was revealed by anavidin conjugated with Alexa Fluor 488(Molecular Probes). For each sample,negative controls were also performed byomission of primary Ab, substitution ofprimary Ab with normal IgGs and substi-tution of primary Ab with indifferent rab-bit and goat primary Abs. Immunochem-ical stain was interpreted by assessing theintensity of staining. Cytoplasmic and/ormembrane immunoreactivity was consid-ered positive.Samples were then observed and pho-tographed using a Zeiss LSM 5 DUO laserscanning microscope. All images were dig-

2449

Technologies

italized at a resolution of 8 bits into an ar-ray of 2048 x 2048 pixels. Optical sectionsof fluorescence specimens were obtainedusing HeNe laser (543 nm) and Argon laser(458 nm) at a 1-min 2-sec scanning speedwith up to eight averages 1.50-µm-thicksections were obtained using a pinhole of250. Contrast and brightness were estab-lished by examining the most brightly la-beled pixels and choosing settings that al-lowed clear visualization of structural de-tails while keeping the highest pixel inten-sities close to 200. The same settings wereused for all the images obtained from theother samples that had been processed inparallel.Digital images were cropped and figuremontages prepared using Adobe Photo-shop 7.0.

2.3 Statistical AnalysisA statistical analysis of data was made us-ing the GraphPad Prism (version 4.0) pack-age. Analysis of variance among groupsof fishes was established by ANOVA fol-lowed by Bonferroni’s test. Furthermore,an unpaired t-test was employed to assessstatistical differences in fin length betweenEPO-traited fish and controls at establishedtimes (immediately after cutting: T-0, 15days after cutting: T-15 and 30 days aftercutting: T-30). All results were consideredsignificant when p≤0.05.

3 The main results

3.1 Fin growth in EPO-treatedand untreated fish

During the whole observational period, astatistically significant trend in fin growthwas noticed in both groups (ANOVA vari-

ance for EPO-treated: p = 0.01; for EPO-untreated p = 0.04). Furthermore, EPO-treated fishes showed an increased growthrate compared to those untreated (ANOVAvariance: p = 0.01; F = 1.8 vs p = 0.04; F =1.0). By analyzing fin length at establishedtimes, EPO-treated fishes always showedan increased length compared to untreated(T-15: p = 0.03, F = 1.0; T-30: p = 0.01, F= 1.4).Finally, exogenous EPO administration in-duced an enormous increase in EPO-bloodlevels at each time of observation (T-15: p< 0.001, F = 12.41; T-30: p<0.001, F =9.42); on the contrary, predictably, theselevels remained quite unmodified in un-treated fishes. A resume of these findingsis reported in Figure 2 and Table 1.

3.2 Histochemical analysesAnalyses were performed by confocal laserscanning microscopic observations.Hematoxylin and eosin examination ofthe stump of both EPO-treated and un-treated groups showed a structure com-posed by a completely restored epidermalstructure with a distinct apical cap. Un-derneath the cap a well vascularized meso-dermic tissue, composed of fusiphorm cellswith the appearance of fibroblasts-like andendothelial-like cells was detectable. Nosigns of inflammation or fibrosis were rec-ognizable (Figure 3a and 3b).At immunofluorescence, EPOr was widelydistributed in the epithelium, above all inthe basal layer of the epidermis. More-over, inside the blastema, some blood ves-sels were positive.Immunodetection of PECAM-1 allowedthe visualization of blood vessels and en-dothelial cells in the mesodermic blastema.In particular a higher number of PECAM-1 positive structures and fusiphorm cells,

2450


Figure 2: Differences in fin length (cm) between EPO-treated fish and controls at estab-lished times. * P: 0.03; ** P: 0.01. Modified from Buemi et al. [2].

FIN

lenght

EPO

levels

FIN

lenght

EPO

levels

FIN

lenght EPO

levels FIN

lenght EPO

levels

Treated (n 5) 2.2 ± 0.4 16.7 ± 1.8 0.1 ± 0.0 16.8 ± 1.6 1.1 ± 0.2a 2240 ± 210c 1.9 ± 0.3b,d 2340 ± 190c

Controls (n 5) 2.1 ± 0.4 17.0 ± 1.7 0.1 ± 0.0 16.9 ± 1.8 0.7 ± 0.2 16.7 ± 1.8 1.2 ± 0.2 17.1 ± 1.9

Time Basal T=0 T=15 T=30

aP: 0.03 vs untreated; bP:0.01 vs untreated; cP < 0.01 vs untreated; dP for trend (ANOVA): 0.01; eP for trend (ANOVA): 0.04.

Basal, starting values; T=0, immediately after cut; T=15, 15 days after cut; T=30, 30 days after cut.

Table 1: Differences in length (cm) and EPO-blood levels (mU·mL−1) between EPO-treated fish and controls at established times. (Modified from Buemi et al. [2]).

most likely endothelial cells, were de-tectable, by visual inspection, in treatedfish compared to untreated ones. Microves-sels were distributed along the mesoder-mic blastema showing an increasing den-sity near the apical epidermal cap (Figure3c and 3d).

4 Conclusions and futureperspective

Results from this study suggest that EPOmay exert important regenerative effects,

even in an experimental fish model. Thiscan be evidenced by the fact that EPOtreated fish subjected to amputation of thecaudal fin showed an increased growth ratecompared with those untreated, thus allow-ing us to hypothesize that this hormonemay be directly involved in stimulatingthe regenerative and angiogenic processes.The presumed direct action of EPO on fingrowth is also supported by hystochemicalanalyses, which demonstrated an increasedexpression of EPO-receptors and PECAM-1 (a marker of vessel sprout) in the regen-erating tissue, whereas on the contrary, no

2451

Technologies

Figure 3: A-B. Confocal laser scanning microscopic observations of EPO-treated fish(A) and controls (B). A well vascularized mesodermic tissue composed of fusiphormcells with the appearance of fibroblast-like and endothelial-like cell is detectable. Nosigns of inflammation or fibrosis were present. (C-D). Immunofluorescence of PECAM-1. A high number of PECAM-1 positive structures are detectable in EPO-treated fish (C)compared with the controls (D). Modified from Buemi et al. [2].

signs of inflammation or fibrosis were de-tectable.The regeneration of organs and appendagesafter injury occurs in diverse animal groupsand the goal of regenerative medicine isto restore cells, tissue and structures thatare lost or damaged after disease, injury orageing.The use of fish has been born out by pre-vious by previous studies aimed to evalu-ating anomalies in angiogenic mechanismsand disorders genetically determined bythe cardiovascular system.

Among teleost fish, zebrafish, is taking anincreasingly dominant role in the study ofthe possible mechanism that underlines theprocesses of tissue regeneration. In factin previous studies, it has shown the re-markable capability to regenerate their fins,optic nerve, scales heart and spinal cord[16, 17, 15, 18, 19].In our study, we have convalidated an otherexperimental model using specimens of D.labrax, which being bigger than zebrafish,are probably more suitable for morpholog-ical analysis.

2452


Moreover, fish appear to be a particularlyappropriate model for evaluated the ery-thropoietic and extra-erythropoietic effectsof EPO. It is in fact well known that thesolubility of oxygen in water is just 1/30of that in the air, and the rate of oxygendiffusion in water is only 1/10000 than inair. Therefore, fish would be expected topossess a sensitive and rapid mechanismto regulate the level of erythrocytes. Fishhave indeed been shown to maximize oxy-gen uptake and economize oxygen expen-diture, under hypoxic condition, by a va-riety of measures such us increased venti-lation, bradycardia, cardiorespiratory syn-chrony, constriction of peripheral bloodvessels and increased release of catheco-lamines. Blood erythrocyte levels are in-creased, initially, due to release from thespleen and then, subsequently, due to ery-thropoiesis in response to the hormoneEPO, produced by kidney [20, 21], organin which the same erythropoiesis occurs.Our data confirm the capacity of EPO todirectly stimulate angiogenic capacity oftissue, promoting an adequate regenerativeactivity after amputation.However, it is likely that the efficacy ofEPO observed in our and other model sys-tems depends on EPO’s key role in mul-tiple protective pathways, including an in-hibition of apoptosis, restoration of vas-cular autoregolation, attenuation of in-flammatory responses and augmentation ofrestorative functions, including the directrecruitment of stem cells [22], as shown bythe fact that regenerative tissue in animaltreated with EPO do not reveal signs of fi-brosis or phlogosis.

To conclude, there are some questionsposed by Regenerative Medicine on the ta-ble of experimentation. One of these is:“What are molecular differences that per-mit tissue regeneration in fish and makemammalian tissue recalcitrant to regener-ation?”.One aspect of adult wound healing in mam-mals that has been discussed in relation tothe curtailment of regeneration is the oc-currence of fibrosis, and also of immuneand inflammatory responses.Fish recover the lost caudal fin tissue afteramputation through a process of epimor-phic regeneration, and this occurs in a step-wise manner with the formation of an ep-ithelial wound cap, followed by blastemaformation and finally the regenerative out-growth. This complex regenerative processis orchestrated by sequential interactionsbetween biomolecules and cells in a spa-tiotemporal manner.In animals treated with erythropoietin, tis-sue regeneration occurs through a prema-ture formation of angiogenic sprouts andwithout an important fibrotic or phlogisticcomponent.Although erythropoietin has a significantangiogenic effect similar to the effect of theother angiogenic factor bFGF, this is prob-ably only one of the mechanisms throughwhich the hormone exercises its protectiveaction on tissues.Further studies of regenerative mecha-nisms induced by erythropoietin could per-haps new perspectives on the possibility ofregenerative medicine in more evolved an-imals.

References[1] N. DeWitt. Regenerative medicine. Nature, 453:301, 2008.

2453

Technologies

[2] M. Buemi, A. Lacquaniti, G. Maricchiolo, D. Bolignano, S. Campo, V. Cernaro,A. Sturiale, G. Grasso, A. Buemi, A. Allegra, V. Donato, and L. Genovese. Re-generative medicine: does Erythropoietin have a role? Current PharmaceuticalDesign, 15(17):2026–36, 2009.

[3] S.S. Watowich, H. Wu, M. Socolovsky, U. Klingmuller, S.N. Constantinescu, andH.F. Lodish. Cytokine receptor signal transduction and the control of hematopoieticcell development. Annual Review of Cell and Developmental Biology, 12:91–128,1996.

[4] A. Anagnostou, Z. Liu, M. Steiner, K. Chin, E.S. Lee, N. Kessimian, and C.T.Noguchi. Erythropoietin receptor mRNA expression in human endothelial cells.Proceeding of the National Academy of Science of the United States of America,91:3974–3978, 1994.

[5] M. Stoskopf. Anaesthesia. Aquaculture for veterinarians. Fish husbandry andmedicine, Ed. Brown L. Pergamon Press, New York, USA, pages 161–167, 1993.

[6] H. Wu, S.H. Lee, J. Gao, X. Liu, and M.L. Iruela-Arispe. Inactivation of erythro-poietin leads to defects in cardiac morphogenesis. Development, 126:3597–3605,1999.

[7] M. Ogilvie, X. Yu, V. Nicolas-Metral, S.M. Pulido, C. Liu, U.T. Ruegg, and C.T.Noguchi. Erythropoietin stimulates proliferation and interferes with differentiationof myoblasts. The Journal of Biological Chemistry, 275:39754–761, 2000.

[8] C.J. Parsa, J. Kim, R.U. Riel, R.U. L.S. Pascal, R.B. Thompson, J.A. Petrofski,A. Matsumoto, J.S. Stamler, and K.J. Koch. Cardioprotective effects of erythropoi-etin in the reperfused ischemic heart: a potential role for cardiac fibroblasts. TheJournal of Biological Chemistry, 279(20):20655–662, 2004.

[9] P. Carmeliet. Angiogenesis in health and disease. Nature Medicine, 9(6):653–660,2003.

[10] C.L. Stoick-Cooper, R.T. Moon, and G. Weidinger. Advances in signaling in ver-tebrate regeneration as a prelude to regenerative medicine. Genes & Development,21(11):1292–315, 2007.

[11] C.Y. Chu, C.H. Cheng, G.D. Chen, Y.C. Chen, C.C. Hung, K.Y. Huang, and C.J.Huang. The zebrafish erythropoietin: functional identification and biochemicalcharacterization. FEBS Letters, 581:4265–4271, 2007.

[12] S.N. Wickramasinghe. Erythropoietin and the human kidney: evidence for an evo-lutionary link from studies of Salmo gairdneri. Comparative Biochemistry andPhysiology, 104(1):63–65, 1993.

[13] J.C.C. Lai, I. Kakuta, H.O.L. Mok, J.L. Rummer, and D. Randall. Effects of mod-erate and substantial hypoxia on erythropoietin levels in rainbow trout kidney andspleen. Journal of Experimental Biology, 209:2734–2738, 2006.

2454


[14] D.A. Rojas, D.A. Perez-Munizaga, L. Centanin, M. Antonelli, P. Wappner, M.L. Al-lende, and A.E. Reyes. Cloning of hif-1alpha and hif-2alpha and mRNA expressionpattern during development in zebrafish. Gene Expression Patterns, 7(3):339–45,2007.

[15] K.D. Poss, M.T. Keating, and A. Nechiporuk. Tales of Regeneration in zebrafish.Developmental Dynamics, 226:202–210, 2003.

[16] R.R. Bernhardt, E. Tongiorgi, P. Anzini, and M. Schachner. Increased expression ofspecific recognition molecules by retinal ganglion cells and by optic pathway gliaaccompanies the successful regeneration of retinal axons in adult zebrafish. Journalof Comparative Neurology, 376:253–264, 1996.

[17] T. Becker, M.F. Wullimann, C. G. Becker, R.R. Bernhardt, and M. Schachner. Ax-onal regrowth after spinal cord transection in adult zebrafish. Journal of Compara-tive Neurology, 377:577–595, 1997.

[18] A.L. McDowell, L.J. Dixon, J.D. Houchins, and J. Bilotta. Visual processing of thezebrafish optic tectum before and after optic nerve damage. Vision Neuroscience,21:97–106, 2004.

[19] R.J. Major and K.D. Poss. Zebrafish heart regeneration as a model for cardiac tissuerepair. Drug Discovery Today: Disease Models. Disease Models, 4:219–225, 2007.

[20] G.H. Satchell. Physiology and form of fish circulation. pages 8–22, 1991.

[21] D.J. Randall and S.F. Perry. Catecholamines in fish physiology. Vol. XII B: thecardiovascular system (Ed. Hoar and Andrandall). Academic Press, New York, USA,pages 5–33, 1992.

[22] G. Grasso, A. Sfacteria, A. Cerami, and M. Brines. Erythropoietin as a tissue-protective cytokine in brain injury: what do we know and where do we go? Neuro-scientist, 10:93–98, 2004.

2455

Technologies

2456

A Powerful Versatile Data Acquisition and Trans-mission System for Environmental Monitoring

G. ZappalaInstitute for Coastal Marine Environment, CNR, Messina, [email protected]

Abstract

In recent years particular attention has been given to marine offshore and coastalenvironments. Offshore measurements can be used to detect and understand largescale phenomena, while coastal monitoring is useful to preserve and protect thosecoastal areas where multiple interests converge (related to tourism, recreational orproductive activities) and which could be heavily damaged from local anthropic ac-tivities or remote events (e.g. an oil spill from a ship accident or tank washing).All modern instruments are designed to be interfaced to a computer system, runninga data acquisition program often provided by the instrument manufacturer, but usu-ally neither flexible nor remotely manageable.Connecting together instruments from different manufacturers might be complex.To overcome these limitations an integrated hardware-software system was devel-oped, that can be customized in various versions to fit the different needs of dataacquisition and transmission both on buoy-based and on ship-based instrumentation.The system uses modular IEEE 696 boards implementing a PC-like architecture; ac-cording to the measurement needs a variable number of serial ports, digital I/O andpower outputs, analog input and outputs can be connected; GPS receivers, satelliteand/or cellular phone modems can be interfaced.The embedded macro-commands allow to easily write mission programming se-quences, that can be also remotely modified without interfering with normal activity.

1 Introduction

The need for advanced systems to surveywater quality and meteorological parame-ters, both at medium and long term scales,stimulated in recent years the developmentof different kinds of coastal and offshorebuoys, platforms and networks [1, 2, 3, 4,5, 6, 7, 8, 9]. Anthropic pressure on theenvironment, together with the risk of pol-lution deriving from accidents (oil spills,failure in waste water treatment plants. . . )stressed the importance of (near) real timedata to assess water quality and to activate

proper measures in emergency situations toprotect both the environment and the re-lated activities (aquaculture plants, fishing,recreational uses. . . ).To obtain a good synopticity of observa-tions both in spatial and temporal domains,it is necessary to complement traditionalship observations with measurements fromfixed stations (buoys moored in sites cho-sen to be representative of wider areas, orto constitute a sentinel against the arrivalof pollutants), satellite observations, use ofships of opportunity and of newly devel-oped instruments.


Technologies

An ideal monitoring system should berugged, cost effective and simple to man-age in various and changing operating con-ditions.This paper focuses on the development ofan universal data acquisition and transmis-sion system, developed at CNR-IAMC inMessina, able to satisfy environment moni-toring needs from buoys, ships or fixed sta-tions [10].

2 Integrating softwareand hardware: designrules

The system design must face several differ-ent (and somehow contrasting) needs:• The CPU must offer high elaboration

speed and low power consumption: re-ally, power consumption vs speed is acompromise, sometimes managed usingsleep (halt) states;

• The software must be portable amongdifferent hardware platforms: this meansit should be written in a high-level lan-guage, but high-level languages are notas efficient as low-level ones, so requir-ing higher elaboration speed (and morememory, which increases power con-sumption);

• The software must be very efficient: thismeans it must be strictly tailored on thehardware, but this assumption contrastswith former one;

• Mission programming (i.e. setting ofkind and frequency of measurements)must be performed “on the fly”, with-out recoding the managing software; thiscan be obtained using sequence files ofmacro-commands;

• New instruments must be added to thesystem easily and quickly.

3 The firmwareThe software architecture was designed tobe as modular as possible; it is based onthree main modules, the “event machine”,the “sequence manager” and the “parser”, anumber of “device” and “instrument” mod-ules to fit the installed instruments and thedata communication hardware, some aux-iliary functions modules, written in vari-ous high and low level languages to runin ROMDOS. During inactivity periods,the system enters if required a low-power“sleep” mode from which it is wakenedup synchronously each hour by the se-quence manager or asynchronously by thereception of a communication interrupt. Awatch-dog timer automatically resets thesystem in case of hanging.

3.1 The Event MachineThe event machine runs in an endless loop,waiting for one of the following events:- arrival of a character from the communi-

cation device or from the keyboard- occurrence of a timed or “position” event

3.2 The Modem Input ManagerThe arrival of a character activates a rou-tine scanning the input string looking forthe “Z” character, that is recognised as thestart of the remote command sequence; fol-lowing characters up to the “LF” are thenconsidered the command and sent to theparser to be interpreted.

3.3 The Keyboard Input Man-ager

The Keyboard Input Manager allows to lo-cally interact with the data acquisition sys-tem during setup and test phases, issuing

2458


local commands, running a DOS shell, ex-ecuting a macro-command sequence, ac-cessing the communication menu, runninga terminal program.

3.4 The Sequence ManagerThe Sequence manager is activated whenan event occurs; according to the type ofevent, it opens the proper sequence filecontaining the macro-commands for thatevent, reads it line by line, sending eachline to the “parser” to be interpreted andexecuted.

3.5 The ParserThe “Parser” receives in input a string sup-posed to contain a valid macro-command,that, if found and recognised, is executed.The macro-commands (that can be com-bined into sequences using a simple texteditor) enable to fully control the system,the connected instruments, and the datatransmission. Conditional branch com-mands are also available; this feature canbe very useful in case of partial operativityof the system due, for instance, to low bat-tery level or failure of some instruments.

3.6 The “Device” modulesThese modules are written in assembly lan-guage to fit the peculiarities of the com-puter hardware and are installed as Inter-rupt Service Routines.

3.7 The “Instrument” modulesThese modules are written to interface thecomputer with various measuring instru-ments, taking into account their peculiar-ities. To obtain the maximum flexibility,parameters can be passed to the modules

to specify the instrument address and thecommand to be performed.

3.8 The data transmission rou-tines

The system can receive remote commands;it is possible to choose among differentdata transmission systems, using either ter-restrial or satellite modems. The firstmethod uses a service offered by some tele-com providers enabling to send e-mails en-coded in a SMS; this solution is usuallyquite expensive and it should be used onlywhere it is impossible to obtain an Inter-net connection. The second method simplyconnects the measuring station to Internetusing a tcp-ip protocol, so allowing to di-rectly send e-mails. The data path was de-signed to be fault tolerant:

- A copy of collected data is locally storedin the measuring station;

- A copy of the sent e-mail is also locallystored in the measuring station;

- In case of failure of the receiving mailserver, the sending server (managed bythe telecom company) will retry the de-livery for some time, so allowing to re-pair the defective server without loss ofmessages;

- All the mail messages are numbered, soenabling to detect undelivered ones andask the remote station for retransmission;

- As soon as they are received, the mailmessages are immediately assimilatedinto the data bank, and a copy remainsfor some time in the mail server;

- Periodic back-up operations are per-formed on the data base.

The command-data flow is described inFigure 1.

2459

Technologies

Figure 1: The command-data flow

4 The hardware

The data acquisition and transmission sys-tem is conceived to implement a PC-likearchitecture running a ROMDOS (a ROMbased operating system similar to MS-DOS) environment; this choice was dueto the availability of both industry-gradeboards and of native software developmenttools in high and low level languages. Theboards comply with IEEE 696 (PC104)standard. The system meets with somevariations and exceeds the capabilities ofthe ARES architecture [11]. Usually, theboard set could comprise:- A CPU board including RAM and ROM

(usually rewritable) memory, Disk-on-chip or Disk-on-module memory host-ing, keyboard and video interface, diskinterface, serial port interface (up to 4),parallel port interface, other interfaces(USB, audio, mouse. . . ) not used in thisapplication

- Serial ports expansion board (usually 8ports/board)

- Analog to digital converter (up to 16single-ended input channels) with 12 bit(for system diagnostic) or 16 bit (formeasurement purposes) resolution.

- Remote communication device board(GSM-GPRS-UMTS modem) often inte-grated with a GPS module; sometimesthis board is substituted with devicesconnected to serial ports obtaining bet-ter performances (higher data transmis-sion speed, satellite communication, tcp-ip integrated stack. . . )

- Digital I/O board- Relay board to switch on and off the

connected instruments, or semiconductorpower interfaces driven by the digital I/Oboard or by the parallel printer interfaceusually included in the CPU board.

According to the measuring needs and topower supply availability, the hardware canbe fully or only partly implemented.

2460


Figure 2: The sequence generator screen

5 Utility software

Softwares were developed in Visual Ba-sic to run in Windows environment to helpmanage the systems.

5.1 The Control Terminal

This program integrates a Terminal pro-gram with some utilities: an “applicationsequence generator”, a ”launch schedulingmodule”, an “HEX file generator”, a “fileupload manager” and a ”remote file man-ager module”. From the main terminalwindow it is possible to customize and usethe terminal program or to activate the spe-cific utilities.

5.2 The Application SequenceGenerator

The ”application sequence generator” is amenu-driven utility that generates the ”se-quence” files containing the macro com-mands for mission programming. Ascreenshoot is displayed in Figure 2; thetop frame is used to choose the kind of op-eration to be performed (instrumental mea-surements, device management, programmanagement).

5.3 The Launch schedulingmodule

This module was developed to enable set-ting probe release parameters in operationfrom ships of opportunity.

2461

Technologies

Figure 3: A buoy near Messina. The wind generator is evident on the top, under it theradar reflector, the solar panels, the pumping system on the deck, on the left, the meteostation

5.4 The HEX file generatorIt is possible to transfer files to the dataacquisition and transmission system cod-ing them in a specifically designed format,somehow similar to an ascii coded hex-adecimal dump, with checksum and trans-mission order informations.

5.5 The File Upload ManagerThis utility was designed to perform auto-matic uploads of ”HEX” files to the dataacquisition and transmission systems in thebuoys; it can work in a “manual”, “auto-

matic”, or “blind” mode.

5.6 The Remote File ManagerModule

This module gives remote control of themeasuring station file system offering DIR,COPY, DELETE and RENAME facilities.

6 Some applications

Various versions of the data acquisitionand transmission system were built: the

2462


Figure 4: The buoy control electronics; on the left the power control board, on the rightthe watertight connectors

first one was used in a network of coastalmonitoring buoys; a second one equips anautomatic multiple launcher for expend-able probes; a microcontroller-based ver-sion was also designed to equip small in-struments. Some examples of data ac-quired with the system here described areshown in this same volume in the papersby Zappala [12] and Zappala et al. [13]

6.1 The “buoy” version

This release was used at the beginning ofthe century to equip a network of coastalmonitoring buoys in the South of Italy([14]), one of which is shown in Figure 3;the control electronics is shown in Figure4.

6.2 The “Launcher” versionA main goal of a Ship Of Opportu-nity Programme (SOOP) is the provi-sion of sea temperature profiles in (near)real time. The use of commercial shipsand expandable probes allows the reduc-tion of costs, in comparison with re-search ship cruises. An autonomous mul-tiple launcher was developed in the frame-work of the Mediterranean ForecastingSystem - Toward Environmental PredictionProject (MFSTEP), able to automaticallycollect eight temperature profiles, using asoftware-programmable sampling strategy[15, 16, 17]. The device is shown in Figure5 and its control electronics in Figure 6. Inthe “launcher” version, more than waitingfor an hourly event, the program monitorsthe position of the hosting ship, comparingGPS data against the programmable set oflaunch points-times.

2463

Technologies

Figure 5: The launcher on the deck of Magnaghi Ship; the eight launch tubes are evidenton the right, together with the boxes containing the electronics and the electropneumaticvalves

7 Discussion

The availability of data to satisfy the policy,stakeholders, environment managers andscientists relies on continuous advance-ments in technological equipments in or-der to obtain cost-effective and reliable sys-tems to assess and monitor environmen-tal status. In this framework the sys-tems here described constitute an improve-ment and application of the know-how ac-quired in the last decade of past century[18, 19, 20, 21], and were funded by theItalian (“SAM” Sistemi Avanzati di Moni-toraggio - Advanced Monitoring Systemsand “PI-CNR” Potenziamento Infrastrut-ture - Infrastructure Increase, both partof the Cluster 10 of the Italian Ministryfor University and Research) and Euro-pean (“MFSTEP” Mediterranean Forecast-

ing System Towards Environmental Pred-ition) programmes. The first applicationhas been the network of coastal moni-toring buoys of the SAM program [14,22]. Further developments were obtainedin the framework of the EU-funded MF-STEP program in which the computer sys-tem was refitted with a new, more power-ful CPU, a Wavecom modem managing anembedded TCP-IP stack and a new set ofprograms to control an automatic launcherfor expendable probes. The observing sys-tems can perform meteorological observa-tions and measurements of physical, chem-ical and physico-chemical parameters char-acterizing sea water state and quality, cur-rent speed and direction; the measuring de-vices range from “static” sensors to col-orimetric analyzers for nutrients, a watersampler, taking samples for further labora-

2464


Figure 6: The launcher control electronics; the GSM-GPRS modem is fixed on the backof the watertight box door (left), the electronics with the signal conditioning board is onthe right.

tory determinations, may also be includedin the systems to complete the series ofmeasurable parameters. Water measure-ments can be carried out in situ at fixeddepths, on samples pumped from variousdepths into a measurement chamber, us-ing expendable profiling probes, or profil-ing instruments, like the SAVE [23, 24],that adopted a microcontroller reduced ver-sion of the “launcher” electronics, with asoftware subset. New releases of the soft-ware and of the sequences are uploadableto the system without suspending its nor-mal activity and without opening the water-tight computer box. The macro-commandsenable to manage the data acquisition andtransmission, the mission programming,the system hardware and the measuringinstruments. Thanks to the hardware-software architecture, it is easy to upgrade

the system to more powerful processorswithout the need to completely rewrite thesoftware, which can be easily coded us-ing standard development packages. Theuse of e-mail enables to allocate the datacentre in the most convenient place, with-out any need of proximity to the measur-ing station. The flexibility in mission pro-gramming, given by the execution of sim-ple “sequence” files, allows to fit the mea-surements to the different and varying re-quirements of scientific research and envi-ronment monitoring (e.g. early warning orfollow-up of pollution phenomena). Thestandard interval between measurements isone hour (buoy system) or the occurrenceof a launch event (launcher system), but itis possible to program the sequence files(one for each hour plus the launch ones)to obtain higher or lower frequency, the

2465

Technologies

only limit deriving from the time needed bythe measuring instruments and from power

consumption. The systems have proven tobe flexible, reliable and cost effective.

References[1] A.H. Carof, D. Sauzade, and Y. Henocque. Arcbleu, an integrated surveillance

system for chronic and accidental pollution. Proc. of OES-IEEE OCEANS ’94Conference, III:298–302, 1994.

[2] K. Grisard. Eight years experience with the Elbe Estuary environmental survey net.Proc. of OES-IEEE OCEANS ’94 Conference, I:38–43, 1994.

[3] C.C. Eriksen. Instrumentation for physical oceanography: the last two decades andbeyond. NSF APROPOS Workshop, Ailomar, CA, December 15-17, 1997.

[4] G. Griffiths, R. Davis, C. Eriksen, D. Frye, P. Marchand, and T. Dickey. Towardsnew platform technology for sustained observations. Proc. of OceanObs 99. Onlinewww.bom.gov.au/OceanObs99/Papers/Griffiths.pdf, 1999.

[5] W. Paul. Buoy Technology. Marine Technology Society Journal, 35(2):54–57,2001.

[6] H. Seim, F. Werner, J. Nelson, R. Jahnke, et al. SEA-COOS: SouthEast AtlanticCoastal Ocean Observing System. Proc. of MTS-IEEE OCEANS 2002 Conference,I:547–555, 2002.

[7] K. Nittis, C. Tiavos, I. Thanos, P. Drakopoulos, V. Cardin, et al. The MediterraneanMoored Multi-sensor array (M3a): system development and initial results. Ann.Geophys., 21:75–87, 2003.

[8] N. Pinardi, I. Allen, E. Demirov, P. De Mey, G. Korres, et al. The MediterraneanOcean Forecasting System: first phase of implementation (1998-2001). AnnalesGeophysicae, 21:3–20, 2003.

[9] H.W. Jannasch, L.J. Coletti, K.S. Johnson, S.E. Fitzwater, et al. The Land/OceanBiogeochemical Observatory: A robust networked mooring system for continu-ously monitoring complex biogeochemical cycles in estuaries. Limnol. Oceanogr.:Methods, 6:263–276, 2008.

[10] G. Zappala. A versatile software-hardware system for environmental data acqui-sition and transmission. Computational Methods and Experimental MeasurementsXIV, pages 283–294, 2009.

[11] P. Lessing and D. Henderson. Architecture and development of an environmentalacquisition and reporting system. Proc. of MTS-IEEE OCEANS ’99 Conference,II:785–788, 1999.

2466


[12] G. Zappala. Development and test of an automatic multiple releaser for expandableprobes. Marine Research at CNR, 2011.

[13] G. Zappala, F. Azzaro, and G. Caruso. Coastal water monitoring from automaticoceanographic platforms. Marine Research at CNR, 2011.

[14] G. Zappala, G. Caruso, and E. Crisafi. The SAM integrated system for coastalmonitoring. Coastal Environment IV, pages 341–350, 2002.

[15] G. Zappala and G. Manzella. An automatic multiple launcher for expendableprobes. Operational Oceanography: Present and Future, pages 188–191, 2006.

[16] G. Zappala, F. Reseghetti, and G. Manzella. Development of an automatic multiplelauncher for expendable probes. Ocean Sciences, 3:173–178, 2007.

[17] G. Zappala. Development of advanced instrumentation for operational oceanogra-phy. Computational Methods and Experimental Measurements XIII, 46:841–850,2007.

[18] E. Crisafi, F. Azzaro, G. Zappala, and G. Magazzu. Integrated automatic systemsfor oceanographic research: some applications. Proc. of OES-IEEE OCEANS ’94Conference, I:455–460, 1994.

[19] G. Zappala, L. Alberotanza, and E. Crisafi. Assessment of environmental conditionsusing automatic monitoring systems. Proc. of MTS-IEEE OCEANS ’99 Conference,II:796–800, 1999.

[20] G. Zappala. Advanced technologies: equipments for environmental monitoring incoastal areas. Science-technology synergy for research in marine environment -Developments in Marine Technology, pages 261–268, 2002.

[21] G. Zappala, G. Caruso, and E. Crisafi. Design and use of advanced technologydevices for sea water monitoring. Operational Oceanography. Implementation atthe european and regional Scales, pages 273–280, 2002.

[22] G. Zappala. A software set for environment monitoring networks. Envirosoft 2004,pages 3–12, 2004.

[23] M. Marcelli, A. Di Maio, V. Piermattei, G. Zappala, and G. Manzella. Develop-ment of new technologies for the high variability phenomena data acquisition inthe MFSTEP-VOS project. Operational Oceanography: Present and Future, pages184–187, 2006.

[24] G. Zappala, M. Marcelli, and V. Piermattei. Development of a sliding device forextended measurements in coastal waters. WIT Transactions on Ecology and theEnvironment,, 111:187–196, 2008.

2467

Technologies

2468

Development and Test of an Automatic MultipleReleaser for Expendable Probes

G. ZappalaInstitute for Coastal Marine Environment, CNR, Messina, [email protected]

Abstract

In operational oceanography great amounts of data are required to validate andfeed numerical models predicting environmental trends. To avoid the high cost oftraditional oceanographic cruises, for some applications, Ships of Opportunity (com-mercial ships periodically returning on the same route) have been used to collect andtransmit in (near) real time water profiles measured by expendable probes (XBTsfor temperature, XCTDs for temperature and conductivity, T-FLAPs for temperatureand fluorescence). Costs can be further reduced using automatic devices requiringminimum personnel effort (one person only, possibly a member of the ship’s crew).Funded by the “Mediterranean Forecasting System-Toward Environmental Predic-tion” Project (MFSTEP FP5), a device allowing to release eight expendable probes,collecting and transmitting the measured profiles was designed. The system in-cludes the mechanical hardware (the tubes in which the expendable probes are fitted),electro-pneumatic actuators to release the probes, an electronic circuit to interface theexpendable probes, the control computer and communication devices. The measure-ment and data transmission strategy is easily programmable. The data acquisitionsystem can be remotely controlled in every functionality, and data can be transmittedby GSM-GPRS or satellite telephone systems.Some examples of measurements are reported. The device was granted the Italianpatent and is under examination at the European Patent Office.

1 Introduction

The assessment of environmental condi-tions requires a series of measurements,with a good spatial and temporal resolu-tion. The high cost and limits of traditionaloceanographic surveys stimulate the use ofnew techniques to obtain a proper cover-age. So, traditional moorings and oceano-graphic cruises are complemented with au-tonomous devices (e.g. drifters and gliders)and with the use of ships of opportunity.As a part of the “Mediterranean oceanForecasting System Pilot Project” (MF-

SPP), a Voluntary Observing Ship (VOS)program started in 1999 to collect temper-ature XBT profiles along seven Mediter-ranean transects, designed to study, in eachof the sub-basins (the Algero-Provencal,the Tyrrhenian, the south Adriatic, the lo-nian and the Levantine), the variability ofthe main circulation features [1, 2, 3, 4, 5,6, 7, 8].In the framework of the EU funded“Mediterranean ocean Forecasting Sys-tem Toward Environmental Prediction”(MFSTEP), several experiences were per-formed both in the use and in the develop-


Technologies

Figure 1: A close-up view of a launch cylinder with the pneumatic actuators during theassembly. The small pneumatic cylinder at the bottom locks the launch door, the verticalbigger one operates it.

ment of advanced instrumentation, includ-ing the design and test of the automatic de-vice this paper is about, also called “auto-matic multiple launcher” [9, 10, 11].Although less expensive than dedicatedoceanographic cruises, the use of commer-cial ships to launch XBT probes requireshowever the presence of one or more tech-nicians on board the ship, with the relatedcosts. In order to obtain cheaper opera-tions, an automatic system was designed,to release unattended up to eight probes,easy enough to be recharged by a memberof the crew.


An XBT probe has the shape of a smallmissile, contained in a plastic canister withan overall length of 36 centimetres and ex-ternal diameter of 7 centimetres; its weightis about 1200 grams.The temperature sensor is a precision ther-mistor enclosed in the lead nose; the elec-trical wire to connect it to the measuringsystem is contained in two spools, one inthe plastic body of the probe and the otherone in the canister, that also hosts on itsback the electrical connections towards thedata acquisition system.

2470


Figure 2: The inboard version of the multiple launcher in laboratory.

Traditional use of XBT (expendable bathy-thermograph) probes implies the presenceof at least one technician on board the shipto manually perform the “launch” (or, bet-ter “release”) of the probe using an hand-held “gun” where the plastic canister is in-serted; pressing the trigger the probe is re-leased; while the probe falls into water, thewire inside it starts unreeling. The wireunreels also from the spool in the canis-ter, compensating the movement of the shipand allowing the probe to freefall from thesea surface down to several hundred metersdepth.A data acquisition board connected to acomputer controls the measurement whenthe wire breaks, the profile is completedand the system can be prepared for anotherlaunch. The nominal accuracy of the probeis 0,1 °C. The depth is estimated as a func-tion of time, using a formula of the kind:Z(t) = At−Bt2.

3 The new deviceThe new system is an integrated set of me-chanical and electronic hardware and soft-ware programs offering the maximum offlexibility.

3.1 The mechanical hardwareHeart of the mechanical hardware is thelaunch tube, built in AISI 316 steel, inwhich the probe is fitted with its envelope.An upper cap, holding the electrical con-nections of the probe, closes the launchtube; opening the lower door the probe isreleased and falls into seawater.Two pneumatic cylinders control the door:a small one keeps it closed acting as asafety lock also in case of pressure loss, abigger one moves it. In the actual MFS-VOS design, the system assembles on aframe eight launch tubes with their pneu-matic actuators (Figure 1). A small com-pressor supplies compressed air to the sys-tem, keeping a reserve of 5 litres at 6 bar,

2471

Technologies

Figure 3: The watertight box containing the electro-pneumatic valves feeding the pneu-matic actuators.

a filtered pressure regulator is mounted toprevent damages to the pneumatic system.Two versions of the launcher were built:the first to work hanged outboard the ship,the second to stay inboard, on a deck. Fig-ure 2 shows “inboard” version, mounting afunnel with a pipe to drive probes outboard.Two watertight boxes host respectively theelectro-pneumatic valves feeding the cylin-ders (Figure 3) and the computerized con-trol system (Figure 4).

3.2 The control computer hard-ware and software

The control computer hardware and soft-ware are enhanced versions of those for-merly designed to be used in environmentalmonitoring network [12, 13].

3.2.1 The electronic hardware

All operations are coordinated by an in-dustrial grade Pentium computer, basedon IEEE 696 compliant boards, interfacedwith GPS, data acquisition and communi-cation devices (Figure 5). Both analog anddigital interfaces are available, to collectdata coming from the most various devices(passive and active expendable probes, me-teo sensors. . . ).Remote communications are performed us-ing a GPRS modem with an embeddedTCP-IP stack; a serial port is available toconnect other communication devices (e.g.satellite modems) but also other communi-cations systems could be used.A balanced source circuit (Figure 6) wasdesigned to interface standard passive tem-perature probes with 12 or 16 bit Analog toDigital Converters (ADC). Two equal cur-

2472


Figure 4: The watertight box containing the control computer; the vertical bottom boardhosts the XBT probe interface.

rents are injected in both the wires com-ing from the probe and the potentials VA

and VB are measured: the circuit is closedby the sea water to which the circuit is“grounded” through the ship’s hull; theswitch Sw1 (a relay contact) allows to testthe continuity of the probe circuit beforethe launch.The use of sea water as a third wire isnecessary because of the high variabilityin the wire resistance also in probes be-longing to the same production lot. Ap-plying Ohm’s law, we obtain the voltageacross the wires, the thermistor and thefictitious resistor constituted by the seawater multiplying their resistance by theflowing current. Assuming I1=I2=I andRwire1=Rwire2=Rwire, we can write:

• Vsea = Rsea x 2 I

• Vw1= Rwire x I• Vw2= Rwire x I• Vw1 = Vw2 = Vw

• Vth= Rth x I• VA = Vsea + Vw

• VB= Vsea + Vth + Vw

• VB= VA + Vth

So, VA is the sum of the voltage acrossRwire1 and Rsea (the resistance of sea wa-ter), VB is the sum of the voltage acrossRwire2, Rthermistor and Rsea. The voltageacross the thermistor is obtained subtract-ing VA from VB .To avoid any perturbation to the measure-ment, a multiple stage amplifier was usedto adapt the signal coming from the probeto the need of the ADC; the circuit wasbuilt using high quality precision instru-mentation amplifier ICs. The first stage

2473

Technologies

CONTROL ELECTRONICS AND DATA ACQUISITION SYSTEM

GSM -GPRS MODEM(S) GPS RECEIVER

POWER ACTUATORS

MULTIPLE LAUNCHER

OPERATOR CONSOLE

LAPTOPPC

SATELLITE MODEM

ANALOG AND DIGITAL INPUTS

FROM PROBES AND SHIP SENSORS

Figure 5: The system block schematics.

uses unity gain configuration, having highinput and low output impedance; the sec-ond stage is a differential amplifier withgain = 2; the third stage is a low out-put impedance low-pass filter. Ten timesa second, 16 measurements are performedand their mean is calculated; the resistanceof the thermistor is obtained using Ohm’slaw, or, better, using the regression coeffi-cients or the conversion table obtained af-

ter calibration of the circuit against a setof standard high precision resistors (Figure7). The measured temperature is finally ob-tained using the standard formula [14]:

T = −273.15+1/[A+(B× ln R)+C× (ln R)3],

where: A = 1.29502 x 10−3, B= 2.34546x 10−4 and C = 9.9434 x 10−8. Theconstants were determined from laboratorytests of XBT thermistors [15].

2474


to

ADC

A B

Rsea

A

Rwire1

Rwire2

Rthermistor

I1

I2

Rballast

Rballast

Sw 1

x 2

B

Ax 1

x 1A B

Rsea

A

Rwire1

Rwire2

Rthermistor

I1

I2

Rballast

Rballast

Sw 1A B

Rsea

A

Rwire1

Rwire2

Rthermistor

I1

I2

Rballast

Rballast

Sw 1A B

Rsea

A

Rwire1

Rwire2

Rthermistor

I1

I2

Rballast

Rballast

A B

Rsea

A

Rwire1

Rwire2

Rthermistor

I1

I2

Rballast

Rballast

A BB

Rsea

A

Rwire1

Rwire2

Rthermistor

I1

I2

Rballast

Rballast

Sw 1

x 2

B

Ax 1

x 1

x 2

B

Ax 1

x 1

x 2x 2

B

Ax 1

x 1B

Ax 1

x 1B

Ax 1

x 1B

Ax 1

x 1

Figure 6: The XBT probe interface board schematics.

Figure 7: Calibration of the conversion circuit against a set of high precision resistors.

2475

Technologies

3.2.2 The software

A local and a remote control programswere developed, able to manage 96 launchevents. Launch options are:• northern than a defined latitude• southern than a defined latitude• eastern than a defined longitude• western than a defined longitude• far from a previous launch• at GPS time• at PC time.Collected data are locally stored and can betransmitted as e-mails. The local controlprogram was written in Microsoft Com-piled Basic v. 7.1 and runs in DatalightDOS environment. Assembly languageroutines are used to manage the Analog toDigital Converter. This program is able tofully control the instrument functions, i.e.real time and position acquisition, compar-ison against set points-times, probe release,data acquisition and transmission, ancillaryfunctions.Every hour and at every launch event,a “sequence manager” starts a macro-command sequence, that can be differ-ent for each time and is remotely repro-grammable; new releases of the softwareand of the sequences are uploadable tothe station without suspending its normalactivity. The macro-commands enable tomanage the data acquisition and transmis-sion, the mission programming, the stationhardware and the measuring instruments.The system can be connected to a computer(local laptop or remote desktop), using aremote control program (Figure 8), writ-ten in Microsoft Visual Basic, running inWindows environment. This program en-ables to set up all the launcher functionali-ties, prepare launch event sequences, trans-fer files to and from the launcher, and, ifneeded, to take control of all the launcher

operations, including the time-position ac-quisition and comparison.

4 Tests

The system was tested in laboratory andduring two short cruises. The in situ testswere performed in March 2006 using theItalian Hydrographic Institute vessel Mag-naghi (Figure 9) and in June 2006 with theENEA boat S. Teresa, in the Ligurian Sea— Northwestern Mediterranean. Duringthese tests, data was recorded with the au-tomatic system to verify the efficiency ofits hardware and software, and, for com-parison, XBT probes were launched witha standard Sippican System (hand launcherLM 3A, MK21 read-out card).Figure 10 shows a profile obtained by themultiple launcher (A) together with a pro-file measured simultaneously by the Sip-pican system (B); comparison of valuesdemonstrates the high quality of the auto-matic system.

5 Discussion and conclu-sion

The multiple launcher development aimedat improving the cost-effectiveness of anoperational observation system. The pos-sibility to have a light, simple and robustsystem allows easy use by ship personnel.This instrument requires a simple powerconnection and very limited space (about1 × 0.5 m). As usual, the best solution isto have a place on a ship side, possibly lee-ward. The profiles recorded by the multiplelauncher system show temperature valuescongruent with other measurements. Com-pared to the existing hand launcher system,

2476


Figure 8: Two screenshots of the remote control software.

the improvement is evident. The multi-ple launcher allows automatic launches ofa certain number of probes (eight XBTs, inthe actual arrangement) without human in-tervention. With respect to the commercialTSKA AL-12 multiple launcher, the newlydeveloped device has a minor amount ofelectric or mechanic parts, increasing thereliability of the entire system. Signifi-cant improvement is achieved by the lighterweight (about 30 kg instead of about 120kg), also due to the absence of electric mo-tors with related protections from sea wa-ter. A further advantage is the possibility

to manage the system remotely. The multi-ple launcher (just like the Sippican manualone) requires the manual removal of the se-curity pin from each xbt canister, whereasin the TSKA AL-12 the pin is removed by amotor, but this functionality causes weightand cost.Another advantage of the multiple launcheris that it is an “all-in-one” system. Thetransmission system, GPS, data logger andcomputer are all housed inside the appara-tus, therefore there is no need for the ex-ternal PC required by Sippican and TSKAand the acquisition time can be easily set

2477

Technologies

Figure 9: The multiple launcher installed on the deck of the Magnaghi ship.

by the operator.The small number of electric and mechaniccomponents reduces the risks of malfunc-tions; the remote control capability en-hances the flexibility of the instrument andallows real time operations.The system proved its reliability and isopen to further developments and im-provements to be used with other expend-able probes (XCTDs, T-FLAPs...) simplyadding new interface boards.A final remark is devoted to cost reductionthat can be achieved by using a multiplelauncher.Working in the Mediterranean Sea withcommercial ships (cruise speed greaterthan 20 knots), it is necessary to launcha probe every 20-30 minutes to obtain agood spatial resolution of the mesoscalephenomena on sections several hundredsof miles long, lasting up to 24-36 hours.So, using the manual launcher, at least twotechnicians are needed, with a cost up to

2000 USD for each travel; using the auto-matic launcher, it is possible to avoid thiscost, or at least halve it, engaging one per-son only.The cost of a multiple launcher is about15000 USD, that can be recovered in 10-20 travels by engaging only one person.The device was granted the Italian patent0001359525 and is intended to be grantedthe European Patent EP 1886197.

6 AcknowledgementsThe activity was funded in the frame ofthe MFSTEP project, 5th EU FP. The Au-thor acknowledges the support provided byN. Pinardi and G. Coppini. The work oftechnicians during these years has providedthe necessary information for the design.The Author thanks G. Manzella and F. Re-seghetti for the useful suggestions and A.Baldi and F. Conte from ENEA, for theirtechnical support.

2478


Figure 10: Temperature profiles measured using the multiple launcher (A) and the Sippi-can system (B).

References[1] A. Hecht and I. Gertman. Physical features of the eastern Mediterranean resulting

from the integration of POEM data with Russian Mediterranean cruises. Deep-SeaResearch, (48):1847–1876, 2001.

[2] G. Fusco, G. M. R. Manzella, A. Cruzado, M. Gacic, et al. Variability of mesoscalefeatures in the Mediterranean Sea from XBT data analysis. Ann. Geophys., 21:21–32, 2003.

[3] E. Ozsoy, A. Hecht, U. Unluata, S. Brenner, et al. A review of the Levantine Basincirculation and its variability during 1985-88. Dyn. Atmos. Oceans, 15:421–456,1991.

[4] E. Ozsoy, A. Hecht, U. Unluata, S. Brenner, et al. A synthesis of the Levantine

2479

Technologies

Basin circulation and hydrography, 1985-1990. Deep-Sea Res., 40:1075–1119,1993.

[5] N. Pinardi, I. Allen, E. Demirov, P. DeMey, et al. The Mediterranean ocean Fore-casting System: first phase of implementation (1998 -2001). Ann. Geophys., 21:3–20, 2003.

[6] POEM Group. General circulation of the Eastern Mediterranean Sea. Earth SciRev., 32:285–309, 1992.

[7] C. Millot and I. Taupier-Letage. Additional evidence of LIW entrainment acrossthe Algerian Basin by mesoscale eddies and not by permanent westward-flowingvein. Progress in Oceanography, 66:231–250, 2005.

[8] G. Zodiatis, P. Drakopoulos, S. Brenner, and S. Groom. Variability of the Cypruswarm core Eddy during the CYCLOPS project. Deep Sea Res., 52:2897–2910,2005.

[9] G. Zappala and G. Manzella. An automatic multiple launcher for expendableprobes. Operational Oceanography: Present and Future, pages 188–191, 2006.

[10] G. Zappala, F. Reseghetti, and G.M.R. Manzella. Development of an automaticmultiple launcher for expendable probes. Ocean Sci., 3:173–178, 2007.

[11] G. Zappala. Development of advanced instrumentation for operational oceanog-raphy. Computational Methods and Experimental Measurements XIII, pages 841–850, 2007.

[12] G. Zappala. A software set for environment monitoring networks. Envirosoft 2004,pages 3–12, 2004.

[13] G. Zappala. A versatile software-hardware system for environmental data acqui-sition and transmission. Computational Methods and Experimental MeasurementsXIV, pages 283–294, 2009.

[14] J.S. Steinhart and S.R. Hart. Calibration curves for thermistors. Deep-Sea Res.,15:497–503, 1968.

[15] D.T. Georgi, J.P. Dean, and J.A. Chase. Temperature calibration of expendablebathy-thermographs. Ocean Engineering, 7:491–499, 1980.

2480

Atomic Force Microscopy as a Tool to Investi-gate the Cytotoxicity in Marine Organisms: De-tection of Emerging Nanostructured and Metal-Based Pollutant

M. Girasole1, G. Longo1, G. Pompeo1, A. Cricenti1, P.F. Moretti21, Institute of Structure of Matter, CNR Research Area of Tor Vergata, Roma, Italy2, Department of Earth and Environment, CNR, Roma, [email protected]

Abstract

The Atomic Force Microscopy (AFM) is a well established technique to studythe surface of samples. After its invention the technique has diffused in different sci-entific communities and has been largely applied in several research fields includingsurface and material sciences, solid state physics and biology. Such a spreading re-flects the advantages of AFM compared to more traditional microscopes: AFM pro-vides quantitative 3-dimensional morphological images and, simultaneously, othernon-morphological data on the material properties. These latter can be the surfacefriction (related to the sample’s chemistry), hardness, elasticity, hydrophobicity etc.and can all be probed on the nanoscale and mapped in non-topography images. More-over the imaging is totally non-destructive, requires minimal or no sample prepara-tions and the AFM can be operated in various environments (including physiologicalbuffers).In this paper we present a novel AFM-based approach capable to detect metal-based and/or nanostructured pollutants in marine organisms. This environment- andtoxicology-relevant research subject has many intriguing aspects that can also be pre-liminary studied in vitro. Our approach consists in an operative method combiningdifferent microscopies (including AFM and SEM) to detect the consequences of theoccurrence of nanostructured pollutants at different level of the hierarchic organiza-tion of the body of marine natural bioaccumulators (Mytilus edulis).

1 Introduction

Scanning Probe Microscopies (SPM) [1]are a class of versatile techniques, themost popular being Scanning TunnelingMicroscopy (STM) [2] Atomic Force Mi-croscopy (AFM) [3, 4, 5] and ScanningNear-field Optical Microscopy (SNOM)[6, 7], whose invention has remarkablychanged and modernized the concept of

microscopy. Historically, the SPM tech-niques have been introduced in the eight-ies, when the development of the STM de-served a Nobel price to the inventors Bin-nig and Roher [8]. In the following years,the concepts underlying the developmentof STM have been generalized and noveltechniques, such as the AFM and, later, theSNOM, were introduced.The main idea behind these techniques is


Technologies

the use of a sharp probe, driven in sub-nanometric steps over a sample by a piezo-electric actuator, while maintained at verysmall distance from the surface, to lo-cally investigate topography and additionalproperties. Indeed the SPMs, initially de-veloped as pure imaging techniques, allowto record magnified images of the samplewith a resolution well below the limit ofthe conventional optical microscopy and,at the same time, can produce maps ofphysical quantities not always linked tothe bare sample’s morphology. The detec-tion of these properties and their correla-tion with the sample morphology are a pre-cious method to solve many scientific prob-lems that could not be addressed by a puremorphological analysis [9, 10, 11]. Theadditional physical quantities that can becollected through SPMs can differ, strictlydepending on the nature of the probe em-ployed (i.e. on the specific SPM em-ployed). For instance, the AFM probe isan extremely sharp, pyramidal (or conical),silicon (or silicon nitride) tip mounted atthe apex of a very flexible cantilever todetect, through the deflection of the can-tilever, the magnitude of the forces be-tween the tip and the sample’s. In the mag-netic force microscopy (MFM), the choiceof a magnetic probe enables the map ofmagnetic domains at a high spatial reso-lution. Similarly, in the STM, the probeis a conductive tip brought very close toa conductive sample in such a way thatthe quantum tunneling current between thetwo can be collected and monitored, ob-taining information on the local electronicstate of the sample. In the SNOM, finally,the probe is a tapered optical fiber used toilluminate and/or to collect light from thesample in the near-field regime enablingthe investigation of the local optical prop-erties well-below the diffraction limit.

Obviously, the choice of a specific probeallows the investigation of definite proper-ties of the sample and determines the in-trinsic sensitivity of the related microscopyas well as the spatial resolution that can beachieved in an experiment. Namely, the de-pendence of the force vs. distance typicalof the AFM (following, according to thetheoretical models, a d6 behavior, whered is the probe-sample distance), as wellas the d4 of the SNOM, makes such tech-niques particularly appropriate for molecu-lar scale studies, while the exponential de-pendence from the distance of the tunnel-ing current makes STM the most sensitiveSPM technique, enabling studies down tothe atomic scale. In technical terms, theprobe size and shape are the most impor-tant factors affecting the spatial and non-topographic resolution that can be achievedin an SPM experiment. To improve thisresolution, novel setups, such as the tap-ping mode operation for SNOM and AFM,have been developed in the laboratory ofISM-CNR [12].The present work is focused on applica-tions of the AFM. Indeed, its ability todetect the normal and lateral forces be-tween tip and sample can be exploited tocollect high-resolution, quantitative, three-dimensional topographies while probingthe local chemical properties (maps of sur-face friction) of the sample. These ca-pabilities are unique tools in many re-search fields, ranging from the tribology(the study and application of friction, lu-brication and wear) that, with AFM, canbe studied on the nanoscale [13] to the cel-lular or molecular biology. In particular,the biological applications of AFM are re-markably interesting because the techniqueallows non destructive imaging that canbe performed in different environments in-cluding physiological buffers [14] enabling

2482


high-resolution studies of the time evolu-tion of biomolecules or entire living sys-tems [15].Thus, the AFM peculiarities makes such atechnique, not only a valuable tool to char-acterize the morphology of the samples onthe nanoscale in a quantitative and non de-structive way, but also a powerful instru-ment to study the material properties anda valid support to the development of ap-plications in a variety of different researchfield.In this paper we describe the potential ofinnovative investigation techniques basedon the use of AFM in the study of aquaticorganisms and in the analysis of emergingtopics of environmental and toxicologicalinterest. In particular, we will give exam-ples of the capability of AFM to detect thedirect presence and the biological conse-quences of the exposure of biosystems tometal-based and nanostructured pollutants.Nanostructured pollutants [16] can be ofnatural origin or anthropogenic. These lat-ter, mainly produced by the industry ofnanotechnology and increasingly releasedinto the environment, owe most of theirproperties to the small size and to chem-ical functionalization. Such treatmentscan change the surface properties of thenanoparticles enhancing their specific reac-tivity. Once released into the environment,their fate will be a probable dissolution inrainwater and groundwater that could leadthem in the freshwater and, finally, intothe sea. While several studies have fo-cused on the interaction of nanosized pol-lutants with terrestrial organisms, there arestill very few experimental data on their in-teractions with the marine flora and fauna.This is partially due to the extra degree ofdifficulty imposed by the aquatic environ-ment, in which nanostructured pollutantsmay interact with pre-existing surfactants,

changing their surface chemistry and en-hancing the properties and the bioavailabil-ity of individual pollutants. Cytotoxicitydue to different types of nanoparticles hasbeen observed in various freshwater andmarine organisms (including algae, bacte-ria, small fish and crustaceans). Naturally,different types of nanoparticles or differentorganisms show a different degree of toxic-ity. In the mussels, we cite the induction ofoxidative stress by gold nanoparticles [17],reduction of phagocytosis after administra-tion of cadmium-containing nanoparticlesand the detection of apoptosis and damageto the membranes after exposure to variouscontaminants [18, 19, 20]. Overall, unfor-tunately, the research conducted on mus-sels or other shellfish, are rather scarce,and, thus, the effects of new pollutants onthe marine organisms are still largely un-known.The rationale of the present work consistsin the description of the abilities of AFM(and, in particular, of the instruments de-veloped at CNR) in the detection of cyto-toxic effects due to different heavy met-als administered to cells in culture, first inthe form of salts and then as nanoparticles.This approach, based on a morphologicalanalysis conducted in a highly controlledway, allows to adequately highlight the po-tential of the methodology and to point outthe information that can be obtained. Sub-sequently, an approach to the analysis ofmarine organisms is shown. The methodcombines data from conventional opticalmicroscopy, AFM and SEM and allowsto detect the presence and the effects ofnanostructured pollutants at different lev-els of the hierarchical organization of theorganism.

2483

Technologies

2 Materials and Methods

2.1 AFM: Instrument and datacollection

The microscopy group at ISM-CNR has along term experience in the developmentof new AFM, STM and SNOM micro-scopes for a variety of applications. Thedesign includes mechanical and electroniccomponents and the development of ded-icated software for image collection anddata analysis. As consequence of such ac-tivities, several patents and a continuousimprovement in the performances and inthe range of application of the microscopeshave resulted over the years.The home-built AFM instrument employedin this study was described in detail else-where [21] and has already been used in thestudy of different kind of biological sam-ples [22, 23, 24].Briefly, the instrument consists of a ∼20cm long stainless-steel unit made oftwo separable hollow cylindrical supportsequipped with a vibration isolating system.The lower unit contains the sample holder,that is mounted at the top of a piezoelec-tric scanner that allows a maximum scansize of 40 um x 40 um x 8 um. Addition-ally, an X-Y-Z motor controlled translatoris used for larger motion and allows to se-lect suitable areas of the sample. The up-per unit contains the cantilever holder, themirror deflection system and a four sectorposition sensitive photodiode used as de-flection detector. An electronic feedbackloop is used to integrate the optical signalallowing to maintain a constant cantileverdeflection during the data acquisition. Thenominal capabilities of the microscope in-clude a typical lateral resolution ≤10 nmand 0.1 nm in the Z direction (for a sampleplaced on a flat surface).

The friction images arise from the record-ing of the lateral torsion of the cantilever.These are the consequence of forces ex-erted at the base of the tip (i.e. the contactpoint) during the scan and can be relatedto differences in the friction coefficient ofdifferent zones of the sample under investi-gation. In other terms, the lateral force im-ages can be related to the chemical compo-sition of the specimen and are particularlysensitive to contrasts between soft (biolog-ical) and hard (metallic) matter.Regarding the AFM images presented inthis work, cconstant force and lateral fric-tion images were acquired simultaneouslyin air, (at room temperature and controlled30% relative humidity) with a typical scanrate of 3-4 sec/row and with the microscopeworking in the weakest possible regime ofrepulsive forces (less than 1 nN from zerocantilever deflection). Gold coated Si3N4

cantilevers with a spring constant of 0.03N·m−1 and a statistical apical radius ofabout 10 nm were used. The raw datawere treated only by a background subtrac-tion and, when required, a plane alignment.The reproducibility of the data (includingthe absence of sample damage due to themeasurement procedure) was carefully andsuccessfully tested.

2.2 Sample preparation

2.2.1 In vitro experiments: Heavymetal doping and metal nanopar-ticles doping

The protocol for handling and doping thecultured cells can be summarized as fol-lows: at confluence, the pancreatic cellsINS-1, cultured at 37°C and 5%CO2, wereseeded on poly-lysine treated glass coverslides. After 24h the medium was re-moved and the cells were incubated for 3

2484


h, at 37°C, with complete RPMI medium(Gibco laboratories; Paisley, Scotland UK)in the presence of 100 uM of metal com-pounds (CdCl2; ZnCl2 or PbCl2). A con-trol sample not exposed to the metal wasprepared as well. Then, the cells werefixed in para-formaldehyde (2.5 % for 20min), washed twice with PBS (not contain-ing Ca+2 or Mg+2 ions) and twice withdistilled water to eliminate metal sedimentsfrom the cell surface. Finally, the sampleswere air dried, to allow a detailed AFManalysis spanning over several days. Allreagents (from Carlo Erba or Fluka, Milan,Italy) are of analytical grade.A mixture of Hematite (Fe2O3) and Mag-netite (Fe2O3) nanoparticles (Nanostruc-tured & Amorphous Materials Inc., LosAlamos, New Mexico, USA) were used.The nominal mean size of the chosen par-ticles was 60 nm (± 30 nm, rms) forhematite and 35 nm (± 15 nm, rms) formagnetite. Before use, all particles werede-pyrogenated at 210°C in glass tubes forabout 2 hours. The particles were, then,stored in a mother solution of sterile Dul-becco’s Phosphate Buffered Saline (Invit-rogen Corporation, CA USA) at the con-centration of 1mg/ml at 4°C temperature.For each test, a different concentration ofparticles was used in order to detect thethreshold toxic concentration.NHI 3T3 mouse fibroblast cells were cul-tured in Dulbecco’s Mem (GIBCO tm,U.K.), supplemented with 100 IU/100ug·ml−1 Penicillin/Streptomycin, 2 mML-glutammin, 1 mM Sodium piruvate,10% foetal bovine serum and incubated at37°C in presence of humidified atmosphere(95%) and CO2 (5%). After this prelimi-nary incubation, the 3T3 cells were seededin 96-well microplates (NUNC Micro wellplates, USA), at a concentration of 7x104

cells per 100 ul/well. After a 24 hours incu-

bation, the different nanoparticles were in-troduced in the wells and the cells were in-cubated for 24 hours. The chosen nanopar-ticle concentration was varied in each welland for each material: 15 mg, 3 mg, 0.6mg, 1.2 x 10−1 mg, 2.4 x 10−2 mg, 4.8 x10−3 mg (concentration values calculatedx 106 cells). A well containing the pristinecells acted as control.

2.2.2 Marine organisms

The experiments on sea-life creatures wereperformed on mussels, that are naturalbioaccumulators and, thus, particularly ap-propriate for studies of environmental pol-lution. We collected different musseltissues, mostly digestive glands and go-nads, that have been investigated afterhistological sample preparation. Mus-sels (Mytilus edulis) were collected in thecentral Tyrrhenian inshore. The organsof interest were excised and immediatelyfixed. Specifically, two fixation proto-cols were chosen: for the preparation ofthe AFM and SEM samples, a Bouin’ssolution (picric acid: formalin: aceticacid=15:5:1 v/v) was used, while for thesections to be investigated by conventionalmicroscopy, a Methacarn-based fixativewas preferred (methanol:chloroform:aceticacid=5:3:1 v/v). The sample preparationrequires dehydration, embedding in paraf-fin blocks and, when necessary, staining ofthe sections.Dehydration was performed according tothe following: immersion in 70% ethanol(several steps of 2 hours), 80% ethanol(2 hours) 90% ethanol (1,5 hours), 95%ethanol (1 hour) 100% ethanol (2 x 15 min-utes each). For the samples fixed withMethacarn, the dehydration was much sim-pler, consisting only in two baths in 100%ethanol.

2485

Technologies

Figure 1: The Atomic force microscope employed in the present study, designed andbuilt at ISM-CNR.

After dehydration, the tissues were treatedwith a solvent (toluene or xylene) that pre-pares the sample to the paraffin embeddingprotocol: immersion in 100% Histoclear(2 x 15 minutes); 50% paraffin and 50%Histoclear (1 hour at 60 °C); 100% liquidparaffin (2 x 1,5 hours at 60 °C); inclu-sion in a new paraffin block. Subsequently,the samples were cut with a rotary micro-tome (Leitz 1512) and the sections weredeposited on a substrate coated with bovinealbumin or poly-lysine to promote adhe-sion and, finally, dried overnight at 37 °C.The staining gives no benefits for SEM orAFM analysis, yet its use allows a use-ful preliminary screening of the sectionsthrough conventional optical microscopy.The protocol used requires a preliminaryimmersion in Histoclear, followed by theinverse path of the dehydration. Thisprocedure is required in order to get ridof the paraffin and to prepare the sam-ples for the true Ematossilina/Eosina based

staining, performed as follows: immer-sion in Hemalum for 4-6 minutes; rinsingin distilled water; immersion in eosin for1 minute; rinsing in 95% ethanol; 100%ethanol 2 x 5 minutes; Histoclear, 2 x 5minutes. Finally the sections were treatedwith a protective agent (Eo-Kit), coveredwith a glass slide and dried.

3 Results and DiscussionAt first, we will describe the potential ofAFM to investigate the effects of the invitro exposure of cultured cells to heavymetals [11] administered as either chlorinesalts or nanoparticles.Heavy metals are key components of pro-teins and enzymes, nevertheless, the ef-fect on biological systems of the exposureto heavy metals is an important issue dueto possible implications on human healthcaused by the increasing environmentalpollution [25]. The uptake of heavy met-

2486


Figure 2: AFM topography, cross sections and lateral force images of, from top to bot-tom, control cells, cadmium, zinc and lead-doped cells. Pristine cells have a typical domeshape, with higher quotas in the nuclear region, that can change upon exposure to exoge-nous metals. The most evident effect is the appearance of strong contrasts in the lateralforce images after doping the cells with cadmium or zinc.

als in many biosystems, indeed, results indifferent effects depending on the charac-teristics of the metal and of the cell line.The most important aspects of the interac-tion between a generic cell and an heavymetal are governed by the bioavailability(related to concentration, exposure time,temperature) and the nature of the metal(intrinsic toxicity and cell uptake). As con-sequence of these interaction parameters,the intrinsic toxicity of the considered met-als follows the order Cd > Pb ≥ Zn. Todescribe the cell response to heavy metals,

in vitro cultured cells were exposed to 100uM chlorine salts of Cd+2, Zn+2 or Pb+2

and the effect of such administration werecharacterized by AFM.The data reveal that two major classes ofeffects appear on the cultured cells: overallshape variation and occurrence of frictioncontrasts. In particular all the metal treat-ments induce a change of the cell shapethat become more spherical. Besides that,Cd and Zn produce an increase in the cellheight while Pb behaves oppositely. Thesemorphological patterns distinguish Cd and

2487

Technologies

Figure 3: AFM topography (left) and lateral friction (center) images of iron nanoparticlesin a cell culture. Right panel shows human 3T3 cell in the presence of iron nanoparticlesthat can be detected above and around the cell surface. A detailed analysis of the imagesallows to understand if the nanoparticles have been uptaken by the cell.

Figure 4: Maps of gonads slices from the mussel Mytilus edulis show the arrangementof different tissues in the organ. For instance, in the left image, filamentous matrixes inthe upper and fibrous tissue in the lower part can be recognized. Some secretive glandscan also be seen in the image. The right panel shows a different organs whose texture isdominated by secretive cavities.

Zn behaviour from Pb, that seems quiteeffective in killing the cells by inducingnecrosis.More impressive differences are revealedby the analysis of the lateral force images.Indeed, in both the Zn and Cd-doped sam-ples strong friction contrasts were observedat the edge of the cells, as well as on theirplasma membrane. The contrasts observedat the edge of the cells are likely to be as-sociated to the direct detection of metalaggregates: the AFM tip feels dramati-cally different lateral interactions when incontact with a cell surface or with a metalagglomerate. On the other hand, the fric-

tion contrasts on the plasma membrane aremore deeply related to the cell metabolism.Indeed, these latter contrasts reveal thatthere are hot spots on the plasma mem-brane whose chemical characters changesafter the metal incorporation. A deeperanalysis of the observed friction contrastsreveal minor differences between what canbe observed in the cases of Cd and Zn.For instance, differences in the localizationof the friction effects on the membrane orother qualitative differences between thetwo cases are likely to reflect different bio-chemical signals occurring in a commondegradation pathway. In the same context

2488


Figure 5: Normal digestive gland: optical microscopy reconstruction and SEM imaging.

the absence of friction contrasts observedfor lead doped cells corresponds, in bio-logical terms, to a qualitatively differentsituation.As a whole, these data demonstrate thatAFM can detect the direct accumulationof metal clusters in the extracellular ma-trix. This is a non-trivial result because theheavy metals, in the presence of biosys-tems, are expected to be chemically neu-tralized by the cells in such a way thattheir detectability may be altered. More-over, our study shows that a quantitativeAFM analysis of morphological variations,coupled with the detection of chemical in-homogeneity on cells membrane, allows todiscriminate the different pathways of celldamage induced by Cd, Zn and Pb on cul-tured cells. In the case of Pb the observedmorphological pattern unequivocally sug-gest that this metal, basically, kills the cellsby necrosis. The consequence of Cd orZn administration are more subtle: theydisturb the cell morphology and the homo-

geneity of the plasma membrane suggest-ing that a more complex biochemical pat-terns, for some points resembling the earlyapoptotic response, takes place. Next, wewant to exploit the high-resolution of theAFM to study the interaction of nanos-tructured materials of metallic nature withbiosystems. Indeed, nano-sized pollutantsinteract in different ways with living organ-isms: their effects originate both from theirchemical composition and from their sizeand shape. This makes them a new kind ofpollutants that must be closely studied.In our in-vitro experiments, cultured cellswere exposed to iron-based nanoparticlesand the consequence of the exposure wasevaluated by AFM. Typical images col-lected on these samples are reported in Fig-ure 3, where the potential of the techniqueto detect the nanoparticles, both alone andin the presence of cells, is evident. Thepeculiarities of AFM, in this case, can beused to understand the basic interactionsbetween the nanostructures and a biosys-

2489

Technologies

Figure 6: A partially destructured digestive gland: optical microscopy reconstruction andSEM imaging.

tem, as well as to understand the long-term consequences of these interactions.Indeed, for instance, a careful observa-tion of the AFM and of lateral force im-ages allows to understand whether or notthe particles are interacting with the outerleaflet of the cell membrane or have beenincorporated, presumably by endocytosysor pinocitosys, into the cell, leaving justa weak morphological fingerprint on theplasma membrane. Moreover, a series ofexperiments performed at different timesafter the administration of the nanomate-rials can couple the uptake of exogenousmaterials with the evolution of the cellmorphology and the development of cel-lular disorders.The above reported in vitro experimentshave introduced the reader to the capabil-ity of the AFM to study the interactionbetween metal-based nanostructured mate-rials and biosystems.With this essential knowledge we can ap-

proach another topic of this work: theapplication of AFM to the study of real-life marine organisms to detect the finger-prints of nanostructured pollutants. A duepremise is that in vitro experiments involv-ing cell culture from mytilus or other ma-rine organisms would be highly envisagedin this scientific context too.The experiments on marine organismswere carried out on mussels of the speciesMytilus edulis collected in the inshore ofthe middle Tyrrhenian sea. Indeed, thebioaccumulation ability of these animalsmakes them ideal candidates for this re-search. Two organs of the mussels, thoughtto be among the most sensitive to envi-ronmental agents, were considered: thedigestive gland and the gonads. The firstscreening were performed through a recon-struction of the slices through conventionaloptical microscopy at relatively low mag-nification (typically 200x). An example isgiven in Figure 4 where the reconstruction

2490


Figure 7: AFM topographies (left images), cross section and lateral force images (rightside images) collected on digestive glands of M. edulis. The contrasts in the lateral forceimages clearly evidence presence and exogenous nature of the nanomaterials.

of a gonad’s slice from an healthy subjectis shown. These optical maps identify themost interesting samples for further anal-ysis, are useful to provide information onthe tissue’s texture and on the occurrenceof extensive or localized alteration at thesub-tissue level.The most interesting sections were subse-quently investigated by Scanning ElectronMicroscopy (SEM) at a much larger mag-nification (up to 5000x). These measure-ments allow to describe in greater detail thefine texture of selected areas of the organs(for instance specialized domains) and toanalyze the differences in the microscopicstructure of normal or altered tissues oc-

curring in a given organ. The SEM and theconventional microscopy data, have beencoupled providing a joined morphologicalcharacterization of the sample. Interestingexamples are shown in Figures 5 and 6.Among the analyzed samples, there are fewcases where anomalies in the tissue’s tex-ture were identified. Such destructured fea-tures have been recognized mostly in thedigestive glands that, hence, will be takenas illustrative examples for a more detaileddiscussion. Indeed, comparing the opti-cal images of the two reported digestiveglands (Figures 5 and 6), the most promi-nent structure in both the slices are theducts and the secretive glands for the nutri-

2491

Technologies

ents’ assimilation (the segmented irregular-shaped clusterized light area). Despite that,a comparison of the two slices clearly ev-idences a much higher degree of tissueorganization and of functional domain seg-regation in the case of the normal section(Figure 5), while a large number of small,hollow grain-like structures are detectablein the case of the tissue of Figure 6. Moredata on the tissues and on the observablestructures is provided by the correspondingSEM images.A deeper morphological characterizationhas been provided by AFM. Indeed, inthis work, AFM has mostly been used torefine the information on sections of par-ticular interest This is mostly because, de-spite AFM characterization can produceextremely high spatial resolution quanti-tative images, such analysis is limited toquite small areas of the samples (maxi-mum scan size is 40 um x 40 um) and is,therefore, not the best technique to providean overview of the slices. Hence, the AFMmeasurements have been performed onlyon selected samples and revealed quiteinteresting data. Indeed, while the mor-phology of the normal-shaped digestiveglands is basically featureless, several clus-ters of nanostructured exogenous materialhas been observed, through AFM topogra-phy and lateral force imaging (see Figure7), in the partially destructured digestivegland of Figure 6.The topographies collected on the partiallydestructured digestive gland clearly showthe presence of nanomaterials as clustersor as single sparse spots, with sizes in thetens of nanometer (FWHM) scale. Suchspots seems to be mostly embedded in thetissue texture or, sometimes, protrudingfrom the cell surface rather than incorpo-rated into the cell body. Their presence iseven more clear in the lateral-force images

(see Figure 7), where friction contrasts,similar to those observed in the case of thein vitro heavy metal doping of cells, areshown. The chemical composition of thenanostructures cannot be determined withaccuracy, although the images acquired inthe friction mode indicate that the nanopar-ticles’ chemistry is different than their sur-roundings. No evidence of the presenceof nanostructured materials incorporatedin the tissues has been reported after ex-tensive AFM analysis of normal-shapedglands. A preliminary interpretation of to-pography and lateral force data suggeststhat the characteristic of the friction con-trasts are compatible with a metallic naturefor the nanoparticles that apparently hasnot been transformed by the organism.This relevant finding fully demonstrates thepotential of our approach to detect and cor-relate tissue anomalies to nanostructuredmaterials in the organs of marine species.The advantages of the presented methodcompared to the standard investigationsbased on biochemical methods can be eas-ily pointed out. Indeed, the standard meth-ods of investigations are, mostly, basedon the evaluation of biomarkers that arespecific molecules, mobilized in the pres-ence of one, or several, external stimuli.Thus, by analyzing such molecules, onecan demonstrate that animals have beensubjected to environmental stresses, butit is quite difficult to directly trace downthe specific cause of the stress or to iden-tify the chemical mechanisms that causedthe stress [19, 25, 26]. In this context,our study proposes an integrated approachbased on convergence of various morpho-logical techniques that can address theproblem at different level of the hierar-chic organization of the marine organisms.In particular, it can add information onthe localization, preliminary identifica-

2492


tion and transformation or neutralizationof the nanomaterials and on their correla-tion with the occurrence of morphologicalalterations of tissues.

4 ConclusionsAtomic force microscopy is an high reso-lution, non destructive, technique that pro-duces intrinsically three-dimensional andquantitative topography images while, si-multaneously, probing the local chemicalproperties (maps of surface friction) of thesample.The ability of AFM to evaluate the ef-fects of an external stimulus on a biosystemhas been highlighted by an in vitro studyof cultured cells treated with heavy met-als administered as either chlorine salts oras nanoparticles. This approach providesa method to help understanding the basicmechanisms of interaction in reproducibleand controlled conditions. The AFM inves-tigation succeeded both in the direct local-ization of the pollutants and in the identi-fication of different paths of cell degrada-tion determined by the uptake of the vari-

ous metals.Moreover, a more sophisticated morpho-logical approach devoted to the detectionof nanostructured pollutant in the tissuesof a marine organisms has been described.The method integrates conventional opti-cal microscopy, SEM and AFM to addressthe problem of the presence of exogenousnanomaterials at different levels of the hi-erarchic organization of the organism.The reported data showed that the couplingof optical images and SEM approach isvery promising as it contains the potentialto describe tiny features at high resolutionand to complement their occurrence withthe morphological status of the entire tis-sue. The implementation of the AFM, giveto this approach the capability to detectthe presence of exogenous material on thenanoscale, thus providing the basic infor-mation for a comprehensive understandingof the pollutant-to-biosystem interaction.In conclusion, the proposed approach,based on the implementation of AFM withmore conventional microscopies, has beenproved an interesting and novel investiga-tion method in the study of the impact ofnovel and emerging nanostructured pollu-tant on marine species.

References[1] E. Meyer, H.U. Hug, and R. Bennewitz. Scanning Probe Microscopy: The Lab on

a Tip. 2003.

[2] G. Binnig and H. Rohrer. In touch with atoms. Review of Modern Physics, 71:S324–S330, 1999.

[3] H.G. Hansma and L. Pietrasanta. Atomic force microscopy and other scanningprobe microscopies. Current Opinion in Chemical Biology, 2:579–584, 1998.

[4] F.J. Giessibl. Advances in atomic force microscopy. Reviews of Modern Physics,75:949–981, 2003.

2493

Technologies

[5] K.D. Jandt. Atomic force microscopy of biomaterials surfaces and interfaces. Sur-face Science, 491:303–332, 2001.

[6] C. Dunn. Near-Field Scanning Optical Microscopy. Chemical Reviews, 99:2891–2927, 1999.

[7] S.F. Wu. Review of near-field optical microscopy. Frontiers of Physics in China,3:263–274, 2006.

[8] G. Binnig, H. Rohrer, C. Gerber, and E. Weibel. Surface Studies by ScanningTunneling Microscopy. Physical Review Letters, 49:57–61, 1982.

[9] G.J. Simpson, D.L. Sedin, and K.L. Rowlen. Surface roughness by contact versustapping mode AFM. Langmuir, 15:1429–1434, 1999.

[10] S.N. Magonov, V. Elings, and M-H. Whangbo. Phase imaging and stiffness intapping-mode atomic force microscopy. Surface Science, 375:L385–L391, 1997.

[11] M. Girasole, A. Cricenti, R. Generosi, G. Longo, G. Pompeo, S. Cotesta, andA. Congiu-Castellano. Different Membrane Modifications Revealed by AtomicForce/Lateral Force Microscopy After Doping of Human Pancreatic Cells With Cd,Zn, or Pb. Microscopy Research and techniques, 70:912–917, 2007.

[12] M. Girasole, G. Longo, and A. Cricenti. An Alternative Tapping Scanning Near-Field Optical Microscope Setup Enabling the Study of Biological Systems in LiquidEnvironment. Japanese Journal of Applied Physics, 45:2333–2336, 2006.

[13] E. Tocha, H. Schonherr, and G.J. Vancso. Quantitative Nanotribology by AFM: ANovel Universal Calibration Platform. Langmuir, 22:2340 –2350, 2006.

[14] C. Bustamante, C. Rivetti, and D.J. Keller. Scanning force microscopy under aque-ous solutions. Current Opinion in Structural Biology, 7:709–716, 1997.

[15] Y.F. Dufrene and G.U. Lee. Advances in the characterization of supported lipidfilms with the atomic force microscope. Biophysica Biochimica Acta, 1509:14–41,2000.

[16] A. Vaseashta, M. Vaclavikova, S. Vaseashta, G. Gallios, P. Roy, and O. Pummakarn-chana. Nanostructures in environmental pollution detection, monitoring and reme-diation. Science and Technology of Advanced Materials, 8:47–59, 2007.

[17] S. Tedesco, H. Doyle, G. Redmand, and D. Sheehan. Gold nanoparticles and oxida-tive stress in Mytilus edulis. Marine Environmental Research, 66:131–133, 2008.

[18] M. Fafandel, N. Bihari, L. Peric, and A. Cenov. Effect of marine pollutants on theacid DNAse activity in the hemocytes and digestive gland of the mussel Mytilusgalloprovincialis. Acquatic Toxicology, 86:508–513, 2008.

2494


[19] D.M. Lowe, V.U. Fossato, and M.H. Depledge. Contaminant-induced lysosomalmembrane damage in blood cells of mussels Mytilus galloprovincialis from theVenice Lagoon: an in vitro study. Marine Ecology Progress Series, 129:189–196,1995.

[20] M.N. Moore. Do nanoparticles present ecotoxicological risks for the health of theaquatic environment. Environment International, 32:967–976, 2006.

[21] A. Cricenti, R. Generosi, and M. Barchesi. A multipurpose scanning near-fieldoptical microscope: Reflectivity and photocurrent on semiconductor and biologicalsamples. Review of Scientific Instruments, 69:3240–3244, 1998.

[22] M. Girasole, A. Cricenti, R. Generosi, A. Congiu-Castellano, G. Boumis, andG. Amiconi. Artificially Induced Unusual Shape of Erythrocytes: An Atomic ForceMicroscopy Study. Journal of Microscopy, 204:46–52, 2001.

[23] G. Pompeo, M. Girasole, A. Cricenti, F. Cattaruzza, A. Flamini, T. Prosperi, J. Gen-erosi, and A. Congiu-Castellano. AFM characterization of solid-supported lipidmultilayers prepared by Spin-coating. Biochimica Biophysica Acta, 1712:29–36,2005.

[24] G. Longo, M. Girasole, G. Pompeo, A. Cricenti, G. Andreano, F. Cattaruzza, L. Cel-lai, A. Flamini, C. Guarino, and T. Prosperi. An AFM investigation of oligonu-cleotides anchored on unoxidized crystalline Silicon surfaces. Biomolecular Engi-neering, 24:53–58, 2007.

[25] P.A. Olsvik, P. Gundersen, R.A. Andersen, and K.E. Zachariassen. Metal accu-mulation and metallothionein in brown trout, Salmo-trutta, from two Norwegianrivers differently contaminated with Cd, Cu and Zn. Comparative Biochemistryand Physiology, 128:189–201, 2001.

[26] D.R. Livingstone. Contaminant-stimulated Reactive Oxygen Species Productionand Oxidative Damage in Aquatic Organisms. Marine Pollution Bullettin, 42:656–666, 2001.

2495

Technologies

2496

A new Versatile Reconfigurable AcquisitionBoard: an Application to Marine Acoustic SignalsMonitoring

M. Zora1,2, G. Sottile2, G. Nunzio Gallı2, S. Aronica1, A. Bonanno1,2, G.Buscaino1,2

1, Institute for Coastal Marine Environment, CNR, Capo Granitola (TP), Italy2, Daimar s.r.l. Innovation for Science (IAMC-CNR Spin-off), Mazara del Vallo (TP),[email protected]

Abstract

Make a multi-channel data acquisition system is often complex, expensive andcumbersome in marine applications. A new compact system for acoustic data acqui-sition and elaboration has been made by the Hardware Development Team of CapoGranitola IAMC-CNR.The system, is based on a custom board with an FPGA (Field Programmable Gate Ar-ray), four parallel acquisition chains with charge amplifier, variable gain, selectablehigh-pass filter and Analog to Digital converter up to 1.33 MS·s−1 with a 16 bit sam-ple resolution. All in a PCB measuring just 10cm x 7.5cm. All electronic signals andcontrol lines, flow in a single FPGA chip that manages the low level I/O operationsof the whole board, so system operation can be modified changing only the firmwareof the FPGA. Today, high performance, small size and low cost FPGAs, allow re-searchers to build reconfigurable instruments for various scientific applications.Preliminary tests results are presented.

1 Introduction

In the last decade, marine scientific instru-ments have been in high demand, partic-ularly autonomous instruments that are ca-pable of staying under water for a long timewhile performing continuous data acquisi-tion and storage. This paper introduces aninstrument, under development, based ona concept of reconfigurable measurementinstrument. The “reconfiguration” term isreferred as hardware reconfiguration, per-formed via software.Development of such equipment is a jointeffort between different disciplines, such as

physics, electronics, marine biology etc.Long Term monitoring of marine ecosys-tems is today a challenge, logistically diffi-cult and expensive.Many instruments are capable of measur-ing a wide range of environmental pa-rameters but only a few of these canrecord biological activity such as acous-tic data. Acoustic monitoring can improvethe knowledge of marine environment, be-cause sounds can be an indicator of back-ground noise and health status of biologicalhabitat [1].Some marine species use underwatersounds to help them navigate, locate food

mailto: [email protected]

Technologies

and communicate with other animals.For example, dolphins make many types ofsounds: whistles, clicks and burst pulses.Some Whistles are used like a signatureto serve as identification of individual dol-phins. The approximate frequency rangeof bottlenose dolphin whistles is 200 Hz to24 kHz [2]. Clicks are used exclusively forecholocation and are produced in rapid se-quence. The frequency range for echolo-cation clicks is 200 Hz to 150 kHz [2].Dolphins use lower repetition rate of clickswhen echo-locating an object at a great dis-tance, and higher when they move closer toan object, increasing spatial resolution ofecholocation.Man-made noise such as sonar, explosivesand ships traffic, can have an adverse affecton marine mammals and ecosystems. Sev-eral scientific studies are needed to inves-tigate this issue to preserve the safety andprotection of marine mammals and theirenvironment.Scientific instruments used to record ma-rine acoustic data, are often build by re-searchers assembling many different de-vices in order to obtain each time thebest experimental results. These assemblysome times are cumbersome, high powerconsuming and expensive. To meet un-derwater requirements, marine researchersneed small size instruments and low powerconsumptions user friendly devices.A new compact, low power, high per-formance reconfigurable electronic systemhas been developed by IAMC-CNR hard-ware development team in Capo Granitola(TP, Italy).The Digital Acquisition System namedDAQsys, is a four channel reconfigurablehardware for real-time marine acousticdata acquisition and elaboration. This pro-totype is a FPGA-based board that man-ages up to four high frequency signals si-

multaneously. The acquisition system wastested in the laboratory under controlledconditions.

AbbreviationsRefer to appendix for further acronyms andabbreviations.

2 System DescriptionAn overview of the DAQsys board isshown in Figure 1. This board containseverything an engineer needs to create andimplement a complete, small size, instru-ment for acoustic measurements.The board is made up of three main sec-tions: analog to digital front-end, data-management and data-storage.Analog to digital front end includes fourprogrammable gain amplifiers, four pro-grammable high-pass filters and four ADCdevices (Analod Devices AD7983).Data management is absolved by a centralFPGA device that is reprogrammable, flex-ible, low cost and low power. Thanks to itsreprogrammability, the non recurring engi-neering costs are removed from prototyp-ing and testing stages and, in addition, theplatform can be reconfigured for future ap-plications.Data-storage section includes a MemoryCard interface and an expansion bus, whichcan be connected to a PC through an USBadapter.

3 HardwareField Programmable Gate Arrays aresemiconductors devices containing pro-grammable logic components, memory el-ements and programmable interconnects.

2498


Figure 1: DAQsys board overview. All components are mounted on a small card (100mmx 75mm x 13mm).

Programming the interconnections and thelogic components, is possible to implementsimple logic gates such as AND, OR, XOR,NOT or more complex combinatorial func-tions and sophisticated micro-controllers.Current FPGAs contain up to tens of thou-sands of basic building blocks, with thisnumber increasing rapidly. The main dif-ference between FPGA and CPU is theability to perform parallel computations.CPU performs operations in a strictly se-quential mode and needs several steps tocomplete a task whereas in the FPGA thesame task can be often solved in only onestep building hardware components thatwork in parallel. This last solution in somecases can significantly reduce the execution

time compared to a pure software imple-mentation.All input and output devices are connecteddirectly to the FPGA in a star topology [3]and in order to take full advantage of thisarchitecture, the features are implementedin the form of “system-on-a-chip” withinthis single electronic component. This al-lows you to control all activities in parallelwith response times of a few nanoseconds.

FPGA

The FPGA chip employed is the AlteraCyclone II EP2C5T144C6N [4]. In or-der to implement the needed logic func-tions, the FPGA must be programmed with

2499

Technologies

Figure 2: DAQsys architecture overview.

a Hardware Description Language such as.VHDL, Verilog, etc. .. or with a schematicdesign. The firmware can be developed andtested with software such as Altera QuartusII and Mentor Graphics ModelSim.Many complex logic functions are pro-vided for free by FPGA vendors or areavailable on the market in the form of in-tellectual property (IP core), then the de-velopment of an application can be fast andflexible.Among them all, an interesting IP core isthe so-called soft-processor. It is a micro-processor with reconfigurable architecturethat can be programmed through the high-level language C / C + +. By implement-ing this component inside the FPGA, userscan build complex solutions using appro-priate combinations of hardware and soft-

ware with a simpler approach.Altera’s Embedded portfolio offers thebroadest selection of soft processor coresin the industry. From Nios® II, the FPGAindustry’s 1o soft processor, to popular ar-chitectures from ARM and Freescale.During the testing and debugging, theFPGA is programmed via the standardJTAG interface (boundary-scan test de-vices [5]). When the firmware is completeand ready for use, it is stored in an EEP-ROM and every time you restart the sys-tem, the FPGA loads its configuration fromthe memory and executes it.

Analog to Digital Front End

DAQsys card can handle up to fourdata acquisition chains, each consisting

2500


of a charge amplifier with high inputimpedance (100 M Ω) and programmablegain (6dB, 20dB, 32dB, 40dB and 50dB),a first-order High-Pass filter (cutoff fre-quency: 16 kHz, 5 kHz, 1.6kHz, 160Hz,16Hz) and a low power 1.33MS·s−1 - 16bit ADC converter. The input impedanceof the charge amplifier can be set also to 50Ω via a jumper.The gain factor and the cutoff frequencyof the filter are selected via serial interfacefrom the FPGA.The high impedance, high gain and lownoise front-end makes the system suitablefor interfacing with piezoelectric sensorssuch as hydrophones.The ADC voltage reference is 2.5V and itmatches the full scale voltage (FS).The voltage gain AvdB in dB is:

AvdB = 20 logVout

Vin,

where Vout and Vin are respectively theoutput and the input voltage of the chargeamplifier.

Vout = 10AvdB/20 · Vin;

substituting the values and assuming a gainof 50dB:

Vout ' 316 · Vin

The maximum theoretical sensitivity of thesystem, is given by:

Sx =FS

2N ·Av,

where N is the ADC resolution and Av isthe voltage gain. For a 16 bit resolution weobtain:

Sx =2.5

216 · 316= 120nV.

As we will see in the section of Prelimi-nary Results, we must take into account the

Effective Number of Bits of a system, be-cause this is what gives the true resolution.A useful feature of DAQsys board is theability to synchronize the four ADC with asingle clock or to target a single input to allthe ADC and then use four sampling clocksignals with a mutual phase shift of T / 4,where T is the sampling period.In this way, the analog signal can be over-sampled at up to 5.3 MS·s−1 to signifi-cantly improve signal to noise ratio.

Real Time Clock (RTC)

DAQsys comes with a Dallas Semicon-ductor RTC [6] mounted on board. Thisdevice is a low cost, extremely accurateSPI bus real time clock with an integratedtemperature compensated crystal oscillator(TCXO). The accuracy is ±2 ppm from0oC to 40oC and device maintains seconds,minutes, hours, day, date, month and yearwith leap year compensation valid up to2099.It allows to join precise time informa-tion with collected data or other interest-ing events. DAQsys card is equipped witha secondary connector to power the RTCwith a back-up battery even when the mainpower of the card is turned off, so as tokeep in memory the time data. This inte-grated circuit gives information about tem-perature as a 10-bit code with a resolutionof 0.25oC and temperature accuracy withinthe range of ±3oC, allowing temperaturemonitoring of the system.

I/O peripherals and connectors

The DAQsys board has many pins used forI / O, expansion and programming inter-face and all are linked directly to the FPGA

2501

Technologies

• One 14bit-12MHz DAC digital to ana-log converter with SPI interface [7] and acouple of strobe (out) and trigger (input)available on 50Ω SMB connectors whichare useful to generate analog signals orto synchronize DAQsys with other elec-tronic boards.

• One standard JTAG (Join Test ActionGroup) interface is used for: ProductionTest, Prototype Debug and In-circuit Pro-gramming.

• One 9 pin multimedia connector allowsdata storage into a multimedia card (orSD card) or loading FPGA firmwarefrom it.

• One 50 pin - 32bit and a 12 pin - 4bitexpansion connectors are used to inter-facing with 3.3V TTL compatible exter-nal devices. Currently, USB interface isavailable to connect the DAQsys board toa standard PC through the 50 pin connec-tor.

4 Data Format

DAQsys saves data in RAW format,even though a simple compression algo-rithm can be implemented in the FPGA’sfirmware.RAW data are sequences of 16 bit unsignedwords coming from the ADC converters inthe little-endian format.To take full advantage of the connected PC,we perform a real time compression of theacoustic data using the free FLAC codecsavailable on the web. This application con-vert RAW data into FLAC files or into stan-dard WAV format. Researchers can there-fore use the preferred tools for analysis andelaboration of acoustic data.

FLAC conversion

FLAC means Free Lossless Audio Codec.In short, it is a way of reducing a very largeacoustic file size into something more man-ageable. It is similar to making a ZIP file.An acoustic file can be reduced in size by50% to 60%. As the conversion is lossless,the audio quality is preserved.FLAC stands out as the fastest and mostwidely supported lossless audio codec, andthe only one that at once is non-proprietary,is unencumbered by patents, has an open-source reference implementation, has awell documented format and API, andhas several other independent implementa-tions.FLAC supports tagging, cover art, and fastseeking. FLAC is freely available andsupported on most operating; systems, in-cluding Windows, “unix” (Linux, *BSD,Solaris, OS X, IRIX), BeOS, OS/2, andAmiga.There are many programs and devices thatsupport FLAC, but the core FLAC projecthere maintains the format and providesprograms and libraries for working withFLAC files. Notable features of FLAC:• Lossless: The encoding of audio (PCM)

data incurs no loss of information, andthe decoded audio is bit-for-bit identicalto what went into the encoder.

• Fast: FLAC is asymmetric in favor ofdecode speed. Decoding requires onlyinteger arithmetic, and is much lesscompute-intensive than for most percep-tual codecs. Real-time decode perfor-mance is easily achievable on even mod-est hardware.

• Hardware support: FLAC is supportedby dozens of consumer electronic de-vices, from portable players, to homestereo equipment, to car stereo.

• Seekable: FLAC supports fast sample-

2502


a b

Figure 3: a) DAQsys board, VIA PICO-ITX and hard-disk assembled. b) Bottom viewof DAQsys board with memory card

Condition DAQsys+USB current @5V PC current @12V Total Powerwait-state 330mA 751mA 10.7Wexecution 604mA 800mA 12.6W

Table 1: Total power consumption during wait-state and execution. During executiononly one ADC is active and works at a rate of 750 KS·s−1.

accurate seeking. Not only is this usefulfor playback, it makes FLAC files suit-able for use in editing applications.

• Error resistant: Because of FLAC’sframing, stream errors limit the damageto the frame in which the error occurred,typically a small fraction of a secondworth of data. Contrast this with someother lossless codecs, in which a sin-gle error destroys the remainder of thestream.

A Custom C++ software has been writtento take advantage of FLAC compression al-gorithm before saving data on hard disk.FLAC allows an optimization of the stor-age space without loss of information,soincreasing the maximum acquisition time.

5 Preliminary Results

Some tests have been performed and acomplete data acquisition system has beenassembled.The first experimental set-up consists ofDAQsys connected to a PC via USB cardinterface.We built a simple application in C++ run-ning on the PC, which collects data fromthe DAQsys board and save it on hard disk.The second experimental set-up consists inthe DAQsys alone that saves data directlyto memory card installed on board.In the following measurements, the samplerate was set to 750 kS·s−1. To facilitatethe analysis of acoustic data, the applica-tion reads a continuous stream of data andcreates a file of about 4 MB each time.

2503

Technologies

a

b

Figure 4: a) The complete embedded acquisition system. b) Power consumption dur-ing execution. Lower power is referred to DAQsys + USB interface whereas the Higherpower is refereed to the PICO-ITX PC.

Power Consumption

Power consumption is been measured inwait state and during execution with theAgilent DC Power Analyzer [8]. In waitstate, the DAQsys and the PC are poweredbut no acquisition is made. In execution,data is acquired continuously and savedon hard-disk.Table 1 shows averaged totalpower in the wait-state and with a singlechannel acquisition.Power considerations are important if wewant to make a complete battery powered,

reduced size system for underwater mea-surements. We are currently testing sucha system with several different lithium-ionbatteries.Figure 4 shown the complete system layoutand the power consumption.As we can see, the most part of powerconsumption is due to the PC, the USBinterface and the hard-disk. In order toreduce power consumption, the DAQsysboard must save data directly on the mem-ory card (MMC). In these preliminary tests

2504


Condition DAQsys current @5V MMC max current @3.3V Total Powerwait-state 153mA 150µA 0.76Wexecution 160mA 80mA 1.06W

Table 2: Power consumption of the DAQsys alone. “MMC max current” is the maximumcurrent absorbed by the MMC during read/write cycles.

we have measured power consumption ofDAQsys in two conditions: the first withall the ADC off, the second with a singlechannel running at 750kS·s−1. Table 2summarizes the results.A typical battery pack of small size, has anenergy capacity of 14.5V ·5Ah = 72.5Whin only 700g weight.Using first set-up, the embedded PC canwork uninterruptedly for about 5.7 hours,with the second set-up DAQsys can run un-interrupted for 68 hours.We are investigating other solutions, to fur-ther reduce the power required by the sys-tem, overcoming the typical memory limi-tations of the MMC. One of these possiblesolutions is the “mixed mode operation”.In this mode, just DAQsys board must bealways powered on, and data is saved onthe MMC only during significant acousticevents. DAQsys will turn on the pc as soonas the MMC memory will be almost fullthen the PC will copy the entire contentsof memory on the hard disk, and turns offagain.

Signal to Noise Ratio Measure-ments

In handling real-world signals, analog cir-cuitry sometimes becomes the limiting fac-tor on overall accuracy, for this reason,a dynamic test is needed to show exactlywhat you can expect from your data acqui-

sition system.There are several methods used to take in-formation about accuracy of a system. Wewant estimate the Effective Number of Bit(ENOB) of the ADC and the signal-to-noise-ratio.Shorting to ground a channel of theDAQsys, the value sampled should be the-oretically 0V. What we get instead is aset of points distribuited around a costantvalue. The total noise can be estimated asthe mean of the squared deviation of thesepoints from the mean value.The variance is defined as:

σ2x =

1N

N−1∑i=0

(xi − µ)2,

where N is the number of measurementsand xi is the i-th measurement. We havesetted the amplifier’s gain of the acquisi-tion channel to +6db and the high-pass fre-quency to 5kHz. A LabView application,has been built to find the value of σ2

x. Pre-liminary result on a dataset of about 500points give:

σ2x = 6.3.

The on-board ADC converters have a 16bit resolution (N) and a Full-Scale (FS) of2.5V, so the minimal theoretically voltagestep δV that can be detected, due to thequantization of the analog signal, is:

δV =FS

2N=

2.5V216

= 38µV.

2505

Technologies

GaindB σ2x ENOB SNRdB

6 6.3 14.7 90

50 110 12.6 77.6

Table 3: ENOB (Equivalent Number Of Bit) and SNR (Signal to Noise Ratio) when gainis 6dB, 50dB and input is shorted.

This is also the conversion factor betweendigital samples and voltage. Now we cal-culate the standard deviation as the squareroot of σ2

x and multiply it by the conver-sion factor to obtain the minimum effectivevoltage step ∆V detectable by our systemunder these experimental conditions:

∆V = σx · δV

substituting the values:

∆V = 95µV.

Being ∆V the minimum effective voltagestep, it is related to the effective number ofbit ENOB by:

∆V =FS

2ENOB

and by simple manipulations:

ENOB = log2

( FS∆V

)(1)

ENOB = 14.7bit

The signal-to-noise-ratio can be computedby the following expression:

SNRdB = ENOB · 6.02 + 1.76 (2)

substituting the values:

SNRdB = 90dB.

The previous calculation was repeated with50dB gain setted on amplifier. The results

are summarized in Table 3. As verifica-tion of these results, the short to groundwas removed from the input. A sinusoidalsignal with frequency of 20kHz and am-plitude of 500mVpp, generated by Agi-lent 81150A function generator, has beensent to the DAQsys input channel and ac-quired to compute the Mean Squared Errorof the measured data. The input impedanceof the charge amplifier is matched to theimpedance of the function generator (50Ω).The resolution of the function generator is14bit and this value definitely will affectthe measurement of the ENOB. The gainand the cut-off frequency of the high-passfilter have been setted respectively to +6dBand 5kHz.The Mean Square Error (MSE) for two setof samples xi and yi is defined as:

MSE =1N

N−1∑i=0

(xi − yi)2.

We can use the MSE to compare a set ofsoftware generated samples xi with the ac-quired ones yi. The measurements willbe affected by random noise, but we knowthat the acquired signal is a sinusoid withknown frequency and amplitude, then wecan find a sinusoidal function that best fitwith our experimental data and take theMSE as a noise estimation. If we changeamplitude, offset or phase of the softwaregenerated sinusoid, the value of MSE will

2506


increase or decrease according to these val-ues. For noise estimation we must considerthe minimal value of MSE.Preliminary result on a dataset of about 500points give:

MSE = 44.18.

In analogous way, we calculate the squareroot of MSE (RMSE), to obtain the effec-tive minimal voltage step ∆V in this con-dition:

∆V = RMSE · δV =√MSE · δV.

and substituting the values for RMSE andprevious calculated δV , we obtain:

∆V = 0.25mV.

The ENOB of DAQsys ADC is computedas before and then from (1) we have:

ENOB = 13.27bit.

With function generator that feeds the in-put, the ENOB is smaller than that evalu-ated with input channel shorted to ground.This value reflects in some way, the resolu-tion of the function generator which is just14 bit, so we can’t expect more.Again, using the equation (2) of SNR andENOB we have:

SNRdB = 81.6 dB.

This calculation was obtained manuallychanging parameters of the generated sinu-soidal function to obtain the minimal MSE,so it is only a coarse value for the SNR andit is underestimated.

Time Delay AccuracyTo evaluate accuracy in time delay mea-surement we have arranged the system as

follow:Two analog signals having the same am-plitude and frequency with a precise timedelay between them (“true delay”) are gen-erated by the function generator and feedtwo inputs of DAQsys board. These in-puts are setup with the same gain and samecut-off frequency respectively of 6dB and5kHz. The two signals are sampled syn-chronously and, to avoid ADCs saturation,their amplitude was kept below 600 mV.Two comparators was used to compare thesampled signals to a threshold and gener-ate two trigger events, one to start and an-other to stop a counter, whose count will beproportional to the delay between the twosignals. The clock used to increment thecounter is faster than sampling clock, andit is exactly 16 MHz (fcount). Under theseconditions the number of counts multipliedby the clock period (1/16MHz = 62.5 ns)gives the time delay between the two inputsignals (“measured delay”).The Firmware on the FPGA has been mod-ified to include these comparators and thefast counter.The indetermination on a sample time is re-lated to the sampling period of the AD con-verters. Therefore given a sampling rate fsof 750kS·s−1, the indetermination ∆t can’tbe lower than:

∆t =1fs

=1

750kS · s−1= 1.33µs;

if no other parameters affect the measure-ment, ∆t would also be the measurementindetermination. In real situation severalfactors must be taken into account.Measured values and expected ones areshown in Table 4.For each true delay, we found two differ-ent counter values: a minimum count and amaximum count.The measured delay is ob-tained by:

2507

Technologies

Figure 5: Schematic block diagram illustrating the architecture used to measure timedelay.

measured delay =

=max.count+min.count

2· δt,

where δt is:

δt =1

fcount=

116MHz

= 0.065µs

We can define delta error as:

δ error = (max.count−min.count) · δt

Thus it is evident that the time samplingof ADC is the main indetermination on themeasure. Other noise effects have no rele-vant influence.

6 Application ExamplesMany different applications can be per-formed with the DAQsys board changingonly the firmware on the FPGA and, ifneeded, the software running on the con-nected PC.There are two main examples that empha-sise the DAQsys flexibility. One regard the

acquisition and storage of marine acousticdata and the other concerns passive detec-tion of underwater acoustic sources.

Data Acquisition

Record the sound environment of marinehabitats for extended periods of time, re-quires large amounts of resources for datastorage. For example at a sampling rateof 500 kS·s−1, data flow is 1MB per sec-ond, will therefore be used 3.6GB of stor-age space per channel for each hour ofrecording. Today are commercially avail-able MMC with memory capacity up to 32GB, so we can record up to 10 hours ofcontinuous acoustic data using an MMC.To increase the recording time there are atleast three solutions: the first is to save onlythe interesting parts of a signal and discardthe others, the second is to compress datawith some appropriate algorithm and thethird is to expand the amount of storagespace.The first solution has been investigated byimplementing a discrimination circuit withadjustable threshold into the FPGA. This

2508


true delay µs max. count min. count measured delay µs delta error µs

2 42 21 1.97 1.312.5 42 21 1.97 1.31

5 85 64 4.66 1.3110 171 149 10 1.3850 811 789 50 1.3860 960 959 59.97 0.06

100 1600 1599 99.97 0.06300 4800 4799 299.97 0.06500 8000 7999 499.97 0.06

1000 16000 15999 999.97 0.06

Table 4: The true delay is setted on the function generator. When the true delay is muchgreater than the sampling period, the uncertainty on the measured delay is negligible.

circuit is essentially a real-time 16 bit com-parator, which enables data memorizationonly when the ADC output is greater than aselected threshold. Further improvementscan be achieved by adding a band-pass dig-ital filter in the acquisition chain, so as torecord only the signals of interest. Reducethe number of memory accesses, also helpsto reduce energy consumption.The second solution requires more FPGAresources and an increased developmenttime to be satisfactorily achieved.We also investigated the third solution ex-ploiting an external PC. It was connected tothe DAQsys board through an USB adaptercard. The PC used in our tests was an em-bedded VIA PICO-ITX based on WindowsXP operating system. A custom softwarewas written to acquire data and save onhard disk.To meet the requirements of a real-timesystem for data acquisition and storage itwas necessary to design an interface mod-ule inside the FPGA between the serialADC converters and the USB bus, bothoperating at different clock frequencies.

We compared two architectures, with dif-ferent clock for read and write operations:one based on a FIFO (First In First Out)and the other based on a double-bufferedRAM. Both architectures have allowed tomantain a sustained data streaming on thehard-disk at the maximum sampling rate of5.3MS·s−1.

Passive Detection

Multilateration, also known as hyperbolicpositioning, is the process of locating anobject by accurately computing the timedifference of arrival (TDOA) of a signalemitted from that object to four or more re-ceivers.If a sound pulse is emitted from a source, itwill arrive at slightly different times at twospatially separated receivers, TDOA is dueto the different distances of each receiverfrom the source. Given two receivers po-sitions and a known TDOA, the locus ofpossible emitter positions in a 3D spaceis a two-sheeted hyperboloid. To obtain a

2509

Technologies

Figure 6: Four hydrophones can be connected directly to the DAQsys. Data can be savedon hard disk exploiting an embedded PC.

unique position of the source in a 3D space,we need at least four receivers and calcu-late the hyperboloids intersection.Four hydrophones can be connected di-rectly to the four high impedance chan-nels of the DAQsys board and synchronoussampling the four signals, we can measureTDOA with high precision.We have performed some investigationsabout the ability to get a measure of timedifference between two signals and wehave evaluated the maximun error. Thesetests are reported in the Preliminary Resultssection.

7 ConclusionThe results show that the new DAQsysboard provide an effective combination ofhardware and software for acquisition andelaboration of acoustic signals.Thanks to its small size and low power con-

sumption, DAQsys is very useful in marineunderwater applications where high perfor-mance are required.Preliminary tests have revealed that theDAQsys board is very helpful to devel-oping a long-term recorder of biologicalsounds.The high speed of data acquisition up to1.33 MS·s−1 per channel, suggests the useof DAQsys to study the sounds made bydolphins.High system accuracy is proven by anequivalent number of bits between 12.6 and14.7, that is a remarkable results in marineelectronics.We have achieved also an excellent accu-racy in time delay measurements, valuableparameter to develop instruments for pas-sive acoustic localization.The main characteristic of this new hard-ware is the use of reconfigurable FPGA tomanage all functions of the board instead

2510


of conventional CPU. The system can becustomized by researchers for new futureapplications and experiments, giving life toconcept of Lab On Module, similarly to theconcept of Computer On Module widelyused in the electronics field.

Future WorksWe are working on various projects andmany applications are under developmentto make use of DAQsys easier in the nextfuture.An advanced firmware for TDOA passivelocalization will be developed to allowDolphins localization.Power consumption will be reduced fol-lowing differents strategies to give powersupply only to essential circuitry, disablingthe unused peripherals.For example, when PC is needed to exploitdata-storage space on HDD, its power sup-ply will be switched on from DAQsys onlyduring storage operations.We must study data compression algo-rithms in hardware to improve efficiency indata storage and test the system under realconditions to evaluate the results.

AbbreviationsDAQsys = Digital Acquisition System,with reconfigurable hardware.FPGA = Field Programmable Gate Array.HDL = Hardware Description Language.JTAG = Join Test Action Group. In the1980s, the Joint Test Action Group de-veloped a specification for boundary-scantesting that was later standardized as theIEEE Std.1149.1 specification.ADC = Analog to Digital Converter.DAC = Digital to Analog Converter.RTC = Real Time Clock.MMC = Multimedia Memory Card.FIFO = First In Fist Out. A memory arrayworking like a shift register.TDOA = Time Difference of Arrival.RAW = Data coming from the Analog toDigital Converter without any elaborations.FLAC = Free Lossless Audio Codec. Sim-ilar to ZIP file but optimized for acousticdata.WAV = Standard audio format.PCM = Pulse Code Modulation.API = Application Programming Interface.

References[1] M. Lammers and R. et al. Brainard. An ecological acoustic recorder (EAR) for long-

term monitoring of biological and anthropogenic sounds on coral reefs and othermarine habitats. Acoustical Society of America, 2008.

[2] J.E. Reynolds and S.A. Rommel. Biology of Marine Mammals edited by J.E.Reynolds and S.A. Rommel. Smithsonian Institution Press, Washington, page 578pp., 1999.

[3] Imaging & Microscopy. Scanning Probe Microscope Control. New Electronic Con-cept for Utmost SPM Demands. G.I.T. Imaging & Microscopy, GIT VERLAG GmbH& Co. KG, Darmstadt, Germany, 1:52–55, 2007.

2511

Technologies

[4] Altera. Cyclone II Device Handbook, Volume 1. Altera Corporation(http://www.altera.com), 1, February 2007.

[5] Altera. IEEE 1149.1 (JTAG) Boundary-Scan Testing in Altera Devices. Altera Cor-poration, ver. 6.0 Application Note 39, June 2005.

[6] Dallas Semiconductor. Extremely Accurate SPI Bus RTC with Integrated Crystaland SRAM: DS3234. Dallas Semiconductor - MAXIM integrated Products. Rev. 2,10, 2008.

[7] Analog Devices. 12-/14-Bit High Bandwidth Multiplying DAC with Serial Interface:AD5444/AD5446. Analog Devices, Inc. Rev. A, 2006.

[8] Agilent Technologies. Agilent DC Power Analyzer Models: N6705A, N6715A,N6731B-36B,N6741B-46B, N6751-54A,N6761A-62A, N6773A-76A. AgilentTechnologies, Inc., January 30, 2008.

2512

http://www.altera.com

Phytosterols from Marine Microalgae Dunaliellatertiolecta and D. salina: a Potentially Novel Indus-trial Application

M. Francavilla1, P. Trotta1, R. Luque2

1, Institute of Marine Sciences, CNR, Lesina, Italy2, Department of Organic Chemistry, University of Cordoba, Cordoba, [email protected]

Abstract

Sterols have been extracted and analysed from Dunaliella tertiolecta and Dunaliellasalina, in order to evaluate a potentially novel industrial exploitation of these microal-gae as source of phytosterols. The effect of salt concentration on sterols yields hasbeen studied varying the quantities of NaCl into culture medium. Twelve sterolswere identified by Gas chromatographic MS/MS analysis for both algal strains. Themost abundant phytosterols were (22E,24R)-methylcholesta-5,7,22-trien-3β-ol (er-gosterol) and (22E,24R)-ethylcholesta-5,7,22-trien-3β-ol (7-dehydroporiferasterol).The whole sterol fraction consisted mainly of phytosterols (C28 and C29 sterols).Good yields of total sterols were achieved at lower salt concentration (1.3% and0.89% of dry weight in D. tertiolecta and D. salina, respectively, at 0.6M NaCl),while an increase in salt concentration resulted in a significant decrease in totalsterols yield.The exploitation of D. tertiolecta and D. salina (at 0.6M NaCl) as source of phytos-terols might be an achievable target taking into account that microalgae biotechnol-ogy has reached high levels of development; biomass productivity for microalgae ismuch higher than that of land-based crops; the production of phytosterols could beconsidered as an added value for Dunaliella biomass applying the conceptual modelof biorefinery; and that phytosterols have grown in popularity and market.

1 Introduction

Algal biotechnology has made major ad-vances in the past three decades, andseveral microalgae like Botryococcus,Chlorella, Dunaliella, Haematococcus andSpirulina are cultivated for the produc-tion of proteins, astaxanthin, β-carotene,glycerol, liquid fuels, pharmaceutical for-mulations and also for fine chemicals [1].Among these algae, those of the genusDunaliella, especially D. salina and D. ter-

tiolecta, are microalgae most studied formass culture [2].Dunaliella is a unicellular, bi-flagellate,naked, green alga (Chlorophyta, Chloro-phyceae) which is widely distributed andmay be found in fresh water (e.g. D. flagel-late, D. chordate, D lateralis, D. paupera),in mixed and euryhaline waters (e.g. D. ter-tiolecta, D. bioculata, D. primolecta) andin hypersaline waters (e.g. D. salina, D.minuta, D. parva, D. virdis) [2]. Dunaliellaspp. are grown as a food source in aquacul-


Technologies

ture and D. salina is the richest algal sourceof β-carotene and glycerol [3, 1]. D. salinacan produce β-carotene up to 14% of itsdry weight under conditions of high salin-ity, light and temperature, as well as nutri-ent limitation. Dunaliella β-carotene hasan increasing demand and a wide varietyof market applications: colouring agent infood industries; component in pharmaceu-tics, cosmetics and health foods; dieteticindustries; diagnostics and biomedical re-search.Among chemical constituents of microal-gae, sterols are of increasing interest since:a) the presence of these natural products inmicroalgae determines their food values;b) these compounds are useful biomarkersfor identifying sources of organic matterin sediments [4]; c) and they are not onlyessential components of biomembranes,but function in cell proliferation and sig-nal transduction of microalgae, and othereukaryotic organisms, modulating the ac-tivity of membrane-bound enzymes [5].In contrast with higher plants, algae (in-cluding microalgae) contain a larger di-versity of different sterols [6]. Althoughsome sterols are widespread in many taxaof algae, sterol profiles can be sometimescharacteristic of a particular class, family,genus, or even species of microalgae, andso are often used for chemotaxonomic andphylogenetic comparisons [7, 8].Wright [9, 10] identified in D. terti-olecta by GC-MS and 13C NMR a mix-ture of 24-methyl (C28) and 24-ethyl(C29) ∆7, ∆7,22, ∆8,14, ∆5,7,22 sterols.He found that two major sterols were(22E,24R)-methylcholesta-5,7,22-trien-3β-ol (trivial name ergosterol, C28) anda closely related C29 trienol, (22E,24R)-ethylcholesta-5,7,22-trien-3β-ol (trivialname 7-dehydroporiferasterol), but hedid not quantify them. Patterson et

al. [8] identified in the same microalgatwo rare tetraene sterols, (22E,24R)-methylcholesta-5,7,9(11),22-tetraen-3β-oland (22E,24R)-ethylcholesta-5,7,9(11),22-tetraen-3β-ol but even in that study, sterolswere not quantified. In two other scientificworks [11, 12], authors made reference tosterol composition of D. salina, but theyreported the most abundant sterols only(ergosterol and 7-dehydroporiferasterol)and did not analyse the sterols percentagein relation to dry weight.Nowadays the phytosterols, C28 and C29sterols, are playing a key role in nutraceuticand pharmaceutical industry because theyare precursors of some bioactive molecules(eg.: ergosterol is a precursor of Vita-min D2 and it is also used for the pro-duction of cortisone and hormone flavoneand have some therapeutic applications totreat hypercholesterolemia. In particular,phytosterols have been shown to lower to-tal and LDL cholesterol levels in humansby inhibiting cholesterol absorption fromthe intestine. High serum concentrationsof total or low-density-lipoprotein (LDL)-cholesterol are major risk factors for coro-nary heart disease, a major cause for mor-bidity and mortality in developed countries[13]. It is understandable that efforts aremade to minimize such risks, and reduc-tion of serum cholesterol is a feasible ap-proach, since a risk reduction of coronaryheart disease of about 3% can be achievedthrough a 1% decrease in total choles-terol [14]. In addition to their cholesterollowering properties, phytosterols possessanti-inflammatory and anti-atherogenicityactivity and may possess anti-cancer andanti-oxidative activities [15].Today, consumers are actively seekingproducts containing health-promoting in-gredients to improve their health and well-being. In keeping with this trend, nutraceu-

2514


ticals such as phytosterols, that have a ben-eficial impact on heart health, have grownin popularity.Currently, higher plants are the main indus-trial sources of phytosterols. Phytosterolsisolation in large scale is based in two ma-jor raw materials, vegetable oils and tall oil[13].In this study we analyzed and quantifiedsterols in two species of Dunaliella, D. ter-tiolecta and D. salina (already studied butnot for quantitative sterol analysis). Weevaluated the effect of salt concentrationof medium culture on quali-quantitativecomposition of sterols in order to find opti-mal conditions for major yields of phytos-terols. The aim of this work is to evaluatethe feasibility of using the two strains ofDunaliella as a new commercial source ofphytosterols.The findings of this study could lead to anew industrial exploitation of Dunaliellabiomass for phytosterols production in thelight of a forecast increase of EuropeanPhytosterols market (from $103.9 millionin 2005 to $196.7 million in 2012) [16].


2.1 Biomass preparation

Monoxenic strains of D. tertiolecta andD. salina collected from local algal strainsbank (Fitoteca of ISMAR-CNR Lesina)were grown in 1 L flasks containingWalne’s medium (modified from [17]).Cultures were grown at constant temper-ature (20 ± 0.5°C), under artificial lightwith an intensity of 150 µ E·m−2 · s−1 anda light periodicity of 12/12 light/darkness.Cultures were under air bubbling agitation(1 L min 1) with a CO2 concentration of1% v/v. D. tertiolecta and D. salina were

grown at three different salt concentrations(0.6, 1.4 and 2.1 M NaCl), obtained vary-ing the amount of NaCl dissolved in theWalne’s culture medium.After cultures reached the end of loga-rithmic growth phase (approximately 7-10days), D. tertiolecta and D. salina werecollected in a pyrex glass container andthen centrifuged at 4500 rpm for 10 min-utes. The harvested biomass was washedwith 0.5 M NaCl and distilled water to re-move non-biological material such as min-eral salt precipitates and finally it wasfreeze-dried and stored at -30°C until lipidanalysis.

2.2 Lipid extractionLipids from the cell containing pellets wereextracted according to Bligh and Dyer[18]. The mixture was centrifuged andthe solid residue resuspended in a chlo-roform/methanol mixtur. The combinedsupernatants were cleaned using a satu-rated NaCl water solution and the chloro-form phase was recovered using a sepa-ratory funnel. The chloroform phase wascombined and dried with sodium sulfateovernight. Purified lipids were weighed af-ter the solvent was removed on a rotaryevaporator.

2.3 Saponification of lipids andsterols isolation

Lipids (50 mg under N2) were saponifiedby refluxing in 20 mL of a 5% (w/v) KOHmethanol/water (4:1, v/v) solution for 2 h.Unsaponified was extracted with n-hexaneand then combined, dried with sodium sul-fate overnight, filtered and evaporated.Total sterols from unsaponified materialwere isolated by preparative thin layerchromatography (TLC 20 × 20 cm, silica

2515

Technologies

gel 60 A, layer thickness 500 µm) devel-oped in one dimension in n-hexane /ethylacetate 8:2 (v/v). Sterols bands, whichwere large and single bands, were identi-fied on TLC plate according to the Rf val-ues of standards and visualized with iodinevapour. Such bands were scrapped off andthe silica was eluted with a mixture of chlo-roform/methanol (1:1, v/v). The mixturesolvent/silica was filtered off and the or-ganic phase was evaporated under vacuumand then the purified sterols fraction wasweighed and quantified (mg sterol per g ofdry weight-d.w.-algae).

2.4 Analyses by gaschromatography-mass(tandem) spectrometry(GC-MS/MS)

Purified sterols fraction of D. tertiolectaand D. salina were analyzed by gaschromatography-mass spectrometry. AVarian Saturn 2200 GC/MS/MS ion trap(Varian Analytical Instruments, WalnutCreek, CA) was used. The GC-MS wasoperated in the electron ionization (EI) andchemical ionization (CI) mode over a massrange of 50-650 m/z. The chemical ioniza-tion mode was used to confirm the molec-ular weight (M+1) of sterols. The analysisof sterols was also performed using the tan-dem mass spectrometry (MS/MS), where atarget compound ion is isolated from ma-trix and then fragmented to generate veryunique spectra.Identification of sterols was based on thecomparison of their retention times rela-tive to authentic standards, mass spectraof authentic standards, and available spec-tra in NIST05 and Wiley 07 mass spec-tral libraries. Sterol standards used foridentification include cholesterol (C27 ∆5)

and ergosterol (C28 ∆5,7,22). The iden-tification of fungisterol (C28 ∆7) and 7-dehydroporiferasterol (C29 ∆5,7,22) wasestablished based on mass spectra de-scribed in literature [19, 8, 20].

3 Results

Figure 1 shows values of yields in respectto dry weight of total sterols extracted fromD. tertiolecta and D. salina and purified bymeans of preparative TLC.Yield of total sterols in D. tertiolecta grownin a medium at salt concentration of 0.6MNaCl was 1.13% of dry weight, whereasthe yield in D. salina was slightly lower(0.89% d.w.). In the algae grown at 1.4Mand 2.1M NaCl we found a decrease in to-tal sterols yield. In particular, the sterolsyield in D. tertiolecta decrease to 0.49%d.w. at 1.4M NaCl, and to 0.58% d.w. at2.1M. Total sterols amount in D. salinawas 0.41% d.w. at 1.4M NaCl and 0.28%d.w. at 2.1M NalCl.Trend of sterols yield (Figure 1) in D.salina was almost linear with increasingsalinity, whereas the trend in D. tertiolectawas almost hyperbolic and showed a dras-tic decrease in sterol yield between saltconcentration of 0.6M and 1.4M, and anysignificant variation between 1.4M and2.1M NaCl. Therefore, at 2.1M NalCl,D. tertiolecta had a higher concentrationof total sterols compared to that one in D.salina, whereas at 0.6M and 1.4M NaClconcentrations of total sterols in two al-gae strains were statistically comparable,although the sterol concentration tendedto be higher in D. tertiolecta than in D.salina.Gas chromatography method used in thepresent study separated twelve sterolsfrom D. tertiolecta and D. salina.Sterols that we identified were the

2516


Figure 1: Trend of sterols yield in Dunaliella tertiolecta and D.salina.

same ones in the two algae strains:cholesta-5-en-3β-ol (St1), (22E,24R)-methylcholesta-5,7,9(11),22-tetraen-3β-ol (St2), (22E,24R)-methylcholesta-5,7,22-trien-3β-ol (St3), (22E,24R)-methyl-5α-cholesta-7,22-dien-3β-ol (St4),(24ξ)-methyl-5α-cholesta-8(14)-en-3β-ol (St5), (22E,24R)-ethylcholesta-5,7,9(11),22-tetraen-3β-ol (St6), (24S)-methyl-5α-cholesta-7-en-3β-ol (St7),(22E,24R)-Ethylcholesta-5,7,22-trien-3β-ol (St8), (22E,24R)-ethyl-5α-cholesta-7,22-dien-3β-ol (St9), (24ξ)-ethyl-5α-cholesta-8(14)-en-3β-ol (St10), (24ξ)-ethylcholesta-5,7-dien-3β-ol (St11),(24S)-ethyl-5α-cholesta-7-en-3β-ol(St12). We did not identify stereoisomeryof C-24, and, when possible, we assignedC-24 orientation by means comparisonwith published data.

Figure 2 shows the yield (mg·g−1 d.w.)of twelve detected sterols in D. tertiolecta(Figure 2a) and D. salina (Figure 2b) atthree salt concentrations (0.6, 1.4 and 2.1MNaCl). St8 and St3 were the most abundantsterols in both algae strains, followed bySt7, St9 and St12. These sterols were themost abundant in each of salt concentrationof medium we assayed.Concentration of St3,7,8,12 in D. terti-olecta grown at 0.6M NaCl was higherthan that one we found in same alga grownat 1.4 M NaCl. In particular, the concentra-tion of St8 at 0.6M NaCl was 5.18 mg·g−1

d.w. and decreased to 1.98 mg·g−1 d.w.at 1.4M NaCl. Concentration of St3 de-creased from 2.94 mg·g−1 d.w. to 1.27mg·g−1 d.w., whereas concentrations ofSt7 an St12 decreased from 1.49 and 0.78mg·g−1 d.w. to 0.68 and 0.33 mg·g−1 d.w

2517

Technologies

Figure 2: Yield of the most abundant phytosterols detected in a) Dunaliella tertiolectaand b) D.salina at three salt concentrations.

respectively. There seemed to be no sig-nificant variations in concentration of otherminor sterols with varying salinity.At 2.1M NaCl, concentration of St3,7,8,9in D. tertiolecta slightly increased in re-spect to concentration in same alga strainat 1.4 M NaCl. In fact the concentrationof St8 increased from 1.98 to 2.32 mg·g−1

d.w., and St3 increased from 1.27 to 1.47mg·g−1 d.w., whereas concentrations ofSt7 and St9 increased from 0.54 and 0.27mg·g−1 d.w. to 0.86 and 0.48 mg·g−1 d.w.

respectively. We did not find changes inquantities of minor sterols.St3,7,8,9,12 were also the main sterols inD. salina in each salinity condition we as-sayed. Increase in salt concentration ofmedium from 0.6M to 1.4M NaCl causedin D. salina as well as in D. tertiolecta a de-crease in concentration of main sterols. St8decreased from 3.86 to 1.99 mg·g−1 d.w.,St3 decreased from 2.01 to 1.02 mg·g−1

d.w. and St7 from 1.26 to 0.52 mg·g−1

d.w., whereas concentrations of St9 and

2518


Figure 3: Comparison of the most abundant phytosterols in Dunaliella tertiolecta and D.salina at three salt concentrations: a) 0.6 M NaCl, b) 1.4 M NaCl, c) 2.1 NaCl.

St12 decreased from 0.58 and 0.59 mg·g−1

d.w. to 0.19 and 0.22 mg.g d.w respec-tively. Even in this case, we did not foundquantitative variation in minor sterols.Increase in salt concentration from 1.4Mto 2.1M NaCl caused a further decrease ofsterols amount in D. salina. In particular,at 2.1M NaCl concentrations of two mostabundant sterols (St8 and St3) decreased to1.38 and 0.69 mg·g−1 respectively. Con-centrations of St7,9,12 also decreased to0.29, 0.10 and 0.15 mg·g−1 d.w respec-tively, whereas there were not significantvariations in minor sterol amount.

In order to evaluate the salt effect on yieldof each sterol, we compared sterol com-position of two algae strains at three saltconcentrations (Figure 3). At 0.6M NaCl(Figure 3a), concentrations of St8 and St3were significantly higher in D.teriolectathan in D. salina. In fact, the concentrationof St8 was 5.15 mg·g−1 d.w. in D. terti-olecta and 3.86 mg·g−1 d.w. in D. salina,whereas the concentration of St3 was 2.94and 2.00 mg·g−1 d.w. in D. tertiolecta andD. salina respectively. Concentrations ofother sterols in two strains were statisti-cally not different. At 1.4 M NaCl (Figure

2519

Technologies

3b), concentrations of all twelve sterolsdetected in D. tertiolecta and in D. salinawere comparable, whereas at 2.1 M NaCl(Figure 3b) concentrations of St3,7,8,9,12were higher in D. tertiolecta than in D.salina.Some characteristic mass fragments for thedetected sterols are summarized in Table 1.The most abundant sterol found in all sam-ple (St8) showed a molecular ion (M+) ofm/z 410 in the mass spectrum, while in themass spectrum of ergosterol (St3, the sec-ond in abundance), the molecular ion (M+)occurred at 396 m/z. Principal peaks ofthe mass spectrum of St3 can be explainedby the same fragmentation pattern of St8.The identity of the mass peaks for the twosterols after removal of the side chain (m/z271, 253, 211, 145, 143) confirm that thetwo sterols differs only in their side chains.The mass spectrum of St9 was similar tothat of homologous, St4, with two charac-teristic peaks for ∆7 sterols at m/z 255 and229. St7 showed a molecular ion at m/z400 which was also the base peak and thehomologous C29, St12, showed a similarmass spectrum with fragments which weredifferent in 14 m/z units.

4 Discussion

Dunaliella is a unicellular green alga thatcan adapt to an extremely wide range ofsalinities, from 0.1 to 5.5M NaCl. Varia-tion in the salinity of medium can inducechanges in metabolic pathways that leadto a quali-quantitative variation of metabo-lites in algal cells.Synthesis and accumulation of glycerolin Dunaliella spp. are certainly the mostknow effects induced by increasing salin-ity. The role that glycerol plays in thesalt adaptation of Dunaliella was firmly

established by the studies of Ben-Amotzand Avron (1973). Liska at al. [21], in aproteomic study found that salinity stressup-regulated: key enzymes in the Calvincycle, starch mobilization, and redox en-ergy production; regulatory factors in pro-tein biosynthesis and degradation and ahomolog of a bacterial Na+-redox trans-porters.Effect of salt concentration on the accu-mulation of lipids and triacylglyceride inDunaliella was investigated by Takagy etal. [22] in order to get high liquefactionyield from marine algae cell mass to fueloil. They obtained that the increase ofNaCl concentration from 0.5 to 1.0M re-sulted in higher lipid and triacylglyceridecontent (from 60% and 41% to 67% and57% respectively).Salt concentration can affect also thecarotenoids content in Dunaliella. Gomezet al. [23] working on Dunaliella bardawilfound the highest carotenoids contents percell at extreme salinity and reported an in-crease of 9-cis/ all-trans β-carotene ratiowith increase in salt concentration. Fazeliet al. [24] found that although high salinityfavoured total carotenoid production by D.tertiolecta on a cellular basis, it negativelyaffected on a per volume basis because cellgrowth was repressed at elevated salt con-centrations.Among secondary metabolites, sterols areundoubtedly very important components.They are essential components of the mem-branes of all eukaryotic organisms. Inparticular, the amount of sterols in thestructure of membrane cell and the sterol /phospholipid molar ratio rule the flexibilityand permeability of membrane itself [25].Sterols can function also in cell prolifera-tion and signal transduction of microalgae,and other eukariotic organisms, modulat-ing the activity of membrane-bound en-

2520


Sterol Structure RRT Molecular

weight

MS-data

St1 C27 5 1.00 386 386(57), 371(32), 368(86), 353(83), 301(100),

275(60), 255(50), 213(60), 145(41)

St2 C28 5,7,9,22 1.50 394 394(vis CI), 376(41), 361(12), 251(100),

209(10), 69(14), 55(13)

St3 C28 5,7,22 2.03 396 396(43), 378(18), 363(100), 337(5), 271(9),

253(27), 211 (9), 145(7), 143(6)

St4 C28 7,22 2.19 398 398(7), 380(3), 365(8), 300(16), 271(100),

255(40), 253(22), 229(15) 213(16), 147(20)

St5 C28 8 2.51 400 400(100), 385(71), 367(31), 273(36), 255(26),

213(31), 147(24)

St6 C29 5,7,9(11),22 2.57 408 408 (vis CI), 390(44), 375(9), 347(3), 251(100),

83(10), 69(6), 55(13)

St7 C28 7 3.03 400 400(100), 385(61), 367(20), 273(36), 255(93),

213(37), 147(18).

St8 C29 5,7,22 3.25 410 410(57), 392(24), 351(5), 377(100), 271(9),

253(29), 211(9), 145(7), 143(6)

St9 C29 7,22 3.42 412 412(8), 397(7), 379(4), 369(10), 351(7),

300(12), 273(16), 271(100), 255(32), 253(19),

213(11), 147(15)

St10 C29 8 3.67 414 414(100), 399(65), 381(26), 287(12), 273(31),

255(30), 213(38), 147(24).

St11 C29 5,7 3.94 412 412(40), 397(23), 379(100), 353(32), 271(61),

253(21), 213(14), 145(21)

St12 C29 7 4.21 414 414(100), 399(58), 381(21), 287(9), 273(35),

255(96), 229(22), 213(35), 147(21).

Table 1: Relative retention time (RRT) and some characteristic mass fragments of de-tected sterols.

zymes [5].In our experiments, we purified from D.tertiolecta and D. salina a sterol fraction(total sterols) which consisted of twelvesterols (Figure 2). The sterol profiles oftwo algal strains was the same, accordingto Volkman [7] and Patterson [8] who as-serted that sterol profiles can be sometimescharacteristic of a particular class, family,genus, or even species of microalgae, andso could be used for chemotaxonomic andphylogenetic comparisons.Comparing our results with those reportedby other authors [9, 10, 8], we foundsome differences, however, only in mi-nor sterols (Table 2). In particular, we did

non find (22E,24ξ)-methyl-5α -cholesta-7,9(11),22-tetraen-3β-ol, 24(ξ)-methyl-5α-cholesta-8,14-dien-3β-ol, 24(ξ)-ethyl-5α-cholesta-8,14-dien-3β-ol and 24(ξ)-ethyliden-5α-cholest-7-en-3β-ol detectedby Wright, whereas we confirmed thepresence of two tetraenols (22E,24R)-methylcholesta-5,7,9(11),22-tetraen-3β-oland (22E,24R)-ethylcholesta-5,7,9(11),22-tetraen-3β-ol detected by Patterson et al.[8]. Furthermore, in this study we foundin D. tertiolecta and D. salina traces ofcholesterol and (24ξ )-ethylcholesta-5,7-dien-3β-ol that were not reported in pre-vious studies. The analytical techniqueswe used allow us to do not have great un-

2521

Technologies

Sterol Stucture Our study Wright

(1979, 1981)

Patterson et al.

(1992)

Cholesta-5-en-3β-ol

C27 5 D (St1) N.D. N.D.

(22E,24R)-Methylcholesta-5,7,9(11),22-tetraen-3β-ol

C28 5,7,9,22 D (St2) N.D. D

(22E,24R)-Methylcholesta-5,7,22-trien-3β-ol

C28 5,7,22 D (St3) D D

(22E,24R)-Methyl-5-cholesta-7,22-dien-3β-ol

C28 7,22 D (St4) D D

(24)-Methyl-5-cholesta-8(14)-en-3β-ol

C28 8 D (St5) D N.D.

(22E,24R)-Ethylcholesta-5,7,9(11),22-tetraen-3β-ol

C29 5,7,9(11),22 D (St6) N.D. D

(24S)-Methyl-5-cholesta-7-en-3β-ol

C28 7 D (St7) D D

(22E,24R)-Ethylcholesta-5,7,22-trien-3β-ol

C29 5,7,22 D (St8) D D

(22E,24R)-Ethyl-5-cholesta-7,22-dien-3β-ol

C29 7,22 D (St9) D D

(24)-Ethyl-5-cholesta-8(14)-en-3β-ol

C29 8 D (St10) D N.D.

(24)-Ethylcholesta-5,7-dien-3β-ol

C29 5,7 D (St11) N.D. N.D.

(24S)-Ethyl-5-cholesta-7-en-3β-ol

C29 7 D (St12) D D

(22E,24)-Methyl-5-cholesta-7,9(11),22-tetraen-3β-

ol

C28 7,9,22 N.D. D N.D.

24()-Methyl-5-cholesta-8,14-dien-3β-ol C28 8,14 N.D. D N.D.

24()-Ethyl-5-cholesta-8,14-dien-3β-ol C29 8,14 N.D. D N.D.

24()-Ethyliden-5-cholest-7-en-3β-ol C29 7,28 N.D. D N.D.

D: detected; N.D.: Not detected.

Table 2: Comparison of sterol composition of Dunaliella tertiolecta achieved in this studywith those reported by[9, 10] and [8].

certainty in data achieved. These smalldifferences between our data and those re-ported in previous studies, but betweenpublished data also (e.g., Wright’s resultsare partly different from those of Pattersonet al. [8]), can probably be due to differ-ences in culture conditions [5], but also todifferences in extraction and purificationmethods and in sensitivity of analyticaltechniques which have been used.We observed a significant decrease in to-tal sterol amount in D. tertiolecta and D.salina with the increase of salt concen-tration from 0.6M to 1.4M NaCl (Figure1). At 0.6M NaCl concentration, D. terti-olecta and D. salina showed a similar con-centration of total sterols (not statisticallydifferent). At 1.4M, the concentration oftotal sterols in both strains decreased sig-

nificantly of about 55% in respect to totalsterols concentration at 0.6M NaCl. At2.1M NaCl concentration, the amount oftotal sterols in D. tertiolecta remained (sta-tistically) unchanged in respect to sterolsamount obtained at 1.4M NaCl, whereastotal sterol amount in D. salina further de-creased of 69% and 32% in respect to sterolconcentration at 0.6M and 1.4M NaCl.There are previous studies reporting theeffect of salt concentration on sterol com-position of Dunaliella salina using differ-ent approaches. Firstly, Peeler et al. [11]studied the lipid composition of plasmamembranes of D. salina grown under vary-ing salinities. They reported that sterolswere the major components of membranefractions (accounting for 55% of the to-tal lipid content), comprising of ergosterol

2522


(St3) and 7 dehydroporiferasterol (St8).The sterol/phospholipid molar ratio wasapproximately constant (1.7) at three saltconcentrations (0.85, 1.7 and 3.4 M NaCl).The relative amount of individual sterolsin the plasma membrane fraction changedonly slightly with a change in NaCl con-centration. However, they did not reportthe quali quantitative variation of totalsterols with respect to dry weight.Comparatively, Zelazny et al. [12] reportedthat sterols were essential for volume re-covery, and therefore for glycerol synthe-sis, following hyperosmotic shock. In par-ticular, they observed that the enzymaticinhibition of sterol biosynthesis suppressedthe volume recovery of cells of D. salina,whereas the volume recovery was found tobe fully restored upon addition of exoge-nous sterol. Consequently, they claimedthat a hyperosmotic shock caused an in-crease of sterol content in plasma mem-branes.This data seem to be in disagreement withthat obtained by Peeler et al. [11] whodid not observe any significant variationsof sterol/phospholipid molar ratio and/orcontent of individual sterols in the plasmamembrane at varying salinities. However,Peeler et al. [11] studied cells in osmoticequilibrium conditions (at long times af-ter osmotic shocks). They also reporteda decrease in phospholipid content (from75.7% to 64.5% of polar lipids) at increas-ing salt concentrations (from 0.85M to3.4M). A reduction of phospholipid con-tent implies an intrinsic reduction of sterolcontent in the plasma membrane as thesterol/phospholipid molar ratio was con-stant at the different investigated salt con-centrations.Combining all experimental findings re-ported above, we hypothesize that the sig-nificant decrease in total sterol amount in

D. tertiolecta and D. salina with increas-ing salt concentrations can be explainedby a two-stage process in which the sterolconcentrations are differently affected:1. In a first stage (at time of hyperosmotic

shock), sterols (already present in theplasma membrane and in cytoplasmiclipid vesicles) function as a fast systemto increase the low lipid membrane or-der, also inducing glycerol synthesis (inagreement with [26] and [12]). The in-creased quantities of sterols at low saltconcentration (0.6 M) is consistent withthis function as the cells have to accu-mulate and make available greater quan-tities of sterols (to quickly develop ahigh ordered lipid membrane) to face apossible hyperosmotic shock.

2. In a second stage (with increasing in-tracellular glycerol concentration andupon reaching a new state of os-motic equilibrium), the sterol contentdecreases (in agreement with resultsreported by [11]) and the new sta-tus of membrane lipid order (higherthan that prior to hyperosmotic shock,[26]), is maintained by major concen-trations of diacylglyceryltrimethylho-moserine (DGTS) and sulfoquinovosyl-diacylglycerol (SL) and by a decreasedunsaturation degree of fatty acids (inagreement with reported results by [11].Indeed, at higher salt concentration (1.4M and above), the cells do need to pro-vide only small quantities of sterols tofurther increase the lipid membrane or-der.

St3, St7, St8, St9 and St12 were the mainsterols that changed significantly in con-centration at varying salinities. St7, St9and St12 might be either biosynthetic inter-mediates of St3 and St8 or rapidly turning-over sterols [12]. The different responsesobserved for D. tertiolecta and D. salina at

2523

Technologies

2.1 M salt concentration (in terms of totalsterols amount) may be related to differentadaptation capacities to increasing salin-ity (D. tertiolecta is an euryhaline species,while D. salina is a hypersaline species).Actually, phytosterols isolation in largescale is based in two major raw materi-als: vegetable oils and tall oil [13]. Thetotal contents of phytosterols in vegetableoils are very variable and range from nearly8g · kg−1 in corn oil to 0.5 g · kg−1 in palmoil, with intermediate levels being found incommonly used oils. Another major sourceof phytosterols is tall oil, the fat-solublefraction of the hydrolysate obtained fromtrees during the pulping process. The mostnotable characteristic of tall oil is its highstanol concentration (approximately 20%),and fibre oils such as maize.The exploitation of sterols from microor-ganisms for biotechnological or nutritionalpurposes is in its infancy apart from theergosterol production by fermentation ofSaccharomyces cerevisiae [27] and the useof sterol-containing microalgae as natu-ral feedstocks in aquaculture [5] for crus-taceans and some molluscs that lack denovo sterol-synthesizing ability. Volkman[5] suggested that a limitation in the useof microalgae in phytosterols productionwas the low content of sterols. In di-atoms, sterols represent 0.06-0.57% of celldry weight, whereas in dinoflagellates totalsterols are usually more abundant (e.g. 1.5and 3.1% dry wt). These data are in agree-ment with our experimental data (about 1%dry wt for Dt and Ds at 0.6M NaCl).However, we believe that, even if sterolyields do not seem to be particularly high,it needs to do some remarks: a) the biotech-nology of microalgae has reached high lev-els of development due to new produc-tion technologies which allow to obtainhigh biomass productivities [28] and ma-

jor yields of extracted organic compounds[29]; b) the biomass productivity (in termsof dry metric tons·ha−1 · yr−1) for mi-croalgae is much higher than productivityfor land-based crops [3]; c) the productionof phytosterols could be considered as anadded value for Dunaliella biomass whichcould be cultivated to produce also β-carotene and biofuels as well as to reduceatmospheric CO2 (CO2 biomitigation) ap-plying the conceptual model of biorefinery;d) today, consumers are actively seekingproducts containing health-promoting in-gredients to improve their health and well-being. In keeping with this trend, nutraceu-ticals such as phytosterols, that have a ben-eficial impact on heart health, have grownin popularity.In the light of these remarks, we believethat exploitation of Dunaliella tertiolectaand D. salina (grown at 0.6 M NaCl)as commercial source of phytosterols maybe an achievable hypothesis. The wholesterols fraction extracted from two strainsof Dunaliella could be used in nutraceu-tic and pharmaceutical industry because itconsist mainly of phytosterols which havetherapeutic applications to treat hyperc-holesterolemia.In view of these premises, we believe thatthe exploitation of D. tertiolecta and D.salina (grown at 0.6 M NaCl) as com-mercial source of phytosterols may be anachievable target. The whole sterol fractionextracted from two strains of Dunaliella,comprising of phytosterols involved intherapeutic applications to treat hyperc-holesterolemia, could be used in the nu-traceutical and pharmaceutical industries.Moreover, D. tertiolecta and D. salinacould be potentially utilised as commer-cial sources of ergosterol (St3) and 7-dehydroporiferasterol (St8) as these sterolsreached a relatively high concentration

2524


(0.29% and 0.52% d.w., respectively) in D.tertiolecta grown at salt concentrations of0.6M (Figure 2).Finally, to the best of our knowledge,there are no reported studies on bioactiv-ity of 7 dehydroporiferasterol (St8), themost abundant sterol of D. tertiolecta andD. salina. However, the structural sim-ilarity with ergosterol (with only an ex-tra methyl group present in the molecule)could suggest a bioactivity similar tothis phytosterol. Investigations are cur-rently ongoing in our laboratories to testthe anti-inflammatory bioactivity of 7-dehydroporiferasterol (St8) and we havemanaged to achieve interesting preliminaryresults. If successful, these studies mayalso pave the way to the utilisation and ex-ploitation of Dunaliella species as sourcesof bioactive compounds.

5 ConclusionsWe have reported the isolation and identifi-cation of a range of phytosterols from twomicroalgae species: D. tertiolecta and D.salina. Total phytosterol content in thesealgae was close to 1% dry weight and wasfound to decrease, in general, with increas-ing salinities (from 0.6 to 2.1 M). Thisphenomenon was related to the cell func-tions of these components that are initiallyproduced by the algae at high concentra-tions to build-up a consistent lipid mem-brane (at low salt concentrations) to facehyperosmotic shocks and then produced inlow quantities to support the maintenanceof the membrane structure. In the lightof these remarks, we can conclude thatexploitation of Dunaliella tertiolecta andD. salina (grown at 0.6 M NaCl) as com-mercial source of phytosterols may be anachievable hypothesis.

References[1] R. Raja, S. Hemaiswarya, and R. Rengasamy. Exploitation of Dunaliella for β

carotene production. Appl. Microbiol. Biotechnol., 74:517–523, 2007.

[2] M.A. Borowitzka and L.J. Borowitzka. Microalgal biotechnology. 1998.

[3] G.C. Dismukes, D. Carrieri, N. Bennette, G.M. Ananyev, and M.C Posewitz.Aquatic phototrophs: efficient alternatives to land-based crops for biofuels. Curr.Opin. Biotech., 19:235–240, 2008.

[4] J.K. Volkman, S.M. Barrett, S.I. Blackburn, M.P. Mansour, E.L. Sikes, andF. Gellin. Microalgal biomarkers: a review of recent research developments. Org.Geochem., 29:1163–1179, 1998.

[5] J.K. Volkman. Sterols in microorganisms. Appl. Microbiol. Biotechnol., 60:495–506, 2003.

[6] L.P. Ponomarenko, I.V. Stonik, N.A. Aizdaicher, T.Y. Orlova, G.I. Popovskaya,G.V. Pomazkina, and V.A. Stonik. Sterols of marine microalgae Pyramimonas cf.Cordata (Prasinophyta), Attheya ussurensis sp. nov. (Bacillariophyta) and a spring

2525

Technologies

diatom bloom from Lake Baikal. Comp. Biochem. Physiol., 138(Part B):65–70,2004.

[7] J.K. Volkman. A review of sterol markers for marine and terrigenous organic matter.Org. Geochem., 9:83–99, 1986.

[8] G.W. Patterson. Sterols of algae. page 118–157, 1991.

[9] J.L.C. Wright. The occurrence of ergosterol and (22E,24R)-24-ethylcholesta-5,7,22-trien-3β-ol in the unicellular chlorophye Dunaliella tertiolecta. Can. J.Chem., 57:2569–2571, 1979.

[10] J.L.C. Wright. Minor and trace sterols of Dunaliella tertiolecta. Phytochemistry,20:2403–2405, 1981.

[11] T.C. Peeler, M.B. Stephenson, K.J. Einspahr, and G.A.Jr. Thompson. Lipid Char-acterization of an Enriched Plasma Membrane Fraction of Dunaliella salina Grownin Media of Varying Salinity. Plant Physiol., 89:970–976, 1989.

[12] A.M. Zelazny, A. Shaish, and U. Pick. Plasma-membrane sterols are essentialfor sensing osmotic changes in the halotolerant alga Dunaliella. Plant Physiol.,109:1395 1403, 1995.

[13] P. Fernandes and J.M.S. Cabral. Phytosterols: Applications and recovery. Biore-source Technol., 98:2335–2350, 2007.

[14] S.M. Mel’nikov, J.W.M.S. ten Hoorn, and B. Bertrand. Can cholesterol absorptionbe reduced by phytosterols and phytostanols via a cocrystallization mechanism?Phys. Lipids, 127:15–33, 2004.

[15] D. Platt, D. Pelled, and A. Shulman. Oils enriched with diacylglycerols and phy-tosterolesters for use in the reduction of cholesterol and tryclycerides. 2004.

[16] Frost and Sullivan. Strategic Analysis of the U.S. Phytosterols Market June 6, 2006,MC1397684. 2006.

[17] I. Laing. Cultivation of marine unicellular algae. MAFF Laboratory. Leaflet No.67. 1991.

[18] E.G. Bligh and W.J. Dyer. A rapid method of total lipid extraction and purification.Can. J. Biochem. Phys., 37:911–917, 1959.

[19] M.A. Gealt, J.H. Adler, and W.R. Nes. The sterols and fatty acids from purifiedXagella of Chlamydomonas reinhardi. Lipids, 16:133–136, 1981.

[20] F.L.E. Chu, E.D. Lund, P.R. Littreal, K.E. Ruck, E. Harvey, J.R. Le Coz, Y. Marty,J. Moal, and P. Soudant. Sterol production and phytosterols bioconversion in twospecies of heterotrophic protists, Oxyrrhis marina and Gyrodinium dominans. Mar.Biol., 156:155–169, 2008.

2526


[21] A.J. Liska, A. Shevhenko, U. Pick, and A. Katz. Enhanced photosynthesis andredox energy production contribute to salinity tolerance in Dunaliella as revealedby homology-based proteomics. Plant Physiol., 136:2806–2817, 2004.

[22] M. Takagi, K. Yoshida, and T. Yoshida. Effect of salt concentration on intracellularaccumulation of lipids and triacylglyceride in marine microalgae Dunaliella cells.J. Biosci. Bioeng., 101:223–226, 2006.

[23] P.I. Gomez, A. Barriga, A.S. Cifuentes, and M.A. Gonzalez. Effect of salinityon the quantity and quality of carotenoids accumulated by Dunaliella salina (strainCONC-007) and Dunaliella bardawil (strain ATCC 30861) Chlorophyta. Biol. Res.,36(2):185–192, 2003.

[24] M.R. Fazeli, H. Tofighi, N. Samadi, and H. Jamalifar. Effects of salinity on βcarotene production by Dunaliella tertiolecta DCCBC26 isolated from the Urmiasalt lake, north of Iran. Bioresource Technology, 97:2453–2456, 2006.

[25] D. Chapman. Recent studies of lipid, lipid cholesterol and membrane system. Bio-logical Membranes, 2:91–122., 1973.

[26] C.C. Curtain, D. Looney, D.L. Reran, and N.M. Ivancic. Changes in the orderingof lipids in response to osmotic pressure changes. Biochem. J., 213:131–138, 1983.

[27] X. He, X. Guo, N. Liu, and B. Zhang. Ergosterol production from molasses by ge-netically modified Saccharomyces cerevisiae. Appl. Microbiol. Biotechnol., 75:5560, 2007.

[28] Y. Chisti. Biodiesel from microalgae. Biotechnol. Adv., 25:294–306, 2007.

[29] E. Molina Grima, E.H. Belarbi, F.G. Acien Fernandez, A. Robles Medina, andY. Chisti. Recovery of microalgal biomass and metabolites: process options andeconomics. Biotechnol. Adv., 20:491–515, 2003.

2527

Technologies

2528

Unmanned Surface Vehicles for Coastal and Har-bour Monitoring: the CNR Experience

M. Caccia, M. Bibuli, G. BruzzoneInstitute of Intelligent Systems for Automation, CNR, Genova, [email protected]

Abstract

In the last decade Unmanned Surface Vehicles (USVs) emerged as a potentiallygame-changing technology for naval operations as well as coastal and harbour mon-itoring and surveillance. In particular, the position of the USVs at the air-sea inter-face allows them to play a key role both in relaying radio frequency transmissionsin air and acoustic transmissions undersea, and in monitoring ocean and atmospheredynamics as well as surface and underwater intrusions. As a consequence of theirnetworking capabilities, USVs are naturally seen as a part of flotillas of heteroge-neous vehicles executing large scale surveys and supporting rapid environmentalassessment: the result is that an increasing number of prototype vehicles has beendeveloped for science, bathymetric mapping, defence and generic robotics research.This paper presents an overview of USV development and applications, discussingthe main technological and legal issues related to their use in everyday operations.In this context, attention focuses on the CNR experience in developing vehicles forcoastal and harbor applications. The design, development and exploitation of theCharlie and ALANIS USVs is discussed. In particular, general aspects either infras-tructural, e.g. software and hardware embedded real-time architecture and commu-nication systems, or methodological, e.g. navigation, guidance, control and missionsupervision systems, are clearly distinguished by vehicle-related mechanical designand applications.

1 Introduction

In the last decade Unmanned Surface Ve-hicles emerged as a potentially game-changing technology for naval operations[1] as well as for coastal and harbour mon-itoring and surveillance. After the pioneerdevelopment of prototype autonomous sur-face craft basically used for automated col-lection of bathymetries at the MIT SeaGrant College Program [2] around the midof the Nineties, at the beginning of the newMillennium the development of USV tech-nology was dramatically boosted by the

successful operation of the shallow waterinfluence mine sweeping system (SWIMS)to support the British Royal Navy minecountermeasures operations in Iraq in 2003[1]. In particular, the position of the USVsat the air-sea interface allows them to playa key role both in relaying radio frequencytransmissions in air and acoustic transmis-sions undersea, and in monitoring oceanand atmosphere dynamics as well as sur-face and underwater intrusions. As a con-sequence of their networking capabilities,USVs are naturally seen as a part of flotillasof heterogeneous vehicles executing large


Technologies

scale surveys and supporting rapid environ-mental assessment (REA): the result is thatan increasing number of prototype vehi-cles was developed for science, bathymet-ric mapping, defence and general roboticsresearch.In the following an overview of USV de-velopment and applications is presenteddiscussing the main technological and le-gal issues related to their use in everydayoperations. In this context, the CNR ex-perience in developing vehicles for coastaland harbour applications is discussed fo-cusing attention on the design, develop-ment and exploitation of Charlie and ALA-NIS USVs. In particular, general infras-tructures, e.g. software and hardware em-bedded real-time architecture and commu-nication systems, and methodologies, e.g.navigation, guidance, control and missionsupervision systems, will be clearly distin-guished by vehicle-related mechanical de-sign and applications. Moreover, the mostsignificant exploitations of USV technol-ogy will be discussed.

2 Background

2.1 Prototype USVs and appli-cations

As mentioned before, from the Ninetiesof the last century the MIT Sea GrantCollege Program developed a family ofautonomous vessels for education andcivil applications, consisting of the fishingtrawler-like vehicle ARTEMIS, the cata-marans ACES (Autonomous Coastal Ex-ploration System) and AutoCat [3, 2] ,and the kayak SCOUT (Surface Craft forOceanographic and Undersea Testing) [4].These USVs demonstrated the feasibilityof automatic heading control and DGPS-

based way-point navigation, as well as thepossibility of operating autonomously col-lecting hydrographic data. After the inte-gration with the MIT Odyssey class AUVs,from the point of view of human opera-tor interface, mission planning and com-puter architecture [5], the SCOUT kayaksare supporting the development and test ofdistributed acoustic navigation algorithmsfor undersea vehicles.As far as the European research commu-nity is concerned, the concept of USV wasintroduced by the University of Rostock(Germany) with the design and develop-ment of the Measuring Dolphin, devoted tohigh accuracy positioning and track guid-ance and carrying of measuring devices(e.g. depth and current) in shallow water[6]. The idea of a USV acting as commu-nication relay with a companion AUV wasexploited in the European Union fundedproject ASIMOV (Advanced System Inte-gration for Managing the coordinated op-eration of robotic Ocean Vehicles) with thedevelopment of the autonomous catamaranDelfim by the DSOR lab of Lisbon IST-ISR[7].In the wake of these pioneer prototypes,two autonomous catamarans, ROAZ andROAZ II, were developed by the Au-tonomous Systems Laboratory at ISEP, In-stitute of Engineering of Porto, for envi-ronmental monitoring, bathymetry, sciencedata gathering, search and rescue supportand security missions [8]. In the mean-time, the autonomous catamaran Springerhas been developed by the University ofPlymouth (UK) for tracing pollutants [9].The development of the autonomous cata-maran Charlie, originally exploited forthe collection of sea surface microlayer[10], and then upgraded for robotic re-search on autonomous vessels, and of thealuminum Rigid Inflatable Boat-like USV

2530


ALANIS for coastal monitoring by CNR-ISSIA Genova (Italy) will be deeply dis-cussed in the following sections.On the other hand, after the success ofSWIMS, which basically consisted in thedevelopment of a conversion kit to trans-form existing Combat Support Boats, al-ready operated by the British Royal Navy,in remote controlled vessels, numerousUSVs were developed for military pur-poses. Examples are given by the SSC SanDiego, based on the Bombardier SeaDooChallenger 2000 [11], the Israeli Stingraywith a top speed up to 40 knots, and Pro-tector, equipped with electro-optic sensors,radar, GPS, inertial navigation system anda stabilised 12.7 mm machine gun, and theFrench Inspector basically developed forcoastal and port security and mine counter-measures.

2.2 Legal issuesThe main limitations to an extended use ofUSV technology for civil applications, i.e.in areas not restricted to maritime traffic,rely in the lack of rules for the operationsof autonomous vehicles at sea and of a re-liable methodology for anti-collision. Asfar as legal issues are concerned the readercan refer, for instance, to the documentof the Society for Underwater Technology“Issues Concerning the Rules for the Op-eration for Autonomous Marine Vehicles(AMV) - A consultation paper (7 August2006)”, while the first basic steps in thedirection of implementing collision avoid-ance strategies according to the rules of theroad were presented in [12]. Preliminarywork on automatic obstacle detection is be-ing carried out in the military field, e.g. in-tegrating a radar and artificial vision de-vices on the SSC San Diego [11]. Laser-gated intensified CCD (LGICCD) could

represent a dramatic improvement in obsta-cle detection at sea, at least during nightoperations, thanks to the property of theauxiliary laser light source of being com-pletely absorbed by the water. In addi-tion, recently ASTM International has es-tablished the subcommittee F41.90.01 Un-manned Surface Vehicles for addressingstandardisation issues related to this tech-nology.

3 USV Projects @ CNRSince 2003 a couple of prototype USVshave been designed and developed byCNR-ISSIA for operations in coastal andharbour areas: Charlie [13], a fibreglasscatamaran with electrical propulsion, andALANIS (ALuminum Autonomous Nav-igator for Intelligent Sampling) [14], analuminum rubber dinghy like craft with afuel powered outboard motor, funded bythe Parco Scientifico e Tecnologico dellaLiguria scpa. The generic system consistsof a robotic vessel, with on-board com-puting resources for navigation, guidanceand control, and a portable human operatorstation for remote control and supervision.Both Charlie and ALANIS are autonomousfrom the point of view of power supply.The remarkable differences between thevessels, presented in the following, fromthe naval-mechanical, steering, propulsion,and degree of autonomy points of viewallowed the acquisition of a wide spec-trum experience in the development andexploitation of this class of vehicles.

3.1 Charlie USVThe Charlie USV, as shown in Figure 1,has a catamaran-like shape, chosen forstability with respect to roll, redundancy

2531

Technologies

Figure 1: View of the Charlie USV during trials in the Genova Pra harbour (April 2009)

in hull buoyancy, and capability of pay-load transport with respect to the hydrody-namic drag. The vehicle length of 2.40 mand width of 1.80 m were determined bytransportation constraints on support ves-sels and lab trucks. The hull height ofabout 0.60 m, for a weight of about 300Kg in air, gives the possibility of carrying apayload of about 120 Kg.The vehicle is equipped with two pro-pellers, each actuated by a DC motor (330W at 48 V) with corresponding tachome-ter and rotary joint. A set of servo-amplifiers perform proportional-integral-differential (PID) control of the thruster ve-locity on the basis of the error between thesignal of the motor tachometer and the ref-erence speed. After the 2003-04 Antarc-tic campaign, the steering system, origi-nally based on differential propeller rev-olution rate for working basically at verylow speed, was integrated by two rigidly

connected rudders mounted behind the pro-pellers. The rudders are actuated by abrushless motor driven by a standard mo-tion controller able to handle out-of-rangesignals. Magnetic and mechanical switchesgenerate suitable signals in order to keepthe rudder angle module lower than 45 de-grees and to automatically locate the zeroposition when the system is turned on.Power is supplied by a serial package offour lead batteries (42 Ah at 12 V eachone), integrated with a set of four flexiblesolar panels (32 W at 12V).

3.2 ALANIS USV

The ALANIS USV, see Figure 2, based ona commercial boat Mancini Multiprofes-sional 450, has a stern hole with a diame-ter of 0.20 m for deploying and recoveringof scientific instrumentation by mean of awinch. The aluminum hull, 4.50 m long,

2532


Figure 2: View of the ALANIS USV during preliminary trials in the Genova Pra harbour(May 2008)

2.20 m wide and weighting 600 Kg for aload capacity of 800 Kg, has a number ofseparate watertight compartments. A cou-ple of traverse rods are welded on the floorto mount the pilot console, the battery box,the winch and any generic device for futureneeds. The stern hole can be closed with aplug which, when inserted, smoothly con-tinues the hull profile.The propulsion is guaranteed by a 40 HPoutboard motor, which allows human oper-ators lacking in nautical license to pilot thevessel. Steering is performed by rotatingthe motor with respect to its vertical axis.In order to allow a dual, i.e. mannedand unmanned, use of the vessel, a man-ually (dis)connectible electro-mechanicalsystem for servo-actuating steering andthrottle has been designed and developed.A couple of reduction gears connected tothe axis of the steer and throttle respec-

tively are controlled by two servo brushlessmotors. The throttle and steering automa-tion mechanisms can be fast electrically(dis)engaged through a main switch, and,mechanically, acting on a couple of knobs,which (dis)engage the automation gear-boxes (un)screwing the respective threadedbar. The problem of signalling the reach-ing of safety limits, as well as of fixing anabsolute known position during initializa-tion, was solved by introducing a couple ofmicro switches at two ways for each sub-system.The automatic deployment and recoverysystem for scientific instrumentation isbased on a winch actuated by a servobrushless motor piloted by the vehicle con-trol system. The cable, which is 200m long, contains 5 wires for power, 2twisted couples for data, and 1 twistedcouple for video transmission in order to

2533

Technologies

be able to support different payloads. Amicro switch is used to signal the com-plete recovery of the scientific device. Theamount of displaced cable is computed onthe base of signals generated by two mag-netic switches positioned on a pulley.The fuel tank for the outboard motor has acapacity of 65 l, for a vehicle autonomy ofabout 12 hours. It is in aluminium too andembedded in the hull below the floor.A set of 4 lead batteries (42 Ah at 12 Veach one) supplies the control system elec-tronics and sensors.

4 Embedded real-timeplatform

The basic requirement in designing andimplementing the robotic vessel softwarearchitecture was the definition of a com-mon infrastructure for managing an em-bedded system supporting real-time threadscheduling and communications, and thelevels of an intelligent control architecture,including asynchronous modules imple-menting potentially unbounded search al-gorithms.Advances in computing power ofcommercial-off-the-shelf boards and inperformance of free software allowed thedevelopment of a platform based on Sin-gle Board Computers, PC-104 modulesand GNU/Linux. In particular, researchfocused on developing an embedded real-time infrastructure for industrial automa-tion and robotics, constituted by a real-time custom scheduler, which guaranteesan effective scheduling of the control ap-plication threads according to the requiredpriorities and policies, integrated with a setof utilities for thread creation and data ac-cess synchronization and communication.

A detailed description of the developedmethodology can be found in (Bruzzone,2008). Here it is sufficient to remindthat, from the kernel release 2.5.4-pre6,a native fair Unix-like system such asGNU/Linux can be used for real-time ap-plications with strict time requirementsthanks to the so-called pre- emption andlow-latency patches. They guarantee max-imum latency and jitter of the order of afew tens of milliseconds also when par-ticularly demanding activities, which areusually avoided by typical real-time em-bedded system programming techniques,are performed (e.g. video output, dynamicallocation of large amount of memory, pro-cess fork, etc.). Thus it was possible toimplement a real-time custom schedulerto manage real-time threads, both syn-chronous and asynchronous, while nonreal-time ones continue to be scheduledunder the default universal time sharingLinux scheduler policy. As a result theproposed solution allows the scheduling ofall the threads in user space without requir-ing any complex development of kernelmodules.

5 Navigation package

The basic navigation package of CNR-ISSIA USVs is constituted by compass andGPS for the measure of the vessel orienta-tion and position. Pitch and roll measure-ments are provided by clinometers, whilewind conditions are detected by an ultra-sound anemometer.In addition to the basic Charlie system con-stituted by a GPS Ashtech GG24C provid-ing geographic coordinates at 1 Hz, inte-grated with a KVH Azimuth Gyrotrac, ableto compute the True North at 2 Hz on thebasis of the measured Magnetic North and

2534


the GPS-supplied data, devices of differ-ent classes were tested and evaluated. Re-sults demonstrated that standard commer-cial GPS are sufficient for basic line-of-sight guidance through way-points, simplyheading the vessel bow towards the currenttarget. In the case an accurate localisationof the vehicle is required, for instance forcoordinated motion control or bathymet-ric surveys, high positioning GPS systemsexperimentally demonstrated satisfactorilyperformance without requiring any struc-turing of the operating environment. Onthe other hand, due to the relatively fastdynamics of the yaw motion of the con-sidered vessels, high sampling rate head-ing signals, and thus relatively high qualitycompasses, are required for automatic con-trol of the vehicle orientation.

6 Navigation, guidanceand control architecture

6.1 Control architectureA classical hierarchical intelligent con-trol architecture upgraded with a Petri netbased execution control level, as an inter-face between the synchronous execution ofmotion estimation and control modules andthe asynchronous discrete event systemshandling mission control and supervision,was designed and implemented (see Figure3).Drivers, located at the lowest layer, aresoftware modules in charge of handlingInput/Output operations on actuation (e.g.thrusters and rudder) and sensing devices(e.g. GPS and compass).The Execution Level is in charge of thesynchronous execution of a set of ele-mentary navigation, guidance and controltasks communicating among them through

shared variables. It communicates withthe Drivers, sending reference values to ac-tuation devices and receiving instrumentalmeasurements by sensors.The Execution Controller handles the ac-tivation/deactivation of navigation, guid-ance and control tasks, solving conflictsand dependencies on task resources, typi-cally variables, and thus guaranteeing thesystem consistency. As discussed in [13],this capability is based on the definitionand on-line reconfiguration of a controlledPetri net describing the I/O relationshipsof execution level tasks. Differently fromthe Execution Level, which acts at syn-chronous continuous-time layer, the asyn-chronous evolution of the Execution Con-troller is event-driven, i.e. it changes itsstate when receiving a command from an-other architecture component or from thehuman operator.On the other hand, the NGC Monitor inresponsible of the generation of events re-lated to the semantics of the tasks exe-cution, signaling particular interactions ofthe robot with the operating environmentand the state of motion estimation andcontrol tasks. On this basis, the missioncontrol level (or human pilot) dynamicallyschedules the motion control and estima-tion tasks in order to accomplish the de-sired mission.In addition, the Path Planner is in chargeof online computing and handling a col-lision free path to a desired location withadditional constraints on way-points, localpath direction and so on. On this basis itsupplies the guidance modules with the co-ordinates, orientation, curvature of the ac-tual point of the path that has to be tracked.Currently, only functions computing basicpaths such as straight lines, splines for asequence of way-points and circumferencearcs have been implemented. The imple-

2535

Technologies

Figure 3: Intelligent control architecture scheme.

mentation of an on-line path generation,on the basis of position data received froma moving master vehicle, is a fundamen-tal tool for the execution of coordinatedand cooperative control of multiple vehi-cles as, for instance, in the case of vehicle-following.The overall control system architecture ismanaged and supervised by the MissionController, that has not to be consideredlike a raw mission plan executor, but morelike a coordinator between predefined auto-matic operations and unpredictable user in-terventions which can take place wheneverduring mission execution. For this reason,user presence is considered “in the loop” asan intelligent sub-system that interacts withthe mission evolution, usually interruptingautomatic routines to switch to manual op-erations and giving control back to the au-tomatic system. The Mission Controllercollects all the asynchronous discrete-time

event that are generated from the othermodules of the architecture, on the basis ofsuch events reception the state of the mis-sion is updated.

6.2 Motion estimation

As far as motion estimation is con-cerned, conventional Kalman filter tech-niques were applied, providing good re-sults in terms of precision and smoothnessof linear and angular position and speedestimates. In particular, experimental re-sults with the Charlie USV demonstratedthat the virtual acceleration measurementsprovided by an accurate model of the ve-hicle dynamics allow a smooth estimate,and not particularly delayed, of the vehiclespeeds, which can be used by motion con-trollers with relatively high gains.Anyway, it is worth noting some anoma-lies verified in the GPS measurements

2536


Figure 4: Charlie USV GPS raw and filtered data, Genova Pra harbour (December 2006)

as well as the adopted countermeasures.The Ashtech GG24C provides piecewisecontinuous latitude and longitude signals,presenting random-like steps of the typeshown in Figure 4, where the blue and redlines represent the raw and filtered GPS xand y measurements respectively. Except-ing the discontinuity points, the positionsignal is quite smooth with a standard de-viation of the noise of about 0.17 m. Theseresults seem to confirm the fact that al-though the artificial degradation of the sig-nal through the process of selective avail-ability was removed in 2000, the accuracyof GPS, when used in instantaneous stand-alone mode, remains of the order of 10-20m. Thus, the Kalman filter for linear mo-tion estimation was upgraded with an on-

line detector of signal discontinuities. Re-sults are shown by the red path of Figure4.

6.3 Guidance and control

The operating experience with the Char-lie and ALANIS USVs demonstrated thatmany practical applications do not requirethe design and implementation of advancednonlinear controllers. For instance, ba-sic line-of-sight guidance and PD headingcontrol were sufficient for controlling theCharlie motion when collecting water sam-ples from the sea surface micro-layer andimmediate sub-surface during the XIX Ital-ian Expedition to Antarctica in 2004 (seesection 8.1).

2537

Technologies

Figure 5: Charlie USV path-following performance with and without speed adaptation(red and blue line respectively), Genova Pra harbour (December 2008). The target pathis represented with a green line.

Guidance and control research focused onthe design and implementation of high pre-cision autopilots (heading and speed con-trollers) and path-following controllers. Inthis context, a dual nested loop architec-ture decoupling the management of dy-namics and kinematics, i.e. speed and po-sition control respectively, proved its effec-tiveness in the case of the Charlie USV,even using only linear and angular posi-tion sensors, i.e. GPS and compass. Thisstructure facilitates the design of guidancetask functions at the kinematic level, i.e.position controllers generating reference

speed signals, implementing different vehi-cle manoeuvres from basic auto-heading tostraight line and path following. Thanks tothe physical properties of the system, sim-ple position controllers can be defined forexecuting more complex tasks too. This is,for instance, the case of straight line fol-lowing, where sea current and wind distur-bances are naturally compensated by thepresence of a physical integrator betweenthe vehicle yaw rate and the angle of attackbetween the desired line orientation and thevehicle heading.As far as generic path-following is con-

2538


Figure 6: View of the Charlie USV on the deck of the support vessel Malippo, TerraNova Bay, Ross Sea, Antarctica (February 2004).

cerned, the effectiveness of virtual targetbased techniques, developed in coopera-tion with CNRS-LIRMM in the frameworkof an international bilateral research agree-ment, was experimentally proven. The ba-sic idea consists in moving a virtual tar-get on the desired path in such a way asto guarantee the convergence of the vehi-cle, steering according to a suitable algo-rithm, to the target itself [15]. An exampleof the accuracy of the proposed algorithmin tracking a generic path with the CharlieUSV is given in Figure 5, where results ob-tained both at fixed speed and controllingthe vehicle velocity according to the pathshape are presented.On the other hand, automated auto-tuning procedures, based on induced self-oscillations of the system to be controlled,were transferred from the field of automa-tion industry to basic USV guidance and

control, being easy to use and not requir-ing expert knowledge from the USV op-erator. These procedures, developed andtested with the Charlie USV in coopera-tion with the Laboratory for UnderwaterSystems and Technologies of the Univer-sity of Zagreb, Croatia, demonstrated theirfeasibility in designing, implementing andtuning not only auto-heading controllers,but also a straight line-following guidancemodule able to generate suitable headingreference for a generic autopilot [16].

7 Communications andpilot interface

Communications with the remote controland supervision station are guaranteed by aradio wireless LAN at 1.9 Mbps, support-ing robot telemetry, operator commands,

2539

Technologies

Figure 7: View of the Charlie USV sampling sea surface micro-layer, Terra Nova Bay,Ross Sea, Antarctica (February 2004).

and video image transmission. Anyway,due to the poor performances, mainly interms of reliability, offered by commercial,relatively low cost, wireless line-of-sightlinks, the communication system was up-graded with a radio modem working at 169MHz, guaranteeing a safe transfer of com-mands and basic telemetry. In the case ofcooperative control of the two USVs, com-munications between the vessels are sup-ported by a low bandwidth radio modem,working at 436 MHz.The human operator station is constitutedby a laptop computer, running a HumanComputer Interface, implemented origi-nally in C++ and then in Java, and thepower supply system, which integrates acouple of solar panels (32 W at 12V) andone lead battery (100 Ah at 12 V) thusguaranteeing its full autonomy and porta-bility.

8 Exploitation

Among the large spectrum of possible ap-plications of USV technology, the activitiesof research and exploitation of the Charlieand ALANIS prototype vehicles focusedon the study of the sea-air interface, har-bour protection and REA.The possibility of integrating USVs incoastal and harbour underwater anti-intrusion systems was investigated in aproject funded by PRAI-FESR RegioneLiguria in the period 2005-2007, wherebasic NGC techniques presented in sec-tions 6.2 and 6.3 were developed, andwill be exploited in the Industria 2015project ”SLIMPORT - Integrated manage-ment system for logistics and safety of portintermodality”. In the following the ex-ploitation of the Charlie USV for the studyof the sea-air interface and current research

2540


Figure 8: View of the Charlie USV following ALANIS in the Genova Pra harbour (July2009)

in developing basic tools for the support ofREA operations are presented.

8.1 Study of the sea-air inter-face

The first version of the Charlie USV wasbasically designed to satisfy the opera-tional requirement of sampling the seasurface micro-layer and immediate sub-surface water for the study of sea-air in-teractions in the Ross Sea, Antarctica,in cooperation with the CNR Institutefor the Dynamics of Environmental Sys-tems, Venice. In the framework of theSEa Surface Autonomous Modular unit(SESAMO) project, funded by the ItalianNational Program of Research in Antarc-tica, the vessel was equipped with a scien-tific payload consisting of a rotating glassdrum for micro-layer collection, a seawa-

ter intake for immediate sub-surface wa-ter sampling, and a number of teflon mem-brane pumps and three-way valves in or-der to distribute the sampled water in suit-able stocking buckets for organic and in-organic samples (see Figure 6). The sys-tem could stock up to 20 l of organic sam-ples and 10 l of inorganic ones both formicro-layer and immediate sub-surface, foran overall load of about 60 Kg. In order tomaintain about constant the vessel weightduring sampling, a couple of ballast wa-ter tanks were positioned in the hulls to beemptied in the course of the mission. Therotating glass drum was 0.33 m in diame-ter and 0.50 m in length and, actioned byan electrical brushless thruster, could ro-tate between 4 and 10 rpm. The hull waspainted with a special epoxidic resin in or-der to minimise the risk of polluting thecollected inorganic samples.

2541

Technologies

Figure 9: Path of the Charlie USV following ALANIS in the Genova Pra harbour (July2009)

In January-February 2004, during the XIXItalian Expedition to Antarctica, the vehi-cle worked in the proximity of the M. Zuc-chelli station of Terra Nova Bay, Ross Sea.The Charlie USV was deployed and re-covered by the Malippo, a 16 m supportvessel hosting the human operator station.Compatibly with weather and sea condi-tions, the robotics catamaran executed sixmissions (two for communications, navi-gation, guidance and control tests and fourfor water sampling) working for more than20 hours and collecting about 95 l bothof micro-layer and immediate sub-surfacewater samples. A view of the CharlieUSV collecting samples among the Antarc-tic ices is given in Figure 7. A detailed

discussion of system requirements for thestudy of sea-air interface, guidance andcontrol system, and Antarctic exploitationcan be found in [10].

8.2 Towards Rapid Environ-mental Assessment

The 21st century’s scenarios of ma-rine operations, regarding environmen-tal monitoring, border surveillance, war-fare and defence applications, foresee thecooperation of networked heterogeneousmanned/unmanned air, ground and marineplatforms. Examples are given by the Au-tonomous Ocean Sampling Network, inte-grating robotic vehicles and ocean mod-

2542


els to increase the capacity of observingand predicting the ocean behaviour, and theBarents 2020 vision, optimising marine re-sources thanks to historical and real-timeinformation collected by a large network ofcooperating vehicles.In this context, as a consequence of theirnetworking capabilities, USVs are natu-rally seen as a part of flotillas of hetero-geneous vehicles executing large scale sur-veys and supporting REA: in addition tooperate as communication relays for acous-tic communications with companions un-manned underwater vehicles, they can actas force multipliers both extending the useof the same device to a formation of vehi-cles and allowing the displacement of dif-ferent sensors in the same place about atthe same time. Research carried out withCNR-ISSIA USVs focused on the theoret-ical and experimental study of the prob-

lem of a slave vehicle following a mastervessel at a predefined range, constituting,for instance, a force multiplier in executingbathymetric surveys. Preliminary experi-ments, carried out in cooperation with theHydrographic Institute of the Italian Navyin the Genova Pra harbour in July 2009,with logistical support of the personnel ofA.S.D.P.S. Pra Sapello, demonstrated theeffectiveness of an approach relying on vir-tual target based path-following techniquescombined with on-line path generation, onthe basis of position data received from amoving master vehicle. As shown in Fig-ure 8, during preliminary trials the Char-lie USV autonomously followed the ALA-NIS vessel piloted by a human operator.The accuracy of the proposed algorithms,supported by high positioning GPS sys-tems aboard the two vessels, was very goodshown in Figure 9.

References[1] S.J. Cornfield and J.M. Young. Unmanned surface vehicles - game changing tech-

nology for naval operations (in Advances in unmanned marine vehicles). pages311–328, 2006.

[2] J.E. Manley, A. Marsh, W. Cornforth, and C. Wiseman. Evolution of the au-tonomous surface craft AutoCat. Proc. of Oceans’00, 1:403–408, 2000.

[3] J.E. Manley. Development of the autonomous surface craft ”ACES”. Proc. ofOceans’97, 2:827–832, 1997.

[4] J. Curcio, J. Leonard, J. Vaganay, A. Patrikalakis, A. Bahr, D. Battle, H. Schmidt,and M. Grund. Experiments in moving baseline navigation using autonomous sur-face craft. Proc. of Oceans’05, 2005.

[5] J.E. Manley, J. Curran, B. Lockyer, J. Morash, and C. Chryssostomidis. ApplyingAUV lessons and technologies to autonomous surface craft development. Proc. ofOceans’01, 1:545–549, 2001.

[6] J. Majohr and T. Buch. Modelling, simulation and control of an autonomous sur-face marine vehicle for surveying applications Measuring Dolphin MESSIN (inAdvances in unmanned marine vehicles). pages 329–352, 2006.

2543

Technologies

[7] A. Pascoal and et al. Robotic ocean vehicles for marine science applications: theEuropean ASIMOV Project. Proc. of Oceans 2000, 2000.

[8] A. Martins, J.M. Almeida, E.P. Silva, and F.L. Pereira. Vision-based autonomoussurface vehicle docking manoeuvre. Proc. of 7th IFAC Conference on Manoeuvringand Control of Marine Craft, 2006.

[9] T. Xu, J. Chudley, and R. Sutton. Soft computing design of a multi-sensor datafusion system for an unmanned surface vehicle navigation. Proc. of 7th IFAC Con-ference on Manoeuvring and Control of Marine Craft, 2006.

[10] M. Caccia, R. Bono, Ga. Bruzzone, Gi. Bruzzone, E. Spirandelli, G. Veruggio,A.M. Stortini, and G. Capodaglio. Sampling sea surface with SESAMO. IEEERobotics and Automation Magazine, 12(3):95–105, 2005.

[11] J. Ebken, M. Bruch, and J. Lum. Applying UGV Technologies to Unmanned Sur-face Vessels. SPIE Proc. 5804, Unmanned Ground Vehicle Technology VII, 2005.

[12] M.R. Benjamin and J. Curcio. COLREGS-Based Navigation in Unmanned MarineVehicles. IEEE Proceedings of AUV-2004, 2004.

[13] M. Caccia and G. Bruzzone. Execution control of ROV navigation, guidance andcontrol tasks. International Journal of Control, 80(7):1109–1124, 2007.

[14] M. Caccia, M. Bibuli, and G. Bruzzone. ALuminim Autonomous Navigator forIntelligent Sampling: the ALANIS Project. Sea Technology, 50(2):63–66, 2009.

[15] M. Bibuli, G. Bruzzone, andM. Caccia, and L. Lapierre. Path-Following Al-gorithms and Experiments for an Unmanned Surface Vehicle. Journal of FieldRobotics, 26(8):669–688, 2009.

[16] N. Miskovic, Z. Vukic, M. Bibuli, M. Caccia, and G. Bruzzone. Marine vehicles’line following controller tuning through self-oscillation experiments. Proc. of the17th Mediterranean Conference on Control and Automation, 2009.

2544

Design, Development and Exploitation of theRomeo ROV

M. Caccia1, R. Bono1, G. Bruzzone1, G. Veruggio2

1, Institute of Intelligent Systems for Automation, CNR, Genova, Italy2, Institute of Electronics, Computer and Telecommunication Engineering, CNR, Genova, Italy

[email protected]

Abstract

The design, development and exploitation of Romeo, a networked ROV capableof supporting heterogeneous payloads for both robotics research on intelligent vehi-cles and scientific missions in very harsh environment, are presented.The over-actuated open-frame vehicle, designed for high precision maneuvering forbenthic applications minimising the interactions with the operating environment, ischaracterised by: an interchangeable toolsled with common electrical and mechan-ical interfaces for the integration of heterogeneous scientific and technological de-vices a hierarchical control architecture and the definition of logical interfaces be-tween its levels for the integration with advanced mission control systems the con-nectivity to Internet and satellite communication channels for remote teleoperationthe capability of deploying and recovering scientific instrumentation on the seabed.This allowed the exploitation of the vehicle in a number of scientific missions inMediterranean and polar environment, both Arctic and Antarctica, demonstrating itscapabilities in supporting benthic sample and data acquisition as well as robotics re-search.This paper presents the vehicle design, development and exploitation focusing onthe development of payloads interoperable by other ROVs such as IFREMER Victor,Internet based tele-operation of the vehicle in Antarctica and Svaalbard islands fromEurope, and the deployment and recovery of a benthic chamber under the Antarcticice-pack.

1 Introduction

In the last years Remotely Operated Ve-hicles (ROVs) emerged as a consolidatedtechnology for underwater intervention forboth industrial, mainly in the offshore area,and scientific applications. Indeed, al-though the tether connection with the sup-port vessel constrains the vehicle operatingarea, amplifies the disturbance induced bythe sea currents and increments logistic re-quirements, on the other hand it dramati-cally increases the vehicle autonomy, mak-

ing possible the supply of electrical powerfrom the surface and allowing strict in-teractions with the operating environmentthanks to the possibility of transmittinghigh quality live video streams to the op-erator station, in particular with the use offibre optics. The result is the current pro-duction of more than 450 types of ROVsin the world covering a wide spectrum ofapplications, user requirements and physi-cal characteristics ranging from the heavyworking ROVs to the light micro-ROVs.Here, the main technical facts that, in the


Technologies

Figure 1: Romeo ROV mechanical design.

authors’ opinion, have characterised theevolution of research ROVs in the last fif-teen years are discussed.In the mid Nineties of the last century,the marine robotics research communityexamined the possibility of sharing hard-ware and software resources in order to fos-ter cooperation between different researchgroups and to reduce development costs.An example of this discussion is given bythe Proceedings of the IARP Workshop onmobile robots for subsea environment, heldin Monterey, USA, in 1994. The resultwas the adoption of hierarchical control ar-chitectures, the rigorous definition of thelogical interfaces between their continuousand discrete time levels, and of commonelectrical and mechanical interfaces in or-der to facilitate the integration of heteroge-neous scientific and technological payloadson different vehicles. Thus, the concept of

interchangeable toolsled, able to handle thedata acquisition and control of the aboardinstruments, and connected with the ROVonly through standard power supply andcommunication interfaces, was developed.In this way, the integration of additionalinstrumentation did not require any moreto modify the control and system manage-ment software of the vehicle. Victor 6000by IFREMER, France, [1] and Tiburon bythe Monterey Bay Aquarium Research In-stitute, USA [2] are a couple of typical ex-amples of this class of vehicles. Moreover,these vehicles are fully networked, i.e. ableto broadcast the acquired data.The fast development of embedded sys-tems computing power, as well as the in-creasing diffusion of free software and theenhancement of its real-time capabilities,at the beginning of the Millennium ledto the use of free operating systems, in

2546


Figure 2: Romeo ROV networked architecture.

particular GNU/Linux, as embedded real-time platforms for the implementation ofresearch ROV control systems. This is,for instance, the case of Jason II [3], thenew version of the historical Jason ROVby Whoods Hole Oceanographic Institu-tion (WHOI) [4]. Moreover, the high com-munication bandwidth guaranteed by fibreoptics tethers motivated, with respect to in-telligent ROVs, such as Victor 6000 andTiburon, characterised by a high aboardcomputer power, a return to the past ofROV control system located inside the sur-

face station in order to facilitate its upgradeand maintenance.More recently, dramatic developments inmarine fibre optics made possible the im-plementation of the concept of hybrid ROV,i.e. a vehicle that can work both con-nected to the mother ship by a very thin fi-bre optic tether (diameter of about 0.8 mm)when the transmission of large bandwidthdata is required for strict interactions withthe operating environment and, releasingthe tether, as an Autonomous UnderwaterVehicle able to explore and mapping the

2547

Technologies

Figure 3: Plancton sampling configuration of the Romeo ROV, Terra Nova Bay, RossSea, Antarctica (November 1997).

seafloor with optical and acoustic devices.This is the case of Nereus [5], developed byWHOI in cooperation with John HopkinsUniversity and U.S. Navy Space and NavalWarfare Systems Center of San Diego, thatin 2009 reached the deep of about 11000 min the Marianne Trench more than ten yearsafter the Japanese ROV Kaiko [6].In this context, in the Mid of the Nineties,the robotics group of the Institute for ShipAutomation (Institute of Studies for Intel-ligent Systems for Automation from 2002)of the Italian National Research Councildeveloped a mid class networked ROV,named Romeo, for robotics research and

scientific applications, according to thestructure of Victor 6000 and Tiburon. Inthe following, after a short discussion ofthe vehicle architecture, Romeo’s scientificexploitation is presented focusing on itsmain features such as its capability of car-rying heterogeneous payloads and of work-ing as a benthic shuttle, and its full connec-tivity to Internet from harsh and remote en-vironments.

2548


2 Romeo ROV architec-ture

2.1 Mechanical design

Romeo’s mechanical design should satisfybasic scientific needs in terms of manoeu-vring and payload capabilities, and oper-ational requirements in terms of logisticsand maintainability. During typical mis-sions the vehicle had to operate in differentpayload configurations, also in harsh envi-ronment and from vessels of opportunity,as in the case of polar under-ice and opensea applications, in order to execute pelagicand benthic surveys and data sampling.The need of an easily portable system oper-able by a team with few people induced thedesign of a shallow water ROV (up to 500m), weighting about 450 kg in air and about1 m in height, 0.9 m in width and 1.3 m inlength. This allowed the vehicle to be oper-ated from relatively small vessels (at least16 m) or from remote camps by a dedicatedcrew of four people.The main scientific requirement, that is be-ing able to operate different sets of de-vices during the same campaign, was sat-isfied by the possibility of mounting an in-terchangeable toolsled below the vehicle’sbasic frame. As shown in Figure 1, Romeo,which is intrinsically stable in pitch androll, is divided into three sections. The bot-tom section of the vehicle carries an inter-changeable toolsled for scientific devices,which in the standard version is equippedwith two cylindrical canisters for batteriesand payload electronics. The middle sec-tion, the core vehicle, is made up of a basicframe carrying one 100x32 cm (ld) cylin-drical canister for the control system, plantmanagement and communication electron-ics, two 80x15 cm (ld) cylinders for bat-

teries and battery chargers, one 60x15 cm(ld) cylinder for compass, gyro, and in-clinometers, four thrusters for propulsionin the horizontal plane, four for the verti-cal propulsion, and the basic sensor pack-age (CTD, pilot and scientist TV cameras,etc.). The top section, carried by the basicframe, is made up of foam for buoyancy.The thrusters, canisters, and instrumentsare allocated so that the overall structureof the vehicle is symmetric with respect toboth the x-z and y-z planes. The thrustersare arranged two by two in the corners,with the horizontal ones parallel to the di-agonals of the x-y section. This thrusterconfiguration enables full control of the ve-hicle’s motion even in cases of faults inone horizontal or vertical actuator, makingspace available for vision devices in frontof the vehicle and for electronics equip-ment in the middle. In addition, the al-location of the vertical propellers on thetop of the vehicle reduces their pull-up ef-ficiency because of propeller-hull interac-tions, but minimizes the interactions be-tween the boosted water and the sedimentsduring benthic operations, satisfying a ba-sic scientific requirement. The result was avehicle whose dynamics was modelled andidentified as discussed in [7].

2.2 Computing and communi-cation architecture

An architecture for unmanned underwa-ter vehicles basically consisting of threeEthernet LANs (surface, on-board andlab LAN), which can be connected tothe world-wide web for remote controland data access, was designed and imple-mented in order to allow fast system de-velopment through hardware-in-the-loopsimulations and to enhance the ROV ca-

2549

Technologies

pabilities by integrating it with a multi-disciplinary payload, i.e. a robotic andscientific system which can supervise themission and pilot the ROV, coordinating itsmanoeuvres with the acquisition of scien-tific data and samples.As shown in Figure 2, the surface Ether-net LAN connects a connection managercomputer (IPER) and a multi-machine dis-tributed Human Computer Interface (HCI),which allows a number of different oper-ators to interact with the robot at variouslevels of the control system. The conven-tional HCI for scientific applications con-sists of three interfaces: one for the pilot,who tele-operates the vehicle, one for thesupervisor, who supervises plant behaviourand resources allocation, and one for themarine scientist, who examines real-timeimages and sensors data to detect areasof interest. The IPER machine acquiresall the vehicle’s surface sensors as, for in-stance, acoustic positioning system, shipGPS and gyro-compass, and manages com-munications from the human interfaces andthe robot control system aboard the vehi-cle, dispatching telemetry data and collect-ing, while solving conflicts, the user com-mands. The surface LAN can also supportthe connection of local and remote roboticsand multidisciplinary payloads, which canautomatically supervise and manage thevehicle missions.The on-board Ethernet LAN connects acontrol system computer (SUB), whichexecutes the basic navigation, guidanceand control (NGC) tasks and managesthe vehicle plant and on-board scien-tific/multidisciplinary payloads. Due tothe limited computing power of the SUBmachine, consisting of a VMEbus with aMotorola 32 MHz 68040 CPU MVME162board, providing an Ethernet interface, car-rying four IP piggy-back modules for dig-

ital and serial I/O, and an additional boardfor analog I/O, an additional MVME162board was added for managing the auto-matic coordination of navigation, guidanceand control tasks. On these boards runsthe Romeo’s control system, a real-timeapplication developed using the VxWorksoperating system.The surface and on-board Ethernet LANswere originally connected by two twistedpairs in the 600 m long tether. In orderto overcome the problem lying in the factthat Ethernet cannot operate with twistedcables longer than 100 m, a modem anda router were added on each side of thecable for the modulation/demodulation onthe tether and the dispatching of Ether-net packets. The main drawback of thissolution was in limiting to approximately2 Mbps the maximum speed of Ethernetobtainable through the tether. However,in 1999 the system modularity allowed aneasy upgrade to a fibre optic link support-ing Ethernet at 10 Mbps, 5 RS-232 serialchannels at 115 Kbps, and 5 RS-422 serialchannels at 250 Kbps and 4 video chan-nels.At last, the lab Ethernet LAN supports theUnderwater Virtual World facilities allow-ing hardware-in-the-loop simulation of thevehicle 6 d.o.f. dynamics, environment andsensors [8].As far as the vehicle navigation, guid-ance and control system, performing auto-heading, auto-depth, auto-altitude andvision-based motion control, as well asexecution control of basic tasks, the readercan refer to [9, 10, 11], and [12].

2550


3 Integrability with het-erogeneous payloads

The Romeo ROV capability of carryingtoolsleds with heterogeneous devices wasexploited already during its first Antarc-tic campaign, performed during the australsummer 1997-98 at the Italian station ofTerra Nova Bay (TNB), Ross Sea, in theframework of the project PRISMA, fundedby the National Program of Research inAntarctica (PNRA). As a consequence ofthe success of that experience, Romeo wasemployed for collecting samples and datain the proximity of the Antarctic coast in anumber of further expeditions, when otherimportant vehicle characteristics, in termsof benthic and networked operation capa-bilities, were exploited too. Anyway, it isworth remembering the integration in thevehicle toolsled of: i) a LIDAR (XXI expe-dition, 2001-02), developed by ENEA, forthe detection of the chemical compositionof the substances dissolved in the sea in or-der to detect different algae chromophoregroups, water impurities and to analyse theplankton photosynthesis; ii) a fish eggs col-lection pump for the study of Antarctic sil-verfish reproduction conditions and habi-tat.Anyway, the integrability of the data ac-quisition and control system of the RomeoROV payload was fully demonstrated inthe EC MAS3 “Advanced ROV packagefor Automatic Mobile Inspections of Sed-iments (ARAMIS)” project, where a sci-entific and technological toolsled were in-tegrated with both the Romeo ROV and alarger toolsled mounted below Victor 6000.

3.1 PRISMA project

During the Italian XIII Expedition toAntarctica, which took place from Oc-tober 1997 to February 1998 at TerraNova Bay Station (Lat. 74.41’.42” S,Lon. 164.07’23” E), Romeo supported ma-rine biology and oceanographic research inthe framework of the PNRA - IntegratedRobotics Project for the Study of Antarc-tic Sea (PNRA-PRISMA). Romeo’s activ-ity was organized in three legs. In the firsttwo legs, from the end of October to midJanuary), the vehicle was operated from ahole in the ice-pack to collect data aboutunder-ice biological processes and to eval-uate the under-ice performances of a setof acoustic devices (echosounders, high-frequency pencil beam profiling sonar,Doppler speed-meter) preparatory to thestudy of SARA, the Italian Antarctic AUV.In the third leg (from mid January to endof February) Romeo was operated froma small boat to monitor the benthic envi-ronment in the Marine Antarctic SpeciallyProtected Area established nearby the Ital-ian TNB Station. The implementation ofa serial tunnelling mechanism between thesurface and on-board Ethernet LANs en-abled scientific end-users to transparentlycontrol instruments mounted on the ROVdirectly from the surface.During the first leg Romeo was oper-ated to support the Project 2.b.3 Ecol-ogy and Bio-geo-chemistry of the South-ern Ocean about the paths of the materialsubstances and flow of energy in the cryo-systems. The scientific payload was con-stituted by a multi-parametric gauge (con-ductivity, temperature, depth, oxygen, flu-orimeter), and a “microness” (zooplanctonsampler), mounted on the toolsled. In addi-tion, as shown in Figure 3, a spectral irradi-ance meter was mounted vertically in front

2551

Technologies

Figure 4: ARAMIS adaptation of the Romeo ROV networked architecture.

of the vehicle. Romeo’s principal missionsconsisted in zoo-plankton sampling withthe microness in an area of 200 meters fromthe hole, by means of horizontal traversesat different depths between 3.5 m and 20 mat the constant speed of 25 cm/s, and hori-zontal and vertical traverses to collect datawith the spectral irradiance meter.During the second leg of the expeditionRomeo supported the Project 4.a Roboticsand Telescience in Extreme Environment.In order to test under-ice behaviour ofacoustic devices, Romeo was equippedwith two Tritech ST-200 echo-sounders at200 kHz, a high frequency pencil beamprofiling sonar at 1.25 MHz, and a RDI

Workhorse Navigator Doppler velocime-ter at 300 kHz. The capability of theecho-sounder and profiling sonar in track-ing the ice-canopy at different angles ofincidence was tested, and interference tri-als were performed with the two echo-sounders mounted in upward/downwardlooking configuration. The feasibility ofthe upward/downward looking sonar con-figuration to execute the task of automat-ically following the ice-canopy and map-ping the seabed were also demonstrated.A number of radial traverses from the de-ploying ice-pack hole were executed, al-lowing the mapping of the bathimetry andice-canopy profile of the area.

2552


Figure 5: View of the Romeo ROV over the thermal vents of the Milos island (September2000).

During the third leg Romeo collected ben-thic data (water samples, TV images andphotos) in the Marine Antarctic SpeciallyProtected Area established nearby the Ital-ian TNB Station. A high number of tra-verses at constant range from the seabedwere performed in order to collect highquality and stable video images useful forthe counting and sizing of the organismsliving in the explored area. The vehiclewas equipped with the complete suite ofacoustic instruments plus a set of scien-tific devices consisting of two Niskin bot-tles for water sampling, still camera andstrobe lights, and calibrated laser pointermatrix for the sizing of the observed fea-tures of interest. Moreover, the keel of theCampbell Ice Tongue, reaching the depthof 160 m, was explored collecting watersamples below it at a depth of 216 meters.

3.2 ARAMIS project

In the period 1997-2000, ROMEO and Vic-tor 6000 were used as carrier ROVs inthe EC funded ARAMIS project. Thisproject aimed at developing a scientific andtechnological system to be integrated withtypical mid class existing ROVs to forma highly automated and robotised scien-tific tool for carrying out benthic investi-gations both in shallow and deep waters.The ARAMIS system consisted of a skidequipped with technological instruments,e.g. altitude echo-sounder, obstacle avoid-ance sonar, and quantitative image cam-eras, and scientific devices, e.g. water sen-sors, small sediment sampler, mini multi-purpose multi-sampler, object recognitioncamera, and coring monitoring camera.The system was completed by a data acqui-

2553

Technologies

Figure 6: View of the Romeo ROV pilot interface during coring operations over thethermal vents of the Milos island (September 2000).

sition and control system of the ARAMISinstruments, located on-board the toolsled,and a surface controller, constituted by anetwork of computers devoted to the man-agement of scientific devices and pilot in-terfaces and to acoustic and video imageprocessing tasks. In particular, the ROV,making available a conventional auto-pilot,was guided by the ARAMIS supervisor.The integration of the ARAMIS modules inthe Romeo networked architecture, in thatcase including a fibre optic link betweenthe surface and the vehicle, is shown in Fig-ure 4.The operability of the ARAMIS system in-tegrated with the Romeo ROV with respectto scientific specifications was demon-strated in a test campaign carried out inSeptember 2000 in the Aegean Sea in thehydrothermal vent areas near the island ofMilos. Seven dives were performed overa hydrothermal vent area, where the seabottom was constituted by loose sand with

ripples, some sea grass patches and bacte-rial/colloidal silica mats (the white stripeclearly visible on the seabed in Figure 5).It is worth noting that Romeo was ableto support direct interactions of devicesmounted in the toolsled with the seabedduring sediment probing (with four sen-sors: temperature, pH, sulphide, oxygen)and coring. In particular, the ROV agilityin manoeuvring allowed, together with thedetailed view of the scene provided by thecoring monitoring camera, the ARAMIShuman operator to position the sedimentprobes right over a thermal vent. A view ofthe ROV pilot station during coring opera-tions is shown in Figure 6, where the im-ages taken by the Coring Monitoring cam-era are clearly visible.

4 Benthic shuttleIn the wake of the demonstration of ma-noeuvring accuracy shown in the ARAMIS

2554


Figure 7: View of the ABS benthic chamber during mission preparation, Terra Nova Bay,Ross Sea, Antarctica (January 2004).

project and of the experience of SIRENE,the IFREMER AUV for the deployment ofbenthic stations at sea [13], Romeo was ex-ploited as benthic shuttle for the deploy-ment and recovery of a benthic chamberover the sea floor below the ice-canopy dur-ing the XIX Italian expedition to Antarc-tica in 2003-04. Thanks to the video cam-eras mounted aboard the ROV, the scien-tific end-user could choose with high ac-curacy the place where releasing the in-strumentation and recovering it after 24hours with the vehicle equipped with a tele-operated automated hooking system. Ben-thic module localisation was simply per-formed by tracking the ROV and benthicchamber positions with a SSBL acousticpositioning system, whose transducer wasmounted in a fixed location, i.e. a hole inthe ice canopy, in the ice field.The project Antarctic Benthic Shuttle

(ABS) was carried out in cooperation withDIPTERIS-UNIGE (Department for theStudy of the Territory and its Resources- University of Genoa) to study the struc-ture and long-term variations of the ben-thos in the Protected Marine Area of BaiaTerra Nova, from Road Cove to AdelieCove, as required by the ASPA and theSCAR CS - EASIZ (Coastal Shelf - Ecol-ogy of the Antarctic Sea Ice Zone) pro-grams. The standard benthic module con-sisted of a benthic chamber for the in-situanalysis of sediment-water interactions indifferent conditions of ice cover, and of atime-lapse image and data acquisition sys-tem. Benthic chambers, isolating a knownvolume of seawater and a known surfaceof sediments, allow the estimation of oxy-gen consumption by different benthic com-partments, e.g. macrofauna, meiofauna andbacteria, assessing their relative contribu-

2555

Technologies

Figure 8: Adaptation of the Romeo ROV networked architecture for Internet-based tele-operation.

tion in different environmental conditions.In particular, the project attention focusedon bacteria, which exert a major role inthe turn-over of organic matter and in re-mineralization processes (microbial loop).In addition, benthic chambers allow directmeasurements of benthic fluxes, taking intoaccount bio-turbation and sediment mix-ing. Particular interest relayed on diage-netic reactions in surface sediments, withchanges in concentrations of organic mat-ter, nutrients, dissolved oxygen at water-

sediment interface. Photographic systems(video-tape and time-lapse photographs)allowed the visual analysis of the evolutionof some proliferative and bioturbative pro-cesses, as well as behavioural observations.A dedicated toolsled, see Figure 7, was de-signed and built for the ABS project, basi-cally consisting of a benthic chamber, i.e. aglass box of 30x30x20 cm (lwh), equippedwith a CTD with oxygen sensor, two cam-corders, which monitored the space insideand the sea floor nearby the chamber, and

2556


Figure 9: Romeo ROV ice camp for Internet satellite tele-operation, Terra Nova Bay,Ross Sea, Antarctica (December 2001).

three bottles for suspended particle sam-pling. The data acquisition system wascontrolled by means of a single board com-puter mounting an Intel 486 CPU and run-ning a software application written usingthe RTKernel operating system.A total of eight missions were performedboth through a hole in the ice-pack (seventrials) and from a support vessel (one trial),with successful recovery of the instrumen-tation.

5 Networked ROVAfter preliminary experiments of Internetbased mission control carried out in coop-eration with the Instituto Superior Tecnicoof Lisboa in 1999, the accessibility of theRomeo ROV through Internet was mainlydemonstrated remotely tele-operating thevehicle in polar regions in the projects E-

robot 1 and E-robot 2 in the period 2001-02.As shown in Figure 8, the Romeo net-worked architecture can support the vehi-cle Internet-based tele-operation with theintegration of a couple of modules, the so-called Remote and Local servers. The Re-mote Server is a computer located in thevehicle operating site handling communi-cations between the ROV and a genericcommunication channel, which can be alsoa point-to-point connection with the Lo-cal Server guaranteed by a radio or satel-lite link. In addition to transmit/receivevehicle telemetry and commands, the Re-mote Server acquires, by means of a framegrabber, the video images provided by theROV cameras coming from Romeo’s videocamera and send them after compressionthrough the communication channel. Onthe other hand, the Local Server, a com-

2557

Technologies

puter usually located in the Genova ma-rine robotics lab, handles Internet commu-nications between the Remote Server andRomeo’s remote users connected to theWorld Wide Web, by means of suitablesoftware applications.

5.1 Internet-based mission con-trol

In Fall 1999 a preliminary demonstra-tion of Internet-based mission control ofunmanned underwater vehicles was per-formed in cooperation with the InstitutoSuperior Tecnico of Lisboa [14]. Duringthe experiment, the CORAL robotics mis-sion planner running remotely in Lisboncoordinated the activity of the Romeo ROVin a test pool in Genoa, in order to exe-cute a user-defined mission. Although theproblem of mission control of autonomousvehicles through the Internet network wasmainly addressed, the success of the exper-iment showed that a ROV tele-operation bymeans of Internet was possible and that itwas worth addressing research in that di-rection.

5.2 E-robot 1 projectOn December 2001, during the XVII Ital-ian Expedition to Antarctica, in the frame-work of the E-robot 1 project, in coopera-tion with CNR-SRT, several experiments ofInternet-based satellite tele-operation withRomeo were carried out [15]. To allowthe connection to Antarctica, Romeo’s con-trol system was integrated with a satellitecommunication system. Several users con-nected to the World Wide Web and, in par-ticular, on December 18, 2001, from theCNR headquarters in Rome, had the pos-sibility of remotely operating Romeo im-merse in the sea in the Marine Antarctic

Specially Protected Area nearby the Ital-ian Terra Nova Bay Station. As shown inFigure 9, where the surface field is shown,the Romeo ROV was operated from a sur-face station located on the ice-pack. Theseexperiments were, to authors’ knowledge,the first example of satellite tele-operationof a ROV directly usable by users surfingon the web. The main purpose of thesetests was to evaluate the possibility of ex-ecuting reliable Internet tele-operation ofremote robotic systems in actual operatingconditions. Given the preliminary phase ofthat research, the experiments focused onverifying the possibility of piloting the re-mote ROV from a generic Internet interfaceby non-specialised end-users rather than onthe execution of an Internet-controlled sci-entific mission.

5.3 E-robot 2 projectIn September 2002, in the framework ofthe E-Robot2 project in cooperation withCNR-SRT, Romeo was remotely operatedthrough Internet while operating in theArctic sea, more precisely in the Kongs-fjord in Norway, nearby the Italian Station“Dirigibile Italia” situated in Ny-Alesund,Svalbard islands [16]. The ROV was oper-ated by means of a support working boat,on which was installed the Remote Serverand connected to the Italian ground sta-tion of Ny-Alesund with a wireless link at11 Mbps (IEEE 802.11b standard). Theconnection between the Italian Station andthe Local Server located in the Genovalab consisted of 4 ISDN lines at 64 Kbps. Of the 256 Kbps available communi-cation band, one half was used to set upa standard H323 video-conference systemthat allowed the marine scientist to com-municate with Romeo team operating inKongsfjord during experiment operations

2558


and the other half was reserved for ROVcommands, data telemetry and video im-ages. Operating in this way it was possibleto be world-wide connected with the un-derwater vehicle to follow the experimentsand even to pilot it by means of high levelcommands. This allowed, for the first time,the execution of a number of missions un-der the control of a few teams of marinescientists from different European coun-tries (the Scottish Association for MarineScience, the Norwegian College of Fish-ery Science of University of Tromsø, theInstitute of Protein Biochemistry and theEnzymology and Institute of Biomolecular

Chemistry of Naples) who, without mov-ing from their laboratories, had the oppor-tunity to conduct real-time experiments inthe underwater environment in the Kongs-fjord. The campaign in the Arctic Regionreached its two main goals: i) to carry outa set of preliminary dives in various sitesin the Kongsfjord giving the marine scien-tists a preliminary set of data to plan a moresystematic investigation in the future years;ii) to test and demonstrate the effective-ness of the remote control of robots usedin extreme environment for tele-science ap-plications, along with the methodology al-ready used in the E-robot 1 project.

References[1] J.L. Michel et al. Victor 6000: design, utilization and first improvements. Proc. of

the 13th International Offshore and Polar Engineering Conference, 2003.

[2] J.B. Newman and D. Stokes. Tiburon: development of an ROV for ocean scienceresearch. Proc. of Oceans ’94, 2:483–488, 1994.

[3] L.L. Whitcomb, J.C. Howland, and D.A. Smallwood. A new control system forthe next generation of US and UK deep submergence oceanographic ROVs. Proc.of the 1st IFAC Workshop Guidance and Control of Underwater Vehicles, pages137–142, 2003.

[4] D.R. Yoerger, J.B. Newman, and J.E. Slotine. Supervisory control system for theJason ROV. IEEE Journal of Oceanic Engineering, 11(3):392–400, 1986.

[5] Fletcher et al. Proof of concept demonstration of the Hybrid Remotely OperatedVehicle (HROV) light fiber tether system. Proc. of Oceans’08, 2008.

[6] S. Takagawa. Advanced technology used in Shinkai 6500 and full ocean depth ROVKaiko. J. of Mar. Techn. Soc., 29(3):15–25, 1995.

[7] M. Caccia, G. Indiveri, and G.Veruggio. Modelling and identification of open-framevariable configuration unmanned underwater vehicles. IEEE Journal of OceanicEngineering, 25(2):227–240, 2000.

[8] G. Bruzzone, R. Bono, M. Caccia, and G. Veruggio. A Simulation Environment forUnmanned Underwater Vehicles Development. Proc. of MTS/IEEE Oceans 2001,2001.

2559

Technologies

[9] M. Caccia and G. Veruggio. Guidance and control of a reconfigurable unmannedunderwater vehicle. Control Engineering Practice, 8(1):21–37, 2000.

[10] M. Caccia, G. Bruzzone, and G. Veruggio. Bottom-following for remotely operatedvehicles: algorithms and experiments. Autonomous Robots, 14:17–32, 2003.

[11] M. Caccia and G. Bruzzone. Execution control of ROV navigation, guidance andcontrol tasks. International Journal of Control, 80(7):1109–1124, 2007.

[12] M. Caccia. Vision-based ROV horizontal motion control: near-seafloor experimen-tal results. Control Engineering Practice, 15(6):703–714, 2007.

[13] J. Opderbecke, V. Rigaud, and D. Semac. Autonomous navigation of the free-swimming vehicle SIRENE. Proc. of Oceans’98, 3:1650–1654, 1998.

[14] G. Bruzzone and et al. Internet mission control of the Romeo unmanned underwa-ter vehicle using the CORAL mission controller. Proc. of MTS/IEEE Oceans’99,3:1081–1087, 1999.

[15] G. Bruzzone and et al. Internet-based satellite teleoperation of the Romeo ROV inAntarctica. Proc. of 10th IEEE Mediterranean Conference on Control and Automa-tion, 2002.

[16] G. Bruzzone, R. Bono, M. Caccia, P. Coletta, and G. Veruggio. Internet-basedtele-operation of the Romeo ROV in the Arctic region. 6th IFAC Conference onManoeuvring and Control of Marine Craft, 2003.

2560

Proteolytic Marine Enzymes and their Applica-tions

M. Salamone1, A. Cuttitta1, G. Seidita2, L. Tutino1, T. Masullo1, S.Rigogliuso3, F. Bertuzzi4, G. Ghersi31, Institute for Coastal Marine Environment, CNR, Capo Granitola (TP), Italy2, Department of Biopathology and Biomedical Methodology, University of Palermo,Palermo, Italy3, Department of Cell and Developmental Biology, University of Palermo, Palermo, Italy4, Hospital System “Niguarda Ca’ Granda”, Milano, Italy

[email protected]

Abstract

The use of proteinases from marine organisms has several advantages over com-mercial proteinases since their optimal temperatures and other enzymatic charac-teristics differ from homologous proteinases obtained from warm-blooded animals.High activity of these enzymes at low temperatures is interesting for many industrialapplications.Enzymes used in tissue digestion to isolate the islets of Langherans for transplan-tation are extracted from Clostridium histolyticum bacteria; they have strong col-lagenolytic activity compared to vertebrate collagenases. However, several impedi-ments persist in human islet success probably due to the variability in compositionand concentration of collagenases used during the digestion phase and to the toxicityeffect of the enzyme blend. Also the temperature used during tissue dissociation is acritical step since all the endogenous human proteases are active at 37°C.We analyse different batches of commercial enzymes, used in Langherans islets pu-rification, both in proteins composition by SDS-PAGE electrophoresis and in enzy-matic activities by zimography analyses. We propose an innovative evaluation essayto select enzymes useful in islet pancreas isolation procedure. Using this approach,we characterized and tested marine enzyme from hepatopancreas of different marineinvertebrates. We show that marine enzymes have a stronger proteolytic activity,also at low temperature on several interesting substrates, compared to the commer-cial ones.

1 IntroductionEnzymes are molecules involved in severalprocesses; they have the capability to ac-celerate a reaction, activate or inhibit func-tions, and are involved in life processes.They are an essential target for the adapta-tion of an organism to a cold environment.

In recent times, several studies have beencarried out in order to understand the rulesgoverning their molecular adaptation tolow temperatures. These fundamental as-pects are strongly connected with a strongbiotechnological interest in the exclusiveproperties of these molecules [1]. They


Technologies

present important advantages regardingtheir specific activity, lower stability andunusual specificity. Enzymes with thesecharacteristics are well represented in seaorganisms.Innumerable tons of fish are harvested eachyear, about 70% of which, including vis-cera, head, skin, bone and some muscletissue, are considered as waste during fish-ery products processing. These are actu-ally discarded and few attempts to recov-ery them are done; recently, more severeecological policies have pushed severalseafood producers to find alternative waysin using secondary raw materials [2, 3].Fish viscera, about 5% of the total, are richpotential sources of many enzymes thatmay have some exclusive interesting prop-erties for both basic research and industrialapplications.Main digestive proteinases detected infish hepatopancreas are trypsin and chy-motrypsin. Trypsin has been extracted,purified and characterized in several fishspecies including the sardine (Sardinopsmelanostictus) and the arabesque green-ling (Pleuroprammus azonus) [4] fromthe pyloric ceca of jacopever (Sebastesschlegelii), yellowfin tuna (Thunnus al-bacores) [5], skipjack tuna (Katsu-wonus pelamis) [6] and Monterey sardine(Sardinops sagaxcaerulea) [7].Moreover, with the new molecular biol-ogy technology it is possible to clone sin-gle genes and to produce marine organ-ism enzymes by in vitro synthesis systems.For example, enzyme with collagenoliticactivity are synthesized from Paralith-odes camtschaticus and cystein proteasescathepsin has been produced from Metape-naeus ensis [8].One of the first applications of cold-adapted enzymes, such as described pro-teases, lipases, alfa-amylases and cellu-

lases, were their use as additives in deter-gents. They were used also in bioremedi-ation; in fact, the high catalytic efficiency,together with the unique specificity at lowand moderate temperatures, should renderthem suitable for bioremediation purposes[9].Furthermore, the use of cold-adapted en-zymes in the food industry are several.For example, in the milk industry, beta-galactosidase is used at low temperature toreduce the amount of lactose responsiblefor severe induced allergies; infant formulais an artificial substitute for human breastmilk, destined to infant consumption. Gen-erally, these enzymes are obtained as hy-drolysates from bovine milk proteins inorder to match with infant nutritional re-quirements. However, whey protein andcaseins may induce the formation of spe-cific immunoglobulin E antibodies in chil-dren that develop an allergy to cow’s milk.alfa-lactoglobulin is considered to be oneof the main allergens in bovine whey.Generation of bioactive peptides is an-other important application of marine en-zyme. For example, bioactive peptideswith antioxidant properties derived fromvarious proteins by enzymatic hydrolysishave become a topic of great interest forpharmaceutical, health food and process-ing/preservation industries. Antioxidantactivity has been reported for protein hy-drolysates obtained from different sourcessuch as whole capelin, squid skin Pacificwhiting [10]. Furthermore, controlled en-zymatic hydrolysis of proteins may pro-duce a series of small polypeptides whichcan modify and even improve the func-tional properties of proteins.Gelatin is produced by heat denaturation ofcollagen; fewer studies have been done onthe potential uses of gelatin derived fromaquatic animal skin compared to mam-

2562


malian. Kim et al. [11] showed that fishskin gelatin can be modified into biolog-ically active peptides by protease treat-ments; in particular some purified peptidesexhibited the potential to act as inhibitorsof angiotensin I converting enzyme and an-tioxidants against peroxidation of linoleicacid. Moreover, biologically active pep-tides by protease treatments jumbo squid(Dosidicus gigas) [12] showed also antiox-idative activity via radical scavenging andby retarding lipid peroxidation. Regardingfish gelatin-derived peptides, there are onlya few reports dealing with their antioxidantproperties [11] and no information can befound about functional properties of thesepeptides.In the past, decapods enzymes obtainedfrom hepatopancreas were used in wouldhealing cure; in fact, their capability toremodelling the fibrin clot during woundhealing process have been reported, suchas to induce tissue repair in burnings, ul-cerations and other [13].Several families of proteases are involvedin proteins digestion/remodeling and theywere well described in mammalian sys-tems. They can be distinguishable on thebasis of their proteolytic behavior, biologicroles and structural organization. Manyproteolytic enzymes were characterized;there are:- the matrix metalloproteinases (MMPs),

including members associated to plasmamembrane (MT-MMPs);

- the ADAM (a family of disintegrin andmetalloprotease);

- the meprins;- the secretases (also termed sheddases or

convertases);- the collagenases;- the serine-peptidases.They exert their role prevalently on extra-cellular proteins; differently by cysteine-

and aspartic- proteases that, in eukaryoticcells, generally play their functions insidethe cell (cytoplasm compartment), as regu-lators of cell cycle and apoptotic processes.In our work, we principally focused onthree different classes of proteolytic en-zymes: collagenases, MMPs and serineproteases, and their role in cell extractionfrom tissues/organs for cell therapy andbiomedical applications.

2 Proteolytic enzymes

A generic subdivision of proteolytic en-zymes in mammalians puts them in fourmain classes: Aspartic-, Cysteine- MatrixMetallo- and Serine- proteases. Those be-longing to the first two groups are com-monly involved in apoptosis and cell pro-liferation processes, even if there are someexceptions. The other two are basically in-volve in cell migration/invasion processes,such as in ECM remodeling, proteins acti-vation, digestion processes, and are com-monly used in cell extraction from tis-sues/organs.Some members of proteolytic enzymes be-longing to MMPs and Serine-proteasesfamilies, directly or indirectly involved inproteins digestion, are listed in Table 1.Generally, soluble forms of MMPs andSerine-proteases are all secreted in inac-tive forms and extracellulary enzymaticallyactivated. A similar activation mechanismis observed for the plasma membrane as-sociated forms of MMPs, the MT-MMPs,which are also activated by the action ofspecific enzymes. While, an activationdissimilar way is adopted by the integralplasma membrane serine peptidases; sev-eral of them are synthesized in monomericinactive forms beginning in active formswhen associated to form a dimer to the site

2563

Technologies

of proteolysis [14].Differently, collagenases produced fromprocaryotic organisms are secreted just inactive forms [15].Plasma membrane associate proteolytic en-zymes exert their function both on ECMcomponent digestion, such as in proteoly-sis cascade activation, and in chemokinesand factors activation.

3 Collagenases

Due to its three-dimensional organization,only a restricted number of proteases candigest collagen. Because of the varietyin the collagen structure, it is not easy todiscriminate true collagenases from othercollagenolytic proteases and gelatinases,which just digest gelatin (denatured col-lagen). Moreover, the thermodynamicallyfavored conformation of collagen at bodytemperature is a random coil rather than anα-helix [16]. This specifies that a varietyof denatured forms of collagen are nor-mally present in the body. It is feasible thata good number of collagenases can digestboth collagen and gelatin. Moreover, asis frequently the case with collagenolyticproteases, collagenases are considerablyactive in hydrolyzing other protein sub-strates.Mammalian matrix metalloproteinases(MMPs) have been explored better thancollagenolytic proteases from bacteria.Only the collagenases ColH and ColG fromClostridium histolyticum, have been wellstudied [17, 18, 19]. As reported in Figure1, the collagenases of bacteria are a simpleorganization compared to mammalians andcollagen-binding regions have been identi-fied in the structures of collagenases ColGand ColH from C. histolyticum and colla-genase ColA from Clostridium perfringens

[17] but not in others. However, sequencescomparison between mammalian and bac-terial collagen-binding sites seem to bedifferent [20]. C. histolyticum ColG colla-genase possesses tandem collagen-bindingdomains that interact with all types of col-lagen fibrils in spite of their diameter; thus,collagenase are said to show broad sub-strate specificity [21]. This is in strikingcontrast to the strict specificity of mam-malian MMPs toward various types of col-lagens [22].

4 Matrix Metallopro-teases

MMPs belong to a family of about thirtymembers zinc-containing endopeptidaseactivities and have the capability to digestmany ECM constituents [19]. All mem-bers of the family are generated as inac-tive forms, pro-enzymes; they must be en-zimatically modified to be active. Thefamily members have been divided in dif-ferent groups on the basis of their struc-ture and/or digestive substrates (Table 1).The MMPs are structured in different do-mains, that are present in diverse numberand composition inside to the family (Fig-ure 2). The MMP3, MMP10 and MMP11are stromelysins, they are very similar tothe collagenases, but they have a broadspectrum substrate specificity and degradecomponents as proteoglycans, fibronectin,and laminin.

5 Serine proteasesAlmost 40 members belong to serine-protease family having different functions.The members of the family are highlyconserved within the human genome and

2564


Figure 1: Structural organization of procariotic collagenases. Typical collagenolytic pro-teases from bacteria are organized with a pre domain followed by catalytic domain, com-mon sequence and collagen-binding domains.

Figure 2: Structural organization of MMPs. All members of MMPs family are struc-turally organized in the way to have a signal peptide, the pro-peptide and the catalyticsite containing the zinc-binding site. The MT-MMPs are the members of the family as-sociated to the plasma membrane; they are connected to the plasma membrane troughta trasmembrane domain followed by a cytoplasmic tail, having signals function or via aglycosylphosphatidylinositol (GPI) membrane anchor.

amongst species. The enzymes of thisfamily seem to have a common chemicalmechanism of peptide substrate hydroly-sis. The members of this family enclose en-zymes involved in several functions; diges-tive enzymes, blood coagulation and fibri-nolytic enzymes, Several of them are solu-ble molecules, they are secreted in inactiveform to be enzymatically activated (Figure3). Other are plasma membrane associatemembers and belong to the TTSP family;the prototype member could be considereddipeptidyl peptidase 4 (DPP4) (Figure 3).

The individual serine proteases show a sur-prisingly ample diversity in their targetspecificities. For example, plasmin has avery small broad specificity and it is a veryefficient enzyme; its activity is analogousto trypsin, cleaving C-terminally Lys andArg aminoacids. On the other hand, theuPA and tPA, activators of plasminogen,are tremendously precise and a very fewnumber of substrates for these enzymeswere identified.An emerging class of serine protease be-longing to TTSPs family are seprase and

2565

Technologies

Table 1: Extracellular matrix metallo-proteases and serin-proteases.

DPP4; it has been demonstrated that theyare directly involved in new vessel for-mation [23]; they are inducible, specificfor proline containing peptides and macro-molecules, and active on the cell surface.In particular, DPP4 has prolyl exopep-tidase activity; some of its natural sub-strates are: neuropeptide Y [24], substanceP [25] and beta chemokines such as eo-taxin, SDF-1 (stromal derived factor) andRANTES (regulated on activation normalT-cell expressed and secreted) with eitherL-proline, L-hydroxyproline, or L-alanine

at the penultimate position [26].

6 Enzymes in tissue diges-tion and transplantation

End-stage organ failure is a major causeof death worldwide that can occur in pa-tients of all ages; transplantation is the cur-rent standard of care for chronic end-stagedisease of many organs. Despite the suc-cess of organ transplantation, it is becom-ing clear that organs available through do-

2566


Figure 3: Structural organization of Serine proteases. The soluble forms of serine pepti-dases are organized with a catalytic site common to all members, some has one kringleand one growth factor domains, a Zn2+ finger domain, as too some vitamin K-dependentCa2+ binding domains. A different organization was observed in transmembrane serineproteases. A- In all members of the family the catalytic site and the transmembrane do-main are present than several different domain in number and compositions are present.The frizzled-like cysteine-rich repeats (FRZ); the LDL receptor repeats (LDLR), themeprins-like domain (MAM), the SEA repeats found in sea urchin – enteropeptidase –and agrin (SEA), the CUB repeats found in complement – urchin embryonic growth fac-tor – and bone morphogenic protein (CUB), the scavenger receptor cysteine-rich repeat(SR). B- shows the organization of TTSPs that are functionally organized in dimmers.

nation will never be enough to meet theincreasing demand. In addition, organtransplantations are basically affected byan high rate of complications, sometimessevere. The past decade’s rapid advance-ment in cell production, stem cell biology,and tissue engineering generated an explo-sive outburst of reports that gave rise tocell therapy. Examples of active cell ther-apy include Langerhan islets, hepatocyte,condrocyte transplantation, but many othernew approaches were proposed.Hepatic progenitors were isolated from

healthy liver [27].Moreover, recent studies elucidated theability of cardiomyocytes to regenerate inthe event of myocardial injury [28]. In var-ious tissues of the adult body, multipotentstem cells exist, they can give rise to newstem cells (self-renewal) and differentiatedcells necessary for maintenance or for therestoration of damaged tissue.Despite of these premises, one of the fac-tors inhibiting the widespread clinical ap-plication of these new approaches is thelack of standardization of procedures for

2567

Technologies

cell production, in particular for those re-garding tissue digestion and cell harvest-ing.Consequently, a standardization requestedfor optimizing results, making them repro-ducible and for being fully compliant to theupcoming EU regulations on cell handlingfor clinical purpose.One of the first successful example of celltherapy is represented by islet transplan-tation since the early ’90. Clinical islettransplantation is a relatively straightfor-ward and safe procedure. However, theprocedure of human islet isolation is com-plex and requires a great deal of experi-ence to yield sufficient number of produc-tive cells to be used in transplantation ap-proaches.Current isolation methods involve the de-livery of collagenase enzymes blend intothe pancreatic duct, in order to deliver en-zymes to the islet-exocrine interface fol-lowed by purification of them from ex-ocrine tissue based on their different den-sity. However, it is still difficult to recoverthe entire quantity of islets contained ina pancreas. A major obstacle in success-ful human islet isolation is the variabilityof the collagenases used in the digestionphase.Ricordi and colleagues introduced a tis-sue dissociation chamber and recirculationof enzyme solution; they termed it the”automated method” [29]. They placedthe collagenase-distended pancreas insidea stainless steel chamber and mechani-cally dissociated the pancreas by gentle ag-itation. The principle behind this com-bined enzymatic and mechanical approachis to enable minimal physical trauma to theislets by collecting the islets as they are lib-erated from the digestion chamber. Today,the modified “Ricordi chamber” is the uni-versal device used in human islets isolation

process.A practical advantage was the culture of C.histolyticum to produce a large amounts ofenzymes. Enzymes derived from C. his-tolyticum are called ”crude collagenase”,and its biologic and enzymatic propertieswere extensively studied.In addition to the collagenases, the crudepreparations also contain several other pro-teolytic enzymes. Due to the heterogene-ity composition in lytic enzymes, the activ-ity of crude preparations is extremely vari-able among different ”crude collagenase”batches.This variability was recognized as the ma-jor obstacle to successful human islet pu-rification. The introduction of a novel col-lagenase blend, Liberase HI (Roche Ap-plied Science, Indianapolis, IN), was in-tended to reduce the lot-to-lot variabilityof enzyme effectiveness. However, this en-zyme blend still exhibits significant lot-to-lot and even intra lot variability. Moreover,recent studies showed that Liberase is notmore efficient than ”crude collagenase”.To optimize proteolyses digestion, it isimportant to understand the structure ofthe islet-exocrine interface and the natureof substrate at this interface. Collagensare the major fibrous proteins constitutingislet- exocrine interface. Interactions ofcells with their environment play a criti-cal role in maintaining tissue integrity inadult life. The use of collagenases for celltissue extraction need to be evaluated caseto case, in order to optimize the methodto isolate progenitive cells from differenttissues. The optimization of the extractivemethod is crucial for cell viability and forcell therapy application.The most important enzymes used inthis application were collagenases obtainedfrom the Clostridium histolyticum humanpathogenic microorganism. These en-

2568


zymes have a very strong enzymatic activ-ity, but always present some toxic effectsreducing the efficiency in cell purification.To optimizate tissue digestion we pro-duce recombinant col G e Col H withshow a better performance comparedwith the commercial (Italian Patent N°RM2009A000661 15-12-2009).Moreover we partially characterized ma-rine enzymes extracted form hepatopan-creas of Octopus vulgaris, PalinurusElephas and Eriphia Verrucosa, MajaSquinado. Special attention was paid foranalyzing the capability of extracted en-zymes to digest native collagen (paper inpreparation).We compared the protein/proteolytic pat-terns of some commercial enzyme usedin islets pancreas isolation, to the col-lagenolitic enzymes in different marine or-ganisms. These comparisons were per-formed by SDS-PAGE and Zimographyanalyses. To better characterize the bio-physical/functional qualities of analyzedenzymes we valued their stability/activity,using different incubation temperatures fordifferent times.Temperature is a very important factor forthe success of cell extractive process; infact, host endogenous proteases present inthe pancreas, or in other extractive organs,can be activated during the process andcontribute non-specifically in ECM disgre-gation/cells extraction; this effect could betoxic.To optimize the enzyme blend and the tem-perature of tissue dissociation, we used anew in vitro method of analyses; we gener-ated type-I collagen matrix substrates, in-side which we cultured human endothelialimmortalized cells. Endothelial cells weregrown inside the type-I collagen three-dimensional matrix in 96-well plates toform vessel-like structure, mimic the in

vivo tissue. After two days of culture, thetime needed to reach vessel-like structures,each well was treated with diverse blendenzymes at different concentrations and di-verse incubation temperatures. Endothe-lial cells were dissociated from vessel-likestructures by enzymes’ action; using a con-focal microscopy approach we valued themorphology/vitality of matrix network re-leased cells. These data allowed us to eval-uate how much and how long to use differ-ent enzyme blend in cell tissue extraction.Moreover, this approach showed the pres-ence of toxic components in the enzymeblends actually used in prancreas islets ex-traction (Salamone M. et al. submittedto Transplantation Proceeding); moreover,preliminary data on the marine enzymesextracted from epatopancreas of differentcrustacean showed very high efficiency incell extractive process.

7 Conclusions

Future advances in enzyme technology, to-gether with a better understanding of thecharacteristics of the extracellular matrixof the human pancreas, will make it possi-ble to optimally liberate islets using “tailor-made” enzyme blends for each individualdonor pancreas. Understanding the com-plex enzyme milieu of the dissociation pro-cess has the potential either of creating di-rect benefit to the islet transplant commu-nity, or to offer substantial benefits to thescientific community’s understandings oftissue dissociation as a whole. Using newcollagenase from marine organisms will bean evident opportunity to optimize the pro-cedure of tissue dissociation applications.Cell therapy was successfully used to re-build damaged cartilage in joints, repairspinal cord injuries, strengthen a weakened

2569

Technologies

immune system, treat autoimmune diseasessuch as AIDS, and help patients with neu-rological disorders such as Alzheimer’sdisease, Parkinson’s disease, and epilepsy.Further uses showed positive results in thetreatment of a wide range of chronic condi-tions such as arteriosclerosis and congeni-tal defects. The optimization of islet purifi-

cation from rat and porcine pancreas, suchas other cell types from other tissue/organs,using new enzyme at different ratio will beapplied to other organs in order to isolateprogenies cells for cell therapy application.Moreover, this technique could be suitablein the emerging technology of scaffold en-gineering medicine.

References[1] C. Gerday, M. Aittaleb, M. Bentahir, J.P. Chessa, P. Claverie, T. Collins,

S. D’Amico, J. Dumont, G. Garsoux, D. Georlette, A. Hoyoux, T. Lonhienne, M.A.Meuwis, and G. Feller. Cold-adapted enzymes: from fundamentals to biotechnol-ogy. Trends Biotechnol, 18:103–107, 2000.

[2] H.G. Kristinsson and B.A Rasco. Biochemical and functional properties of Atlanticsalmon (Salmo salar) muscle proteins hydrolyzed with various alkaline proteases.J. Agric. Food Chem, 48:657–666, 2000a.

[3] H.G. Kristinsson and B.A. Rasco. Fish protein hydrolysates: production, biochem-ical, and functional properties. Crit Rev. Food Sci. Nutr, 40:43–81, 2000b.

[4] Kishimura, H, Hayashi, K., Miyashita, Y., Nonami, and Y. Characteristic of trypsinsfrom the viscera of true sardine(Sardinopsm melanosticus) and the pyloric ceca ofarabesque greenling (Pleuroprammus azonus). Food Chemistry, 97:65–70, 2006.

[5] S. Klomklao, S. Benjakul, W. Visessanguan, H. Kishimura, B.K. Simpson, andH. Saeki. Trypsins from yellowfin tuna (Thunnus albacores) spleen: purificationand characterization. Comp Biochem. Physiol B Biochem. Mol. Biol., 144:47–56,2006b.

[6] S. Klomklao, S. Benjakul, W. Visessanguan, H. Kishimura, and B.K. Simpson.Purification and characterization of trypsin from the spleen of tongol tuna (Thunnustonggol). J. Agric. Food Chem, 54:5617–5622, 2006a.

[7] F.J. Castillo-Yanez, R. Pacheco-Aguilar, F.L. Garcia-Carreno, and M.L Navarrete-Del Toro. Isolation and characterization of trypsin from pyloric caeca of Montereysardine Sardinops sagax caerulea. Comp Biochem. Physiol B Biochem. Mol. Biol.,140:91–98, 2005.

[8] G.N. Rudenskaya, Y.A. Kislitsin, and D.V. Rebrikov. Collagenolytic serine proteasePC and trypsin PC from king crab Paralithodes camtschaticus: cDNA cloning andprimary structure of the enzymes. BMC. Struct. Biol, pages 4– 2, 2004.

2570


[9] R. Margesin and F. Schinner. Biodegradation of diesel oil by cold-adapted mi-croorganisms in presence of sodium dodecyl sulfate. Chemosphere, 38:3463–3472,1999.

[10] R. Amarowicz and F. Shahidi. Antioxidant activity of peptide fractions of capelinprotein hydrolysates. Food Chemistry, 58:355–359, 1997.

[11] S.K. Kim, Y.T. Kim, H.G. Byun, K.S. Nam, D.S. Joo, and F. Shahidi. Isolationand characterization of antioxidative peptides from gelatin hydrolysate of Alaskapollack skin. J. Agric. Food Chem, 49:1984–1989, 2001.

[12] E. Mendis, N. Rajapakse, H.G. Byun, and S.K. Kim. Investigation of jumbo squid(Dosidicus gigas) skin gelatin peptides for their in vitro antioxidant effects. LifeSci, 77:2166–2178, 2005.

[13] G.N. Rudenskaya, A.M. Shmoilov, V.A. Isaev, A.V. Ksenofontov, and S.V. Shvets.Aminopeptidase PC from the hepatopancreas of the Kamchatka crab Paralithodescamtshatica. Biochemistry (Mosc. ), 65:164–170, 2000.

[14] W.T. Chen, T. Kelly, and G. Ghersi. DPPIV, seprase, and related serine peptidasesin multiple cellular functions. Curr. Top. Dev. Biol, 54:207–232, 2003.

[15] K. Watanabe. Collagenolytic proteases from bacteria. Appl. Microbiol. Biotechnol,63:520–526, 2004.

[16] E. Leikina, M.V. Mertts, N. Kuznetsova, and S. Leikin. Type I collagen is thermallyunstable at body temperature. Proc. Natl. Acad. Sci. U. S. A, 99,:1314,1318, 2002.

[17] O. Matsushita, K. Yoshihara, S. Katayama, J. Minami, and Okabe andA. Purifi-cation and characterization of Clostridium perfringens 120-kilodalton collagenaseand nucleotide sequence of the corresponding gene. J. Bacteriol., 176:149–156,1994.

[18] O. Matsushita, C.M. Jung, J. Minami, S. Katayama, N. Nishi, and A. Okabe. Astudy of the collagen-binding domain of a 116-kDa Clostridium histolyticum colla-genase. J. Biol. Chem, 273:3643–3648, 1998.

[19] O. Matsushita, C.M. Jung, S. Katayama, J. Minami, Y. Takahashi, and Okabe andA.Gene duplication and multiplicity of collagenases in Clostridium histolyticum. J.Bacteriol., 181:923–933, 1999.

[20] K. Yoshihara, O. Matsushita, J. Minami, and Okabe andA. Cloning and nucleotidesequence analysis of the colH gene from Clostridium histolyticum encoding a col-lagenase and a gelatinase. J. Bacteriol., 176:6489–6496, 1994.

[21] T. Toyoshima, O. Matsushita, J. Minami, N. Nishi, A. Okabe, and T. Itano.Collagen-binding domain of a Clostridium histolyticum collagenase exhibits abroad substrate spectrum both in vitro and in vivo. Connect. Tissue Res.,42:281,290, 2001.

2571

Technologies

[22] L. Ravanti and V.M. Kahari. Matrix metalloproteinases in wound repair. Int. J.Mol. Med, 6:391–407, 2000.

[23] G. Ghersi, Q. Zhao, M. Salamone, Y. Yeh, S. Zucker, and W.T. Chen. The Pro-tease Complex Consisting of Dipeptidyl Peptidase IV and Seprase Plays a Role inthe Migration and Invasion of Human Endothelial Cells in Collagenous Matrices.Cancer Res, 66:4652–4661, 2006.

[24] G. Ghersi, W. Chen, E.W. Lee, and Z. Zukowska. Critical role of dipeptidyl pep-tidase IV in neuropeptide Y-mediated endothelial cell migration in response towounding. Peptides, 22:453–458, 2001.

[25] J.S. Russell, H. Chi, L.E. Lantry, R.E. Stephens, and P.E Ward. Substance P andneurokinin A metabolism by cultured human skeletal muscle myocytes and fibrob-lasts. Peptides, 17:1397–1403, 1996.

[26] P. Proost, M. De, I, D. Schols, S. Struyf, A.M. Lambeir, A. Wuyts, G. Opde-nakker, E. De Clercq, S. Scharpe, and J. Van Damme. Amino-terminal trunca-tion of chemokines by CD26/dipeptidyl-peptidase IV. Conversion of RANTES intoa potent inhibitor of monocyte chemotaxis and HIV-1-infection. J. Biol. Chem,273:7222–7227, 1998.

[27] C. Fougere-Deschatrette, T. Imaizumi-Scherrer, Strick-MarchandH., S. Morosan,P. Charneau, D. Kremsdorf, D.M. Faust, and M.C. Weiss. Plasticity of hepatic celldifferentiation: bipotential adult mouse liver clonal cell lines competent to differ-entiate in vitro and in vivo. Stem Cells, 24:2098,2109, 2006.

[28] A.P. Beltrami, K. Urbanek, J. Kajstura, S.M. Yan, N. Finato, R. Bussani, B. Nadal-Ginard, F. Silvestri, A. Leri, C.A. Beltrami, and P. Anversa. Evidence that humancardiac myocytes divide after myocardial infarction. N. Engl. J. Med., 344:1750–1757, 2001.

[29] C. Ricordi, P.E. Lacy, E.H. Finke, B.J. Olack, and D.W. Scharp. Automated methodfor isolation of human pancreatic islets. Diabetes, 37(4):413–420, 1988.

2572

Antifouling Coatings in Submarine Structures

V. Romairone, I. Trentin, G. LucianoInstitute of Marine Sciences, CNR, Genova, [email protected]

Abstract

It’s well known that metals are used in underwater structures (i.e. the hulls ofships or different types of boats, offshore marine platforms, the probes of tools forthe detection of oceanographic data) are subject to quick deterioration due both tochemical reaction of the inorganic salts contained in sea water and to biochemicalaction caused by settlement of marine organisms (vegetal organism and animals).The present paper describes the most common protection systems, developed forcenturies, through the use of based copper anti-fouling paint.This element, thanks to its toxicity, inhibits the settlement of larvae of more macro-scopic organisms considered the most responsible for the functional deterioration ofthe structure used in the marine field.

1 Introduction

Sea water is a universe for the multitude oforganisms, plants and animals that live init and are part of the ”pelagic domain” orplankton. When wood, metal or otherwise,are immersed in water and come into con-tact with such biological entities there is afast deterioration. A clear example may bethe hulls of boats, offshore piers, marineplatforms, oceanographic probe data etc.Many of these organisms settle and startimpairing the functionality and integrity ofthe structures affected by their chemical-biological action.This phenomenon is related to the life cycleand biological processes that involve thepassage from larval stage to adult benthic.The ”benthic domain” is the aspect of ma-rine biology which includes all organismsthat proliferate and bind to natural and arti-ficial hard surfaces.Fouling involves the establishment of theseorganisms to artificial substrates in marine

structures and in a wide range of degreesdepending on the severity of several pa-rameters of the environment under exam-ination.One of the most important aspects is re-lated to seasonal changes.Seasonal changes involve an alteration ofschedule on the periods of alternatinglight-dark cycles, gradients of temperature,salinity and human activity impact.It’s also important to study the effect con-nected to climatic differences of the vari-ous areas of the world which are related tothe type of organisms.Eutrophic state of waters and the bathymet-ric gradient, through which one has the ab-sorption of light rays, are more related tothe quantity of the organism we can find inthese environments.The formation of fouling on structures as awhole occurs almost exclusively when theyare in terms of static or slight motion of nomore than 2 knots [1].


Technologies

Figure 1: On the left: pelagic larvae of benthic crustaceans (Barnacles). On the right:establishment and growth of fouling on unprotected samples (Barnacles, Polychaetes,Mussels, Oysters).

Figure 2: Fouling problems highlighted in Shipbuilding.

The problems caused by fouling are varied,with significant repercussions on securityand cost management systems.

A crusted hull means weight gain andchanging in the frictional sliding in wa-ter, with negative influences on navigationspeed and fuel consumption.For an offshore platform, the fouling on theestablished pillars of support, can pose aserious threat to its stability; increases inweight and volume of the structure, affect-ing resistance to waves and extreme situa-tions and can lead to its collapse of the en-tire structure.Fouling inevitably causes damage to toolsthat rely on sensors causing alteration todata gathering. The optical parts of thesensors can, for example, be taken out of

service for a quick opacification, while theformation of a thick limestone barrier foul-ing alters the conductivity of electrodes, orblocks the movement of moving parts.The most common way to protect fromsuch organisms is the use of chemicals(biocides) that inhibit larval settlement offouling level. The scheduled mechanicalcleaning of areas affected by biofouling isalso a good practice to protects submersedmanufacts.Antifouling paints are an effective means,especially in the shipbuilding sector, tofight this problem. They contain foulingbiocides, including copper and its deriva-tives (one of the former most used and pop-ular biocides).

2574


Figure 3: Leaching rates of paints containing various amounts of cuprous oxide plottedagainst the time of immersion in the sea. The graph refers to the cuprous oxide contentas percentage dry weight of the paint film.

2 History and Methods

The formulation of antifouling paint ismade by mixing a polymer matrix, solvatedpigment, biocides and other minor ingredi-ents. Biocides, after drying of the protec-tive film, dried, are able to free themselvesand gradually dissolve in the sea.The dissolving allows the formation of atoxic barrier on the surface of the liquid-solid contact that prevents organic settle-ments.How the biocides diffuse, depending on thephysical and biological factors of water,hydrodynamism, temperature, pH and thecomposition of the biofilm (bacteria andmicroalgae) is an intense field of study.The more traditional economic antifouling

paints were usually prepared by dispersionof the toxic substances (mainly cuprous ox-ide) in a mixture of vinyl resins (insoluble)and rosin (resin plant soluble).Cuprous oxides, in contact with sea water,are quickly converted in the bivalent form,the chlorinated and soluble one, and moretoxic for marine organisms [2]:

12Cu2O + H + 2Cl− = CuCl−2 +

12H2O

Such paints do not show long lastingperformances since the biocide exhaustsrapidly due to the irregular fast film con-sumption.The graph of Figure 3 reported from thebook: “Marine fouling and is prevention”[3] summarized the performance of theseproducts: in the starting and ageing test

2575

Technologies

months, a high speed rate of copper releaseis observed and 6 months later the productslost their antifouling efficacy; the obviousconsequence was the strong impact on theenvironment and the short durability of thepaint.The following generation of paints wasdeveloped about 40 years ago using newresins which were reinforced by more hy-drophobic polymers in order to decreasethe water permeability and obtain then athinner layer, less roughness of the exter-nal surface and a higher biocide dispersioncontrol in the water.In addition to traditional copper derivativebiocideans, tin can be used as a biocide. Itcan be employed in a similar way to cop-per.Tin paints releases their charge via a hy-drolysis reaction with sea water (slightlyalkaline). The efficacy of such productswas also made to last for several years.Resins, can drive the gradual release of bio-cides by hydrolytic reaction.They generally consist of acrylic polymersand are endowed with an ester bond withan organo-metallic group.These products are particularly effectivedue to their relatively low cost, reduce thecosts of plant maintenance and in particu-lar those of ships berthed in dry dock.Every new product design must also beevaluated depending on any negative con-sequences that may result from its use, as-sessing the possible risks to the environ-ment. Organic-compounds, introduced inpaints around ‘70s, used till ‘90s and nowbanned, caused severe damage to the eco-system due to the lack of environmentalimpact in their design.Finally, in the latest generations of an-tifouling paints, the dissolution speed con-trol of copper ions takes place in syn-ergy with the hydrolysis of the resin, gen-

erally consisting of acrylic polymers en-dowed with an ester bond with organo cop-per group.Afterwards, the formulations based onorgano-silyl-acrylate polymers showing ahigher hydrophobia and a more gradual hy-drolysis process were further developed.These products should be more durablesince this is proportional to the applied filmthickness consumption speed rate.Nevertheless, there are still some doubtsabout their efficacy in those areas whichare sensitive to a dramatic fouling attack.In fact, their matrix organic radicals do notshow toxicity which is comparable to theorgano-tin ones of the old self-polishingpaints. These antifouling paints are knownas self-polishing tin-free.Currently, to increase the effectiveness ofthese latest generation paints, other bioci-dal products have been developed whichare generally organic in nature; they arecalled stiffeners. Before being put on themarket, these products follow a strict tox-icological protocol, the ultimate goal is tocertify the suitability of the biocide to beeffective only on the target organisms. Theenvironmental impact is therefore limiteddue to the different characteristics requiredsuch as to be readily degradable.

3 Experiments

Copper and its derivatives therefore con-tinue to be the major active ingredientsused in many anti-fouling paints. How-ever, increased sensitivity to the environ-ment requires a particular attention to this.Copper is present in microscopic amountsin living organisms and plays functionalroles in specific biological processes, butin higher concentrations, it becomes toxicand its accumulation in marine sediments,

2576


Matrix type Initial 15 days immersion 7 months laterCu++ thickness Cu++ thickness

µg·cm−2 · d−1 loss µm µg·cm−2 · d−1 loss µm1) Ablative 65.7 10.4 44.6 19.72) Ablative 52.2 10.3 28.1 18.53) Hydrolized 53.9 0.8 22.5 6.04) Insoluble 139.8 0.0 64.8 0.0

Table 1: Ageing test of the paints on some samples immersed in the sea, the parameterstaken into account are: the biocide released and the loss of the film thickness.

Figure 4: Testing of antifouling paints, coated specimens are immersed in the sea for thenatural ageing of the products under examination.

along with other heavy metals could causeserious environmental problems [4, 5]. Itsuse in antifouling paints must therefore berestricted and controlled through carefulmonitoring.The evaluation of the copper ions, releasedin the water as a function of time, is there-fore, an extremely important parameter forexperiments of antifouling products.Quantitative evaluation is important bothat the very beginning of the immersion ofthe specimens and during the various dura-bility, but also to evaluate in quantitativeterms, the potential impact which such abiocide may have on the marine environ-

ment.Currently, after the well known IMO direc-tive on the ban of organo-tin biocides, it ispossible to increase the use of copper basedpaints with a high occurrence rate of this el-ement in the marine ecosystem.As was said above, even though the latestgenerations of paints should no longer re-lease the same content rate of this elementas those which were used a long time ago,the copper quantity introduced in the sea,also from other polluting sources in the ur-ban and industrial discharges, are remark-able. Particularly as for the sea coast andharbour areas, the antifouling paints are an

2577

Technologies

important impact source. From surveyscarried out at the CNR-ISMAR laboratoryin Genoa, it was observed how some dailyused soluble matrix paints (ablative) re-lease in the sea every day in the first sevenservice months, amounts of copper rang-ing from 65.7 to 22.5 µg per surface squarecentimetre; the paints had been applied onageing tested specimens by a static immer-sion in the sea. The quantities of such an el-ement, as shown in Table 1 are rather highin comparison with the threshold valuesaccounting for 10µg·cm−2 · day−1 where,via experiments, in the past, the thresholdof the settlement possibility of the foulinghad been remarked.With longer paint ageing times, generally,such values further decrease, maintainingmore constant till the total exhaustion ofthe applied film (this happens in the caseof paints soluble, self-eroding and self-polishing based matrices. In the case of theinsoluble matrices till the total exhaustion,due to the escaping of the biocides insidethe film through the pores of the polymerstructures).The threshold values which have been ob-served more recently, result in an increasedaccounting for about 20 µg·cm−2 · day−1.Such an increase could let one think of ahigher and higher resistance or selectionof copper-resistant organisms due to thehigher occurrence of this element in the en-vironment.If this were true, the future of these prod-ucts could deem a reduction of the efficacyand/or a stronger biocide impact on the en-vironment.In fact, it often happens that the prod-ucts designed and applied to the hulls withsuch a thickness last more than three years,and six months later, they lose their effi-cacy, they have not exhausted all the cop-per rate in the formulation, neither has the

film thickness disappeared. In this case, thecauses can depend either on some formula-tion mistakes or on the adhesion of very re-sistant micro-organisms, which hamper thewater chemical and physical processes inthe solid-liquid interface; in both cases, thetoxic barrier, needed to avoid the biologicencrustation is lacking.When such cases happen, lengthening thedry-docking operations is always difficult,so, if the substrate is still solid enough,the new antifouling paint is usually laidover the old one, but in the end, layer bylayer, the thicknesses will be more andmore noticeable, with a greater externallayer roughness and loss of solidity due towater hydration; sooner or later, at best,they will have to be removed at high costsdue to their disposal as toxic wastes.The evaluation of the release speed rate ofone antifouling active principle is carriedout during specific ageing steps of the paintapplied on the specific panels. To evalu-ate the copper content rate released by thepaint, the specimens are taken and tempo-rary immersed in given volumes of syn-thetic or natural sea water; then the testgoes on according to the given methods [6].The specimens can be subjected to the nat-ural weather test by static exposure in thesea (long times, Figure 4) or to the accel-erated test (short times) in lab basins or di-rectly in the sea using test equipment ableto provide a strong hydrodynamism on thespecimens themselves (rotor tests).Defence against fouling, on the pylons ofan offshore platform is generally done withregular mechanical cleaning, in order tooptimize the work and lengthen the timebetween cleanings [7].There are other systems such as the use ofvarious biocides, chlorine developed elec-trolytically from sea water, underwater ap-plication of special anti-fouling paints or

2578


Figure 5: Offshore platform for the extraction of methane.

other, but while enjoying an undoubted sci-entific interest, they are almost never im-plemented due to various problems such ashigh costs and environmental impact.The probes or measuring instrumentswhich operate immersed in the sea re-quire cleaning from fouling. Organic anti-fouling coatings can be used but for others,even if this type of transparent barrier mayinterfere with the measurement.Therefore, the most popular system isthe periodic mechanical cleaning, the fre-quency obviously depends on the densityand distribution of spores or larvae of or-ganisms. In oligotrophic and deep sea div-ing the short time scale of the phenomenonmay be less obvious.Where possible the use of copper and itsalloys, such as: copper-nickel, copper- tinand copper-zinc less prone to fouling issuggested [8]. The settlements are notcompletely inhibited but merely sloweddown, also no form of galvanic protectionmust be activated on the system and itemsmust be in a position to release toxic ionsin water.

4 Conclusions

As was said at the beginning of this paper,antifouling paints are copper based ones;but at present, although several pollutionphenomena have been highlighted, and al-though it is thought that among the varioussources, antifouling paint is rather mean-ingful, a total ban forbidding the use of thisbiocide in antifouling paints is not feasi-ble [9]. It might be possible, on the con-trary, to aim at controlling and limiting itsrelease rate in the environment. In fact,some countries, such as Canada, have al-ready established the threshold accountingfor 40 µg·cm−2 · day−1, while Sweden, asfor antifouling paints for the pleasure boatsector, issued a total ban in the local seasand the threshold of 200 and 350 µg·cm2

respectively for the external seas during theinitial 14 to 30 immersion days.Many of the paints which were examinedat ISMAR-CNR, show a good antifoul-ing performance, but some show a cop-per impact which gives rise to some doubts(e.g. sample no. 4, Table 1). The speci-mens treated with this product prove to bethe best ones but one could ask whether

2579

Technologies

it is worthwhile to face all the possibleenvironmental problems in view of whatis reported in the international literature[10, 11].A general rule should consist of keepingthe release speed rate slightly higher thanthe threshold values, and, of course, thesecan vary according to the geographic char-acteristics of the various regions, to the cli-matic conditions, to the quality and quan-tity of the fouling, to their food availabil-ity and to the polluting sources themselveswhich can select the most resistant microorganisms and fouling species.Finally, antifouling paint performanceshould be evaluated keeping clearly inmind three well known properties, thesebeing performance, durability and environ-

mental impact, but to this latter factor it isnecessary to devote the greatest attentionand bear the increase in the related costs.The feasible alternatives based on “green”systems, which avoid the release of toxicsubstances are tested with good results.The most promising is the use of protec-tive polymers with low surface free energysuch as silicone resins where the antifoul-ing effect is exercised through physical ac-tion to prevent a strong adhesion encrust-ing organism. The cleaning effect is furtherenhanced by friction with the water whenthe structure so protected begins to move[12]. The development of these antifoulingproducts, however, requires further investi-gations.

References[1] A. Mollica and A. Trevis. La prevenzione del fouling marino alla superficie delle

condotte per adozione di opportune velocita di corrente. L’energia elettrica, 10:654,1972.

[2] C.O’D. Iselin. The Physical Chemistry of Compounds of Copper and Mercury andTheir Interactions with sea Water. Marine Fouling and its Prevenction. Woods HoleOceanographic Institute, Massachusetts. United States Naval Institute Annapolis,Maryland. 15:270, 1952.

[3] C.O’D. Iselin. Mechanism of Release of Toxics from Paints. Marine Fouling andits Prevenction. Woods Hole Oceanographic Institute, Massachusetts. United StatesNaval Institute Annapolis, Maryland. 16:278, 1952.

[4] A. Katranitsas, J. Castritsi-Catharios, and G. Persoone. The effects of a copper-based antifouling paint on mortality and enzymatic activity of a non-target marineorganism. Marine Pollution Bulletin, 46:1491–1494, 2003.

[5] J. Warnken, J.K. Dunn Ryan, and P.R. Teasdale. Investigation of recreational boatsas a source of copper at anchorage sites using time-integrated diffusive gradientsin thin film and sediment measurements. Marine Pollution Bulletin, 49:833–843,2004.

[6] ASTM. Standard Test Method for determination of Copper Release Rate fromAntifoulig Coating Systems in Artificial Seawater. Designation: D 6442-03. 2003.

2580


[7] G. Relini, S. Geraci, M. Montanari, and V. Romairone. Variazioni stagionali delfouling sulle piattaforme off-shore di Ravenna e Crotone. Bollettino di Pesca, Pis-cicoltura e Idrobiologia, 31(1/2):227–256, 1976.

[8] C.O’D. Iselin. Marine Fouling and its Prevenction. Woods Hole Oceanographic In-stitute, Massachusetts. United States Naval Institute Annapolis, Maryland. 21:350,1952.

[9] A. Helland and T. Bakke. Transport and sedimentation of Cu in a microtidal estuary,SE Norway. Pollution Bulletin, 38(6):448–462, 2002.

[10] B. Jones and T. Bolam. Copper speciation survey from UK marinas, harbours andestuaries. Marine Pollution Bulletin, 54:1127–1138, 2007.

[11] R.F. Brady Jr. and I.L. Singer. Mechanical factors favoring release from foulingrelease coatings. Biofouling, 15(1-3):73, 2000.

[12] L.D. Chambers, K.R. Stokes, L.C. Walsh, and R.J.K. Wood. Modern approachesto marine antifouling coatings. Surface & Coatings Technology, 201:3642–3652,2006.

2581

Technologies

2582

Corrosion Behaviour of Al Based Materials for theNaval Industry

P. Traverso, G. Luciano, P. LetardiInstitute of Marine Sciences, CNR, Genova, [email protected]

Abstract

This paper shows the results of the investigation on the corrosion behaviour of in-novative Al- based composite materials. These classes of materials are generally usedfor engineering applications because of their technological properties as well as rea-sonable cost. Free corrosion and electrochemical tests were carried out in quiescentnatural sea water for different exposure times, with the aim of evaluating corrosionattack amount and morphology, general corrosion resistance, as well as susceptibilityto pitting corrosion. The corrosion layer was also characterised by different surfaceanalysis techniques.

1 Introduction

Aluminium based materials are often usedfor engineering applications because oftheir density, hardness, corrosion resis-tance properties as well as thermal andpower conductivity, light and heat re-flectance, non toxicity and non magnetismas well as their reasonable costs.Their use in the shipbuilding industry,which in the beginning was limited toequipment components, is now spreadingto the field of structural parts and boat hullswhich are expected to be very light for per-formance reasons (competition motorboatsand hydrofoils) as well as for their specialuse (lifeboats). Often the hulls and veryseldom the critical areas are not coated, forexample, in the vicinity of the propeller,they are provided with cathodic protectionby zinc sacrificial anodes. Pure aluminiumshows low elongation and hardness resis-tance, so, even though the sea corrosion re-sistance is high, except for special applica-

tions, in practice it is found under the formof an alloy together with other elementswhich, nevertheless can influence the cor-rosion behaviour.The resistance of the aluminium alloys tothis failure is due essentially to a chemi-cally inert layer formation on the exposedsurface. In sea-water, because of the ag-gressiveness of the chlorine ions this pro-tective film tends to break more easily inthe preferential attack sites such as defectsor heterogeneity spots giving rise to local-ized damage.The objective of this work was:

a) To carry out a study of the corrosionbehaviour of some technologically ad-vanced aluminium based materials so asto highlight which of them shows thehighest resistance to failure occurrencedue to the sea environment;

b) Once a class of material has been se-lected, being by principle alternative totraditional ones, to point out the rele-vant parameters (composition, produc-


Technologies

tion process, etc.) which are able to playa pivot role in corrosion resistance.

Especially, with regard to technologicallyadvanced materials, we tested the Al-Li8090 alloy (Al-2.5 Li-1.5 Cu) and Al 6061T6/Al2O3p and Al 359/SiCp compositematerials (Table 1).Al-Li 8090 T81 is widely used in theaeronautical and aerospace fields [1, 2]as it shows very high physical-mechanicalproperties. The addition of lithium toaluminium, due to their highly differentatomic mass (6.49 over 26.98) causes a 3%density reduction (up to max 4%) and a 6%elasticity increase [3, 4], by each percent-age point added by weight, but generally itresults unfavourable electrochemically be-cause of its highly electronegative prop-erty. This alloy also contains a copper con-tent rate accounting for about 1.5% whichgreatly improves the mechanical character-istics but makes it more sensitive to corro-sion, especially the inter-granular type [5].We should always bear in mind that the cor-rosion phenomena, with the same compo-sition, depend on the microstructural situa-tion of the alloy which is strictly related tothe thermal treatments undergone.The metal based composite materials con-sist of heterogeneous systems where anon-metal phase is dispersed in an alloy,in this specific case an aluminium basedone. Such composites having a lower den-sity than the metallic alloy matrix, gen-erally show comparable mechanical prop-erties [6, 7], which make them economi-cally competitive for special uses (for ex-ample in the aeronautical and automotivefields). In these applications indeed, adecrease in density, obviously leading tolower fuel consumption, plays a specialrole. However, the heterogeneity of thematerial could compromise the service life,as it may spur its failure taking place af-

ter the corrosion [8, 9] or a mechanicalstress. The corrosion resistance of com-posites is strictly linked to the non-metalphase type, dispersed in the matrix and toits capability by diffusion of forming inter-metal compounds at the matrix-particle in-terface which prevent its peeling. For ex-ample, the alumina particles in those com-posites having a metal matrix based onaluminium alloys do not affect the local-ized corrosion sensitivity of the binder be-cause the corrosion products formed on thesurface are aluminium oxides. Since thelatter have the same electrochemical po-tential as the alumina particles, they can-not give rise to microgalvanic elements be-tween the matrix and the reinforcing ma-terial. On the contrary, the SiC particlesprovided with a nobler equilibrium poten-tial than the aluminium matrix [10], makea galvanic micro circuit with the matrix it-self, bringing about its dissolution. Havingobtained encouraging results with the Al6061 T6/Al2O3p type composite materialswe changed some characteristic parameterswithin this class (reinforcement rate, pro-duction process) and tested the influence ofthis parameter on the corrosion behaviour[11].

2 Materials and experi-mental methods

The samples used consisted of 15 mm di-ameter and 1.5 mm thick discs, polishedup to 800-grade emery paper and degreasedwith acetone before dipping them into thecorrosive solution.Some specimens used for the microstruc-tural analysis were polished up to diamond-paste (1 micron) and subjected to met-allographic attack to highlight their mi-

2584


Specimen Al Si Fe Cu Mg Cr Zn Ti Li Mn Other % reinforcing v/v (particles)

Al-Li 8090 T81 (#) rem 0.10 0.15 1.50 1.50 0.05 0.12 0.05 2.50 0.05 0.18 _ Al 6061 T6 /Al2O3 rem 0.90 1.10 0.25 1.00 0.20 ---- --- --- --- --- 10%Al2O3(*) Al 359 /SiC rem 9.50 0.20 0.20 0.65 ---- ---- --- --- --- --- 20% SiC

Table 1: Nominal composition of the alloys and composite materials tested.(#) = This alloy was tested and also subjected to thermal treatment T3.(*) = This composite material was tested by two reinforcement content rates (10 and 20%v/v), obtained from different production processes (cast and extrusion).

crostructure (grain size and differentphases of the metal binder).In all the tests carried out (both for suchtest samples and the surface treated ones)we used a natural sea corrosive water solu-tion (pH = 8.2, Mohr chlorine content rate21.4 %, dissolved oxygen = 7.5 ppm) at T= 25°C, P= 1 atm and static flow-dynamicconditions.The experimental trials were electrochemi-cal tests and free corrosion ones with vari-ous corrosion times.The electrochemical tests consisted of po-tentiodynamic DC polarizations (anodicand cathodic) through which the corrosionaverage speed rate and the localized corro-sion sensitivity were evaluated.The free corrosion tests were performedfor time ranges from 120 to 1440 hours.Through this type of test, the weight losswas measured by summing the content rateof the aluminium in the solution (underthe form of soluble corrosion products)with the amount of Al corrosion productsadhering to the metal substrate and dis-solved using a proper method which al-lows one to keep the metal matrix steady[12]. The quantity determination of thealuminium was carried out via spectropho-tometry by atomic absorption and usinga Perkin Elmer spectrophotometer (AAna-

lyst 100 model).The corrosion products were characterisedvia X-ray photoelectron spectrometry andAuger spectrophotometry using a VG spec-trophotomer (Escalab 210 model). TheXPS analyses were carried out using MgKa radiation (E=1253.6 eV) as excitationsource and the spectra were obtained ata vacuum better then 10−7 mbar and de-tection angle perpendicular to the surface.The C1s peak (Binding energy, B.E. =284.8 eV) accounted for the internal stan-dard which persisted as it was due to thesurface contamination, which was used toevaluate and compensate the charging ef-fects of the specimens.Some Auger mapping was also carried outon corroded test samples to model the sur-face corresponding to the areas where peel-ing or localized corrosion had occurred.Via free corrosion tests we obtained in-formation concerning the corrosion mor-phology and intensity using a Leica met-allographic microscope (DM/RME model)combined with an image analysis system;the occurrence of segregations, disbond-ments of the metal matrix following thelocalized corrosion or peeling of reinforc-ing particles in the composite materialswas tested through X-ray micro diffrac-tometry using a Rigaku DC-MAX microd-

2585

Technologies

iffractometer. Other data were obtainedby means of an atomic force microscope(AFM; Burleigh Metris) working in con-tact mode with an Si3N4 tip on a 70x70 µmarea.


3.1 Corrosion behaviour of in-novative materials based onAl

Al-Li 8090 T81 (and T3 thermal treat-ment) alloyWith reference to the study of corrosion be-haviour in the sea environment of the Al-Li 8090 T81 alloy compared with conven-tional ones, we can state that the corrosionresistance proves that it cannot match thehigh physical-mechanical properties whichcharacterise it [13]. The addition of an ele-ment with a much more negative equilib-rium potential such as lithium (the maincomponent of the alloy), but above all, thehigh copper content and the microstruc-tural heterogeneity favoured localized ag-gression particularly the interstitial type(crevices).Free corrosion tests showed that after aboutone month’s exposure to sea water, thecrevices can even become perforating ontest samples about 1.5 mm thick (Figures1 and 2). In this area red crystals weredetected microscopically which turned outto be metal copper by microdiffractometricanalysis. The complex form of the peakshighlights additional components corre-sponding to larger sizes of the crystallo-graphic pure copper cell which may be re-lated to a new plating after dismutation ofthe copper solution on the pre-existing oneon the substrate [14]. The highly localizedsurface copper (inter-metal phases, inclu-

sions) can give rise to galvanic microcellswhich can concentrate and increase the cor-rosion attack. The sensitivity of the local-ized attack is lower for the Al-Li8090 inthe T3 thermal phase as no ageing ther-mal treatment favours the steady diffusionof the addition elements [15, 16].The Al-Li 8090 T81 also highlighted thepreferential dissolution of the lithium; theLi/Al ratio between lithium and the alu-minium in oxidised solution is higher thanthe Li/Al ratio found in the alloy as such.Actually, a large part of lithium may sol-vate because of its very low standard elec-trochemical potential (-3.045 V; no nobleelement) depleting the metal binder andhampering the protection corrosion prod-ucts based on aluminium; such an effectprevails for short exposure periods (up toabout 100 hours).The marine corrosion resistance of the tra-ditional Al alloys (especially for the Al5000 type ) was higher, although all ofthem are sensitive to pitting or interstitialcorrosion.In order to reduce the grafting occurrencerate of the crevice it is possible to inter-vene in the piece moulding (avoiding in-dentation, slots or porous welding), and inthe assembly operations (avoiding crackswhich favour infiltrations), as well as inthe shielding phenomenon (of the precip-itated corrosion products or possible exter-nal agents).The analytical determination, performedby several analytical techniques, enlight-ened hydroxy, hydroxy-oxide and alu-minium oxide as prevailing compounds onthe surface passivating layer.Al 6061 T6/Al2O3p composite materialThe analysis of electrochemical resultsshows by increasing the exposure time, thecorrosion kinetics of the compound de-pends not only on the chloride concentra-

2586


Figure 1: Macro photo of a corrosion crevice on sample of Al-Li 8090 T81 alloy after 1month of exposure in sea water.

tion acting at the metal-solution interface,but also on other parameters such as thetype of corrosion products which changesas a function of the composition of the cor-rosion solution.In sea water, specimens can undergo lo-calized corrosion. The likelihood of cor-rosion site formation increases with im-mersion time, as from the perfect passiv-ity range value which takes more and morenegative values. The longer the immer-sion times, the higher the passivity rangeof the composite material in the sea waterreaching values higher than those of saltywater, an environment where the passivityrange keeps steady. Such a range, althoughnot very significant, is slightly higher thanthe passivity range shown by the traditionalAl 6061 T6 alloy free of the reinforcementmaterial (always lower than 50 mV) [17].This makes the use of the composite overthe massive alloy competitive.Comparing the weight loss results of thecomposite material in water with thosewhich were obtained from the previous

studies of the Al 6061 alloy (metal matrixof the composite) there is evidence that theweight loss of the alloy is slightly higher(≈10%) than the composite’s. Thereforethe weight loss results confirm that Al2O3

reinforced composite can be consideredcompetitive compared to the massive al-loy. The optical microscopic analysis of thespecimens subjected to the electrochemicaltests and to free corrosion showed, for thespecimen corroded in seawater, deeper andlarger localized corrosion areas. To explainthe origin of such sites better, analyses viaAuger spectroscopy in the surrounding ar-eas of them were performed and it wasfinally concluded that localized corrosionhad not occurred at the metal matrix andthe alumina particle interface, which didnot separate from the metal matrix. On thecontrary, the localized corrosion graftingsites are related to the existing microstruc-tural heterogeneities of the metal binder,for example the Al-Cu phases (since cop-per is one of the minor constituents of thebinder), and they are precipitated as inter-

2587

Technologies

Figure 2: Detail of perforating morphology on same sample (Point A in Figure 1) afterremoving corrosion products.

metal compounds and they are also ca-thodic with respect to the Al alloy and so,they are able to bring about galvanic micro-cells.Al 359/SiCp composite materialElectrochemical parameters obtained fromthe potentiodynamic polarization curvescarried out on the specimens after differ-ent immersion periods characterize the cor-rosion behaviour of the composite over itsgeneral and localized corrosion resistance[18]. From the obtained data it can be con-cluded that:

a) Al 6061 T6/Al2O3 composite. The cor-rosion behaviour could be related to theSiC non-metal phases, as they are ca-thodic to the matrix;

b) the longer the exposure time, the lowerthe composite average corrosion rate insea water;

c) in sea water the composite dissolutioncould be governed by the diffusion pro-cess of the ions through the corrosionproducts whose protective strength in-creases as much as the exposure time fol-

lowing the formation of hydroxides ormagnesium oxycarbonates;

d) there is no passivity range for the com-pound studied throughout the whole ex-posure time. As for the Al 6061T6/Al2O3p, in the case of Al359/SiCpcomposite the likelihood of corrosionsite formation in both corrosion solu-tions increases with the immersion time,as from the PPd perfect passivity rangevalue taking more and more negative val-ues.

The weight loss tests confirm that the Al-SiC composite has an average corrosionspeed rate much higher than the compos-ite with Al2O3 particles. The Auger map-ping carried out on non-corrosive speci-mens has highlighted areas rich in silica(reinforcement particles) and aluminium(metal matrix) but also a phase at thematrix/reinforcement interface, where boththe aluminium and the silica are found to-gether. The analyses executed via X-raydiffractometry gave evidence of the alu-minium silicate phase occurrences (ASTM

2588


card 6-221) at the SiC particles/matrix in-terface (ASTM card 22-1319). The lat-ter consists of a phase rich in Al (ASTMcard 4-787) and of another phase rich inSi (ASTM card 5-565). The aluminiumsilicate phase could be the matrix rein-forcement binder. On the contrary, theAuger mapping, concerning the aluminiumand the silicate respectively, carried outon a specimen exposed for 360 hours insea water, points out two different areas,one rich in aluminium and the other insilica, definitely separated at the matrix-reinforcement interface. The aluminiumsilicate as previously observed at the inter-face was not detected either by the Augeranalysis or by macro diffractometry.The analysis of the corrosion products werecarried out by XPS by which it was possi-ble to obtain a quality and quantity char-acterization of the corrosion film compo-nents. From the qualitative analysis it re-sulted that the surface layer of the compos-ite is made up of aluminium and silica com-pounds. The specimens exposed to sea wa-ter have a high content of magnesium co-precipitated with aluminium oxidised com-pounds (oxides and oxichlorides), Table2 shows the atomic percentage values ofAl, Mg and Si obtained carrying out thequantitative analysis of the surface layersof the composite before and after the cor-rosion attack. From the atomic ratio ofthe different elements the Al/Si ratio wascalculated and reported in the Al/Si ratioon the surface of the specimens exposedchanges with respect to the non immersedmaterials by changing the exposure time.Indeed, by increasing the exposure time,such a ratio increases greatly (by about 30times) on the surfaces of the specimensdipped for 360 hours in sea water. Theincreased Al/Si ratio can be related to thedecrease in the average silica content rate

on the specimen surface. Such a decreaseis due to the disappearance of the com-pounds with Si-O-Me bonds binding en-ergy,(B.E. = 101.9 eV) [19, 20] and belong-ing to the aluminium silicate inter-metalphase at the SiC and matrix interface. Fur-ther on, the drop in the Si-C bond signal(B.E. = 98.0 eV) in the non metal phase ofthe silica carbon could be due to removalfrom the metal matrix following the disso-lution of the carbon-matrix aluminium sil-icate phase. The appearance of the Si-Obond (B.E. = 102.9 eV) of the silica dioxide(SiO2) on the corroded samples shows thata part of the silica belonging to the phaserich of Si (B.E. = 99.0 eV) of the matrixwas oxidised following the corrosion attackin sea water. The removal of the silica car-bon particles and the decrease in its surfaceconcentration can lead to a high homogene-ity of the corrosion products of the com-posite. Furthermore, the magnesium com-pounds make them more massive and pro-tective and able to decrease the corrosionspeed rate of the composite by increasingthe exposure time as was observed in theweight loss tests. The decrease in the Al/Siratio can be related to the selective dissolu-tion of the aluminium which does not affectthe SiC particles and the aluminium silicatephase at the metal matrix interphase.Actually the XPS analysis shows that thesilica is under the form of SiC carbon (B.E.= 98 eV), of a silicate (B.E. = 102.9 eV),and of metal Si (B.E. = 99.4 eV) as well asof SiO2 (B.E. = 103 eV). The latter com-pound could be formed through the oxida-tion of the metal silica following the corro-sion attack.

2589

Technologies

Corrosion solution

Exposure time (hours)

% at Al

% at Mg

% atSi

Al/Siratio

None ----- 69.00 1.36 25.64 2.69 Sea water 120 47.17 51.36 1.18 39.83 360 48.00 49.43 0.50 95.04

Table 2: Al, Mg, Si atomic percentages and Al-Si/Si ratio on Al359/SiC detected on thecomposite surface before and after 120 and 360 hours of exposition in natural sea water.

3.2 Effect of the reinforcementcontent rate and of the pro-duction process on corrosionof the Al 6061 T6/Al2O3p insea water.

With reference to the study of corrosion ofthe Al 6061 composite material in sea wa-ter all the experimental measurements car-ried out allow to know the following con-clusions:

a) by increasing the particle content rate(from 10 to 20% v/v) the corrosion attackis lower (and approximately analogous tothe cast and extruded material);

b) with regard only to the specimen contain-ing 10% reinforcing agent, the extrusionprocess improves the material corrosionresistance;

c) the sensitivity of the composite materialto the localized corrosion tends to de-crease as the reinforcement increases andtogether with the extrusion process.

Cast specimens (in particular specimenscontaining 10% of reinforcement agents),observed by optical microscope, reveal agreat deal of porosities, macro and micro-cavities and a non homogeneous distribu-tion of the reinforcing phase (agglomeratedparticles were often visible).

AFM images (Figures 3 and 4) show anevident physical defect on the uncorrodedspecimen which can function as a nucle-ation site in localized corrosion phenom-ena.

4 ConclusionCorrosion resistance in sea water of theAl-Li 8090 alloy showed not to be up tothe high physical and mechanical proper-ties which makes it a special material. Theaddition of an element provided with anequilibrium potential much more negativethan the aluminium, such as lithium, (themain component of the alloy), but aboveall the high copper content rate as wellas the microstructural heterogeneity haveinduced a remarkable tendency to local-ized attack, particularly the interstitial one.Such a situation highly improves at the T3thermal phase as any ageing thermal treat-ment favours the even distribution of theaddition element.Corrosion resistance in sea water of theconventional alloys tested globally resultedbetter although they are all sensitive to pit-ting and interstitial corrosion.They did not show dealligation occur-rences; on the contrary, the Al-Li 8090

2590


Figure 3: AFM ”top view” image of uncorroded Al 6061T6/Al2O3p 10% cast compositesample.

T81 showed the preferential dissolution oflithium.The aluminium matrix composite materi-als showed a meaningful trend to local-ized corrosion in seawater environments;such a phenomenon mainly comes from thedishomogeneity of the metal matrix and itis only partially affected by problems con-cerning the matrix reinforce particle inter-face.The corrosion behaviour resulted markedlyaffected by the type of reinforcement be-cause the SiC particles having a more no-ble potential than the aluminium matrix incomparison with the Al2O3 such as the sil-ica carbon, can induce corrosion process bya galvanic effect.

Considering the marine environment as acorrosive one and the alumina particles(10% v/v) as a reinforcing agent and if wecompare the results obtained with those re-lated to the traditional Al 6061 T6 alloy(which was the composite material matrixtested) it was pointed out that the corrosionattack on the non-reinforced alloy is higher(by about 10%).Increasing the alumina particles contentrate (from 10 to 20%) a lower corrosiveattack was observed (approximately analo-gous between the cast and extruded prod-uct), while only for 10% reinforced con-taining product, did the extrusion productseem to improve the corrosion resistance ofthe material.

2591

Technologies

Figure 4: AFM ”3D” image of uncorroded Al 6061 T6/Al2O3p 10% cast compositesample.

It is possible, therefore, that a higheramount of reinforcing phase reduces the lo-calized corrosion diffusion forming a phys-

ical type barrier, while the extrusion pro-cess, making the material more homoge-neous, restricts the number of graft sites.

References[1] D. Webster and C.G. Bennet. A d v. Mater. & Proc. 10:49, 1989.

[2] R.J. Rioja and R.H. Graham. A d v. Mater. & Proc. 6:23, 1992.

[3] K.K. Sankaran and N.J. Grant. Mat. Sci. Eng. 44:213, 1980.

[4] E.A. Starke Jr. and T.H. Sanders Jr. J. Metals. 33:24, 1984.

[5] The Marine Corrosion Handbook. Ed. T.H. Rogers - Mc Graw Hill Comp. ofCanada Ltd, Canada 4, 1960.

[6] P. Appendino and M. Montorsi. La Metallurgica Italiana. 10:743, 1986.

[7] M . Fontana, E. Gariboldi, G. Silva, and M. Vedani. La Metallurgica Italiana.10:743, 1986.

[8] A.J. Sedriks, J.A.S. Green, and D.L. Novak. Met. Trans. 2:871, 1971.

[9] P.P. Trzaskoma, E. Mc Cafferty, and C.R. Crowe. J. Electrochem. Soc. 30:1809,1983.

2592


[10] Handbook of Chemistry and Physics. CRC press, Boca Raton, Florida 61st ed.,D:155–159, 1980-81.

[11] A.M. Beccaria, P. Traverso, and F. Gnecco. Pitture e Vernici European Coatings.14:75–92, 2000.

[12] A.M. Beccaria, E. Mor, and G. Poggi. Werkst. und Korr. 34:236, 1983.

[13] A.M. Beccaria and P. Traverso. Atti 2° Convegno Nazionale PFT2, 29-31 Maggio1995 Genova. 1301, 1995.

[14] M. Pourbaix. Corros. Sci. 14:25, 1974.

[15] A.M. Beccaria and P. Traverso. Werkst. und Korr. 47:261, 1996.

[16] P. Traverso and A.M. Beccaria. Br. Corros. J. 32:255, 1997.

[17] A.M. Beccaria, G. Poggi, D. Gingaug, and P. Castello. Br. Corros. J. 29:65, 1994.

[18] F. Gnecco and A.M. Beccaria. Br. Corros. 34:57, 1999.

[19] Z. Szklarska Smialowska and M. Janik-Czachor. Corros. Sci. 11:901, 1971.

[20] D. Briggs and M.P. Seah. Practical Surface Analysis. J. Wiley & Sons, Chichester,UK, page 487, 1984.

2593

Technologies

2594

Coastline Extraction from SAR Images

R. Cossu, M.M. CerimeleInstitute for Applied Mathematics “Mauro Picone”, CNR, Roma, [email protected]

Abstract

The aim of this work is to develop a software tool to monitor the coastline. Theimplemented procedure is constituted by a set of techniques and methods for the SAR(Synthetic Aperture Radar) image segmentation, in order to investigate the state ofconservation of the coastal environment. The proposed approach is based on the“level set” method applied to SAR images. In this method, an initial curve definedon the image evolves according to a PDE (Partial Differential Equation) model bymeans of a velocity related to the characteristics of the image to be segmented. Thiscurve is deformed until it assumes a steady position at the boundary of the area to beextracted from the image. In our case the initial curve includes the coastal area andit moves until it reaches the line that marks the boundary between land and sea. Wehave pre-processed SAR images in order to remove the speckle noise, implementingan anisotropic diffusion filter. The proposed method has the advantage to extract thecoastline from a single image unlike the methods, based on the classic interferometrictechniques, which require two images of the same area acquired at different times orfrom different points of view. We used images SAR (Precise Image) PRI acquiredduring the ERS2 mission.

1 Introduction

Synthetic Aperture Radar (SAR) uses anactive radar antenna which emits modu-lated microwave pulses. The amplitude andthe phase of the returning signals from anEarth surface area are recorded and thencombined. Since microwaves wavelengthis much longer than the optical one, ahuge antenna would be necessary to getacceptable resolutions. This problem hasbeen bypassed by post-processing methodswhich synthesize an antenna of large di-mensions with an antenna of a few metersthat allows us to obtain images of up tosome centimeters of resolution [1].SAR images present advantages and dis-advantages; active microwave energy pen-

etrates clouds, thus the acquired imagesprovide information in inclement weather,during night as well as day. Moreoverthe acquisition in wavelengths outside thevisible and infrared regions of the electro-magnetic spectrum provides informationon surface roughness and moisture con-tent of surface layers of vegetation, sand,snow and on wave properties of sea/ocean[2]. On the other hand SAR images presenta granular effect due to the multiplicativenoise called speckle. Because of sum thecontributions of a large number of scatter-ers, the image pixels have different grayvalues and even for a single surface type.Important gray level variation may occurbetween adjacent pixels producing a gran-ular effect.


Technologies

The aim of this work is to develop a soft-ware tool to observe and monitor the coast-line, i.e. to create a set of techniquesand methodologies for the segmentation ofSAR images in order to investigate the con-servation status of coastal environment. Itis well known that the segmentation is apartitioning of the image into disjoint re-gions and it is a crucial step in field of im-ages interpretation.In our case the developed software is basedon interactive and automatic methodolo-gies for the analysis and processing of re-mote sensing data [3]. The analyzed im-ages are SAR PRI acquired during the mis-sion ERS2 and provided by the Consortiumfor Informatics and Telematics “Innova” ofMatera. ERS2 SAR system is capable of 25m resolution from an altitude of 800 km, atradar wavelength of 5.7 cm.The identification of the coastal profile inSAR images requires a complex method-ology, mainly due to speckle noise, whichmakes the grainy image. Moreover, a fur-ther difficulty arises from the return signalmeasured in the areas of sea. Indeed in thecase of rough sea conditions, the signal de-termines the intensity of SAR image simi-lar to the land, making it difficult to distin-guish the line of coast [4].It is known that, in a planning of the coastalstrip, information on the position of thecoast is a key element. In this context, therealization of works related to the sea alsodue to man-made events can cause changesin the coastline. Hence it is fundamen-tal to monitor the evolution of the coastwith regular checks in order to examine theseasonal variability, the storms, and alsochanges due to human intervention.Currently there is no typical methodologyto automatically detect the coastline. Thisis traditionally remarked in situ by special-ized operators, and it is an expensive pro-

cess and prone to errors.Recently, satellite and /or aerial imageshave been added to control the profile ofthe coast, but traditional methods of photo-interpretation and manual-tracing are af-fected by errors, because they are charac-terized by subjectivity of the result, whichdepends on interpretation of the operator.For this reason, in recent years, the extrac-tion of the coastline has been the subject ofresearch for the development of automaticor semi-automatic procedures aimed at car-rying out that task.In particular, satellite-optical images, beingessentially noise-free images, are usuallyprocessed by procedures which use tradi-tional techniques of image segmentation asthresholding and edge detection.The processing of satellite-optical imagesand aerial yields results in the ascertain-ment of the coastline, but it should be notedthat the acquisition of these images de-pends on sunlight and weather conditions.In this regard, the use of images acquiredby SAR provides a considerable advantageover the use of optical images. This sensor,as above written, is not dependent on sun-light, thus it produces image information atnight, and also can operate in all weatherconditions. This therefore allows the ac-quisition of image information of areas ofland subject to special conditions, such asthe polar areas or areas with low light con-stantly covered by clouds. This feature isvery important because visible changes ofthe coast are found in the presence of in-clement weather.The proposed approach is based on thelevel set method [5, 6] applied to SAR im-ages.In this method, an initial curve definedon the image evolves according to a PDEmodel by means of a velocity related tothe characteristics of the image to be seg-

2596


mented. This curve is deformed until it as-sumes a steady position at the boundary ofthe area to be extracted from the image [7].In our procedure the initial curve includesthe coastal area and it moves until itreaches the line which marks the boundarybetween land and sea.SAR images, due to speckle noise, requireto be pre-processed by using a denoisingfilter. For this purpose, we have imple-mented an anisotropic diffusion filter[4, 8].Finally, the proposed procedure has the ad-vantage to extract the coastline by work-ing on a single image unlike the classicalinterferometric techniques, which requiretwo images of the same area acquired atdifferent times or from different points ofview [2]. In the future we plan to applythe proposed methodology to images ob-tained from the constellation of satellitesCosmo Sky-Med. These satellites are use-ful in monitoring changes in the Earth’ssurface with a very high time resolution,because they can observe the same areaseveral times a day in all weather condi-tions.

2 Coastline extraction

Detecting the coastline from SAR image isa complex depending on the signal comingfrom water and land region. In fact whenthe sea is calm, we have a black region, oth-erwise, if the sea is slightly rough, the emit-ted signal is only partially reflected produc-ing a gray region. For this reason the re-turn signal from the sea can be frequentlyindistinct from one coming from the land.Moreover the presence of speckle modeledas a strong multiplicative noise makes thecoastline detection even more complicated.In recent years, the “level set” method,based on the evolution of curves has been

an important tool for solving image seg-mentation problems. In particular, thismethod describes the evolution of a poly-line (curve), containing the interested area,until it reaches the boundary of the area tobe extracted.The main advantages of the ”level set”method is that complex shaped regions canbe detected and handled implicitly.In order to obtain the governing equationof a front evolution, in [3] we have consid-ered the contour of a region as a zero levelof an implicit function defined as a distancefunction. So we obtained the evolution ofthe curve driven by a speed function de-pending on the image features.In the ”level set” method the choice of thespeed function is fundamental to achieve agood segmentation.In this work we propose an integrated pro-cedure obtained by combining two differ-ent speed functions: one, called average-based speed function obtained by mod-elling the intensity of an image using theGamma distribution, and an other, calledgradient-based speed function, obtained bythe computation of image gradient.The SAR PRI image here analyzed hasbeen acquired during ERS2 mission. Theimages are related to the sea of Europeancoasts.We have examined the behaviour of the al-gorithm by using the average-based speed.From these tests it is evident that the resultsdepend on the initial position of the curve.In particular we have obtained a better re-sult if the initial curve is located near thecoast.Usually the image gradient is used to iden-tify the edges or contours; indeed if the gra-dient of an area is high then the related pix-els correspond to edges. However, as it isknown, the SAR images can look granularbecause they are corrupted by strong noise

2597

Technologies

Figure 1: Integrated procedure pipeline: a) original image, b) SRAD filtered image, c)the initial curve for the gradient-based speed, d) intermediate result, e) initial curve forthe average-based speed, f) final result.

speckle, and in this case the computation ofthe gradient could detect false edges.For this reason in the gradient-basedspeed computation the images have beenpre-processed by means of the SRAD(Speckle Reducing Anisotropic Diffusion)algorithm, which is an extension of Perona-Malik algorithm [8].Unlike performed tests with average-basedspeed, in this case the results are indepen-dent of the position of the initial curves.The ”level set” method, by using theaverage-based speed, detects the coastline

with more precision in terms of pixels. Infact the image is not filtered for the noisereduction, because the average computa-tion is already a filtering. We observed in[3] that the best result is obtained by locat-ing the initial curve as near as possible tothe coastline. This implies that the final re-sult depends on the position of the startingcurve.The result obtained with the speed based onimage gradient is less accurate in terms ofpixels, because it works on the filtered im-age and not on the original one. However,

2598


Figure 2: Example of regular coast: a) original image of Netherlands coast, b) originalimage included in the red square of a), c) starting contour, d) final contour.

this last approach is independent of the po-sition of the initial curve.The integrated procedure we developed al-lows the coastline to be identified automat-ically and independently of the initial loca-tion of the curve. In fact, it combines thetwo speed functions by improving the seg-mentation results. In the original image of150x150 pixels (Figure 1a) SRAD filter isapplied (Figure 1b). Then the initial curve(Figure 1c) evolves until it reaches the re-sult of the “level set” method with gradient-based speed (Figure 1d). Hence the result-ing contour is used as initial curve for ap-

plying the level set method (Figure 1e) tothe original image by using the average-based speed, in order to obtain the final re-sult (Figure 1f).

3 Segmentation results

In this section we show some results, ob-tained by using the proposed integratedprocedure, combining both the average-based speed and the gradient-based speed.The integrated procedure is applied to theregular coast of Netherlands (Figure 2). In

2599

Technologies

Figure 3: Example of indented coast: a) original image of Netherlands coast, b) detailincluded in the red square of a), c) starting contour, d) final contour.

particular, the red square in the original im-age of 1500 x 1500 pixels (Figure 2a) in-dicates the portion of 150 x 150 pixels tobe segmented (Figure 2b). The initial po-sition of the curve (Figure 2c) is modifiedduring the evolution until the final result isreached, identifying three islands and thecoast (Figure 3d).In this example the developed technique isapplied to an indented coast of Netherlandsof (Figure 3). Again, the red square in theoriginal image of 1500x1500 pixels (Fig-ure 3a) is the portion of 150 x 150 pixels tobe segmented (Figure 3b). The initial po-

sition of the curve (Figure 3c) is modifiedduring the evolution until to reach the finalcontour, identifying the coast and the inlets(Figure 3d).Finally we show the result (Figure 4) ob-tained in the case of an ill-defined con-tours coast of Balearic Islands. In thiscase, the red square in the original image of1500x1500 pixels (Figure 4a) is the portionof 150x150 pixels (Figure 4b). The initialposition of the curve (Figure 4c) is modi-fied until the identification of the low con-trast line between land and sea (Figure 4d).Because benchmarks of the SAR images

2600


Figure 4: Example of low contrast coast: a) original image of Balearic Islands coast, b)detail included in the red square of a), c) starting contour, d) final contour.

are not known, in order to verify the accu-racy of the results obtained by the proposedprocedure, in [9] some tests are presentedon synthetic images obtained by copyingthe pattern of the SAR images of two dif-ferent regions (e.g. land and sea). Theseimages were used to have the reference ofthe exact location of the contours.

4 Conclusions

In this work we have faced the problem todetect the coastline from SAR image by the“level set” method.

Two distinct speed evolution functionshave been examined.The former, based on the average inten-sities of the regions, does not need toreduce speckle noise, since the average-based speed already contains a filtering; thelatter, based on the image gradient, filtersspeckle noise by the SRAD technique.We proposed an innovative procedure com-bining the two speed functions in order toimprove the two individual approaches.The developed technique allows the seg-mentation results to be independent of theinitial location of the curve and more-over, it automatically stops when the curve

2601

Technologies

achieves the coastline. More precisely theproposed procedure is such that any differ-ent position of the initial curve that con-tains the area to be segmented produces thesame result.Future development will concern the exten-sion of the segmentation to more regions of

different gray levels and the use of the Cos-moSkyMed images (Constellation of SmallSatellites for Mediterranean basin observa-tion) which are a constellation of 4 satel-lites in order to monitor the entire globeand the Mediterranean area.

References[1] H. Maitre. Processing of Synthetic Aperture Radar Images. 2008.

[2] S. Dellepiane, R. De Laurentiis, and F. Giordano. Coastline extraction from SAR im-ages and a method for the evaluation of the coastline precision. Pattern RecognitionLetters, 25:1461–1470, 2004.

[3] M.M. Cerimele, L. Cinque, R. Cossu, and R. Galiffa. Coastline detection from SARImages by level set model. Lecture Notes in Computer Science, 5716:364–373, 2009.

[4] Yongjian Yu and Scott T. Acton. Speckle reducing anisotropic diffusion. IEEE Trans.on Image Processing, 11:1260–1270, 2002.

[5] J.A. Sethian. Evolution, implementation and application of level set and fast march-ing methods for advancing front. Journal of Computational Physics, 169:503–555,2001.

[6] S. Osher and R. Fedkiw. Level Set Methods and Dynamic Implicit Surfaces. 2002.

[7] I. Ben Ayed, A. Mitiche, and Z. Belhadj. Multiregion level-set partitioning of syn-thetic aperture radar images. IEEE Trans. Pattern Analysis and Machine Intelligence,27:793–800, 2005.

[8] P. Perona and J. Malik. Scale space and edge detection using anisotropic diffusion.IEEE Trans. Pattern Analysis and Machine Intelligence, 12:629–639, 1990.

[9] M.M. Cerimele, R. Cossu, and R. Galiffa. Level set method extension to more re-gions segmentation in SAR images. IAC REport, (190/2010), 2010.

2602

Seafloor Exploration Using the MultiBeam EchoSounder Technology: Some Examples

R. Tonielli1, M. Barra1, G. Di Martino1, F. Foglini2, S. Innangi1, A.Mercorella2, M. Rovere2

1, Institute for Coastal Marine Environment, CNR, Napoli, Italy2, Institute of Marine Sciences, CNR, Bologna, [email protected]

Abstract

Until the late 70’s bathymetry data of the oceans were commonly acquired bymeans of single beam echo sounders, along discontinue track lines with an unevenspatial distribution. More recently, the MultiBeam Echo Sounder (MBES) technol-ogy was made available to the academic research. These systems provide a full seafloor search and a high resolution imaging of the seafloor. The MBES systems pro-vide depth values along a swath instead of a single point value by means of acousticpulses arrays. There are several MBES models characterizing by different technicalequipments in terms of frequencies, beam number and spacing. MBES systems areused in archaeological and engineering fields, where high resolution data in small ar-eas are required, as well as in deep seafloor mapping aimed at large scale structuresdetection. MBES may provide also grain size information besides bathymetric data,recording the backscatter values that result from the interaction of acoustic pulseswith the seafloor. This tool is particularly important to characterize the distributionof some features on the seafloor such as submarine landslides and other elementsthat may play an important role in the assessment of the geological hazard alongthe coasts. Actually some research groups are improving the backscatter analysis togenerate a semi-automatic seabed classification method. This contribution will pro-vide an overview of some recent examples studied within the framework of the CNRresearch projects.

1 Introduction

Over the last 30 years technological ad-vances lead to discover and map the Earth’sseafloor. The acoustic waves are today themain instrument to explore the sea bottom,being easily transmitted through the watercolumn. Their reflection off the seafloor al-lows us to imagine its shape and texture.They are used underwater to detect and lo-cate obstacles and targets, as well as tomeasure the characteristics of marine envi-

ronment, such as seafloor topography, liv-ing organisms, currents and hydrographicstructures.Acoustic waves are the basis of Sonar(Sound Navigation And Ranging). A sonaris a device that uses sound in order to re-motely detect and locate objects in the wa-ter. There are two basic types of sonars: thepassive sonar, that is a “listening” device,i.e. it can only detect sounds emitted by asource; the active sonar, that can generatesound waves in addition to detect them [1].


Technologies

Figure 1: Difference in bottom coverage between singlebeam and multibeam systems.

The origins of sonar are old, likely in 1822,when Daniel Colloden used an underwa-ter bell to calculate the speed of sound inLake Geneva, Switzerland. Later on, in1906, Lewis Nixon built the first sonar-type listening to detect icebergs. A majorbreakthrough came from Paul Langevin,a French physicist; he demonstrated, be-tween 1915 and 1918, that it was possibleto transmit signals, and to actively detectsubmarines giving both their angles anddistances from receiver. Sonar technologyimproved considerably between the twoworld wars, but the period since SecondWorld War has seen enormous advances inmarine geophysics. Rapid progress cameduring the 1950s and 1960s, when gov-ernment agencies financed multi-purposeoceanographic expeditions and gave scien-tists much flexibility in conducting their re-searches [2].Until that time bathymetric data were col-lected using Singlebeam Echosounders.An echosounder is an active sonar, as itmeasures depth by producing an acoustic

pulse and listening for the echoes returningfrom the seafloor. The echosounder’s pro-jector generates a directional pulse, so themain part of the energy is radiated withina solid angle towards the bottom, whereit is reflected and finally detected by theechosounder’s receiver. The Singlebeamprovides a single measurement at the nadirdepth, so data logged along transect linescannot provide continuous coverage nei-ther detailed information about the bottom.In 1964 a new technology, known as SonarArray Sounding System (SASS), was de-veloped for the United States Navy. TheSASS received a large number of side-ways reflected signals, recorded their rangeand bearing and, from the configurationof the transducers, computed off-trackdepths. This instrument was the first Multi-beam EchoSounder (MBES) that achieveda great development during the 1970s.A MBES emits pulses with a beam patternthat is narrow along the ship track and wideacross the ship track, providing a swathof data that, depending on sensor’s speci-

2604


fication, is normally several times the wa-ter depth. A Global Positioning System(GPS), a Motion Sensor, a Sound VelocityProfiler (SVP) and a sound speed probe areused during Multibeam survey to providevessel attitude data and sound speed in thewater in order to produce accurate bathy-metric record (Figure 1).Even though Multibeams are more com-plex and more expensive systems than sin-glebeams, they are largely used within themarine sciences. In fact they are more pro-ductive, as they provide a full sea floorsearch and a high resolution imaging of theseafloor than Singlebeam in a short time.MBES are suitable for different applica-tions according to their specifications; theycan be installed on small vessels to performhigh resolution surveys in shallow water, aswell as on large ships to explore offshoredeep sea areas.Technological development leads to con-tinuous improvements of Multibeam hard-ware and so to a better instrumental per-formance; on the other side, new softwareavailability allows the extraction of furtherinformations from the acoustic signal, suchas backscatter intensity. In fact over thelast years, great attention has been paid tobackscatter analysis, that can be used tostudy the seafloor characteristics.

2 Multibeam EchoSounderSystems

The MBES transducer emits an acousticpulse that travels in the water column to-ward the bottom and detects the reflectedecho. The receive array is normal to thetransmit array and the intersection betweenthe transmit beam and the receive arrayforms a sequence of elliptical areas on the

seafloor, called footprints (Figure 2).The system performs a depth measurementfor each footprint along a swath, comput-ing the range along each beam to the bot-tom by the relation between the Two-WayTravel Time (TWTT) and the sound speedin the water.The TWTT is measured by the maximumbackscatter amplitude of inner beams,where the footprint is small and the echosignal has a prominent peak, and by phasedetection of outer beams, where the foot-print increases and the echo signal mayhave several peaks of comparable ampli-tudes [3].There are several MBES models character-ized by different technical characteristicsin terms of frequencies, beam number andbeam solid angle.The following information outlines the ma-jor specifications collected from a varietyof systems (Table 1).These specifications affect the system ca-pacity in terms of maximum operatingdepth and spatial resolution.Operating frequency is the multibeam mainfeature: it affects the depth capabilities ofthe system and the vertical resolution.Low frequency waves can travel far inthe water, because they propagate along astraight line and they are not attenuated anddo not lose intensity, so they can easilypenetrate the medium. On the other side,high frequency waves are rapidly attenu-ated, but the short wavelength allow themto resolve small object providing high ver-tical resolution images.The beam solid angle specifies the beamwidth and it can be used to estimate theacross-track individual beam footprint ac-cording to the equation:

Footprint = Tan(beamwidth)∗NadirDepth

2605

Technologies

Figure 2: The intersection between total area ensonified by acoustic pulse and area cov-ered by receive array produces the footprints.

Table 1: Multibeam Systems technical specifications.

So, the beamwidth defines the system hori-zontal resolution that is function of the ves-sel speed and the number of pulses per sec-ond (ping rate). When performing a TWTTamplitude detection method, the smallerthe beam angle, the better the system isable to discern true depth and resolve smallfeatures. When performing a TWTT phasedetection method, the horizontal resolutiondoes not depend on beamwidth, becausethe feature detection capability may be sig-nificantly better than the physical beamopening, i.e. the system can resolve targetssmaller than the footprint [4].Recent low frequency MBES operate inlarger depth range than in the past, thanks

to the beam dynamic focusing. It reducesthe distortion of the array beam pattern inthe near field, so the footprint size is nar-rower than predicted and the system oper-ates in shallow water without loss of reso-lution.Over the last few years, some technicalimprovements were introduced in order tooptimize system performances. MBESperforming an equi-distant beam spac-ing mode, in addition to the equi-angularmode, were produced; systems perform-ing the equi-angular mode produce an un-even distribution of beams that graduallydecrease from the swath’s centre to theswath’s outer edges. This reduces the

2606


system productivity by increasing the per-centage of swath overlap required betweensurvey lines. Equi-distant beam spacingmode, selectable by the user, provides uni-form sounding density and maximizes us-able outer swath, increasing system pro-ductivity [5].Some MBES systems perform a high den-sity processing mode to improve the hor-izontal resolution: they generate extrasoundings so that the number of totalsoundings is higher than the number ofacoustic beams. This technique reduces theacoustic footprint thus improving the sys-tem capability to detect objects and otherdetails on the bottom.During a bathymetric survey, a number ofinstruments has to be used in order to per-form a real-time correction of bathymet-ric data for vessel position and attitude.GPS system produces vessel position andspeed data by a triangulation computationfounded on a satellites constellation. Po-sitioning data may accuse errors becauseGPS signal propagates through the atmo-sphere, thus producing data accurate to amatter of meters; better accuracy is pro-vided by other positioning systems, such asDifferenzial, High Precision DGPS, LocalDGPS and Real Time Kynematic DGPS.These systems provide positioning datawith sub-metric to centimetric precision;RTK DGPS also provides real-time correc-tions for the z-axis.Vessel movements such as roll, pitch andyaw, cause swath pointing errors and incor-rect bathymetric data. The gyrocompassand the motion sensor produce very pre-cise measurements of vessel attitude manytimes per second: by integrating these val-ues with the timing of the sonar echo, anaccurate bathymetric record can be pro-duced regardless of the echo path throughthe water.

The water sound velocity is a basic param-eter for Multibeam operations, so a soundvelocity probe is mounted near the trans-ducers; it provides a real-time sound ve-locity measurement in order to perform thebeam-steering, i.e. to compute the echoarrival-time for each beam to each receiverelement.Finally, a Sound Velocity Profiler (SVP)provides the sound velocity profile in thewater column. The sound speed dependson the medium density, so when the acous-tic pulse propagates through a stratified wa-ter column the beam’s path bends becauseof the refraction phenomenon. SVP datastored allow the system to compute the re-fraction coefficient in each water layer andto obtain the right location of the object onthe seafloor relative to the sonar transduc-ers. The SVP is lowered in the water col-umn every 6 - 8 hours according to Intena-tional Hydrographic Organization [6] stan-dards or whenever the swath bends to thebottom (frowning) or to the surface (smil-ing); this event can occur while survey-ing in coastal areas, where a fresh-waterrunoff is present, as well as in offshore ar-eas, where upwelling is present. The ef-fect of these events is an overestimated oran undervalued sound velocity profile thatproduce typical artifacts, such as “ripples”in logged data [7, 8].Multibeams and associated instrumentscan be permanent or semi-permanentmounted on small and large ships. MBESis usually hull-mounted on ships specifi-cally dedicated to bathymetric surveying;this is a stable configuration where themultibeam is integrated with the vessel’shull and the exact physical location of eachsystem component and the distances be-tween them is measured with great preci-sion during the installation process. Pole-mounted transducers are normally used on

2607

Technologies

smaller vessels that will be temporarilydedicated to acoustic surveying. If possi-ble, MBES is installed near the vessel cen-tre of gravity, in order to increase the sys-tem stability and to keep it far away fromvibration sources and from seafoam that af-fects the sound speed. Finally, MBES canbe installed on Remote Operating Vessel(ROV); a shallow water instrument, nor-mally operating at 100 meters depth, canbe lowered in very deep water to performa high resolution bottom detection. Somemultibeam models are high-pressure proofand operate until 6,000 meters depth.

3 Backscatter

The primary purpose of MBES system wasto obtain high-resolution bathymetry maps,but recent researches have focused on uti-lizing MBES backscatter data as seafloorcharacterization tool.When sound waves encounter a roughboundary between two media with differ-ent impedances, waves are reflected, trans-mitted and scattered. The portion of energyscattered back towards the source, knownas backscatter, is typically expressed indecibel value. The backscatter strength de-pends on: wave incidence angle with thebottom, ensonified area, absorption coef-ficient of water column and, most of all,bottom physical proprieties, i.e. rough-ness, density, acoustic impedance and co-efficient reflection of sediment. In gen-eral, as the impedance contrast or rough-ness of a surface increases so does the in-tensity of backscatter; so as logarithmic pa-rameter F of grain size decreases the acous-tic impedance increases and, consequently,the scattering increases [11, 12].Typical values of density, sound speedand resulting acoustic impedance for se-

lected sediment types are given in Table 2.Thus, backscatter has great importance inknowing the grain size by studying differ-ences in bottom acoustical response. Cur-rently, there are two approaches to loggingMBES’s acoustic backscatter data [13]:

a) Forming two additional port and star-board wide angle receive beams to log asidescan-like time series of intensities.

b) Logging a single backscatter value witheach beam, either taken as the maximumintensity at bottom detect, or as averageintensity centered on the bottom in thetime series (called snippet or footprinttime series).

The main difference between snippet andsidescan-like data is that the “sidescan”mode records the backscatter intensityfrom all beams and computes the averageof the data, while the snippet records arange (in the time T0-Tn) of intensity val-ues from each beam, without computingaverage values. This is why it is also calledfootprint time series (Figure 3).The basic principles of measuring theseafloor backscatter strength are similar forlow and high frequency systems. How-ever, the measurement geometry and phys-ical conditions, such as the ensonificationand footprint areas relative to the seafloorroughness scale, Rayleigh parameter, etc.,are significantly different [3].As for now, there are a few software prod-ucts available to process MBES backscat-ter data. Some of the most notable soft-ware packages are: SwathEd, developedby Hughes Clarke, MB systems, devel-oped by Caress and Chayes, SonarScope,developed by the Acoustic & Seismicsdepartment of Ifremer, Geocoder, devel-oped by Fonseca and Calder, Isis (©TritonElics International), Hips and Sips (©2009CARIS) and a Matlab toolbox developedby Gavrilov, Siwabessy and Parnum of the

2608


Media Grain Size(F)

Density(kg·m−3)

SoundSpeed

(m·s−1)

AcousticImpedance

(rayls)

Reflectioncoefficient

(R0)Seawater N/A 1000 1500 1.5*106 N/A

Clay 9 1200 1470 1.764*106 0.0809Clayey silt 7 1500 1515 2.496*106 0.2492

Very fine sand 3 1900 1680 3.192*106 0.3606Course sand 1 2000 1800 3.6*106 0.4118

Rock N/A 2500 3750 9.375*106 0.7241

Table 2: Typical values of density, sound speed and resulting acoustic impedance forvarious grades of sediment [9, 10].

Center for Marine Science and Technology(CMST).Each of these software implements a setof steps in order to process the backscatterdata that can be summarized in the follow-ing list:

1. Conversion from original acquisitiondata format into the software propri-etary data format;

2. Calculation of XYZ position and inci-dence angle for each beam;

3. Calculation of XYZ position and aver-age intensity of backscatter;

4. Removal of the system Time VariableGain (TVG);

5. Radiometric correction (beam pattern);6. Calculation of the surface backscatter

strength.

Any software listed above has its special-ization in the processing of backscatter andthe choice of the most suitable software de-pends on what kind of output is required.For example a software like Isis is suitedfor processing sidescan data, but it does notallow the processing of snippet data; othersoftware, as Caris, process the Time Seriesas well as the Sidescan-like; finally, thereare software, as the Matlab toolbox, thatimplement various processing and analysis

algorithms for processing MBES data.Seafloor classification based on the anal-ysis of MBES backscatter images oftencontains artifacts due to inadequate cor-rection for angular dependence. For thispurposes the correction for incidence an-gle is worthwhile. The relationship be-tween backscatter strength and incidenceangle can be exploited empirically, i.e. us-ing the relative difference between angu-lar responses to distinguish seafloor types[14]. Hughes Clarke [15] identified andused ten features of angular dependencecurves, including the mean and slope fromthree different angular domains, that couldbe used for seafloor classification. In sum-mary we can state that utilizing the angu-lar dependence of backscatter in additionto the mean backscatter strength is a betterapproach to seafloor classification, becauseit provides more informations about mor-phological and physical proprieties of theseafloor [16].

2609

Technologies

4 Multibeam Echosounderapplications: some ex-amples

MBES systems are suitable for differentapplications according to their technicalspecifications. They can be used in shallowwater surveys, where high resolution datain small areas are required, as well as indeep seafloor mapping aimed at large scalestructures detection.The depth range of the survey area, the re-quired resolution and the final output affectthe choice of Multibeam model and of as-sociated instruments. An overview of someexamples studied within the framework ofthe CNR research projects are shown.

4.1 ArchaeologyMultibeam systems are used in archaeolog-ical field for different purposes, such asancient ruins surveying or wreck search-ing. Very high resolution data are requiredfor these surveys, in order to generate 20centimeters grid cell or less, as well ashigh precision position system; thus, highfrequency MBES and HP DGPS or LocalDGPS are used. An example of an archae-ological application of multibeam systemis shown in Figure 4.The bathymetric survey was carried out in2006, on board M/N “Red Fish”, a ves-sel 8 meters long, in order to navigate invery shallow water; a pole-mounted ResonSeabat 8125, a Landstar HP DGPS, an IxeaOctans 3000, a Reson SPV-C and a Nav-itronic SVP15 were used. The Tin Modelshows Roman ruins of the ancient city ofBaia (Naples). Some walls and a room lo-cated in very shallow water, are recogniz-able in the snapshot. The structures lie onthe seafloor at about 6 meters depth and

they are about 1 meter high; the room isincluded by 20 meters and 10 meters longwalls.

4.2 EngineeringIn engineering field very high resolutiondata of anthropical structures, such asdocks and wharf, are required. The targetof these surveys is to perform a structuresmonitoring as well as to plan new projects;so, high frequency MBES and sub-metricprecision GPS are required for these bathy-metric applications. Some examples areshown in Figure 5, Figure 6 and Figure 7).The structure in Figure 5 is located in Bag-noli, Gulf of Naples. It is a 650 mt.long wharf aimed to goods loading and un-loading operations; the wharf is includedin Ex-Italsider dismissed steelwork indus-trial area. Supporting piers and protectionblocks located at the wharf’s head are vis-ible. The bathymetric survey was carriedout in 2006, on board M/N “Red Fish”, us-ing a pole-mounted Reson Seabat 8125, aLandstar DGPS, an Ixea Octans 3000, aReson SPV-C and a Navitronic SVP15.The shaded relief in Figure 6 is producedfrom very high resolution bathymetric dataof Pozzuoli tourist harbour. Blocks sur-rounding the docks, mooring blocks andchains are visible as well as other dis-missed anthropic structures such as elec-tric lattices. The bathymetric survey wascarried out using a small vessel in order tonavigate around the docks; a pole-mountedReson Seabat 8125, a Landstar DGPS, anIxea Octans 3000, a Reson SPV-C and aNavitronic SVP15 were used. An exampleof high resolution data collected in Naplesharbour is shown in Figure 7. Raw data arevisible at the highest instrumental resolu-tion: all correct logged beams are in factvisible showing an outer breakwater por-

2610


tion, 30 meters long, and the blocks placedin front of it. Many processing softwareprovide a detailed data view in order tomanually remove incorrect beams. Bathy-metric data were collected using a pole-mounted Reson Seabat 8125 on board M/N“Red Fish”, a Landstar DGPS, an Ixea Oc-tans 3000, a Reson SPV-C and a NavitronicSVP15.

4.3 Hydrography and Morpho-bathymetry

Many of the human activities are strictlybound to the sea. The knowledge ofseafloor morphology allows optimizationof these activities, especially the ship navi-gation. Bathymetric charts, therefore, rep-resent an obvious resource for all the mar-itime activities and their development. Theuse of the Multibeam allows a continuousmap updating, aimed to bottom variabilitystudy within the marine geohazard field.Two examples of this application areshown. The bathymetric chart in Figure 8was realized within the mapping project ofBasilicata Region. Bathymetric data werecollected from 2004 to 2007 by IAMC,during different oceanographic cruises onboard R/V “Urania”, R/V “Thetis” andM/N “Napoli”. Different Multibeam mod-els have been used in order to realize thischart depending on the operational depth(Seabat 8125 between 5 and 50 m depth,Seabat 8111 and Seabat 8160 between 50and 600 m depth).The morphological map of Straits ofMessina in Figure 9 was produced fromBathymetric data collected by IAMC dur-ing different oceanographic cruises, from2001 to 2003, using Reson Seabat 8111(100 kHz), hull-mounted on R/V “Thetis”.

4.4 Marine Geohazard

In recent years high-resolution seafloormapping is increasingly employed formonitoring and assessing natural im-pact and hazard on coastal and ma-rine areas. High-resolution multibeambathymetry represents a fundamental toolfor a first characterization of geologicalfeatures that represent potential geohazardssuch as: volcanic vents, active faulting,submarine canyon activity, bedform migra-tion, gas seepage, fluid escape, slope ero-sion and failures. Here below some exam-ples of possible geohazards represented bya field of pockmarks in the Tyrrhenian Seaand by submarine slides affecting the conti-nental slope of Gela Basin Sicily Channel).The multibeam data in Figure 10 have beenacquired with a Kongsberg EM12 modelin 1999 by ISMAR aboard R/V Strakhov.The aim of the survey was to investigate theTyrrhenian sea at water depths deeper than400 m. The primary positioning sensor wasthe RACAL DGPS SKYFIX SPOT BEAMwith a 12 channel Trimble receiver. In typ-ical ocean depths a sounding spacing ofabout 50 m across and along is achievable.The data acquisition was performed to con-stantly match a 20 % overlap. The datawere post-processed using the Kongsberg-Neptune software, with a standard proce-dure, including positioning and depth cor-rection, manual and statistical cleaning ofthe data. The shaded relief produced frombathymetric data shows pockmarks with adiameter of 500 m.Twin slides are shown in Figure 11. Theyare located along the western side of theMalta Plateau. When exposed at seafloor,failed masses show an extreme morpholog-ical complexity with massive slide blocks,pressure ridges and hummocky surface.The multibeam data have been acquired by

2611

Technologies

a Reson Seabat 8160 (50 kHz) and Kons-berg EM300 (30 kHz) on board R/V Uraniaand R/V Odin Finder. The data acquisitionwas performed to constantly match a 20%overlap. The aim of the survey was to in-vestigate the continental slope, in order todetect widespread mass failures along theWestern side of the Malta plateau.

4.5 BackscatterThe MBES backscatter data provide grainsize information and characterize the distri-bution of seafloor features such as subma-rine landslides and other elements that mayplay an important role in the assessmentof the geological hazard along the coasts.Overviews of some examples of processedMBES backscatter data are in Figure 12and in Figure 13.A shallow water survey was performed by aReson Seabat 8125 in order to collect highresolution morphobathymetric data alongMaratea coast (Figure 12).The sidescan-like data were processed by Isis of TritonElics: the navigation data were smoothed,TVG adjustment was performed, radiomet-ric (beam angle and grazing angle) and ge-ometric (slant range) corrections for eachline were applied; then the lines were as-sembled in a mosaic. The snippet data wereprocessed by the Matlab toolbox of CMST.This software removes all artifacts due toangular dependence and returns a file foreach line, containing Easting, Northing and

the decibel value of each beam. So it ispossible to achieve a grid of snippet datathat can be further processed (i.e. interpo-lation). Snippet grid contains more infor-mation about lithology than a sidescan mo-saic, but there is no morphological infor-mation.The Time Series Mosaic in Figure 13 wasproduced from backscatter data processedby Caris Hips and Sips. It is noteworthythat the backscatter mosaic points out thelithological facies.

5 ConclusionsNowadays, multibeam surveys represent abasic tool for marine research projects. Inthis review some MBES applications, de-veloped by CNR research scientists, wereprovided. The progressive improvementof multibeam echosounder systems makesthese instruments suitable for a new rangeof applications. In particular, geologicalhazard, sea bottom facies characterization(including benthonic assemblages), hydro-carbon deposits acoustic response, wa-ter column acoustic imaging, quantitativestudies of demersal resources, are possibleapplication of this technology. In conclu-sion, it is clear that multibeam echosoundersystems provide the fundamental knowl-edge of the seafloor for a effective devel-opment of research projects in marine en-vironment.

2612


Figure 3: PDS2000 software windows of backscatter data from MBES, respectivelySnippet window (3a) and Sidescan-like window (3b). Figure 3(c) - shows the CentralUnit of MBES framework: the green line is the profile of the bottom ensonified byMBES. Snippet data logging horizontal range is the same as for MBES data logging(A), while Sidescan-like data logging horizontal range encloses the water column (D)and all range of acquisition (B), including null data (C).

2613

Technologies

Figure 4: Archaeological structures - Tin Model, 3D view.

Figure 5: Wharf 3D view - Grid cell 50 cm.

2614


Figure 6: Pozzuoli tourist harbour 2D view Shaded relief - Grid cell 20 cm.

Figure 7: Naples harbour, beams 3D view - Outer breakwater.

2615

Technologies

Figure 8: Bathymetric Chart and Shaded Relief of Lucania Tyrrhenian coast: Scale1:25000, isobaths spacing 5 m, projection UTM33N, Datum WGS84, depth range 5 -600 m.

2616


Figure 9: Straits of Messina shaded relief - Grid cell 10m - Depth range: 10 - 700 m.

Figure 10: Pockmarks field on top a structural high in the Tyrrhenian Sea - DTM cellsize: 50 m, datum WGS84. Projection Direct Mercator on 40° N.

2617

Technologies

Figure 11: Twin slides on the continental slope of Gela Basin (Strait of Sicily).

Figure 12: Bathymetric data (a), sidescan-like data (b) and snippet data (c) collected inAcquafredda’s Bay (Maratea City).

2618


Figure 13: Time Series mosaic from Multibeam Simrad EM3002.

2619

Technologies

References[1] L3 Communications SeaBeam Instruments. Multibeam Sonar Theory of Operation.

2000.

[2] E.J.W. Jones. Marine Geophysics. page 474, 1999.

[3] I.M. Parnum. Benthic habitat mapping using multibeam sonar systems. page 208,2008.

[4] J. Miller, J.E. Hughes Clarke, and J. Patterson. How Effectively Have You CoveredYour Bottom? Hydrographic Journal, (83):3–10, 1997.

[5] National Oceanic, Atmospheric Administration (NOAA), and Office of Coast Sur-vey. Field Procedures Manual. 2010.

[6] IHO International Hydrographic Organization. IHO standard for Hydrographic Sur-veys - Special Publication no. 44. page 28, 2008.

[7] J.E. Hughes Clarke. Dynamic motion residuals in swath sonar data: Ironing out thecreases. International Hydrographic Review, 4(1):6–23, 2003.

[8] G. Di Martino and R. Tonielli. Fresh water runoff effects on shallow water Multi-beam surveys. Sea Technology, 51(5):10–13, 2010.

[9] APL. APL UW High Frequency Ocean Enviroment Acoustic Models Handbook.1994.

[10] X. Lurton. An Introduction to underwater acoustic, Principles and Applications.page 347, 2002.

[11] I.M. Parnum, A.N. Gavrilov, P.J.W. Siwabessy, and A.J. Duncan. Analysis of multi-beam data for the purposes of seafloor classification. S.P., 2007.

[12] J.A. Goff, H.C. Olson, and C.S. Duncan. Correlation of side scan backscatter inten-sity with grain size distribution of shelf sediments, New Jersey margin. Geo-MarineLetters, (20):43–49, 2000.

[13] J. Beaudoin, J.E. Hughes Clarke, E. van den Ameele, , and J. Gardner. Geomet-ric and radiometric correction of multibeam backscatter derived from Reson 8101systems. Canadian Hydrographic Conference Proceedings. 2002.

[14] I.M. Parnum, A.N. Gavrilov, and P.J.W. Siwabessy & A.J. Duncan. The effectof incidence angle on statistical variation of backscatter measured using a high-frequency Multibeam sonar. S.P.:433–438, 2005.

[15] J.E. Hughes Clarke. Towards remote seafloor classification using the angular re-sponse of acoustic backscatter: a case study from multiple overlapping Gloria Data.IEEE Journal of Oceanic Engineering, 19(1):112–127, 1994.

2620


[16] G. De Falco, R. Tonielli, G. Di Martino, S. Innangi, S. Simeone, and I.M. Par-num. Relationships between multibeam backscatter, sediment grain size and Posi-donia oceanica seagrass distribution. Continental Shelf Research, pages 1941–1950, 2010.

2621

Technologies

2622

Integrating Meteo-Marine Forecast Data withinHeterogeneous Earth Observation InformationServices

G. Brugnoni1,2, S. Gianfranceschi3, F. Pasi4,2, S. Taddei5,2, A. Orlandi4,2,B. Doronzo5,2, B. Gozzini4,2, F.P. Vaccari21, Institute for Biometeorology, CNR, Livorno, Italy2, LaMMa Consortium, Laboratory of Monitoring and Environment Modelling for theSustainable Development, Sesto Fiorentino (FI), Italy3, INTECS s.p.a., Pisa, Italy4, Institute for Biometeorology, CNR, Sesto Fiorentino, Italy5, Institute for Biometeorology, CNR, Livorno, [email protected]

Abstract

At the Consorzio LAMMA (Laboratory of Monitoring and Environmental Mod-elling for the sustainable development) an operational weather forecasting chaincomposed by numerical models, such as Weather Research and Forecasting (WRF)and the full-spectral third-generation ocean wind-wave model WaveWatch III (WW3).These models run few times a day and produce a few days forecast over the Mediter-ranean area. The Service Support Environment (SSE) platform created within theEuropean Space Agency (ESA) General Support Technology Programme (GSTP)project, provides an advanced test-bed that supports service prototyping and demon-stration processes, allowing the design of automatically executed workflows, with avaluable decrease of the overall development effort.Atmosphere and wave forecast model data has been integrated in the SSE platformexperiencing new weather forecast services prototyping and deliv-ery, sharing ofsuch a data together with major European Earth Observation (EO) and geospatialdata providers, fostering forecast data integration within other EO services and eval-uating new business opportunities.

1 Introduction

Earth’s weather and climates influencedaily human events, activities, leisure, hol-idays, transportation, commerce, agricul-ture, and nearly every aspect of our livesas well as human history.Our fascination with the weather has led to24-hour weather networks, feature-lengthmotion pictures, and an explosion of de-

tailed weather data over the Internet.Due to this great impact on human activ-ities, the study, the observation and statusprediction of weather conditions are funda-mental tasks of any forecasting center. Forthis reason an operational meteo-marineforecasting chain for the whole Mediter-ranean area has been implemented at Con-sorzio LAMMA (Laboratory of Monitor-ing and Environmental Modelling for the


Technologies

sustainable development). The forecastingsystem, operating since 2006, consists ofa meteorological component, based on theWRF-ARW atmospheric model, and a ma-rine component, based on the WW3 wind-wave model [1, 2]. The current configura-tion of the WRF-WW3 chain is the result ofseveral years of study and operational ac-tivity at LAMMA, in the fields of numeri-cal weather prediction and development ofmeteo-marine operational forecasting ser-vices.The great and wide request of forecast dataand the more and more appropriate and de-tailed forecast information availability gen-erates a great interest in defining and pro-viding user taylored forecast information.That means new services and new opportu-nities to exploit forecast data.In recent years (2005) the European SpaceAgency (ESA) made operational the Ser-vice Support Environment (SSE) platform.This platform, born in 2004 from an ESAGeneral Support Technology Programme(GSTP) project - known as Multiple Ap-plication Support Service System (MASS),provides an advanced test-bed that supportsservice prototyping and demonstration pro-cesses, allowing the design of automati-cally executed workflows, with a valuabledecrease of the overall development effort.Due to the fact that SSE is a neutrally man-aged, open and distributed platform that of-fers an exclusive opportunity to integratea wide range of heterogeneous Earth Ob-servation (EO) and geospatial informationservices, including product catalogues, ithas fostered the integration of different ser-vices in the EO and Geography Informa-tion Systems (GIS) domains.In this paper the key elements of the inte-gration of meteo-marine forecasting modeldata into the SSE platform are under-lined. The opportunity provided by such

an open platform, to share forecast datawith the major European EO and geospa-tial providers, to easily create new servicesand to evaluate their business model, is alsoassessed.

2 Numerical modelsThe LAMMA meteo-marine forecastingWRF-WW3 chain runs on a cluster ofLinux PCs hosted in the LAMMA facili-ties in Livorno.The system performs two operational runsevery day, the first initialized at 00:00 UTCand the second at 12:00 UTC. Both runsproduce wave forecasts over the wholeMediterranean sea, at 0.1 degrees of res-olution (namely about 10 km at 45° oflatitude), for 72 hours (72 h, i.e. threedays). Furthermore, the run initialized at12:00 UTC also produces high resolution72h forecasts (about 2 km of resolution)over a nested domain, covering the Lig-urian and high Tyrrenian seas.

2.1 Meteorological modelThe version 2.1.2 of the WRF-ARW modelis used in the LAMMA meteo-marineforecasting chain both for scientific pur-poses and for the regional weather ser-vice. ARW is the dynamical nonhydro-static solver of the Weather Research andForecasting (WRF) system that is devel-oped and maintened by the Mesoscaleand Microscale Meteorology Division ofthe National Center of Atmospheric Re-search (NCAR). The main features ofthe model are: Fully compressible, Eu-ler nonhydrostatic, Conservative for scalarvariables, Terrain-following hydrostatic-pressure, Arakawa C-grid, Time-split in-tegration using a 3rd order Runge-Kutta

2624


Figure 1: Meteorological model domain.

scheme with smaller time step for acous-tic and gravity-wave mode.In the LAMMA configuration, the modelhas an horizontal grid with a resolutionof 0.2 deg (about 20 km) over a domaincovering the whole Mediterranean area(224x136 points), as shown in Figure 1,and a vertical grid of 31 levels unequallyspaced from ground to 100 hPa, with thefirst 10 levels concentrated in the bound-ary layer (about 1.0 km above the groundlevel). The model runs with a 100 s time-step.Initial and boundary conditions, updatedevery six hours, are given by the NCEP- Global Forecasting System (GFS) deter-ministic model at the current resolution ofT382L64 (about 50 km).

2.2 Wave model

The WAVEWATCH III full-spectral third-generation ocean wind-wave model used

in the meteo-marine forecasting chain hasbeen developed at NOAA/NCEP [3] in thespirit of the WAM model. In the LAMMAoperational chain it is implemented the ver-sion 2.2 of WW3, with the following con-figuration:

• Latitude-longitude spherical grid• Discrete interaction parametrization

(DIA)• JONSWAP bottom friction formulation• Linear wind and current interpolation• Seeding of high-frequency energy• Spectral discretization on 30 frequencies

in the interval 0.04-1.1 Hz and 36 direc-tions.

The horizontal resolution of the grid overthe whole Mediterranean sea is 0.1 degreesalong both the meridian and zonal direc-tions, while the resolution of the nested run(see Figure 2) is 0.02 degrees along bothdirections.Wind forcing data are taken from the out-put of the WRF-ARW model. A scaling

2625

Technologies

Figure 2: Domain of the WW3 high resolution grid.

factor of the WRF 10 meters wind speed isused as tuning parameter in order to opti-mize the WRF-WW3 (one-way) coupling.

2.3 Models output dataBoth meteorological and wave models gen-erates data for every hour of forecasts overthe defined domains. The data are stored ona netcdf file. NetCDF (Network CommonData Format) is a machine-independent,self-describing, binary data format stan-dard for exchanging scientific data. Thesefiles are stored on a local well-defined filestructure.

3 SSE platformThe Service Support Environment (SSE)(SSE, 2004) (ICD, 2008) provides areusable Service Oriented Architecture(SOA) for the integration of services in the

Earth Observation and Geography Infor-mation Systems domains. A SOA is com-posed by a set of principles that define aloosely coupled architecture and comprisesservice providers and service consumersthat interact according to well defined in-terfaces. The primary goal of SOA (thusof the SSE) is to expose application func-tions in a standardized way so that theycan be leveraged across multiple domains(EO and GIS for the SSE). This approachgreatly reduces the time, effort and cost ittakes to maintain and expand solutions tomeet business needs.The main entry point of the SSE system isan Internet Portal where users, EO and GISservice providers and EO data providersare integrated in a coherent supply chain.The main components (see Figure 3) ofthe SSE system are: the SSE Portal; theArea of Interest (AOI) tool which allowsthe users to select an area of interest in aconsistent way; the SSE Toolbox which fa-

2626


Figure 3: SSE architecture.

cilitates integration of legacy services notbased on Simple Object Access Protocol(SOAP); an orchestrating workflow engine(the ORACLE BPEL PM has been selectedas Commercial off-the-shelf component).Being a SOA, the SSE is based on a non-proprietary, open, and scalable architecturefor the development, implementation anddeployment of services. It also supports themain OpenGis interfaces.The environment provides various func-tionalities, useful for users as well as ser-vice providers. Its main functions are:

• User and Service registration,• Service directory with browsing and

search functionality,• Request for Quotation and Order man-

agement,• Cross catalogue searches,• Workflow graphical editing and monitor-

ing,• Module for: AOI selection, layered ser-

vice result visualization, integration oflegacy services

• Service and company category manage-ment,

• News item management via browser andemail,

• Notification by email of new serviceavailability,

• Service subscription,• Consultation of order list via a mobile

device.

3.1 Interaction model

The SSE infrastructure does not make anassumption about the time needed by theremote service provider to generate or re-turn a service result. Some services may re-quire days, others such as data access maybe almost instantaneous. The SSE Portalkeeps results of orders (i.e. the XML mes-sage) in a database until the user accessesthe Portal again. The actual data in mostcases remains available a number of dayson the service provider side for the user todownload.SSE supports two interaction models:• Asynchronous operations: when an op-

eration may take too long for a user toget the result immediately displayed inhis browser, the result of the is stored inthe SSE database and is available in theuser’s order list when he logs in the nexttime.

2627

Technologies

• Synchronous operations: when the ser-vice will immediately return the result,the SSE Portal visualizes it on the Webpage.

3.2 Data delivery

Depending on the service, service resultscan be large multi-megabyte files (e.g.raster images), smaller vector images orjust textual information. The SSE does notmake any assumption about the format ortype of a service result. The way informa-tion is presented on the Portal is defined bythe service provider. The SSE support var-ious data (i.e. service result) distributionmechanisms:• data included in SOAP message, textual

or graphical (SVG),• data via FTP (URL in SOAP message),• data via HTTP (URL in SOAP message),• vector data via an OGC WFS server,• raster data via an OGC WCS server.Also a combination of these data deliverymethods can be used.

4 Data integration

Forecast model data integration into theSSE environment has gone through dif-ferent steps and phases scheduled by twoprojects collaboration bewteen INTECSand LAMMA. Within this projects threedifferent services have been taylored andintegrated.

4.1 Meteo-Layer

The Meteo-Layer service [4] provides theend user with a three-hour product describ-ing atmospheric condition forecasts over aspecific area of interest. Based on the AOI

request by the final user the system inter-face with the LAMMA backend to retrievethe information necessary to fulfil the enduser request. The output data, includingonly few out of more than hundred meteo-rological model variables, can be deliveredin different formats according to the userinput either as netcdf or grib files or as jpegmaps.

4.2 Sea State Conditions

The Sea State Condition service [5] pro-vides the end user with data for every hourof sea state condition forecasts over itsarea of interest within the Mediterraneanarea. Based on the AOI request by the finaluser the system interface with the LAMMAbackend to retrieve the information neces-sary to fulfil the end user request. The out-put data can be delivered in different for-mats according to the user input either asnetcdf or grib files or as sea state conditionsjpeg maps.

4.3 Weather along the route

The Weather along the route (WAtR) ser-vice [6] allow the users to define a set oflocation within an AOI. Given this set ofpoints, by means of latitude, longitude andtime of stay in that specific location, a me-teorological bulletin is generated extract-ing and processing data from the latest 72hours of forecast models data.Moreover maps of both winds and wavedata are provided to cover the area of in-terest of the path defined by the user.The output data are delivered to the user viaa bullettin file in pdf format.

2628


Figure 4: Interaction between users, SSE portal, TOOLBOX and LAMMA backend.

4.4 Services flow-chartAll the three services foresee a first phaseto request a quotation (RFQ), even if theservice if free of charge, where the userget a preview of the products. Once theRFQ has been completed the user can pro-ceed to order the product that will be madeavailable for download in the data formatchoosen by the user itself (see Figure 4).As depicted in the Figure 4 the workflow ofthe services is as follow:• The end user connects to the SSE portal,

browses the service catalogue and selectsthe service.

• The user selects either a set of points anddates directly from the map or a set ofdata, depending on the related service,and decides the level of detail he wantsto obtain.

• The end user sends a quotation requestto the service (RFQ). The TOOLBOX in-stallation receives the request and calcu-lates the delivery time according to thedata availability.

• The system extracts the data from themodel and collect all the informationneeded to build the response, creates theresponse and sends it back to the portal.

• The results are displayed on the portal.• The user checks the response and decides

to proceeds with the order.• The Toolbox extracts the information

from the request, looks for the file match-ing the input criteria, invokes the netcdfprocessor to extract the data from themodel data matching the end user inputparameters, publishes the output data fileon the internal FTP server and displaysthe forecasts symbols on the portal.

• The end user checks the results on theSSE pages and downloads the productdata file.

The services previously described exposesthe following SOAP interfaces via the SSEToolbox:

• Request for Quotation (RFQ): it returnsprice, delivery time and payment infor-mation. Moreover it returns some pre-

2629

Technologies

view images to be displayed to the user.• Order: it defines the Order request and

Order result. The result message con-tains an FTP URL, with user name andpassword to allow the user to downloadthe actual result file and the link to theWMS/WCS server to be linked to theArea Of Interest Tool.

In this context the Toolbox use a SOAPwrapper of Shell scripts to extract informa-tion from the model data files. Moreover itadds some logic to the service to calculatevarious parameters needed for the integra-tion.

5 ConclusionsThe availability of the SSE platform andall its components (e.g. TOOLBOX, AOItool) has facilitated the prototyping and de-velopment of the three services in the webservice framework.The integration has allowed the distribu-tion of the metereological data generated

by LAMMA into a web based distributedand chainable architecture. This approachhas carried out many advantages in termsof:• direct accessibility of data according to

the user needs,• possibility to chain meteorological data

with other services to easily create moreand more value added services,

• easy generation of new complete ser-vices without worring of user interface,workflow definition and billing servicedevelopment.

The SSE system allows an easy web en-ablement of manual and semi automaticservices via the Toolbox and an easy userinterface development via the portal andthe AOI tool (by means of configurationfile). Thus the SSE provides the possibilityto publish prototypes and real services witha small effort and high impact on the mar-ket. This is also an opportunity for mete-orological data providers to integrate theirlocal data with major European Earth Ob-servation and geospatial data providers.

References[1] A. Ortolani et al. Implementing an operational chain: the Forence LaMMA labora-

tory. Advances in Global Change Research, 28, 2007.

[2] A. Orlandi et al. Implementation of a meteo-marine forecasting chain and compari-son between modeled and observed data in the Ligurian and Tyrrhenian seas. 2011.

[3] H. Tolman. User manual and system documentation of WAVEWATCH-III version2.22. NOAA/NWS/NCEP/MMAB Technical Note, 222, 2002.

[4] ML. Meteo Layer service. services.eoportal.org/portal/service/ShowServiceInfo.do?serviceId=43804C89, 2008.

[5] SSC. Sea State Conditions service. services.eoportal.org/portal/service/ShowServiceInfo.do?serviceId=8F813580, 2009.

[6] WR. Weather along the route service. services.eoportal.org/portal/service/ShowServiceInfo.do?serviceId=8F814280, 2009.

2630

Merging Physical Oceanographic Observa-tions and Simultaneous Seismic Reflection toStudy Shelf-Slope Processes: the Example ofADRIASEISMIC-09

S. Carniel1, J.W. Book2, R.W. Hobbs3, A. Bergamasco1, M. Sclavo1

1, Institute of Marine Sciences, CNR, Venezia, Italy2, NRL Stennis Space Center, Washington, USA3, Durham University, Durham, [email protected]

Abstract

During the period March 3-16, 2009 an international collaborative field experi-ment of Physical and Seismic Oceanography (ADRIASEISMIC-09) was carried outon board the CNR R/V Urania in the southern Adriatic Sea in order to characterizethe details of the North Adriatic Dense Water (NAdDW) mass structures and test thefeasibility of the seismic approach in shallow basins.The new discipline of Seismic Oceanography is particularly well suited for study ofthe dynamics of bottom-trapped water masses as compared to classic techniques be-cause it measures the full water column at 10 m horizontal resolution, it measuresremotely and measurements are not hampered by a sloping bottom or concerns ofinstrument bottom impact, and it can operate successfully over the entire range from100 m to 1000 m for tracking water-masses evolution down a slope.A fast seismic data processing strategy was used, so that the seismic data were as-sessed for optimum locations of XBT (232), CTD and micro-structure (101) casts.These data of extremely fine detail (order of millimeters) complemented the high lat-eral resolution of the seismic image, allowing to track a thin bottom-boundary layerdescending down the slope near Palagruza sill.

1 Introduction

Holbrook et al. [1] showed that seismicreflection profiles, adequately treated andintegrated by hydrological data along thewater column, are useful to provide infor-mation on the thermohaline structure of thewater masses. This technique is commonlycalled Seismic Oceanography (SO) or Geo-physical Oceanography (GO). Indeed, it isnow possible to observe the oceanic dy-namic and some structures (such as eddies,

double diffusion, thermohaline intrusions,internal waves) at a very high detail.In the field of the SO there exist still sev-eral problems to be solved, such as the def-inition of protocols to allow a better com-parison of acquired data, the developmentof techniques to obtain optimal acquisition,especially in shallow waters, the need ofdevelopment appropriate processing algo-rithms, etc. This will hopefully allow toobtain useful information on the ocean ver-tical microstructure as well.


Technologies

The necessity of testing and developingthese methodologies, as well as the need ofimproving the knowledge of the processesat the water/sediment interface, are at thebasis of the ADRIASEISMIC-09 cruisecarried out using the Italian CNR R/VURANIA. Indeed, the difficulty of acquir-ing synoptic data (e.g. classical hydrology,currentmeter and meteo data, turbulence,etc.) has until now limited the optimal useof modeling potentialities and an adequatesynthesis of results. The availability ofsynoptic observations as those provided bythe seismic oceanography technique couldallow then to monitor much lager regionsin a shorter time and, with the help ofmore classical measurements such XBTsor CTDs tow-yo, together with the supportof turbulence measurements obtained viafree falling probes, to depict and outline thevertical stratification in frontal regions at avery high resolution.The interest at the origin of this proposalwas also shared by several internationalresearch groups, including the Naval Re-search Laboratory (NRL) and the Uni-versity of Durham, that took part in thecampaign. In order to show the feasi-bility and the current state of the art inthe field of the seismic oceanography, wehave selected the southern Adriatic Sea re-gion (close by the Gargano promontory andthe Bari canyon, see Figure 1) for twoweeks at the end of winter 2009. The rea-son for this choice was due to the exis-tence of a legacy in the oceanographic andseismic databases, the presence of differ-ent oceanographic characteristics such asfronts and filaments, and the bottom topog-raphy modified by cascading and overflowprocesses.The benefits resulting from the proposedactivity appear to be multiple: the avail-ability of new seismic-hydrological data

that would be difficult to be collected bya purely national team; if present, the 3Dmapping of the NAdDW (North AdriaticDeep Water) deep water structure when ex-iting from the Palagruza Sill (see again Fig-ure 1) and starting its cascading processalong the slope of the south Adriatic pit);the increase in the distributed knowledgeinside the EU; the use of a large quantity ofseismic data previously acquired; the foun-dation of a national research theme in thefield of the seismic oceanography whereCNR would be among the leading institu-tions.

2 Objectives and deliver-ables

The main objectives of ADRIASEISMIC-09 cruise were:

a) to test and explore benefits and limita-tions of state-of-the-art methods to ac-quire seismic data in a shallow water en-vironment with a relatively light equip-ment.

b) to foster collaboration between the seis-mic and oceanographic community, ad-vancing the knowledge of the watermasses dynamics, such as fronts, inter-nal waves, and of the vertical mixingprocesses on the shelf region, by suc-cessfully exploiting the above mentionedtechniques.

c) to produce a dataset in the area in frontof Bari region in the southern Adriatic,both from the oceanographic and seismicpoint of view.

d) to explore the feasibility of unlockingpreviously acquired data to make themavailable to the oceanographic commu-nity.

e) to contribute to building up a EU exper-

2632


tise on these fields, allowing CNR to bebetter connected with Research Institu-tions that are making cutting-edge sci-ence.

f) to reach a detailed characterization ofphysical properties of the Southern Adri-atic water masses, including turbulentproperties, improving the performancesof turbulent closure models.

g) to improve the understanding of bottomboundary layer processes in the basin.

Several types of products and deliverablesare expected:

a) an observational data-set for seismic re-flectivity in the Adriatic sea region.

b) an observational data set from CTD ac-quisitions, increasing the existing one.

c) an evaluation study of the eddy diffusiv-ity in the southern Adriatic basin.

d) dissemination of ADRIASEISMIC-09data in future studies.

e) joint publications dealing with: insightson the capability of seismic oceanogra-phy techniques, microstructure and tur-bulence measurements, water masses dy-namics and distribution and cascadingprocesses.

3 MeasurementsSeismic dataA fundamental aspect of theADRIASEISMIC-09 cruise was the tightintegration of classical physical oceanog-raphy measurements with the seismic re-flection images of water structure. Thisrequires fast seismic data processing tolocate at best the CTD casts. This extraprocessing work is partly compensated bythe relatively simple average sound-speedwithin the water column. This was the firstattempt to capture seismic images of waterstructure in water depths of less than 400-

500 m, requiring thus testing for the mostsuitable seismic source-receiver configura-tions.

CTD castsAt all the hydrological stations, pressure,conductivity, temperature and dissolvedoxygen concentration were measured witha CTD-rosette system. Salinity and poten-tial temperature were then derived from themeasured parameters. A fluoremeter and alight transmission sensor were also operat-ing. The data were processed on board inorder to correct immediately for the coarseerrors.

Expendable bathy-thermographs(XBTs)The structures in the ocean change on timescales shorter than the time required to de-ploy a seismic system, acquire the data andretrieve the system. Also, once the seismicsystem is deployed the ship can not stopfor a CTD cast. The use of expendablebathy-thermograph (XBT) probes werethen required to provide the backgroundthermal structure and high vertical reso-lution needed to corroborate the seismicdata. We deployed 240 XBTs along theseismic lines in the ADRIASEISMIC-09experiment, of which 232 functioned prop-erly. The XBTs were deployed at timeintervals ranging from 5 to 40 minutes, de-pending upon the complexity of the watermass structure estimated from the last cast.

Lowered acoustic doppler current pro-filer (LADCP)One Lowered Acoustic Doppler CurrentProfiler (LADCP) was used to measure ve-locity profiles. The bin size was 5 m andthe range about 100 m. Ensembles areformed every second, along the down- andupcast. This allowed to obtain a vertical

2633

Technologies

profile of horizontal velocities, and to as-sociate a velocity field to each water mass.Data were processed on board.

Vessel mounted ADCPsThe hydrographic and seismic data set havebeen integrated with direct current mea-surements. During the campaign two VM-ADCPs (75 and 300 KHz) operated alongthe whole ship track. The bin size was16 m and 4 m, the depth range of the twocurrent profilers about 700 m and 110 mrespectively. Data were corrected for er-rors in the value of sound velocity in water,and misalignment of the instrument withrespect to the axis of the ship. Data of thefirst part of the cruise were processed onboard.

Microstructure turbolence profilerMicro-Structure Turbulence (MST) pro-filers was used on the R/V URANIA andprovided simultaneous microstructure andprecision measurements of physical pa-rameters. The MST profiler contained twovelocity microstructure shear sensors, amicrostructure temperature sensor, stan-dard CTD sensors for precision measure-ments and a turbidity sensor. Measure-ments of microscale temperature and ve-locity shear allows for the estimation ofseveral turbulence-related variables includ-ing turbulent kinetic energy and temper-ature variance dissipation rates, and eddydiffusivities. Measured parameters were:Pressure, CTD temperature, Conductivity,Fast temperature, Velocity shear, Turbid-ity. Computed parameters were: Salinity,Density, Sound velocity, Brunt Vaisala-Frequency, Seismic reflectancy, Thorpescale (from density), Ozmidov scale, Coxnumber, TKE dissipation rate, Thermaldissipation rate, Eddy diffusivity fromdissipation, Eddy diffusivity from Cox

number and Eddy diffusivity from Thorpescale. A total of 101 casts were acquiredduring the cruise.

Multibeam and chirpDuring the cruise about 1600 km of CHIRP(Compressed High Intensity Radar Pro-file) profiles have been acquired in the Baricanyon and offshore the Gargano Promon-tory. Along with the CHIRP profiles,multibeam bathymetry was acquired andrecorded. The multibeam system operateswith frequencies between 70 and 100 kHzand owns a resolution of 1 degree by 1 de-gree.

Meteorological dataR/V URANIA was equipped with a fullsuite of meteorological sensors, acquiringwind magnitude and direction, sea surfacetemperature, relative humidity, incomingsolar radiation and ship position.

4 Preliminary results

Previous studies demonstrated the pres-ence of NAdDW in the southern Adri-atic basin in recent years, and the mix-ing down to depths greater than 600 m inthe Southern Adriatic Pit ([2] see Figure1). ADRIASEISMIC-09 showed that theNAdDW was absent in the southern Adri-atic basin in 2009, being the water comingfrom the northern basin anomalously light.From the seismic oceanographic pointof view, the cruise showed how high-frequency and very high-resolution sam-pling is required to illuminate the mixingprocess dynamics, showing that a seismicprofile may provide a consistent image ofdT/dz of approximately 10-15s of meters.In general, while every seismic reflectioncorresponds to a water mass boundary, not

2634


every water mass boundary causes a re-flection. Moreover, high frequency char-acter of water mass boundaries can changeabruptly (10-20 m) and may not be interpo-lated or extrapolated away from point mea-surements.Overall, ADRIASEISMIC-09 was the firstcruise showing a successful use of SOto image water structures close to sea-bed in shallow water; for the first timeseismic data were acquired simultaneouslywith turbulent (micro-structure) ones. Thecruise proved also that high quality datacan be obtained by using a light-weigthseismic system for SO (Source: 2 GI air-guns 45/45 cu in, harmonic mode. Re-ceiver: max offset 1200 m). Preliminaryresults suggest therefore that SO can pro-vide a new and powerful tool for under-standing the detailed horizontal structure ofDense Shelf Water Cascading processes.Figure 2 presents and example of seismicacquisition along the Bari canyon. Thewiggled lines imaged in the plots are re-

lated to density variations along the watercolumn, possibly due to internal waves dy-namics. Figure 3 presents an example ofseismic acquisition along a different lineagain in the Bari canyon, with overlappedreflectivity profiles obtained via 12 concur-rent XBT measurements. Note the qualita-tive matching between XBT (temperature)signature and the seismic patterns.On-going work is focusing on correlatingthe observed ocean fine structures with in-ternal waves dynamics detected.

5 AcknowledgementsThe authors thank to G. Bortoluzzi andK. Schroeder (CNR-ISMAR) for the helpin XBT data acquisition and elaboration.The scientific staff of ADRIASEISMIC-09wishes to thank the Italian National Re-search Council (CNR) Ship Office, whichmade the R/V URANIA available for theCruise.

2635

Technologies

Figure 1: Dense shelf water pathways (from Vilibic and Supic, 2004). The red boxincludes the region of investigation of the ADRIASEISMIC-09 campaign.

2636


Figure 2: Example of seismic acquisition along the Bari canyon. The wiggled lines im-aged in the plots are related to density variations along the water column (possibly dueto internal waves).

2637

Technologies

Figure 3: Example of seismic acquisition along the Bari canyon, with overlapped re-flectivity profiles obtained via 12 concurrent XBT measurements. Note the matchingbetween XBT (temperature) signature and the seismic patterns.

2638


References[1] W.S. Holbrook, P. Paramo, S. Pearse, and R.W. Schmitt. Thermohaline Fine

Structure in an Oceanographic Front from Seismic Reflection Profiling. Science,301(5634):821–824, 2003.

[2] I. Vilibic and N. Supic. Dense water generation on a shelf: the case of Adriatic Sea.Ocean Dynamics, 55(5-6):403–415, 2004.

2639

Technologies

2640

The Infrastructure for Geospatial Information In-teroperability in the SeaDataNet Project

S. Nativi1, G. Manzella2, P. Mazzetti1, M. Santoro1, E. Boldrini11, Institute of Methodologies for Environmental Analysis, CNR, Potenza, Italy2, Marine Environment Research Center “S. Teresa”, ENEA, Lerici, La [email protected]

Abstract

This paper describes an interoperability solution for geospatial resources (i.e.data, services, models and applications) sharing developed in the context of the FP6European Project SeaDataNet. It has been designed taking into account the projectmission: to develop a Pan-European infrastructure for the oceanographic and marinedata and products management, indexing and access.Presently, in the framework of oceanography and marine resources, several interna-tional standards can be implemented in order to access the available resources (e.g.ISO TC211 standards, OGC specifications and INSPIRE Directive implementingrules). At the same time, there exists a number of well accepted community standardssuch as: THREDDS, OPeNDAP, netCDF-CF. The proposed solution is based on the”System of Systems” approach, adopting solutions to make the mentioned specifica-tions interoperable. The resulting infrastructure implements mediation and broker-ing functionalities in order to provide, both scientists and decision makers, with aconsistent and practical way for the discovery, evaluation, and download of hetero-geneous resources needed by SeaDataNet applications. Basing on the conclusionsof this work, the paper presents some recommendations for marine-oceanographicinformation sharing systems.

1 Introduction

The planning and implementation of re-search, and the efficient management ofthe resulting data are still two widely sep-arated realms. Data managers consider thecareful collection, management and dis-semination of research data as essential forthe effective use of research funds. Manyresearchers, on the other hand, considerdata management as a technical issue as-signing a low priority to that. The existinginitiatives in Data Management and Infor-mation Systems are trying to bridge thesetwo different realms.

The concept of data managementhas changed deeply during the lastdecade, moving from the idea ofcentralised-homogeneous systems to thatof decentralised-heterogeneous ones. Thecosts associated with homogeneous andheterogeneous systems implementationare:

• Homogeneous system - a common dataformat for a common application:- High demands on data sharing and use- Low demands on system design, de-

velopment, and maintenance.- High demands on system extension

and interoperability.


Technologies

• Heterogeneous system - multiple dataformats for multiple applications:

- Low demands on data sharing and use- High demands on system design, de-

velop, and maintenance- Low demands on system extension and

interoperability.The centralized-homogeneous data man-agements were important when the infor-mation system technology was based onmainframes and Internet was less devel-oped. However, they have never been at-tractive to oceanography researchers formany reasons:- they were not directly involved in sys-

tems management since the responsibil-ity was lying on data managers only;

- it was not popular to encode data in aunique format, since this was based onthe unrealistic idea that all instrumentsand systems have a common interface;

- data access could be time consuming.During last decade it was also evident thatdata are not only coming from oceano-graphic ‘in situ’ observations, but they arealso generated by different instruments andsystems, such as: remote sensing sen-sors, numerical modelling systems, web-enabled resources, etc.The tradition of creating formal data cen-tres to house and distribute ocean data andproducts has long been shadowed by theparallel existence of direct data downloadsfrom research groups (usually via HTMLpage-links and/or FTP servers). Histori-cally, at least half of the data available toresearchers has been provided outside theIODE/World Data Center System facilities(IODE, 2010) [1]. During the last decadesefforts have been dedicated to create a newsystem of Internet communication (e.g. theData Access Protocol [2, 3]) that would al-low users to access distributed datasets di-rectly, using a family of ”middleware ser-

vices” that overlay datafile caches, allow-ing retrieval of files in desired formats, nomatter what the storage format might be.The web has changed the way we discoverand access data, and mush-up information.The simple concepts sustaining central-ized information systems were overcomeby emerging requirements:- Data are heterogeneous.- Real time data delivery is important for

policy making.- Processing is required to be done outside

of the data archive (e.g. using Web orGrid computing infrastructures).

- Quick and easy access to raw data in itsnative format is required.

- Web-based discovery (i.e. catalogue)services are required.

- Instrument calibrations and informa-tion regarding deployment and recov-ery, should be digitized and made avail-able electronically; people need to be re-moved from this process wherever possi-ble.

- An architecture that supports distributedprocessing and archiving of higher orderdata products is required.

In order to implement these interoperabil-ity requirements, three types of metadatalevels are needed; they support importantsystem functionalities:1. Discovery metadata (Directory func-

tionality): it consists of broad descrip-tions of the contents of data sets used tolocate data sets of potential interest.

2. Evaluation metadata (Inventory func-tionality): consists of a detailed listof granules (e.g., individual satellitepasses, AUV dive, CTD cast) within adata set used to find the data of interest.

3. Use metadata (Data functionality): con-sists of the actual data objects.

To implement loosely coupled interoper-ability and data sharing, data and metadata

2642


encodings should be auto-explicative andauto-consistent. This is accomplished byproviding (or sharing) knowledge on:• Syntactic and lexical encoding: informa-

tion about the data types and structuresused by the encoding often referred to asthe data model.

• Semantics encoding: information aboutthe data content, what the variablesmean, their possible values, etc..

These general concepts are now applied inthe system of systems developed by Sea-DataNet [4]: an EC funded project aimingto create and operate a pan-European, ma-rine data management infrastructure, ac-cessible online through a unique portal.The primary goal of SeaDataNet is to de-velop a system which provides transpar-ent access to marine data sets and dataproducts from 36 countries in and aroundEurope. SeaDataNet aims to insures thelong term archiving to the large numberof multidisciplinary data (i.e. temperature,salinity current, sea level, chemical, physi-cal and biological properties). Hence, theproject is developing a standardized dis-tributed system for managing the large anddiverse datasets collected by the oceano-graphic fleets and the new automatic ob-servation systems. By use of standardsfor communication and new developmentsin Information Technology (IT), the 40 in-situ and satellite marine data platforms ofthe partnership provide metadata, data andproducts as a unique virtual data centre[4]. The core of the SeaDataNet infras-tructure is represented by some discoveryservices (e.g. EDMED, EDMERP, EDMO,CSR, EDIOS) and the Common Data Index(CDI), an XML document allowing the ac-cess to data in a distributed system [5].The middleware service adopted by Sea-DataNet for discovery and query is calledCDI (Common Data Index). However, the

collaboration with other data managementsystems based on THREDDS (ThematicRealtime Environmental Distributed DataServices) [6] required the construction ofa ‘bridge’ between the two technologicalsolutions. In fact, these two technologiesare well adopted by the information sys-tems managed by the SeaDataNet partners.

1.1 CDI

The CDI provides an index (metadatabase)to individual data sets. Furthermore, itpaves the way to direct online data accessor direct online requests for data access orfile downloads; in fact, the primary objec-tive of CDI is to give users a highly detailedinsight in the availability and geographicalspreading of marine data across the differ-ent data centres and institutes across Eu-rope [5]. CDI was initiated in the EU Sea-Search project [7].Currently, it is being further developedand extended by the SeaDataNet project.All the 35 countries participating in Sea-DataNet are going to utilize CDI for de-scribing their marine data [5].

1.2 THREDDS

THREDDS is middleware to bridge the gapbetween data providers and data users. Thegoal is to simplify the discovery and use ofscientific data and to allow scientific publi-cations and educational materials to refer-ence scientific data [6].THREDDS initial focus was to allow datausers to find datasets that are pertinentto their specific education and researchneeds, access the data, and use them with-out necessarily downloading the entire fileto their local system. To achieve this,data providers must be able to publish

2643

Technologies

lists of what data are available and to de-scribe their data enabling discovery anduse: THREDDS introduced the Data In-ventory Catalogs (DICs) concept [8]. Theyare XML documents that describe on-linedatasets. These catalogs can contain ar-bitrary metadata; besides, a standard setof metadata to bridge to discovery cen-ters -like GCMD, DLESE and NSDL-was defined. The THREDDS catalog ser-vice augments traditional MeteoOcean ac-cess services such as OPeNDAP [3] andnetCDF [9] access via HTTP. Moreover,THREDDS implements access services tobridge the MeteoOcean and GIS communi-ties, such the OGC Web Coverage Serviceinterfaces.

2 Distributed Catalog Ser-vices

The interoperability solution we pro-pose consists of utilizing a standard Dis-tributed Catalog Service [10] to: (a) feder-ate the THREDDS/OPeNDAP and CDIresources (i.e. services and data); (b)mediate between THREDDS/OPeNDAPand CDI resources; (c), exposeTHREDDS/OPeNDAP and CDI resourcesthrough a standard interface for geo-spatialinformation discovery and access (i.e. im-plementing international standards speci-fications for catalog and access service ).Figure 1 depicts a schema of the adoptedinteroperability approach. This solutionimplements and publishes the OGC CS-W [11] standard catalog interface for re-sources discovery and access. It is flexibleallowing the federation and mediation ofother interoperability standards (i.e. inter-national standards as well as communitybest practices), such as: (a) OGC access

services (i.e. WMS, WCS, WFS); (b) Bio-diversity community protocols (i.e. theGBIF (Global Biodiversity Information Fa-cility) discovery and access protocols).

3 GI-catThe deployed distributed catalog frame-work is based on the GI-cat technology[10]. The GI-cat supports caching and me-diation capabilities; it can act as a servicebroker component accessing disparate (i.e.heterogeneous and distributed) resourcesand publishing a standard and common in-terface. GI-cat implements a frameworkto federate international standard and com-munity service types, namely: catalog, in-ventory, access, and processing services.As for international standards, GI-cat sup-ports: most of the OGC CS-W applicationprofiles, WCS, WFS and WMS specifica-tions. As to community services, GI-catis able to federate: THREDDS/OPeNDAP,CDI, and GBIF services. GI-cat exposesseveral standard discovery interfaces, in-cluding: (a) the OGC CS-W/ISO inter-face [12], recommended by the Euro-pean Directive INSPIRE; (b) the OGC CS-W/ebRIM-EO interface [13, 14], recom-mended by the GMES/ESA-HMA initia-tive [15]; (c) the OGC CS-W Core[11], rec-ommended by the GEO/GEOSS [16].

4 Mediation Process: GI-cat Accessors

In the GI-cat framework, the geospatial re-sources heterogeneity is addressed by ap-plying a mediation approach. SpecificGI-cat components implement the media-tion services required to interface heteroge-neous and standard (i.e. well-known) ser-

2644


Figure 1: The adopted interoperability solution for CDI and THREDDS.

vice providers. They are called Accessors.They solve the data models multiplicity is-sue by mapping providers data models ontoa more general resource model: the ISO19115 Core profile [17]. The Accessorsalso implement a query brokering function-ality by translating the query requests re-ceived by GI-cat -i.e. queries expressed ac-cording to the interface protocols exposedby GI-cat (e.g. CS-W/ISO)- into the multi-ple query dialects exposed by the resourceproviders. In the SeaDataNet system, theCDI and THHREDDS/OPeNDAP Acces-sors make those resource providers inter-operable.The described interoperability approachresults quite general and flexible en-abling to implement interoperability be-tween CDI and THREDDS/OPeNDAP,

and other provider technologies.

4.1 CDI AccessorFigure 2 shows a UML class diagram rep-resenting a simplified version of the CDIdata model. This is not a hierarchical datamodel. In fact, the data model only dealswith low level granularity not defining highlevel granularity elements, such as “cata-log” or dataset “aggregations”. Hence, aCDI document (i.e. an XML document)describes only a dataset element formal-izing its content and access metadata. Tocarry out CDI data model interoperabilitywith GI-cat, the correct solution is to mapa CDI resource element (i.e. a CDI dataset)onto a GI-cat “catalog” element. Clearly,this approach entails some important scal-ability and management issues.

2645

Technologies

Figure 2: Simplified CDI Data Model for a dataset element.

as well.

4.1 CDI Accessor

Figure 4 shows a UML class diagram representing a simplified version ofthe CDI data model. This is not a hierarchical data model. Actually, thedata model only deals with low level granularity not defining high levelgranularity elements, such as ”catalog” or dataset ”aggregations”. Hence,a CDI document (i.e. an XML file) describes only a dataset formalizing itscontent and access metadata. To carry out CDI data model interoperabilitywith GI-cat, the correct solution would be to map a CDI resource element(i.e. a CDI dataset) onto a GI-cat ”catalog” element. Clearly, this approachentails some important scalability and management issues.

We experimented a CDI data model extension introducing a datasetaggregation level. Thus, data provider can generate and publish an aggre-gation document (encoded as an XML file) which aggregates existing CDIdocuments - CDI XML files. For the aggregation, the CDI datasets are ref-erenced by pointing the XML file they are encoded by. This is an exampleof CDI datasets aggregation document.

<cdiGroup><cdiUrl title="rad1B0BB">http://moon.santateresa.enea.it/SDN_CDI/rad1B0BB.xml

</cdiUrl><cdiUrl title="rad4EF28">http://moon.santateresa.enea.it/SDN_CDI/rad4EF28.xml

</cdiUrl><cdiUrl title="rad8E3EC">http://moon.santateresa.enea.it/SDN_CDI/rad8E3EC.xml

</cdiUrl></cdiGroup>

In this way we would have only one catalog for all the resources publishedby the aggregation document. GI-cat supports the CDI dataset aggregationextension. Table 1 describes the two mapping possibilities described.

Table 1: CDI elements mapping onto GI-cat ones

CDI element GI-cat elementCDI document Catalog containing one

dataset

CDI document (with datasetsaggregation)

Catalog containing one ormore datasets

8

Table 1: CDI elements mapping onto GI-cat ones.

We experimented a CDI data model ex-tension introducing a dataset aggregationlevel. Thus, data provider can generate andpublish an aggregation document (encodedas an XML document) which aggregatesexisting CDI documents. For this aggre-gation, the CDI datasets are referenced bypointing the XML document file they areencoded by.In this way there exists only one catalogfor all the resources published by the ag-gregation document. GI-cat supports theCDI dataset aggregation extension. Table1 describes the two mapping possibilitiesdescribed.

4.2 Querying a CDI Resource

In general, when GI-cat receives a queryrequest, each federated remote providerreceives it through its specific Accessorwhich translates it, opportunately. This isnot the case of CDI providers; to performqueries over CDI resources, at the begin-ning of each session, GI-cat harvests CDIresource (i.e. retrieves and parses the CDIXML document), maps the content ontothe GI-cat data model, and caches it forsubsequent queries. It is possible to per-form a new harvesting during the session.

2646


Figure 3: THREDDS data model.

4.3 THREDDS Accessor

A simplified schema of the THREDDSdata model is shown in Figure 3.The THREDDS Data Server (TDS) imple-ments a catalog service; hence, it definesa hierarchical data model which containsa catalog element that, in turn, may con-tain zero or more datasets. Each datasetelement may contain zero or more sub-datasets. The “catalogref” element allowsa catalog to refer to other remote catalogs.It is noteworthy that THREDDS datamodel supports metadata inheritance.Therefore, when an aggregated dataset(or a “CatalogRef” element) has some in-heritable metadata, they propagate to thenested (or referred) datasets as well. In

this way, it is possible to factor out com-mon metadata avoiding to duplicate thesame information and encouraging to pub-lish datasets characterized by high levelgranularity. This feature, along with theuse of “CatalogRef” element, consents toquery large catalogs in an efficient way.THREDDS data model mapping on theGI-cat model is quite straightforward (seeTable 2).

4.4 Querying a THREDDS Cat-alog

THREDDS publishes “inventory catalogs”(i.e. XML documents) over the Web.Catalogs can reference other catalogsand can also provide links for accessing

2647

Technologies

Figure 4: THREDDS Catalog tree stategy for improving query performance.

Figure 5: THREDDS data model

!

Table 2: THREDDS elements mapping onto GI-cat ones

THREDDS element GI-cat elementCatalog Catalog

CatalogRef DatasetCollection

Dataset (aggregated) DatasetCollection

Dataset (simple) Dataset

4.2.1 Querying a THREDDS Catalog

THREDDS publishes ”inventory catalogs” (i.e. XML documents) over theWeb. Catalogs can reference other catalogs and can also provide links foraccessing datasets. THREDDS services include the Dataset Query Capabil-ities (DQC) which allows users to request a subset of a dataset collection.However, we preferred to harvest THREDDS inventory catalogs, followingthe approach already described for CDI documents. Thus, the THREDDSaccessor implements query requests. Di!erently from CDI, THREDDS ac-cessor must deal with nested structures which make use of the CatalogRefelement. The accessor implements an incremental approach to avoid un-necessary network overhead and improve performances. Once a THREDDScatalog content is requested, referenced catalogs are not accessed until the

10

Table 2: THREDDS elements mapping onto GI-cat ones.

datasets. THREDDS services include theDataset Query Capabilities (DQC) whichallows users to request a subset of a datasetcollection. However, we preferred to har-vest THREDDS inventory catalogs, fol-lowing the approach already described forCDI documents. Differently from CDI,the THREDDS/OPeNDAP Accessor mustdeal with nested structures which make useof the “CatalogRef“ element. The Acces-sor implements an incremental approach toavoid unnecessary network overhead andimprove performances. Once a THREDDScatalog content is requested, referencedcatalogs are not accessed until the contentof those dataset collections is explicitly re-quested.Since THREDDS data model is hierarchi-cal, a THREDDS catalog can be repre-sented as a tree where each internal node is

either a dataset collection or a referencedcatalog; each leaf is a dataset. Every treenode, and leaf, are characterized by its ownmetadata (which may be inheritable) plusinherited metadata. A query is performedvisiting the tree with a depth first strat-egy, reading the metadata of each node andchecking if the query clause is satisfied.When a leaf is reached and its metadata sat-isfy the query constraints too, the leaf (i.e.dataset) is added to the query response.The presence of inheritable metadata al-lows to discard a node along with its chil-dren, if the node metadata do not satisfy thequery clause. This avoids network over-head when the node to be discarded is a“CatalogRef” element.This strategy results also efficient in orderto deal with very large catalogs. In fact, itallows to discard many branches skipping

2648


to read their metadata.

4.5 Improving Query Perfor-mance with THREDDS

Query performances depends on thematching pattern existing between theTHREDDS hierarchical structure and theclause types characterizing most commonqueries (i.e. users common query strategy).Considering most of queries have a space-time clause, datasets should be structuredaccording to their spatial and temporal ex-tension, respectively. To optimize queryperformances with respect to spatial do-main, a possible strategy consists of di-viding Earth spatial domain in finite ar-eas. A valuable example in point are theMARSDEN zones [18] or WMO zones.Following this solution, each node at thefirst level of the THREDDS catalog treecontains datasets belonging to the sameMARSDEN area. Therefore, at the firsthierarchical level, tree nodes are character-ized by spatial information as inheritablemetadata. While, second-level nodes in-herit spatial metadata and have temporalinterval metadata as inheritable one. Fig-ure 4 shows the tree structure strategy im-plementing this approach. Datasets are firstdivided by space (i.e. MARSDEN areas),then by year, subsequently by month, andso on in accordance with the level of tem-poral granularity required by user queries.

4.6 Access Protocol Adaptation

Neither THREDDS nor CDI define anyspecial access protocol; they both use stan-dard protocols (e.g. WMS, WCS, OPeN-DAP and HTTP download). When GI-catreceives a request, it performs the follow-ing actions:

a) it retrieves the available access service(s)as specified in the resource metadata;

b) in case of many available services, itselects the most flexible one -e.g. theservice which supports domain and co-domain sub-settings;

c) it instantiates a correspondent Accessorcomponent (i.e. a service client);

d) it delegates the Accessor to formulatethe appropriate access request, includ-ing the sub-setting clause -if the resourceprovider implements the subsetting func-tionalities.

5 Experimentation

Several experimentations were success-fully conducted using XBT (Expend-able Bathythermograph) data encodedin netCDF (network Common DataForm) format [9] following the CF1conventions [19]. The utilized CF-netCDF data model is shown in Fig-ure 5. The CDI documents describ-ing the XBT resources are available at:http://moon.santateresa.enea.it/SDN CDI/For testing the proposed CDI cat-alog extension supporting files ag-gregation, we worked out an aggre-gated catalog document, available at:http://zeus.pin.unifi.it/projects/cdi registry/registry.xml. The same XBT resources en-coded in CF-netCDF were also utilized totest the interoperability of THREDDS in-ventory catalogs. The THREDDS catalogserver was configured for improving queryperformances applying spatial and tem-poral domain clauses -using the MARS-DEN convention for dividing Earth in in-dexed areas. This catalog is available at:http://apollo.pin.unifi.it:8080/thredds/XBTdemocatalog.xml. The set up systemfor interoperability testing is composed of

2649

http://moon.santateresa.enea.it/SDN_CDI/

http://zeus.pin.unifi.it/projects/cdi_registry/registry.xml

http://zeus.pin.unifi.it/projects/cdi_registry/registry.xml

http://apollo.pin.unifi.it:8080/thredds/XBTdemocatalog.xml

http://apollo.pin.unifi.it:8080/thredds/XBTdemocatalog.xml

Technologies

Figure 5: CF-netCDF Data Model for XBT datasets.

GI-cat catalog (http://zeus.pin.unifi.it/gi-cat-wiki) deployed in front of a CDI anda THREDDS catalog server. Besides,standard OGC Web access services werefederated in order to extend the interop-erability tests. To demonstrate test re-sults, a graphical client application forGI-cat was utilized: GI-go. This is aCS-W/ISO client which is able to bindwith the corresponding interface pub-lished by GI-cat. GI-go can be down-

loaded at: http://zeus.pin.unifi.it/gi-go.Figure 1 depicts the developed systemused for the interoperability experimen-tations. Tests were performed queryingCDI and THREDDS/OPeNDAP catalogswith combinations of spatial, temporal andkeywords clauses.

2650

http://zeus.pin.unifi.it/gi-cat-wiki

http://zeus.pin.unifi.it/gi-cat-wiki

http://zeus.pin.unifi.it/gi-go


6 Conclusions

To make CDI and THREDDS serviceproviders interoperable, we propose a so-lution based on a distributed catalog ser-vice implementing mediation and broker-ing functionalities, namely GI-cat. Thecarried out interoperability tests were suc-cessful. They were conducted in the Sea-DataNet project framework and demon-strated the effectiveness of discovering andaccessing CDI and THREDDS resourcesthrough a common and standard catalog in-terface -the one recommended by the Eu-ropean Directive INSPIRE. In fact, GI-catpublishes standard CS-W interfaces. Theproposed solution is flexible and allow toimplement interoperability also with otherstandard service providers -e.g. OGC ac-cess services.Flexibility was addressed by introducingthe Accessor component: a piece of tech-nology which addresses interoperability is-sues regarding a given type of serviceproviders. By supporting asynchronousand incremental query responses, GI-catachieves scalability as well.The experimentations confirmed that thedata model of remote service influencesGI-cat query and access performances. TheTHREDDS model case was presented andan approach experimented for SeaDataNetwas discussed.For the SeaDataNet CDI model, a datasetaggregation element was proposed in orderto improve query and access performances.This extension is very simple and is go-ing to be adopted by the CDI specificationWorking Group.CF-netCDF data access was also experi-mented (see Figure 5). The implementationof a common data model could improvedata aggregation and interoperability. Weexperimented a CF-netCDF data model for

describing and encoding XBT acquisitions.

7 Recommentadions

Reducing the time required to acquire, pro-cess and analyze data of known quality is amajor goal that requires the development ofan integrated data management and com-munication subsystem. The objectives areto serve data in both real-time and delayedmode and to enable users to exploit multi-ple data sets from many different sources.This requires a distributed network thatworks with common standards, referencematerials and protocols for quality controlto enable rapid access to and the exchangeof data (e.g., metadata standards) and long-term data archival.At Italian level, such a network will de-velop in an incremental way by linking andintegrating existing data centers and man-agement programs. The objective is todevelop a federated data management andcommunications system (System of Sys-tems) that efficiently transmits large vol-umes of multi-disciplinary data (in real-time, near-real-time, and delayed modes)from many sources (in situ measurements,autonomous in situ sensors, remote sen-sors, models’ outputs, etc.) directly tousers for a broad diversity of applications,including data assimilating tools, that pro-cess the measurements into maps, plots,forecasts, and environmental statistics (e.g.climatology). Resource providers shouldbe able to transmit their products into thesystem with a minimum obligation to con-vert them to specialized encoding formats,provided that basic conditions of data qual-ity and metadata standards are met.While it may be possible to provide a sin-gle portal for data users for one-stop ac-cess, this will inevitably be a simplified

2651

Technologies

front end for the data management systemthat supports all the Italian System Of Sys-tems (ISOS). Federated access points willtend to focus first on thematic aspects.A centralized access point which is suit-able for all types of customers, request-ing data from any coast in the world, anddesigned to a common standard, may it-self be an unattainable ideal at this stage.It is more likely that such a contact pointwill direct the user to other more special-ized or regional product delivery centers, orwould be restricted to a limited number ofvariables. The construction of specializedcustomer-related access points can then becarried out by delegated teams of experts.The data management system will providefacilities for two data tracks, one in real-time (or near real-time) and one in the de-layed mode. Both tracks are based on thesame data sources and transmission sys-tems, but the data follow different routesand are processed differently depending onuser requirements. The real-time observedand generated products can also be checkedafter first use for up-grading and additionalquality control to ensure that relevant dataare preserved for final archival. Similarly,climatic data and standardised anomaliesand statistics can be provided to the model-ers from the archived data and time series

in order to compute expected values andanomalies.To summarize, here are some importantrecommendations for implementing a Sys-tem of Systems approach:1. Save all raw data in its native format.2. Attempt to make all data available, even

if it’s at a crude level.3. Archive higher-order products (includ-

ing processed data files) and registertheir existence in a catalog.

4. Catalog these data sets. At minimumthese catalogues must contain:• What format the file is in.• Which sensor captured the data.• What software produced this data set.• Who to contact regarding this data set.• Original capture date/time ”box”.• Latitude/Longitude(/Pressure) ”box”.• General comments about the data/file.

5. Query these catalogs (both with humaninterface and programmatically).

6. Digitize all records including log books(this makes the catalog more useful).

7. Distribute data archives and processingamongst people and machines.

8. Use World Wide Web to provide access.Issues regarding calibration, validation,and metadata capacity are implied but notspecifically addressed by these recommen-dations.

References[1] B.B. Parker. Oceanographic Data Archeology, finding historical data for climate

and global change research. Oceanography, 5(2), 1992.

[2] NASA Earth Science Data Systems Standards Process Group. The Data AccessProtocol - DAP 2.0 [available at http://www.esdswg.org/spg/rfc/ese-rfc-004, ac-cessed on 15 Jan 2010]. 2007.

[3] OPeNDAP. Open-source project for a network data access protocol [available athttp://opendap.org, accessed on 29 Jan 2009]. 2004.

2652

http://www.esdswg.org/spg/rfc/ese-rfc-004

http://opendap.org


[4] SeaDataNet. SeaDataNet Home Page [available at http://www.seadatanet.org/, ac-cessed on 15 Jan 2010]. 2010.

[5] SeaDataNet. Common Data Index [available athttp://www.seadatanet.org/data access/common data index cdi, accessed on15 Jan 2010]. 2007.

[6] B. Domenico, J. Caron, E. Davis, R. Kambic, and S. Nativi. Thematic Real-timeEnvironmental Distributed Data Services (THREDDS): Incorporating InteractiveAnalysis Tools into NSDL. Journal of Digital Information, 2, 2002.

[7] Sea-Search. Your gateway to Oceanographic and Marine Data & Information inEurope [available at http://www.sea-search.net/, accessed on 18 JHan 2010]. 2010.

[8] UNIDATA. NetCDF Documentation [available at http://www.unidata.ucar.edu/software/netcdf/docs/, accessed on 15 Jan 2010]. 2010.

[9] UNIDATA. THREDDS Fact Sheet [available at http://www.unidata.ucar.edu/publications/factsheets/2007sheets/threddsFactSheet-1.doc, accessed on 15 Jan2010]. 2010.

[10] S. Nativi and L. Bigagli. Discovery, Mediation, and Access Services for EarthObservation Data (in publication). IEE Journal of Selected Topics in Applied EarthObservations & Remote Sensing, 2009.

[11] Open Geospatial Consortium Inc. OGC® Cataloguing of ISO Metadata (CIM)Using the ebRIM profile of CS-W (0.1.8). 2007.

[12] Open Geospatial Consortium Inc. OpenGIS® Catalogue Services Specification2.0.2 - ISO Metadata Application Profile, v. 1.0.0. 2007.

[13] Open Geospatial Consortium Inc. OpenGIS® Catalogue Services Specification v.2.0.2 - Revision 1. 2007.

[14] Open Geospatial Consortium Inc. OGC™ Catalogue Services Specification 2.0Extension Package for ebRIM Application Profile: Earth Observation Products(0.2.4). 2009.

[15] European Space Agency. Heterogeneous Mission Accessibility - Context [availableat http://earth.esa.int/hma/context.html, accessed on 15 Jan 2010]. 2007.

[16] Group on Earth Observation. Global Earth Observation System of Systems(GEOSS) 10-Year Implementation Plan. 2005.

[17] International Organization for Standardization. Geographic information - Metadata.2003.

[18] H.E. Sweers. An improved code to classify the location of marine and terrestrialdata Limnology and oceanography. 1970.

2653

http://www.seadatanet.org/

http://www.seadatanet.org/data_access/common_data_index_cdi

http://www.sea-search.net/

http://www.unidata.ucar.edu/software/netcdf/docs/

http://www.unidata.ucar.edu/software/netcdf/docs/

http://www.unidata.ucar.edu/publications/factsheets/2007sheets/threddsFactSheet-1.doc

http://www.unidata.ucar.edu/publications/factsheets/2007sheets/threddsFactSheet-1.doc

http://earth.esa.int/hma/context.html

Technologies

[19] B. Eaton, J. Grogory, B. Drach, K. Taylor, and S. Hankin. NetCDF Climate andForecast (CF) Metadat Convention v. 1.0. 2003.

2654

technologies

Documents

sea ice

polar oceans

polar regionswill

polar ocean basinsare

tosevere ice

aurora borealisteam

unique regions

sea level change