general issue - marine technology society international, interdisciplinary society devoted to ocean...

THE INTERNATIONAL, INTERDISCIPLINARY SOCIETY DEVOTED TO OCEAN AND MARINE ENGINEERING, SCIENCE, AND POLICYVOLUME 39, NUMBER 2, SUMMER 2005

General IssueGeneral Issue

B O A R D O F D I R E C T O R SPresidentJerry StreeterJ P Kenny, Inc.President-electBruce C. GilmanConsultantImmediate Past PresidentTed BrockettSound Ocean Systems Inc.VP—Section AffairsSandor KarpathyStress Subsea, Inc.VP—Education and ResearchCapt. Daniel SchwartzUniversity of WashingtonVP—Industry and TechnologyMark BrownMBARIVP—PublicationsDr. Jerry WilsonFugro Pelagos, Inc.Treasurer and VP—Budget and FinanceJerry BoatmanPlanning Systems, Inc.VP—Government and Public AffairsCdr. Karen KohanowichU.S. Navy

S E C T I O N SCanadian MaritimeDr. Ferial El-HawaryB.H. Engineering SystemsFloridaProf. Mark LutherUniversity of South FloridaGulf CoastCraig CumbeeNAVOCEANHawaiiWilliam A. FriedlCEROSHoustonTerry DaileySchlumberger SubseaJapanProf. Toshitsugu SakouTokai University.Los AngelesJames EdbergConsultantMontereyJon EricksonMBARINew EnglandJames CaseUniversity of New HampshirePuget SoundEdward Van Den AmeeleNOAA Pacific Hydrographic BranchSan DiegoLeonard PoolSidus SolutionsWashington, DCBarry StameyMitretek Systems

P R O F E S S I O N A L C O M M I T T E E SINDUSTRY AND TECHNOLOGYAutonomous Underwater VehiclesJustin ManleyMitretek SystemsOcean EnergyTony JonesoceanUS consultingOceanographic InstrumentationSam KellyRemotely Operated VehiclesDrew MichelTSC Holdings, Inc.Underwater ImagingDonna KocakGreen Sky Imaging, LLCManned Underwater VehiclesWilliam KohnenSEAmagine Hydrospace CorporationDynamic PositioningHoward ShattoShatto EngineeringBuoy TechnologyWalter PaulMooringsJames A. CappelliniMooring Systems, Inc.Ropes and Tension MembersJohn FloryTension Tech International, Inc.Offshore StructuresPeter W. MarshallMHP Systems EngineeringSeafloor EngineeringHerb HerrmannNFESCCables & ConnectorsSteve ThumbeckMacArtney Offshore, Inc.DivingWilliam C. PhoelPhoel Associates Inc.Remote SensingGary MineartMitretek SystemsMineral ResourcesJohn C. WiltshireUniversity of Hawaii

RESEARCH AND EDUCATIONMarine GeodesyDr. Muneedra KumarMarine EducationSharon H. WalkerUniversity of Southern MississippiOcean ExplorationPaula Keener-ChavisNOAA Office of Ocean ExplorationMarine MaterialsGerald LoweFlorida Atlantic UniversityMarine ArchaeologyBrett PhaneufTexas A&M UniversityPhysical Oceanography/MeteorologyDr. Richard L. CroutCNMOC

GOVERNMENT AND PUBLIC AFFAIRSOcean PollutionBrian Broginton-SmithOcean Economic PotentialJames MarshUniversity of HawaiiMarine Law and PolicyCraig McleanNOAAMarine SecurityCapt. Bruce ClarkCEI Maritime, Inc.

S T U D E N T S E C T I O N SFlorida Atlantic UniversityCounselor: Douglas BriggsFlorida Institute of TechnologyCounselor: Eric ThostesonMassachusetts Institute of TechnologyCounselor: Alexandra TechetRoger Williams UniversitySanta Clara UniversityCounselor: Christopher KittsTexas A&M University—College StationCounselor: Robert RandallTexas A&M University—GalvestonCounselor: Victoria JonesU.S. Naval AcademyCounselor: Cecily NatunewiczUniversity of HawaiiCounselor: R. Cengiz ErtekinUniversity of Rhode IslandCounselor: Chris BaxterUniversity of Southern MississippiCounselor: Stephan Howden

H O N O R A R Y M E M B E R SThe support of the following individuals isgratefully acknowledged.

Robert B. Abel†Charles H. BussmannJohn C. CalhounJohn P. Craven†Paul M. FyeDavid S. Potter†Athelstan Spilhaus†E. C. Stephan†Allyn C. Vine†James H. Wakelin, Jr.

†deceased

Marine Technology Society Officers

The Marine Technology Society Journal(ISSN 0025-3324) is published quarterly (spring summer,fall, and winter) by the Marine Technology Society, Inc.,5565 Sterrett Place, Suite 108, Columbia, MD 21044.

MTS members can purchase the printed Journal for$25 domestic and $50 foreign. Non-members andlibrary subscriptions are $120 domestic and $135 foreign.Postage for periodicals is paid at Columbia, MD, andadditional mailing offices.

P O S T M A S T E R :Please send address changes to:

Marine Technology Society Journal5565 Sterrett PlaceSuite 108Columbia, Maryland 21044

Copyright © 2005 Marine Technology Society, Inc.

In This Issue

Volume 39, Number 2, Summer 2005

General Issue

3New Stereo Acoustic Data Logger forFree-ranging Dolphins and PorpoisesTomonari Akamatsu, Akihiko Matsuda,Shiro Suzuki, Ding Wang, Kexiong Wang,Michihiko Suzuki, Hiroyuki Muramoto,Naoki Sugiyama, Katsunori Oota

10An Epibenthic Sledge for Operations onMarine Soft Bottom and BedrockNils Brenke

22A Coupled Asymmetrical MultipleOpening Closing Net with EnvironmentalSampling SystemCedric M. Guigand, Robert K. Cowen,Joel K. Llopiz, David E. Richardson

25Evaluation of the Northern Gulf of MexicoLittoral Initiative Model Based on theObserved Temperature and Salinity inthe Mississippi BightNadya Vinogradova, Sergey Vinogradov,Dmitri Nechaev, Vladimir Kamenkovich,Alan F. Blumberg, Quamrul Ahsan, Honghai Li

39Ship Component in Hull OptimizationKent Davey

47Understanding the Potential Economic Impactsof Sinking Ships for SCUBA RecreationLinwood H. Pendleton

53State of the Art of HVOF Coating Investigations—A ReviewT.S. Sidhu, S. Prakash, R.D. Agrawal

65Groundwater Plume Mapping in a SubmergedSinkhole in Lake HuronSteven A. Ruberg, Dwight F. Coleman, Thomas H.Johengen, Guy A. Meadows, Hans W. Van Sumeren,Gregory A. Lang, Bopaiah A. Biddanda

F R O N T C O V E R I M A G E :Stereo-PIV measurements of axial velocity and vorticityof a propeller’s wake behind a ship model. Imagescourtesy of Guido Calcagno.

MTS Journal design and layout:Michele A. Danoff, Graphics By Design

70Application of Real-time Monitoring BuoySystems for Physical and BiogeochemicalParameters in the Coastal Ocean aroundthe Korean PeninsulaSungHyun Nam, Guebuem Kim, Kyung-Ryul Kim,Kuh Kim, Lawrence Oh Cheng, Ki-Wan Kim,Hyong Ossi, Young-Gyu Kim

81ROV Operation from a Small BoatDavid J. Csepp

90Potential Depth Biasing Using the BiosonicsVBT Seabed Classification SoftwareMichaela Dommisse, Dan Urban, Bruce Finney,Susan Hills

94A Stereo-PIV Investigation of a Propeller’s Wakebehind a Ship Model in a Large Free-surface TunnelGuido Calcagno, F. Di Felice, M. Felli, F. Pereira

103Shipboard Scientific Dive Van—Meeting theStandard for UNOLS Portable Scientific VansMark Gustafson

105Effect of Suspended Sediment on AcousticDetection Using ReverberationPeter C. Chu, Michael Cornelius, Mel Wagstaff

110Remote Video Revisited: A Visual Techniquefor Conducting Long-term Monitoring of ReefFishes on the Continental ShelfC.A. Barans, M.D. Arendt, T. Moore, D. Schmidt

119Book Reviews

B A C K C O V E R I M A G E :Deployment of a Coupled Asymmetrical M.O.C.N.E S.S.off the stern of the R/V F.G. Walton Smith. Thisinstrument was designed and constructed at theUniversity of Miami. Image courtesy of Cedric Guigand.

2 Marine Technology Society Journal

The Marine Technology Society isa not-for-profit, international professionalsociety. Established in 1963, the Society’smission is to promote the exchange ofinformation in ocean and marine engineering,technology, science, and policy.

Please send all correspondence to:The Marine Technology Society5565 Sterrett Place, Suite 108Columbia, Maryland 21044(410) 884-5330 Tel.(410) 884-9060 FAXPublications: [email protected]: [email protected]: [email protected]: [email protected]: www.mtsociety.org

M E M B E R S H I P I N F O R M A T I O Nmay be obtained by contacting the MarineTechnology Society. Benefits include:■ Free subscription to the online Marine

Technology Society Journal, with highlyreduced rates for the paper version

■ Free subscription to the bimonthly news-letter, Currents, covering events, businessnews, science and technology features, andpeople in marine technology

■ Member discounts on all MTS publications■ Reduced registration rates to all MTS and

MTS-sponsored conferences and workshops■ Member-only access to an expansive Job

Bank and Member Directory■ Reduced advertising rates in MTS

publications■ National recognition through our Awards

ProgramsIndividual dues are $75 per year. Life mem-bership is available for a one time fee of$1000. Patron, Student, Emeritus, Institu-tional, Business, and Corporate member-ships are also available.

ADVERTISINGAdvertising is accepted by the Marine Tech-nology Society Journal. For more informationon MTS advertising and policy, please con-tact the managing editor.

COPYRIGHTCopyright © 2005 by the Marine TechnologySociety, Inc. Authorization to photocopyitems for internal or personal use, or the in-ternal or personal use of specific clients, isgranted by the Marine Technology Society,provided that the base fee of $1.00 per copy,plus .20 per page is paid directly to Copy-right Clearance Center, 222 Rosewood Dr.,Danvers, MA 01923.

For those organizations that have beengranted a photocopy license by CCC, a sepa-rate system of payment has been arranged.The fee code for users of the TransactionalReporting Service is 0025-3324/89 $1.00+ .20. Papers by U.S Government employ-ees are declared works of the U.S. Govern-ment and are therefore in the public domain.

The Marine Technology Society cannot be heldresponsible for the opinions given and the state-ments made in any of the articles published.

ABSTRACTSAbstracts of MTS publications can be foundin both the electronic and printed versionsof Aquatic Sciences and Fisheries Abstracts(ASFA), published by Cambridge ScientificAbstracts, 7200 Wisconsin Avenue,Bethesda, MD 20814.

Electronic abstracts may be obtained throughGeobase’s Oceanbase, Fluidex, andCompendex, which is published by ElsevierScience, The Old Bakery, 111 Queen Road,Norwich, NR1 3PL, United Kingdom.Microfishe may be obtained through Congres-sional Information Services, Inc., 4520 East-West Highway, Bethesda, Maryland 20814

CONTRIBUTORSContributors can obtain an information andstyle sheet by contacting the managing editor.Submissions that are relevant to the concernsof the Society are welcome. All papers are sub-jected to a stringent review procedure directedby the editor and the editorial board. The Jour-nal focuses on technical material that may nototherwise be available, and thus technical pa-pers and notes that have not been publishedpreviously are given priority. General commen-taries are also accepted, and are subject to re-view and approval by the editorial board.

Editorial BoardDan WalkerEditorNational Research Council

Scott KrausNew England Aquarium

James LindholmPfleger Institute of Environmental Research

Dhugal LindsayJapan Agency for Marine-Earth Science& Technology

David MindellMassachusetts Institute of Technology

Phil NuyttenNuytco Research, LTD.

Terrence R. SchaffWoods Hole Oceanographic Institution

Edith WidderHarbor Branch Oceanographic Institution

EditorialJerry WilsonPublications Director

Dan WalkerEditor

Amy MorganteManaging Editor

AdministrationJerry StreeterPresident

Judith T. KrauthamerExecutive Director

Emily L. SpeightMembershipCirculation Manager

3Summer 2005 Volume 39, Number 2

A U T H O R STomonari AkamatsuAkihiko MatsudaShiro SuzukiNational Research Institute of FisheriesEngineering, Fisheries Research Agency,Japan

Ding WangKexiong WangInstitute of Hydrobiology,The Chinese Academy of Sciences

Michihiko SuzukiLittle Leonardo Inc.

Hiroyuki MuramotoMarine Micro Technology Inc.

Naoki SugiyamaSuruga Denshi Co. Ltd.

Katsunori OotaIntertec Co. Ltd.

P A P E R

New Stereo Acoustic Data Logger for Free-rangingDolphins and Porpoises

A B S T R A C TTo observe the bio-sonar behavior of dolphins and porpoises, a miniature stereo acous-

tic data logger was developed to record the echolocation clicks of small cetaceans. The‘A-tag’ device is small enough to be attached to a dolphin or porpoise. A-tag can recordthe sonar pulse intensity, precise inter-click-intervals, and time difference between soundsarriving at two different hydrophones. The A-tag works for up to 60 hours continuouslyand allows observation of the sonar target range of free-ranging odontocetes. The time ofarrival at the two hydrophones on the tag allows vocalizations from nearby individuals tobe identified. A less invasive tagging technique using a suction cup was also developed. Amean attachment time of 15 hours was obtained on free-ranging finless porpoises in afreshwater system in China. The A-tag proved to be a useful tool for investigating theunderwater echolocation behavior of odontocetes.

underwater echo sounders are not“colorized”. Therefore, we have lost fre-quency-dependent information about thetarget characteristics in underwater acous-tical surveys.

Compared with manmade sonar, dol-phins and porpoises use wide-band sonar forprey capture and environmental recognition.High-performance dolphin-like sonar haslong been anticipated. An understanding ofhow dolphins and porpoises use their bio-sonar abilities would be beneficial in thedesign of future wide-band electronic sonar.

Although the acoustic characteristics andperformance of the sonar of dolphins andporpoises have been extensively studied (Au,1993; Richardson et al., 1995), the behav-ioral control and use of their bio-sonar sys-tems are not fully understood. Especiallypuzzling is the ability of dolphins and por-poises to avoid jamming during group swim-ming and the mechanism they use to im-prove the signal-to-noise ratio in noisycircumstances. Hence, a study on howodontocetes use their wide-band sonar isneeded. However, three major difficultieshave slowed the progress of such research.

First, a particular vocalizing animal isdifficult to identify, even in captivity. In ce-taceans, sound production does not associ-ate with typical behavior, such as mouth

opening. However, this can be studied us-ing a hydrophone array system, which is apowerful tool for determining the soundsource direction (Au and Benoit-Bird, 2003).With the very short pulse duration of ceta-cean sonar signals, it is relatively easy tomeasure the difference in sound arrival timesbetween several hydrophones forming ashort base-line system. Still, it can be verydifficult to distinguish the phonating indi-vidual within a tight group.

Second, ultrasonic sounds are difficult torecord. The dominant frequency of sonar sig-nals in dolphins and porpoises is in the rangeof several to 150 kHz (Richardson et al.,1995). Full bandwidth digital recording ofsuch sonar signals requires a very high sam-pling frequency. With continuous recording,it is important to avoid missing any signals.This requires very large memory capacity.However, the waveforms of ultrasonic sonarsignals in dolphins and porpoises are stereo-typed (Amundin, 1991), so recording thetransmission waveform is not essential forsonar behavioral studies. Conversely, thesound pressure levels, inter-pulse intervals,and number of sounds produced per unit timeprovide crucial behavioral information, suchas the estimated target range and search ef-fort. Therefore, a pulse event recorder with alow frequency sampling rate to record the

II N T R O D U C T I O N n the last two decades, the bio-sonar system of dolphins and porpoises (Au,1993) has gained attention as a model forfuture underwater echo sounders. Fisheries,ocean engineering, and underwater con-struction and exploitation all require a reli-able sensing technique to examine under-water objects.

A wide-band signal is useful for assess-ing target characteristics, depending on thespecific frequency response of each target.However, most manmade echo soundersuse narrow band signals (Furusawa, 1999)that only sense the echo level of a mono-chromatic frequency component. Al-though several echo sounders use multiplefrequencies to examine target characteris-tics, a wide-band echo sounder, similar tothe bio-sonar of dolphins and porpoises,has not yet been developed. Manmade


sound intensity of sonar signals is needed forthe long-term observation of the sonar be-havior of dolphins and porpoises.

Finally, the recording of sonar signals fromwild odontocetes for a prolonged period oftime is extremely difficult. Dolphins and por-poises move quickly and do not stay in a fixedposition. Moreover, it has been next to im-possible to record individual behavior as wellas the phonation simultaneously for a longperiod of time in the water. Bio-logging tech-niques (Naito, 2004) using data loggers haverecently been used to reveal underwater ani-mal behavior (Hanson and Baird, 1998; Bur-gess et al., 1998; Madsen et al., 2002;Akamatsu et al., 2002; Tyack et al., 2004;Nowacek et al., 2004). Data loggers are verysmall computers with sensors encapsulated ina pressure-resistant case. A data logger can beattached to an animal to record behavioral andphysiological parameters, such as depth, swim-ming speed, body accelerations, the elec-tromyogram, and stomach temperature, andto obtain underwater digital images (Naito,2004). However, a limited number of studieshave focused on the sonar behavior ofodontocetes (Tyack and Recchia, 1991;Madsen et al., 2002, 2005; Blomqvist andAmundin, 2004; Akamatsu et al., 2000, 2005;Johnson et al., 2004; Zimmer et al., 2005).

In summary, an ideal device for observ-ing the underwater sensory behavior of dol-phins and porpoises should (a) have multi-channel hydrophones to identify the sound

source direction, (b) be an ultrasonic pulseevent recorder suitable for long-term re-cording, and (c) be small enough to be at-tached on dolphins and porpoises withoutdisturbing their normal behavior. In theremainder of this paper, we discuss the de-velopment of a miniature stereo acousticdata logger named ‘A-tag’ that meets thesespecifications.

System SpecificationsHardware

The ‘A-tag’ (W20-AS, Little Leonardo,Tokyo, Japan) contains a miniature stereopulse event recorder and a CR123 lithiumbattery cell, encased in a waterproof tube,measuring 22 mm in diameter, 122 mm inlength, and weighing 77 g (Figure 1). A-taghas two miniature ultrasonic hydrophones(System Giken, -210 dB/V sensitivity), oneat each end of the logger. A band pass filter(70 to 300 kHz) is included to eliminatenoise outside the frequency bands of por-poise sonar signals. The CPU is aPIC18F6620 (Microchip, USA) and is usedfor system control and signal processing. Asa storage device, 256 MB flash memory isused. A-tag consumes approximately 10 mAelectric current on average. The total record-ing time is approximately 60 hours, depend-ing on the number of pulses that a dolphinor porpoise produces. The internal clock ofA-tag drifts less than 1 second per day.

Signal ProcessingThe dynamic range of A-tag is between

129 dB peak to peak (reference pressure 1µPa) to 157 dB. The signal output from hy-drophone A is amplified with a gain of 60dB. The noise floor at the output of the hy-drophone is lower than 25 µV

p-p (depend-

ing on the sensitivity variation of each hy-drophone), which corresponds to 118 dB

p-p

re µPa. To avoid thermal or electronic noisecontamination, a hardware detection thresh-old level corresponding to 129 dB re µPa isused. A variable resistor in A-tag sets thehardware detection threshold level. An ana-logue-to-digital converter with 10 bit reso-lution digitizes the peak intensity in every0.5-ms time bin. If the peak intensity ex-ceeds the hardware detection threshold level,the peak intensity is stored in the flashmemory (Figure 2). If the peak intensity isbelow the hardware detection threshold level,the data is deemed null and is not saved inthe memory. A-tag repeats this procedureevery 0.5 ms (2 kHz sampling frequency).

The direction of the sound source is cal-culated from the sound time-of-arrival dif-ference between hydrophones A and B. Thisfeature is added to make it possible to ex-clude vocalizations coming from other in-dividuals. The output of hydrophone B isfed into a second peak holder. If the peaklevels of hydrophones A and B are bothhigher than the hardware detection thresh-old level, then the difference in the sound’sarrival time is stored (Figure 2). The differ-ence in the sound arrival time between hy-drophones A and B should be within ±80ìs, since the inter-hydrophone distance is 120mm. A-tag is programmed to neglect timedifferences outside a ±139 µs time windowto exclude any noise contamination. Thetime difference within this window is quan-tized in 10 bits, corresponding to a resolu-tion of 271 ns. This gives a resolution of 0.4mm in the travel distance difference betweenthe hydrophones from a sound source in linewith the unit. The absolute time of eachpulse can be calculated by integrating theinterval time, such as t2-t1 indicated in thelowest line of Figure 2. This calculation isautomatically done by the software LoggerTools ver. 4.1 (Little Leonardo, Tokyo).

FIGURE 1The frontal and lateral views of the A-tag (W20-AS). Hydrophone A is located at the front of the data logger andHydrophone B with an extension cable is located at the rear of the tag to measure the sound time-of-arrivaldifference between the two hydrophones, revealing the rough direction to the sound source.


AssemblyAs shown in Figure 3, A-tag is assembled

with a VHF transmitter (MM130, Ad-vanced Telemetry Systems, USA), float (ex-panded polyvinyl chloride, Klegecell #55,pressure resistant to 80 N/cm2, Kaneka, Ja-pan) and suction cup (Product# 40-1525-0, 82 mm in diameter, Canadian Tire Corp.,Canada). Suction cups are a less-invasivetechnique for attachment on animals andare commonly used to attach data loggersystems to cetaceans (Hooker and Baird,2001). Although the float consists of closedcells of polyvinyl chloride foam that do notallow the infiltration of water, the surface ofthe float is coated with urethane rubber toprevent mechanical damage and furtherminimize the amount of infiltration. Thefloat is designed to have positive buoyancyfor easier retrieval after the spontaneous re-lease of the suction cup from the animal.The amount of buoyancy is critical for thelong-term attachment and safe recovery ofthe system; greater buoyancy causes prema-

ture detachment because of the larger floata-tion force, while lower buoyancy increasesthe system’s risk of sinking after detachment.The float lost a slight amount of its buoy-ancy in the validation experiment describedbelow. A small amount of infiltration andshrinking due to the repeated change ofwater pressure during successive dives by theanimals is thought to cause this buoyancyloss. A buoyancy margin of 15 to 20 g wasfound appropriate. In order to receive strongradio signals from the VHF transmitter af-ter detachment, the transmitter antennamust be oriented vertically in the air. There-fore, the float is designed to have larger vol-ume in the rear and a smaller volume in frontof the system to make the antenna protrudefrom the water vertically.

To decrease the total weight of the datalogger system, an acrylic supporting plateand acrylic screws were used instead of metalparts to secure the suction cup to the float.These were specially designed for this sys-tem (Suruga Denshi, Japan). The plate is

fixed to the float with epoxy glue (QuickSet 30, Konishi, Japan). A-tag is fixed to theplate by a ribbon sealer (MH 908, MusahiHolt, Japan). The transmitter is situated justbehind the A-tag. We confirmed that radiotransmission did not affect data acquisitionby the data logger. The total weight of thedata logger system including the A-tag, bat-tery cell, transmitter, float, and suction cupis approximately 203 g. When the animal isrespiring, the data logger system is exposedto the air for short periods. In this condi-tion, the data logger remains on the animalbut records splash noises primarily.

AttachmentBasics of Suction Cup Attachment

A preliminary attachment test of the datalogger system was carried out using twofinless porpoises (Neophocaena phocaenoides)kept at the Institute of Hydrobiology, Chi-nese Academy of Sciences, Wuhan, Hubei,China. We drained the water from a kidneyshaped pool (20 × 7 m) and caught the ani-mals one by one. The A-tag was fixed to theright side of the body behind the pectoralfin. This area was the least affected by bodymovements, which ensured a long attach-ment time. A behavior data logger (PD2GT,Little Leonardo, Japan) was attached in asimilar position on the left side of the ani-mal. The behavior data logger had four dif-ferent instruments: a pressure sensor, pro-peller, two-axis accelerometer, andthermometer. It recorded the swimmingdepth, speed, heaving and surging body ac-celerations, and the temperature of the wa-ter where the animal was. The behavior datalogger was 114 mm long, 21 mm in diam-eter, and weighed 59 g. Its size and weightwere quite similar to those of the A-tag anda similar float and attachment system wasused for both types of data logger.

In all, four data loggers were deployedon the two animals. We filled the pool withwater just after attachment and observed theanimals during the day. One data logger sys-tem detached after 4.0 hours on the animal.The other three data loggers were still at-tached after 26.2 hours, when the animalswere re-captured the next day. This was done

FIGURE 2Signal processing in the A-tag. A typical sonar click train of a finless porpoise is shown in the upper trace in thisfigure. Two high-intensity sonar pulses and low-intensity pulse noise are shown in the middle trace waveform.The peak intensities of the signals in each 0.5 ms time bin are held electronically. Once the peak level exceedsthe hardware detection threshold (lower trace: dotted line), the sound pressure level (SPL; SPL1 and SPL2) andtime arrival difference of a sound between the two hydrophones (td; td1 and td2) and the time elapsed from theprevious pulse (interval; t1-t0, t2-t1) are stored in the memory of the A-tag. The absolute time could be calcu-lated by integrating the interval and adding the initial date and time.


to avoid any damage to the skin of thesecaptive animals from the suction cups. Hav-ing three of four devices remain attached tothe animals after more than one day wasconsidered acceptable given the use of thisless-invasive attachment method. Previoussuction cup attachments of data logger sys-tems have been in the range of a few hoursto two days (Hooker and Baird, 2001).

Drag ForceThe drag force of the A-tag could po-

tentially affect the behavior of the taggedanimal. To evaluate this, the drag was mea-sured in the Marine Dynamics Basin at theNational Research Institute of Fisheries En-gineering, Japan (Figure 4). This facility hasa tank (60 × 25 × 3.2 m) with a controlroom above it. The control room can bemoved horizontally in two directions. Analuminum plate (324 × 298 mm) was fixedwith a load cell (FM-6H50S, Izumi Sokki,Tokyo, Japan) beneath the control room.Half of the aluminum plate was submergedin the water. After a calibration run to mea-

sure the drag force of the aluminum platewithout the data logger system, the A-tagassembled with the float and radio trans-mitter was attached to the plate using asuction cup. The A-tag was pulled through

the water at speeds ranging from 0.5 to 2.0m/s in 0.5-m/s increments. The drag forceincreased with the square of the velocity, aspredicted by fluid dynamics theory. At 1m/s, which is similar to the normal cruis-ing speed of finless porpoises (Akamatsu etal., 2002), the drag force of the data loggersystem was 56 gF.

ValidationField Test

The experimental site was an oxbow ofthe Yangtze River, which was cut off fromthe main stream of the river in 1972. Wa-ter still enters the oxbow from the mainstream during the flood season (Wei et al.,2002). This oxbow lake, part of Tian-e-Zhou Baiji National Natural Reserve of theYangtze River, Hubei, China (29°30'-29°37’N, 112°13'-112°48’E), is approxi-mately 21 km long and 1 to 2 km wide. Itwas established by the Chinese governmentin 1992 as a reserve for baiji (Lipotesvexillifer) and finless porpoises. Since 1990,49 finless porpoises have been introducedfrom the main population in the river.These finless porpoises in the lake survivewithout supplemental food and reproduceannually. The environment of the lake isconsidered similar to the natural habitat ofthis species (Zhang et al., 1995).

FIGURE 4Drag force of the data logger system. The drag force was measured in fluid dynamic experiments (squares) andshowed good agreement with a theoretical fitting by a quadratic function (solid line).

FIGURE 3Lateral and top views of the A-tag.


Finless porpoises in the reserve werecaptured for the test in October 2004.Eighteen fishing boats drove finless por-poises from the upper end of the oxbowto the lower end. A net approximately 1km long was used to divide the oxbowtransversely. A fine-mesh net was used toencircle the animals. In the final stage,fishermen wearing life jackets entered thewater and captured the animals individu-ally. In the meantime, 18 boats sur-rounded the seine net and more than 50fishermen carefully watched each sectionof the net to prevent entanglement of theanimals. The water was less than 1 mdeep in the capture area, allowing thefishermen to handle the animals safely.All animals inside the net were capturedand were then temporarily released intoa net enclosure to calm them. In total, 6animals were captured. The enclosure wasestablished close to shore and measuredapproximately 30 × 60 m with a maxi-mum depth of 3.5 m. After the calmingperiod, the animals were fitted with theA-tag (Figure 5) and released back intothe lake.

To ensure retrieval of the data loggers,the radio signals were monitored using twoantennae (RX-155M7/W, Radix, Japan)from the top of a three-story field stationbuilding beside the oxbow. When a con-tinuous radio signal was received, a datalogger was considered to be floating. Re-trieval operations were started six or morehours after release to avoid disturbing theanimals. All of the data logger systems weresafely retrieved and found to be workingwhile they were attached. The average at-tachment duration of the A-tag on the ani-mals was 18.1 hours (range: 2.9 to 29.7hours), which compares favorably with pre-viously reported suction cup attachmentdurations (Hooker and Baird, 2001).

Data ProcessingAs shown in Figure 6, A-tag recorded

the sound pressure (upper trace) andtime difference between the two hydro-phones (middle trace) of the individualsonar signals. In addition, the inter-pulse interval was calculated as indicatedin the lower trace of Figure 6. Low-in-tensity signals below the software detec-

tion threshold level were excluded toeliminate noise contamination. Newlydeveloped pre-processing software writ-ten on MATLAB (The MathWorks,MA, USA) was used for this purpose.The software detection threshold levelcan be selected arbitrarily, as appropri-ate for the purpose of the analysis. Forthis study, a software detection thresh-old level equivalent to 136 dB re µPawas selected for the individual sonar be-havior analysis, and 129 dB was selectedfor the analysis of the water surface re-flection of the sonar signals.

The porpoise changed the sound pres-sure (SP) as well as the inter-pulse inter-val (PI) frequently (Figure 6), which sug-gests that the sonar range of this individualvaried from second to second (Akamatsuet al., 2005a). The time difference betweenhydrophones A and B (TD) was approxi-mately +80 µs, which corresponds to thesound coming from the individual animalcarrying the particular A-tag. Dolphinsand porpoises produce sonar signals fromtheir nasal duct located just below theblowhole (Cranford et al., 1996). The so-nar signal reaches the front hydrophonefirst, travels along the long axis of the A-tag, and is then picked up by the rear endhydrophone. The frequent detection of a+80 µs time difference (middle trace ofFigure 6) indicates that most of the re-ceived sounds came from the sound sourceof the host animal carrying the A-tag.Some of the signals had time differencessignificantly different from 80 µs and thesewere considered to come from other indi-viduals. For the individual sonar behavioranalysis, we excluded sounds with othertime differences to avoid contaminationof the signal by vocalizations of other ani-mals. When there was insufficient inten-sity at the rear end hydrophone, null datawas recorded as the time difference andthat is shown as a zero value in the middletrace. Akamatsu et al. (2000) showed thatmeasurable sound energy can be pickedup at the position of the tag, although themain energy is focused in a narrow beamdirected forward of the animal’s melon(Au, 1993).

FIGURE 5A finless porpoise with the A-tag and behavior data logger attached on either side. This individual was releasedimmediately after attachment.


ConclusionAn acoustic data logger (A-tag) designed

for observing sonar behavior in small ceta-ceans was developed. The A-tag was success-fully used to collect data on free-rangingfinless porpoises in an open-water environ-ment. A less invasive attachment techniquewas also developed. We applied A-tags tofree-ranging finless porpoises and confirmedthat they were capable of collecting relevantsonar behavior data. The underwater sens-ing behavior of this species is being investi-gated using the data obtained by the A-tag(Akamatsu et al., 2005a, 2005b).

AcknowledgmentsX. Zhang, Q. Zhao, Z. Wei, X. Wang,

B. Yu (IHCAS), Y. Naito, K. Sato, A. Kato(NIPR), H. Tanaka (Kyoto University), H.Murakami, T. Kawakami (Kaiyo DenshiCo., Ltd), M. Nakamura, H. Hiruda (Ma-rine World Uminonakamichi), S.Numata, T. Sakai (Oarai Aquarium), T.Tobayama, M. Soichi, H. Katsumata(Kamogawa Sea World), T. Shinke (Sys-tem Intech Co. Ltd.), A. Zielinski, R.Baird, the Institute of Hydrobiology of theChinese Academy of Sciences, Tian-e-zhou Baiji National Nature Reserve Field

Station, the National Institute of PolarResearch of Japan, Kamogawa Sea World,Marine World Umino Nakamichi, OaraiAquarium, and Tamai Kankyo SystemsCo. Ltd., Japan greatly supported our ex-periments. This research was supported byBio-oriented Technology Research Ad-vancement Institution (BRAIN) Promo-tion of Basic Research Activities for Inno-vative Biosciences of Japan, the ChineseAcademy of Sciences (KSCX2-SW-118),the Institute of Hydrobiology, CAS(220103), National Natural Science Foun-dation of China (30170142), and a Grant-in-aid for Scientific Research (B) from theMinistry of Education, Culture, Sport,Science and Technology, of Japan(B09450172).

The field experiments were conductedunder a permit issued by the FisheriesManagement Department of HubeiProvince.

ReferencesAkamatsu, T., Wang, D., Wang, K. and Naito, Y.

2000. A method for individual identification

of echolocation signals in free-ranging finless

porpoises carrying data loggers. J Acoust Soc

Am. 108:1353-1356.

Akamatsu, T., Wang, D., Wang, K., Wei, Z.,

Zhao, Q. and Naito, Y. 2002. Diving behavior

of freshwater finless porpoises (Neophocaena

phocaenoides) in an oxbow of the Yangtze River,

China. ICES J Mar Sci. 59:438-443.

Akamatsu, T., Wang, D., Wang, K., and Naito, Y.

2005a. Biosonar behavior of free-ranging

finless porpoises. Proc R Soc Lond. B,

272:797-801.

Akamatsu, T., Wang, D. and Wang, K. 2005b.

Off-axis sonar beam pattern of free-ranging

finless porpoises measured by a stereo pulse

event data logger. J Acoust Soc Am.,

117:3325-3330.

Amundin, M. 1991. Sound production in

odontocetes, with emphasis on the harbour

porpoise, Phocoena phocoena. Swede Publishing

AB, Stockholm, pp.128.

Au, W. W. L. 1993. The sonar of dolphins.

New York: Springer. 277pp.

Au, W. W. L. and Benoit-Bird, K. J. 2003.

Automatic gain control in the echolocation

system of dolphins. Nature 423:861-863.

Blomqvist, C. and Amundin, M. 2004. An

acoustic tag for recording directional pulsed

ultrasounds aimed at free-swimming bottle-

nose dolphins (Tursiops truncatus) by

conspecifics. Aquat Mamm. 30:345-356.

Burgess, W. C., Tyack, P. L., Le Boeuf, B. J.

and Costa, D. P. 1998. A programmable

acoustic recording tag and first results from

free-ranging northern elephant seals. Deep-Sea

Res. II 45:1327-1351.

Cranford, T.W., Amundin, M. and Norris,

K.S. 1996. Functional morphology and

homology in the odontocete nasal complex:

implications for sound generation. J Morphol.

228:223-285.

Furusawa, M. 1990. Study on echo sounding

for estimating fisheries resources. Bulletin of

National Research Institute of Fisheries

Engineering, 11, 173-249 (in Japanese).

Hanson, M.B. and Baird, R.W. 1998. Dall’s

porpoise reactions to tagging attempts using a

remotely-deployed suction-cup tag. Mar

Technol Soc J. 32(2):18-23.

FIGURE 6The sound pressure (SP) in dB re µPa (upper trace), time difference between the two hydrophones (TD) in ìs(middle trace), and calculated inter-pulse interval of sonar signals (PI) (lower trace) recorded by the A-tag.


Hooker, S.K. and Baird, R.W. 2001. Diving

and ranging behaviour of odontocetes: a

methodological review and critique. Mamm

Rev. 31:81-105.

Johnson, M., Madsen, P. T., Zimmer, W. M. X.,

Aguilar de Soto, N. and Tyack, P.L. 2004.

Beaked whales echolocate on prey. Proc. R.

Lond. B (Suppl.) 271, S383-S386.

Madsen, P. T., Johnson, M., Aguilar de Soto,

N., Zimmer, W. M. X., and Tyack, P.L. 2005.

Bisonar performance of foraging beaked

whales (Mesoplodon densirostris). J Exp Biol.

208:181-194.

Madsen, P. T., Payne, R., Kristiansen, N. U.,

Wahlberg, M., Kerr, I. and Mohl, B. 2002.

Sperm whale sound production studied with

ultrasound time/depth-recording tags. J Exp

Biol. 205:1899-1906.

Naito, Y. 2004. New steps in bio-logging

science. Mem Natl Inst Polar Res Spec Issue.

58:50-57.

Nowacek, D. P., Johnson, M. P. and Tyack, P. L.

2004. North Atlantic right whales (Eubalaena

glacialis) ignore ships but respond to alerting

stimuli. Proc. R. Soc. Lond. B 271:227-231.

Richardson, W.J., Greene, Jr. C.R., Malme,

C.I. and Thomson, D.H. 1995. Marine

mammals and noise. San Diego, New York,

Boston: Academic Press. pp.576.

Tyack, P. L., Johonson, M., Madsen, P. T. and

Zimmer, W. M. 2004. Echolocation in wild

toothed whales. J Acoust Soc Am. 115:2373

Tyack, P. and Recchia, C. A. 1991 A data

logger to identify vocalizing dolphins. J Acoust

Soc Am. 90:1668-1671.

Wei, Z., Wang, D., Kuang, X., Wang, K.,

Wang, X., Xiao, J., Zhao, Q. and Zhang, X.

2002. Observations on behavior and ecology

of the Yangtze finless porpoise (Neophocaena

phocaenoides asiaeorientalis) group at Tian-e-

Zhou Oxbow of the Yangtze River. The Raffles

Bulletin of Zoology, Supplement 10:97-103.

Zhang, X., Wei, Z., Wang, X., Yang, J. and

Chen, P. 1995. Studies on the feasibility of

establishment of a semi-natural reserve at Tian-

e-zhou (swan) oxbow for baiji, Lipotes vexillifer.

Acta Hydrobiologica Sinica 19:110-123.

Zimmer, W. M. X., Johnson, M. P., Madsen, P. T.,

and Tyack, P.L. 2005. Echolocation clicks of

free-ranging Cuvier’s beaked whales (Ziphius

cavirostirs). J Accust Soc Am. 117(6):3919-3927.


A U T H O RNils BrenkeRuhr-Universität Bochum,Department of Animal Morphology andSystematics

P A P E R

An Epibenthic Sledge for Operationson Marine Soft Bottom and Bedrock1

A B S T R A C TA multi-purpose epibenthic sledge, designed for sampling of small benthic macrofauna

in marine habitats and at any navigable depth, is presented. The new epibenthic sledgeoperates reliably on soft sediments in shallow and in open oceanic deep water, as well ason steep slopes, between rocks and glacier moraines as frequently found in Antarcticwaters, and on primary hard substrate. The construction is of high mechanical stabilitywith fully protected nets. In case of damage, parts of the sledge can be replaced or re-paired easily on board. A description of the gear with a detailed construction plan, as wellas parameters for handling in diverse marine habitats, is given. Calculation of the towingdistance and first results with possible sources of errors are discussed.

rock dredge, 1961; Holme, 1964 Müller bagdredge and Robertson bucket dredge). Anchordredges (Forster’s anchor dredge, 1953;MacIntyre, 1964 Box dredge; Carey &Hancock, 1965 Anchor-box dredge; Sanderset al., 1965 Deep-sea anchor dredge) andsledge-dredges (Okelmann’s detritus-sledge,1964; Hessler & Sanders, 1967 WHOIDeep-sea “epibenthic sled”; Salvini-Plawen’ssledge dredge, 1975; Blomqvist & Lundgren’sdredge, 1996; Sneli’s deep-sea sledge, 1998)for sampling the endo- and meiofauna wereemployed nearly at the same time. Thesedifferent dredges sample very close to theseabed. In spite of their robust constructionthey lack a “door” and usually these gearshave only a small opening, short nets andhence low sampling capacities.

At the same time efforts were made tosample organisms above the ground withfinely woven bottom- or near-bottom plank-ton nets (Ekman’s double frame-net dredge andEkman’s tow net, 1911; Mortensen’s sledge,1925; Russel’s combination net, 1928; Colman& Segrove’s tidal-flat net, 1955; Frolander &Pratt’s bottom-skimmer, 1962; Pearcy’s combi-nation net, 1972; Beardsley, 1973 Shrimp sam-pler; Sirenko et al., 1996 Benthopelagic sam-pler). To avoid contamination with planktonicorganisms of pelagic water layers, dredges withopening-closing devices were developed(Riedl’s closing dredge, 1955; Clutter’s self-clos-ing bottom net, 1965).

The difficulty of obtaining adequatequantitative and qualitative samples in de-fined horizons above the sea floor becomesmore challenging with increasing depth. Tocompensate for this, the characteristics ofsledge dredges and bottom plankton netswere combined to create epibenthic sledges.

A remarkable construction was realizedin the unique but impractical Wagennetz byHensen (1895), a net resting on four axesand eight wheels. Unfortunately, a largenumber of epibenthic sledges are sensitiveto structural damage due to the fragile con-struction of the frame (Werner’s automaticbottom shutting net, 1939) and too lightweighted for operation in the deep sea(Beyer’s epibenthic closing net, 1958; see also:Holme, 1964; Hesthagen, 1970; Oug,1977). Furthermore, some existing opening-closing mechanisms for closing epibenthicsledge nets proved to be inefficient(Bossanyi’s apparatus, 1951). The manymoveable parts of opening-closing deviceswere sensitive to damage and therefore notsuited for operation on rough sand and rocks(Wickstead’s spring sampler, 1953; Macerbottom-plankton sampler, 1967). Later con-structions bear more stable frames, horizon-tally divided nets (Brunel et al., 1978 Macer-GIROQ; Brandt & Barthel, 1995), andmore advanced opening-closing mecha-nisms. Aldred et al. (1976) introduced astable sledge (IOS sledge) for sampling

Footnotes1 Result of the expeditions “SEAMEC”(cruise M42/3) and “Diva-1” (cruise M48/1) with RV Meteor.

TIntroduction andHistorical Overview he choice of adequate zoological sampling gear with the right mesh size iscrucial for the success or failure of an expe-dition. Since Edward Forbes’ dredge (1815-1854), various bottom-trawls, plankton-nets, grabs, corers and sledges have beendesigned for biological oceanography. Thevariety of gears reflects the different require-ments that are determined by type, size,abundance and habitat of the organisms tocatch. An overview of the history of marinetechnology and various gear is given byGunter (1957), Thorson (1957), Riedel(1963), Hopkins (1964), Schlieper (1968),Holme & McIntyre (1971), Menzies et al.(1973) and Gage & Tyler (1991).

The benthic boundary layer and its poly-morphic fauna have always claimed specialdemands to the sampling technique. Initially,the benthic fauna was sampled with simpletrawls and dredges as used in commercialfishery (e.g. oyster dredge). Deriving from thisgear, the zoological macrofauna dredges“evolved”, usually operating with larger meshsizes (Steinmann’s triangle netdredge, 1909;Saint-Hilaire’s blade dredge, 1909; Nalwalk’s


macrofauna (see also: Rice et al., 1982;Christiansen & Nuppenau, 1997 IHFFototrawl), but in many of the above men-tioned sledges the nets were not protectedsufficiently from mechanical damage. Dur-ing further development the nets were com-pletely surrounded by the frame (Sorbe,1983; Boyd, 2002; Koulouri, 2002 CEFASFoto-Sled). Some of these sledges have provenreliable during operation on soft bottoms inshallow waters. Nevertheless, the fact thatnone of the epibenthic sledges of the pastcan be employed on different types of bot-tom demanded a modern construction.

In this article, a new multi-purposeepibenthic sledge for catching small benthicmacrofauna (in-fauna and on-fauna, “epi-faunas” sensu [Petersen, 1914:16]) is pre-sented. This new construction is based onthe epibenthic sledge by Rothlisberg &Pearcy (1977) that has been used frequentlyin Europe, although often with slight modi-fications (modified by Brattegard in: Buhl-Jensen, 1986; Brattegard & Fossa, 1991;Brandt & Barthel, 1995). The aims of theconstruction are twofold: (1) good hydro-dynamic properties for stable position inthe water column and (2) a robust con-struction of the frame and opening-clos-ing device to enable operation on primaryhard substrate, among rocks or on slopes.Additionally, the weight of the construc-tion per unit area was to be kept as low aspossible to minimize the digging compo-nent of the gear. The nets were completelysurrounded and secured by the frame. Easyhandling and quick restoration after op-eration or damage were also essential forthe construction.

Design and ConstructionAll parts of the epibenthic sledge are

made of corrosion-proof stainless steel (V4A:1.4571), either welded or screwed. The dif-ferent parts are briefly described, the exactscales are given in construction plans 1 to 5.

General Overview (Figure 4c)The assembled sledge is 3.45 m long and

1.24 m high, 1.3 m wide at the bottom and1.1 m wide at the top. It carries two identi-

FIGURE 1Epibenthic sledge construction plan: net box (a-e) and parts of lever unit (f)

cal closeable nets. The isosceles trapezoidcross-section and the low lying center of gravitygive the sledge a sufficiently stable position inthe water column and keep it from turningupside-down before landing. The weight isapproximately 444 kg in air (386 kg in 35 ‰seawater) without lead balance weights. To

protect the research vessel’s cable, a “crow-foot” (0.85 m; ∅ 2 × 12 mm; max. loadlimit 2.4 t in 45°) is attached between thecable and the sledge with a frontal mooringshackle (max. load limit 8.5 t). A safety cableis fixed to the end of the rear unit (max. loadlimit 6.5 t).


The Frame (Figures 2a-d, 3a-c, 4a-c)The frame consists of steel bars (4 x 4 cm)

and can be dismantled into three subunits,the frontal, center, and rear unit.

The frontal unit (44 kg; Figure 2a) is1.16 m high and 0.65 m wide. In profile,the lower side bar is longer than the upperone and strengthened with runners madefrom 5 mm steel plate to enable sliding overobstacles. On the front-most tip of the fron-tal unit, 0.83 m above the bottom, the hold-ing points for the tow cables are fixed. Be-tween the runners, a removable shield (2 mmthick, 0.25 m high; Figure 2c) is fixed toprotect the center unit’s lever from damage.

The center unit (316 kg; Figures 2b, 2d)is 1.16 m high and 1.80 m long. The centerunit carries the two nets, mounted one abovethe other, and the lever unit (Figure 4a). Itsfloor is covered by a steel plate (2 mm) exceptfor an opening for the lever arm (Figure 3c).

The rear unit (84 kg; Figure 3a) is 1.16m high and 1.0 m long. It serves as protec-tion for the cods of the nets end and forclipping in the cylindrical net buckets (Fig-ure 3b). The floor of the unit is completelyprotected by a steel plate (Figure 3c).

The dismountable construction of theframe allows maximum stability and easy trans-port at the same time, as well as minimumcargo space in freight containers. Furthermore,a quick reestablishment of full functionality ispossible during expeditions by means of simpleexchange of damaged parts. The fastening ofthe units with screws (frontal to center unit: 4M24, center to rear unit: 12 M10) ensures fastand efficient replacement of these components.

For compensation of the weight of thenet boxes in the front, the sledge is equippedwith two 20 kg standardized lead balanceweights (Figure 2f) in the rear unit. In thisconfiguration (standard version: 3.34 m²,444 + 40 kg) the sledge applies a pressure of145 g/cm2. Optionally, the sledge can also beemployed without the rear unit on small re-search vessels in shallow coastal waters. In thiscase the nets are protected by a flexible rub-ber mat attached to the lower posterior trans-verse bar (see Buhl-Jensen, 1986). To keepbalance, 2 x 30 kg lead weights should beapplied to the back of the center unit (shortversion: 2.04 m², 360 + 60 kg, 206 g/cm²).

The Net Boxes (Figures 1a-f)The two net boxes (epi- and

suprabenthic sampler, 2 mm steel plate)form the openings for the nets (mouth area:0.35 m² each; Figure 1b). For easy installa-tion and removal the net boxes are installedinto the horizontal crossbeams of the center

unit with 18 screws each. The net boxes are1.0 m wide, 0.35 m high, and reach from0.25 to 0.60 m (epibenthic sampler) andfrom 0.77 to 1.12 m (suprabenthic sampler)above the sledge floor (Figure 2b). In pro-file, the frontal openings of the net boxesare inclined at a 45° angle. This angle is

FIGURE 2Epibenthic sledge construction plan of the frame: front unit (a, c) and center unit (b, d-f)


important for the position of the lever unit.The flaps (Figure 1e) of the opening-clos-ing device are mounted to the floor platewith four hinges (M10). To prevent the flapsfrom being pushed into the boxes by waterpressure, the sides of the flaps are one centi-meter wider than the boxes. Furthermore,the flaps are strengthened with steel bars

(4 x 4 cm, Figure 1f). Additional support isprovided by a robust grid (steel bars 3 x 3 cm)fitted into the opening of the box (Figure1a). It is also possible to cover the openingswith “input filters” made of wire net to avoidthe catch of undesired objects (rocks, piecesof corals or sponges) that may cause dam-age to the animals collected.

The Lever Unit (Figures 1f, 2e, 4a)The opening-closing device consists of

a simple lever- and spring mechanism. Theflaps of the boxes are pressed against theopenings of the boxes by springs. The steelsprings are attached to the outside of bothboxes. The 1.7 m long and 0.18 m widelever is stiffly fixed to the lower flap of theepibenthic sampler and protrudes backwardsat an angle of 45° (Figure 4a). The connec-tion between the two flaps is provided by afreely rotating steel bar at the upper sides ofthe epi- and suprabenthic samplers (Figure1f). The sledge pushes down on the leverand, overcoming the spring tension by theweight of the sledge, opens the flaps. To com-pensate the distance of the lower net boxfrom the floor (0.25 m), the runner of thelever is thickened at its end.

The Net (Figures 3d-f, 5a-c)The epibenthic sledge is equipped with

two equal cone nets, the lower epibenthic-and the upper suprabenthic net (Figure 4c),aligned horizontally one above the other.The two nets are protected by the frame fromall sides and each one ends in a net bucket.The cone net (Figures 3e, 5a-b) is made of500 µm (mesh aperture) nylon monofila-ment. The mouth area is 1.01 m wide and0.36 m high and therefore slightly larger thanthe net boxes. The front collar consists oftwo layers of strong PVC canvas; it is 0.20m wide and bears holes for attachment. Thenets are pulled over the net boxes and fixedbetween the boxes and the terminal strips(5 mm; Figure 4d). The frontal margin ofthe front collar is thickened by a 4 mm stringto avoid slipping out. The cod end is ring-shaped; it consists of two layers of strongPVC canvas. It is 0.10 m wide and has aninner diameter of 0.12 m. The cod ends arepulled over the net buckets and fixed with aclamp in a locking furrow. The net buckets(Figure 3f) consist of hard PVC pipe of 9mm wall thickness. They are 0.35 m high atan outer diameter of 0.11 m and areequipped with a filter (300 µm nylonmonofilament, 10 x 10 cm). The net buck-ets are mounted in the horizontal center,0.31 m and 0.93 m, respectively, above thesledge bottom (Figure 3b). To avoid loss of

FIGURE 3Epibenthic sledge construction plan of the frame and the net: tail unit (a, b) and bottom view of the mountedframe (c), nets and net bucket (d-f)


Equation 1:

R = filtering area ratio

α = mesh aperture [mm]

φ = Ø of filament [mm]

γ130

= mesh area [m²] for 130 cm long nets

β = porosity

δ = filtering area [m²] = β × γ130

χ = mouth area [m²]

In waters with numerous particles thesledge can optionally also be equipped witha standardized adapter (Figure 4b; 60 kg;0.64 m long) to work with longer nets. Inthis case, to compensate for the higherweight of the net boxes in the front, thesledge should be equipped with two 10 kgbalance weights in the rear unit (long ver-sion: 4.17 m², 504 + 20 kg, 126 g/cm²).

HandlingDuring lowering (manoeuvre phase I) with

a winch speed of 0.5 m/s the ship runs withabout 1.0 kt over ground, depending on thecurrent velocity. This has the effect that the sledgesinks vertically to the ground, keeping the cablestretched and therefore avoiding entangling ofthe latter. The maximum wire length was nor-mally calculated at 1.5 (max. 1.8) times waterdepth. At maximum wire length the winch stopsand the ship runs ahead (manoeuvre phase II)with 1.0 kt over ground for the trawling time.For heaving (manoeuvre phase III), the shipstops shortly before the winch is started (winchspeed 0.5 m/s). Thus, the sledge has nearly thesame speed through the water at each manoeu-vre period until it leaves the ground.

The data provided by Table 1 should beconsidered as values for orientation. The actualparameters have to be adjusted to water cur-rent, quality of the sea floor, wind and waves. Ifthe vessel’s speed exceeds 4.5 kt, the sledge losescontact to the ground. Once on deck, the sledgecan be positioned upright on its rear end, facili-tating recovery of the net buckets and washingof the nets.

material (caused by swell) during heaving, avalve holds back any collected material.

In the standard configuration the sledgeoperates with 1.3 m long nets with amouth area of 0.35 m² and a mesh area ofγ

130 = 2.04 m². The nets are mounted with

some spare space to allow oscillation by tur-bulent flow, which supports self-cleaning of

the mesh. The relation of the mouth area tothe whole area of the mesh pore essentiallyinfluences the catch of the net. A “filteringarea ratio” R of 1:6 is regarded to be a prac-ticable size (Bossanyi, 1951; Riedl, 1955;McGowan & Frauendorf, 1966). Accord-ing to Smith et al. (1968:234, modified),R is calculated with the following equation:

FIGURE 4Epibenthic sledge construction plan of the frame, lever unit and net mounting: frame with lever unit (a); additiveframe (b); net mounting (c-d)


Estimation ofTrawling Distance

For all biodiversity calculations the ac-curacy of the trawl distance recording is ofcrucial importance. The gear starts collect-ing after touching the ground as soon as the

Equation (2): d[m] = vs [kt] × ∆t

ab [h] × 1852

Should vs vary within ∆t

ab it is necessary to divide the calculation of the towing distance

into different velocity sections (Equation 3).

Equation (3): d [m] = (∆tab

× vsab

) + (∆tcd

× vscd

) + (∆txy × v

sxy) × 1852

A winch gives the gear a new (|vw|) and/or an additional (v

s + |v

w|) velocity (Equation 4).

Equation (4): d [m] = (∆tab

× vsab

) + (∆tcd

× |vwcd

|) + (∆txy× (v

sxy+ |v

wxy|))× 1852

d = towing distance in meters vsab

= velocity of vessel within ∆tab

[kt]

vs = velocity of vessel [kt] |v

w| = absolute speed of winch [m/s]

ta = time of start of haul |v

wcd| = absolute speed of winch within ∆t

cd [m/s]

tb = time of end of haul ∆t

ab / ∆t

cd = time difference t

b – t

a/ t

d – t

c [h]

1852 = conversion factor nautical mile in to meters

If only longitude and latitude of start and end of the catch (actually vessel’s position) areknown, a latitude dependent estimation (Equation 5) of the towing distance can be engagedusing the Pythagorean Proposition.

Equation (5): d [m] = ×1852

lat = latitude (max. 90°)long = longitude (max. 360°)∆lat = |lat. start position – lat. end position|∆long = |long. start position – long. end position|

TABLE 1Parameters for each maneuvre period tested in different marine regions (abbreviations: kt: knot; o. g.: over ground: m/s: meters per second).

lowering / phase I trawling / phase II heaving / phase III

vessel speed lowering cable length vessel tow time vessel speed heaving location according too. g. [kt] [m/s] factor × depth o. g. [kt] [min] o. g. [kt] [m/s]

0.5 0.5 1.5 × - 1.8 × 1.0 15 0.0 0.5 Baltic Sea10-25 m N. Brenke

1.0 0.5 2.0 ´ 1.0 15 0.0 0.5 Skagerrak350 m N. Brenke

1.0 0.5 3.0 × – 4.0 × 1.5 10 1.0 1.5 Elbe Estuary N. Brenke10-20 m

0.5-0.1 0.5 1.5 × 1.0 10 - 30 0.0 0.6 Angola Basin A. Brandt5200-5400 m

1.0 0.5 1.5 × 1.0 10 0.0 0.8 Weddell Sea A. Brandt800-6300 m

flaps open and towing begins (ta) and the

sampling ends when the flaps close (tb). The

time tb – t

a is ∆t

ab. Equation 2 is a simple

calculation of the towing distance (d) inmeters assuming that v

s (actual speed of ves-

sel) does not change in ∆tab

.

Note: a nautical mile is equal to thelength of an arc minute on the circle of theearth surface (for the equator and all merid-ians 1 nm = 1 meridian minute = 1852 mand 1° = 60 nm). The circumference of theearth is 40000 km (21600 nm) or 360°.The factor for converting geographic unitsor nautical miles into meters is: 40000km ÷ (60 x 360') = 40000 ÷ 21600 = 1852.

Unfortunately, equation 5 is dependenton the latitude and hence only applicablefor areas close to the equator. The higherthe latitude, the smaller its circumferencebecomes, and hence the divergence of thecalculation increases. To solve this problem,the circumference dependent calculation(Equation 6 after Brattegard, see Brandt,1993b) is used.

Equation (6): d [m] = 1852 ×

= cos of arithmetic mean lat1

and lat2

By insertion of (cos 0° = 1,cos 90° = 0) one receives the correcting fac-tor for each latitude. For an approximatelycorrect calculation the cos of the arithmeticmean of lat

START and lat

END is used:

. .


For simplification, it is also possible to useonly one of the cos lat. Furthermore, suffi-cient accuracy is granted when only using thecos of the geographical degrees (minutes andseconds of the position can be neglected).

If the GPS-position is precisely known andthe timekeeping of the vessel’s towing speed andwinch is accurate, the problem remains that the

GPS-position is only correct for the researchvessel but not for the sledge. This problem isnegligible in shallow waters but increases withdepth. Up to now, no solution for this technicalproblem has been found. Consequently, for cal-culation of the trawl distance equation 2 is ac-curate in shallow waters and equation 4 is ap-proximately exact enough for the deep sea.

ResultsThe sledge has been used on various expe-

ditions during the years 2000-2003 (Figure 6).As expected, the sledge catches endobenthicand epibenthic organisms, mainly Peracarida,Mollusca, Echinodermata and “Polychaeta”.The percentage of hemisessile and non-swim-ming faunal elements (e.g. Bivalvia, Echino-dermata, “Polychaeta”) predominates the per-centage of epifauna capable of swimming (e.g.Peracarida, “Natantia”, Chaetognata). The con-dition of fragile specimens sorted from thesamples was excellent because mechanical dam-age during sampling is minimized. The sledgealso catches meiofauna (sensu Higgins & Thiel,1988:500 - 63 µm), but the abundance of thesespecimens (Harpacticoida and Nematoda) isnot comparable to typical meiofauna samplesbecause of the large mesh size (500 µm).

Trials in shallow waters of the Baltic Sea(Table 2) showed a minimum deviation of thetotal number of caught specimens of 24.8 %between two comparable (depth, sediment,towing distance) stations. For comparison ofthe samples, the number of individuals perstation is standardized for 1000 m2 towingarea (compare Basford et al., 1989; Svavarssonet al., 1990; Brattegard & Fossa, 1991;Brandt, 1993a; Brandt & Pipenburg, 1994).

In catches at seven deep-sea stations inthe Angola Abyssal Basin (South AtlanticOcean) (Table 3) the estimated mean de-viation from the arithmetic mean of all sta-tions is at least 39.1 %. This indicates anincrease of the percentage of error withdepth. The mean towing distance for theseven stations was 3589 m. The number ofcaught specimens of station #320 is notice-ably lower than those of the remaining sixstations. Under the presumption that allsamples (abundance E+S Σ specimen / 1000 m²;Table 3) are normally distributed, it has beenchecked whether the station #320 differs sig-nificantly from the remaining six stations.

Calculating for station #320

shows that the probability of this station tobelong to the ensemble of the other stationsis just 1.07 % (normal distribution; ands will be calculated without the data takenon station #320).

FIGURE 5Epibenthic sledge construction plan of the net: net (a-b); net mounting (c)


Respectively, there is a high probability ofbias in the sampling process on station #320.Excluding station #320 from the statisticalanalysis, the estimated mean deviation fromthe arithmetic mean is only 25.1 %. Thisvalue falsifies the hypothesis of an increas-ing percentage of error with depth.

TABLE 2Invertebrates on two shallow water stations (21 m) in the Baltic Sea, 19.4.2000, for standardized specimenabundances per 1000 m2. (abbreviations E / S: epibenthic- or suprabenthic sampler.)

Another question is whether the netboxes (epibenthic—and suprabenthic sam-pler) sample different communities of theepibenthos in the different horizons abovethe seafloor. Comparing the abundances ofinvertebrates in samples taken during thedeep-sea expedition Diva-1, a STUDENT´s t-

Test (Zöfel, 1992) shows that there is no sig-nificant difference between the samples ofthe two net boxes (t

calc= 1.461 is < tα = 2.179

for df = 12, α = 5%).

DiscussionThe sledge has proved to be successful

for qualitative sampling. The high numbersof caught invertebrate individuals (over12.200 for the Diva-1 expedition) cannotbe achieved with grabs and box cores in areasonable period of time. The extraordinaryquality of the catch provided a deep insightinto this region’s biodiversity. Even on ex-tremely rough bottoms with scattered largerocks as in the Weddell Sea, samples havebeen taken successfully.

The evaluations of the catch of theepibenthic sledge clearly show that thesamples cannot be used for estimation ofabsolute abundances. Furthermore, it is sug-gested that the two net boxes should not beregarded separately, since they do not catchdifferent faunal communities.

Hence, the calculated abundances ofspecimens per m² are only relative values,because the epibenthic sledge operates witha minimum systematic error of 25 %. How-ever, due to the low number of accomplishedoperations, the systematic error is still onlya preliminary approximation.

The sources of error are numerous andsubject to operating depth. One source of

station number #2-E #2-S #3-E #3-S

Mysidacea gen. sp. indet. 1 1 3

Diastylis sp. indet. 1 20 103 771 74

Amphipoda gen. sp. indet. 1 3 7 3

Sagitta sp. indet. 253 74 41

Halacarida gen. sp. indet. 1 1 2 3

Macoma baltica 2

Mya arenaria 1

Cardium fasciatum 4 3

Nudibranchia gen. sp. indet. 2 2

“Polychaeta” gen. sp. indet. 1 158 56 296 51

“Polychaeta” gen. sp. indet. 2 16 116 39

Sipunculida gen. sp. indet. 1 62 547 193 1463

Thyonidium sp. indet. 1 1

Asterias rubens 7 7

Ophiura ophiura 5 2 1

“Osteichthyes” 1

Σ invertebrate specimens E / S 505 810 1402 1675

towing distance [m] 247.0 247.0 463.0 463.0

specimens / 1000 m ² 2045 3279 3028 3618

Σ E+S invertebrate specimens 5324 6646

TABLE 3Abundances of invertebrates on seven deep-sea stations (5100-5450 m) of the Angola-Basin (Atlantic Ocean), Diva-1 M48/1 Expedition. (abbreviations: E / S:epibenthic- and suprabenthic sampler; x: arithmetic mean; s: standard deviation; t: test statistic)

Expedition Diva-1 M48/1 AREA 1 AREA 4 AREA 5 AREA 6 x s t

station number 318 318 320 320 338 338 340 340 344 344 348 348 350 350S E S E S E S E S E S E S E

Σ specimens/1000 m² S 1 6 168 134 449 213 188 192 1461,45

Σ specimens/1000 m² E 487 89 21 489 261 369 281 318 175

abundance E+S 488 95 189 623 710 582 469 509 180Σ specimens/1000 m²

towing distance [m] 2437 2715 3519 4320 4476 4476 3178 x = 3589m


error can be found in the shape of the shieldof the frontal unit. During trawling, thisshield causes turbulences that transportground into the net boxes (compare Buhl–Jensen, 1986:193; Olafsdottir & Svavarsson,2002:608; Linse et al., 2002:5; Urgorri, V.,observations by scuba divers, pers. comm.,2002). The turbulences vary depending onvelocity and on the solidity of the ground.Hence, on stations with sandy or rockygrounds, the volume of a sample is some-times decreased by a factor of ten. So, thisshield produces an incalculable error. Yet,another source of error could be groundcurrents. Because the flaps open at contactto the ground, a current running oppositeto the trawl direction can already transportmaterial into the net. Thus, the calculatedabundances per 1000 m² increase. For ex-ample, according to Zimmermann (1971),the ground currents at 5000 m depth in theAntarctic Bottom Water (AABW) reach amaximum of 16 cm/s. According to Dietrich(1975), a mean current velocity of 3 – 8 cm/sec can be assumed for the AABW. If it isnot possible to correct for current velocitiesthese values imply a possible error of 5 to 31% for a standard trawling speed of 1.0 kt(0.514 m/s).

An improved estimation of the towingdistance and therefore a more accurate quan-tification of the abundances can only beachieved by the employment of additionalmeasuring devices. Unfortunately, the stan-dard deviations of instruments likeflowmeters, tension recorders (Sachs, 1964;Rowe & Menzies, 1967; Wall & Ewing,1967), odometers and pingers (Hersey,1959; Nalwalk et al., 1961; Rowe &Menzies, 1967) also increase with depth anddistance to the vessel. The signal of tensionrecorders, caused by the frictional resistanceof the epibenthic sledge running over theground can be superposed by backgroundnoise caused by the cable, a problem increas-ing with depth and weight of the cable. At atowing speed of only 1.0 kt, flowmeters areat the lower limit of resolution and preci-sion of measurement. Odometers sometimesmeasure too short distances and there canbe differences between measured and calcu-lated trawling distances of up to 60% (com-pare Holme & McIntyre, 1971; Brattegard& Fossa, 1991). The signal of acousticpingers decreases with the square value ofthe distance and can be strongly disturbedby background noise. Nevertheless, if thevessel receives the signals undistorted an ac-curate calculation of the towing distance ispossible with error chances lower than 10 %.

Another source for errors for the calcu-lation of abundances can be the in homoge-neity of the fauna within the habitat. In thespecial case discussed here, the patchiness ofdistribution of organisms in the Baltic Seawill have stronger influence on the samplethan the errors in calculation of the trawldistance, because for operation in shallowerdepths the errors that occur when calculat-ing the towing distance are comparablysmall. On the understanding that the distri-bution of organisms is more homogeneousin the deep sea (preliminary result of theDiva-1 expedition) vice versa the influenceof patchiness is negligible, but the error forcalculation of trawl distance is substantiallyhigher in the deep sea. To minimize the sys-tematic error between the sledges samples,long towing distances should be chosen andthe sledge should be used with constant tow-ing speed.

An estimation of the difference betweencaught and actually existing numbers of in-dividuals is only possible with the help ofreliable instruments for quantitative sam-pling. Corers, multi-corers and grabs provedto be useful for quantitative sampling andare standard equipment in biological ocean-ography (Petersen’s grab, 1911; Petterson’simpact corer, 1928; Thamdrup, 1938 van Veengrab). Thus, the USNEL box corer (Hessler& Jumars,1974) is usually employed fordeep-sea research. Grabs and corers can takequantitative samples from soft ground, butsince the capacity of the gear is very small(usually less than 0.25m2), rare organismscan be underrepresented. On primary hardsubstrate grabs and corers are useless.

The arguments presented by Fossa et al.(1988:299) for an “either-or-release-mecha-nism” are valid, but complex release mecha-nisms such as proposed by Allan (1962),Baker et al. (1973), and Fossa et al. (1988)are too sensitive for rock dredging or not suit-able for use under the high pressure of thedeep sea. The very simple and insensitive le-ver-spring mechanism introduced here hasproven advantageous on all kinds of ground.

Because length and weight of the sledgecan be varied, it can be employed on ships ofdifferent sizes, as well as on various groundsand in various depths. On vessels like the FSPOLARSTERN (118,00 m, 10878,52G.R.T.), FS METEOR (97,50 m, 3990G.R.T.) or MS ALKOR (55,20 m, 999,08G.R.T.) (Reinke-Kunze, 1986), the sledge canbe used without difficulty. On smaller vesselslike the FK UTHÖRN (II) (30,50 m, 254,27G.R.T.), the sledge can well be employed inits short version. But vessels of this size presentthe lower limit for operation.

Depending on the employed version, thesledge weighs 126 g/cm² to > 206 g/cm² andis comparable to the RP77-sledge (~ 185 g/cm²) and the BB95-sledge (~ 136 g/cm²).

In the standard configuration the sledgeis equipped with 1.3 m long nets with R

130

1: 5.34 (compare Sorbe, 1983: R 1:5; Brandt& Barthel, 1995: R 1:6.2; Christensen &Nuppenau, 1997: R 1:4.2). The mesh sizeof 500 µm is a standard mesh size in deep-sea research (Menzies, 1962, 1964; Sorbe,1983). The elongated cylinder cone nets

FIGURE 6The Epibenthic sledge on RV Meteor, side view(Photo: N. Brenke).


(Figure 3d) bear a R194

of 1:9.92 and are thuscomparable to the nets of Rothlisberg &Pearcy (1977: R 1:9). Because the weight,alignment of samplers and configuration ofnets are comparable to those of the RP77-sledge, comparability to other results ofepibenthic research is given (e.g. Brattegard& Fossa, 1991; Brandt, 1993a, 1993b, 1997;Brandt & Piepenburg, 1994; Piepenburg &Juterzenka, 1994; Linse et al., 2002). Theerror rates of all epibenthic sledges are simi-larly high. So far, 27 successful operationshave been conducted in the Atlantic deepsea and on various grounds of the AntarcticWeddell Sea. The sampling efficiency hasbeen exceptionally high because the sledgeconsistently descends right side up and thenets are protected from damage. The con-struction of the epibenthic sledge benefitsfrom its low center of gravity and the slop-ing flaps, which provide advantageous hy-drodynamic conditions during setting outand hauling.

AcknowledgmentsIn a biological department, a project like

this is only realizable with an idealistic divi-sion head, an excellent engineering work-shop and a motivated and innovative spe-cialized engineering staff.

I am grateful to Prof. Dr. J.-W. Wägelefor his valuable suggestions for improvingthe construction of the sledge and for theconfidence shown to me in this project.

Many thanks to W. Dreckmann, H.D.Knoop and M. Pool for their tireless sugges-tions for improvement of the sledge and tothe rest of the workshop staff: Mr. H.U.Goryczka, H. Knoop, H. Pöten, K.H.Sandermann and E. Wuttke.

I would like to thank Dr. D. Piepenburg(IPÖ Kiel, Germany), Prof. A. Brandt, Dr. U.Mühlenhard-Siegel and W. Brökeland (ZIMHamburg, Germany), A. Kremer(GEOMAR Kiel, Germany), Dr. M.Fanenbruck, Dr. B. Hilbig and M. Raupach(University of Bochum, Germany), officers,boatsmen and crew of the RV METEOR,RK UTHÖRN and RV ALKOR for theirhelpful comments and discussions. I wouldlike to thank N. Schulte-Pelkum, A. Mursch

and B. Hilbig for helping to correct themanuscript, and Kelly B. Miller for helpingto correct the English. This work was sup-ported by grants of the DFG under contractNo. WA 530/23-2 and No. WA 530/27-1.

ReferencesAldred, R.G., Thurston, M.H., Rice, A.L. and

Morley, D.R. 1976. An acoustically monitored

opening and closing epibenthic sledge. Deep-

Sea Res. I 23:167-174.

Allan, W. and Bé, H. 1962. Quantitative

multiple opening-and-closing plankton

samplers. Deep-Sea Res. I 2:144-151.

Baker, A. de. C., Clarke, M.R. and Harris,

M.J. 1973. The N.I.O. combination net

(RMT 1+8) and further developments of

rectangular midwater trawls. J Mar Biol Ass.

U.K. 53:167-184.

Basford, D.J., Eleftheriou, A. and Raffaelli, D.

1989. The epifauna of the northern North Sea

(56°-61°N). J Mar Biol Ass. U.K. 69:387-407.

Beardsley, A.J. 1973. Design and evaluation of

a sampler for measuring the near-bottom

vertical distribution of pink shrimp, Pandalus

jordani. Fish Bull. 71:243-253.

Beyer, F. 1958. A new, bottom –living

Trachymedusa from the Oslofjord. Description

of the species, and a general discussion of the

life conditions and fauna of the fjord deeps.

Nytt Magasin f. Zoologi. 6:121-143.

Blomqvist, S. and Lundgren, L. 1996. A

benthic sled for sampling soft bottoms.

Helgoländer Meeresunters. (Hamburg)

50:453-456.

Bossanyi, J. 1951. An apparatus for the

collection of plankton in the immediate

vicinity of the sea-bottom. J Mar Biol Ass.

U.K. 30:265-270.

Boyd, S.E. (ed.) 2002. Guidelines for the

conduct of benthic studies at aggregate

dredging sites. The Centre for Environment,

Fisheries and Aquaculture Science (CEFAS).

Burnham Laboratory, Burnham-on-Crouch

(Essex, UK) 100 pp.

Brandt, A. 1993 a. Redescription of Janiralata

pulchra (Hansen, 1916) (Janiridae) from the

Kolbeinsey Ridge, North Atlantic and

synonymy with Ianthopsis pulchra

(Acanthaspidiidae) (Crustacea, Isopoda).

Ophelia (Helsingör) 37-2:127-141.

Brandt, A. 1993 b. Composition, abundance,

and diversity of peracarid crustaceans on a

transect of the Kolbeinsey Ridge, north of

Iceland. Polar Biol. (Berlin) 13:565-576.

Brandt, A. 1997. Suprabenthic Peracarida

(Crustacea, Malacostraca) sampled at 75°N off

East Greenland. Polar Biol. (Berlin) 17:462-464.

Brandt, A. and Barthel, D. 1995. An improved

supra- and epibenthic sledge for catching

peracarida (Crustacea, Malacostraca). Ophelia

(Helsingör) 43-1:15-23.

Brandt, A. and Piepenburg, D. 1994.

Peracarid crustacean assemblages of the

Kolbeinsey Ridge, north of Iceland. Polar Biol.

(Berlin) 14:97-105.

Brattegard, T. and Fossa, J.H. 1991.

Replicability of an epibenthic sampler. J Mar

Biol Ass. U.K. 71:153-166.

Brunel, P., Besner, M., Messier, D., Poirier, L.,

Granger, D. and Weinstein, M. 1978. Le

traineau suprabenthique MACER-GIROQ:

appareil amélioré pour l’échantillonnage

quantitatif étagé de la petite fauna nageuse au

voisinage du fond. Inter Rev d Ges Hydrobiol

Hydrogr. (Leipzig) 63:815-829.

Buhl-Jensen, L. 1986. The benthic amphipod

fauna of the West-Norwegian continental shelf

compared with the fauna of five adjacent

fjords. Sarsia (Bergen) 71:193-208.

Carey, G. Jr. and Hancock, D.R. 1965. An

anchor-box dredge for deep-sea sampling.

Deep-Sea Res. I 12:983-984.

Christiansen, B. and Nuppenau, V. 1997. The

IHF Fototrawl: experiences with a television-

controlled, deep-sea epibenthic sledge. Deep-

Sea Res. I 44-3:533-540.

Clutter, R.I. 1965. Self-closing device for

sampling plankton near the sea bottom.

Limnol Oceanogr. 10:293-296.


Colman, J.S. and Segrove, F. 1955. The Tidal

plankton over Stoupe Beck sands, Robin

Hood’s Bay (Yorkshire, North Riding). J Anim

Ecol. 24:445-462.

Dietrich, G., Kalle, K., Kraus, W. and Siedler,

G. 1975. Allgemeine Meereskunde. Eine

Einführung in die Ozeanographie. Gebrüder

Bornträger (Berlin, Stuttgart). 3rd ed. 593 pp.

Ekman, S. 1911. Neue Apparate zur

qualitativen und quantitativen Erforschung der

Bodenfauna der Seen. Inter Rev Ges

Hydrobiol. (Berlin) 3:553-561.

Forster, G.R. 1953. A new dredge for

collecting burrowing animals. J Mar Biol Ass.

U.K. 32:193-198.

Fossa, J.H., Larsson, J. and Buhl-Jensen, L.

1988. A pneumatic, bottom-activated, opening

and closing device for epibenthic sledges. Sarsia

(Bergen) 73:299-302.

Frolander, H.F. and Pratt, I. 1962. A bottom

skimmer. Limnol Oceanogr. 7:104-106.

Gage, J.D. and Tyler P.A. 1991. Deep-sea

biology: A Natural History of Organisms at

the Deep-Sea Floor. Cambridge, New York:

University Press. 504 pp.

Gunter, G. 1957. Dredges and Trawls. In:

Hedgpeth, J.W. (ed.) 1957: Treatise on marine

ecology and paleoecology. Vol. 1. Ecology.

Mem Geol Soc Amer. 67(1):73-86.

Hensen, V. 1895. Ergebnisse der Plankton-

Expedition der Humboldt-Stiftung. Methodik

der Untersuchungen. Verlag von Lipsius &

Tischer (Kiel, Leipzig) Bd. I. B. 200 pp.

Hersey, J.B. 1960. Acoustically monitored

bottom coring. Deep-Sea Res. I 6:170-172.

Hessler, R.R. and Jumars, P.A. 1974. Abyssal

community analysis from replicate box cores in

the central North Pacific. Deep-Sea Res. I

21:185-209.

Hessler, R.R. and Sanders, H.L. 1967. Faunal

diversity in the deep sea. Deep-Sea Res. I 14:65-78.

Hesthagen, I.H. 1970. On the near-bottom

plankton and benthic invertebrate fauna of the

Josephine Seamount and the Great Meteor

Seamount. Meteor Forsch.-Erg. (Berlin)

D8:61-70.

Higgins, P.R. and Thiel, H. (eds.) 1988.

Introduction to the Study of Meiofauna.

Washington, DC: Smithsonian Inst. Press. 488 pp.

Holme, N.A. 1964. Methods of Sampling the

Benthos. Adv Mar Biol. (Plymouth) 2:171-260.

Holme, N.A. and McIntyre, A.D. (eds.) 1971.

Methods for the Study of Marine Benthos. 1st

ed. Oxford, Edinburgh: Blackwell Scientific

Publications. 334 pp.

Hopkins, T. 1964. A survey of marine bottom

samplers. Prog Oceanogr. (Oxford) 2:215-256.

Koulouri, P. 2002. Preliminary study of

hyperbenthos in Heraklion Bay (Cretan Sea).

http://www.biomareweb.org/3.6html 3 pp.

Linse, K., Brandt, A., Hilbig, B. and Wegener,

G. 2002. Composition and distribution of

suprabenthic fauna in the south-eastern

Weddell Sea and off King George Island.

Antarctic Science 14(1):3-10.

Macer, C.T. 1967. A new bottom-plankton

sampler. J Cons perm int Explor Mer.

(Copenhagen) 31-2:158-163.

MacIntyre, R.J. 1964. A box dredge for

Quantitative sampling of benthic Organisms.

Limnol Oceanogr. 9(3):460-461.

McGowau, I.A. and Fraundorf, V.J. 1966. The

relation between size of net used and estimates

of zooplankton diversity. Limnol Oceanogr.

10(1):149-152.

Menzies, R.J. 1962. Reports of the Lund

University Chile Expedition 1948-49. 42. The

zoogeography, ecology and systematics of the

Chilean marine Isopods. Lunds Universitets

Arsskrift Avd. 2 (Lund). Bd. 57-11:162 pp.

Menzies, R.J. 1964. Improved Techniques for

Benthic Trawling at Depths Greater then 2000

Meters. In: Biology of the Antarctic seas. Lee

M.O. (ed.) Antarctic Research Series, Vol. 1

American Geophysical Union. (Washington

D.C.) pp. 93-109.

Menzies, R.J., George, R.Y. and Rowe, G.T.

1973. Abyssal environment and ecology of the

world oceans. New York, London, Sydney,

Toronto: John Wiley & Sons. 488 pp.

Mortensen, T. 1925. An Apparatus for

catching the Micro-Fauna of the Sea-Bottom.

Vidensk. Medd. Dansk Naturh. Foren.

(Kjobenhavn) 80:445-451.

Nalwalk, A.J., Hersey, J.B., Reitzel, J.S. and

Edgerton, H.E. 1961. Improved techniques of

deep-sea rock dredging. Deep-Sea Res. I

8:301-302.

Okelmann, K.W. 1964. An improved

Detritus-Sledge for collecting Meiobenthos.

Ophelia (Helsingör) 1-2:217-222.

Olafsdottir, S.H. and Svavarsson, J. 2002:

Ciliate (Protozoa) Epibionts of deep-water

asellote Isopods (Crustacea): Pattern and

Diversity. J Crust Biol. 22(3):607-618.

Oug, E. 1977. Faunal distribution close to the

sediment of a shallow marine environment.

Sarsia (Bergen) 63:115-121.

Pearcy ,W.G. 1972. Distribution and diel

changes in the behavior of pink shrimp,

Pandalus jordani off Oregon. Proc. Natl.

Shellfish. Assoc. (Easton) 62:15-20.

Petersen, C.G.Joh. 1911. Valuation of the sea

I. Animal life of the sea bottom, its food and

quantity. VI: New apparatus for the quantitative

collections of bottom animals. Rep Dan Biol

Stat Board Agric. (Copenhagen) 20-1911:1-81.

Petersen, C.G.Joh. 1914. Valuation of the sea

II. The animal communities of the sea-bottom

and their importance for marine zoogeography.

Rep Dan Biol Stat Board Agric. (Copenhagen)

21-1913:1-44.

Petterson, O. 1928. A new apparatus for

taking of bottom-samples. Svensk Hydrog Biol

Komm Skr., Scr Hydrog. 6:1-7.

Piepenburg, D. and Juterzenka, v. K. 1994.

Abundance, biomass and spatial distribution

pattern of brittle stars (Echinodermata:

Ophiuroidea) on the Kolbeinsey Ridge north

of Iceland. Polar Biol. (Berlin) 14:185-194.

Reinke-Kunze, C. 1986. Den Meeren auf der

Spur. Geschichte und Aufgaben der deutschen

Forschungsschiffe. Koehler (Herford) 169 pp.


Rice, A.L., Aldred, R.G., Darlington, E. and

Wild, R.A. 1982. The quantitative estimation

of the deep-sea megabenthos, a new approach

to an old problem. Oceanol Acta. (Paris)

5(1):63-72.

Riedl, R. 1955. Aufsammlung tiefer

Meeresböden in abgegrenzten Schichten und

Flächen. Arch. f. Hydrobiol. (Stuttgart) 51-

2:189-208.

Riedl, R. 1963. Probleme und Methoden der

Erforschung des litoralen Benthos. Verh. Dt.

Zool. Ges. 1962, Zool Anz. Suppl. (Wien)

26:505-567.

Rothlisberg, P.C. and Pearcy, W.G. 1977. An

epibenthic sampler used to study the ontogeny

of vertical migration of Pandalus jordani

(Decapoda, Caridea). Fish Bull. 74:994-997.

Rowe, G.T. and Menzies, R.J. 1967. Use of

sonic techniques and tension recordings as

improvements in abyssal trawling. Deep-Sea

Res. I 14:271-274.

Russel, F.S. 1928. A net for catching plankton

near the bottom. J Mar Biol Ass. U.K.

15:105-108.

Sachs, P.L. 1964. A Tension Recorder for

Deep-sea Dredging and Coring. J Mar Res.

22(3):279-283.

Saint-Hilaire, C. 1909. Eine Dredge mit

auswechselbaren Messern. Rev d Ges

Hydrobiol u Hydrogr. 2:449-451.

Salvini-Plawen, L.v. 1975. Mollusca Caudofoveata.

Mar Invert Scand. (Oslo) 4:1-55.

Sanders, H.L., Hessler, R.R. and Hampson,

G.R. 1965. An introduction of the study of

the deep-sea benthic faunal assemblages along

the Gay Head-Bermuda transect. Deep-Sea

Res. I 12:845-867.

Schlieper, C. 1968. Methoden der

Meeresbiologischen Forschung. VEB Fischer

(Jena) 322 pp.

Sirenko, B.J., Markhaseva, E.L., Buzhinskaya,

G.N., Golikov, A.A., Menshutkina, T.V.,

Petryashov, V.V., Semenova, T.N., Stepanjants,

S.D. and Vassilenko, S.V. 1996. Preliminary

data on suprabenthic invertebrates collected

during the RV Polarstern cruise in the Laptev

Sea. Polar Biol. 16:345-352.

Smith, P.E., Counts, R.C. and Clutter, R.I.

1968. Changes in filtering efficiency of

plankton nets due to clogging under tow. J

Cons perm int Explor Mer. (Copenhagen)

32(2):232-248.

Sneli, J.-A. 1998. A simple benthic sledge for

shallow and deep-sea sampling. Sarsia (Bergen)

83:69-72.

Sorbe, J.C. 1983. d traineau destine a

l’echantillonnage quantitatif etage de la faune

suprabenthique neritique. Ann. Inst.

Océanogr. (Paris) 59-2:117-126.

Steinmann, P. 1909. Eine Netzdredge für

Tiefenfänge. Rev d Ges Hydrobiol u Hydrogr.

2:452-456.

Svavarsson, J., Brattegard, T. and Strömberg,

J.-O. 1990. Distribution and diversity patterns

of asellote isopods (Crustacea) in the deep

Norwegian and Greenland Seas. Prog

Oceanogr. (Oxford) 24:297-310.

Thorson, G. 1957. Sampling the benthos. In:

Hedgpeth, J.W. (ed.) 1957: Treatise on marine

ecology and paleoecology. Vol. 1. Ecology.

Mem Geol Soc Amer. 67(1): 61-73.

Wall, R.E. and Ewing, M. 1967. Tenssion

Recorder for deep-sea winches. Deep-Sea Res.

I 14:321-324.

Werner, B. 1939. Über ein neues, automatisch

schließendes Bodennetz. Inter Rev Ges

Hydrobiol. (Berlin) 38:368-371.

Wickstead, J. 1953. A new apparatus for the

collection of bottom plankton. J Mar Biol Ass.

U.K. 32:347-355.

Zimmermann, H.B. 1971. Bottom currents

on the New England continental Rise. J

Geophys Res. 76(24):5865-5876.

Zöfel, P. 1992. Statistik in der Praxis. G.

Fischer Verlag (Jena, Stuttgart) 3rd ed. 422 pp.


A U T H O R SCedric M. GuigandRobert K. CowenJoel K. LlopizDavid E. RichardsonUniversity of MiamiRosenstiel School of Marine andAtmospheric Science

T E C H N I C A L N O T E

A Coupled Asymmetrical Multiple OpeningClosing Net with Environmental Sampling System

A B S T R A C TRecruitment levels of fishes are potentially related to the abundance of larval fishes

and their food source. A system that could allow for the concurrent investigation of fine-scale distribution of fish larvae and their potential prey could add significantly to theunderstanding of the early life history of marine fishes. A coupled Multiple Opening Clos-ing Net and Environmental Sensing System (MOCNESS) that combines two sub-systems(1 m2 and 4 m2 net sets) working in synchronization was designed to answer these ques-tions. The mesh size was different on each set of nets allowing the collection of a broadsize range of organisms while optimizing the catch of larger fish larvae and eliminatingunnecessary large samples of zooplankton. Moreover, the system eliminated the need todeploy separate MOCNESS using different mesh sizes, thus reducing ship time costs, andavoiding any aliasing associated with trying to sample the same water mass with separatenets fished sequentially. The system has been used at sea under varying weather condi-tions onboard the R/V F. G. Walton Smith and sampled adequately.

sets of nets with different mesh and mouthsizes. A double, synchronous opening/clos-ing net system within a single frame is thebasis of this sampling gear.

Construction and AssemblyThe coupled MOCNESS system is com-

prised of a 1 m2 and a 4 m2 MOCNESS sys-tem (Figure 1). These two systems were pur-chased from Biological EnvironmentalSampling System Inc. (B.E.S.S. Inc.). Thetwo individual net frames were removed anda new frame which joins the two systemswas designed and built at the RosenstielSchool of Marine and Atmospheric Sciencein Miami.

The frame was composed of five pieces:one bottom beam, one top beam, two sidebeams and a middle beam. The structure wasmade of anodized aluminum H-beam of20.32 cm x 20.32 cm (8x8 inches) for thetop and bottom beams, and 15.24 cm x 20.32cm (6x8 inches) for the side and center beams.The overall size and weight of the frame is3.56 m x 4.26 m and 900 kilograms.

The underwater electronics unit andoptional sensor unit, provided by B.E.S.S.Inc., were attached as on a regularMOCNESS system and controlled the net

opening and closing via two separate togglerelease systems actuated by two step motors.The net response mechanism was only in-stalled on the 1 m2 net side since the 4 m2

net was large enough to create a noticeableangle change when properly released.

Since the net sizes were not equivalenton each side of the frame, the uneven dragapplied on the system when submergedwould have rendered the flight unstable orpossibly caused the spinning of the framearound its vertical axis. To avoid this prob-lem, a drag net was attached under the 1 m2

MOCNESS (Figure 1). The additional net-ted surface area provides the needed addi-tional drag on the side of the smaller netopening. The drag net was made of strongnylon mesh (1/4 inch) fitted with grommets,and was attached to the frame by lacing anylon line through the net grommets andholes drilled in the frame (Figure 1). More-over, the weight distribution on the framewas corrected with an additional lead bars(50 kilograms) on the 4 m2 side.

Double MOCNESS systems have thesame net opening area on each side, and areusually employed to accommodate a highernumber of nets that are triggered in sequence(Wiebe et al., 1985). Our system was de-signed to trigger both nets (1 m2 net and 4

TI N T R O D U C T I O N he Multiple Opening Closing Net and Environmental Sensing System(MOCNESS, Wiebe et al., 1976) was ini-tially based on the Tucker Trawl (Tucker,1951). It is commonly used in biologicaloceanography to investigate the fine-scalevertical distribution of plankton while simul-taneously collecting hydrographic data char-acterizing the water mass being sampled.Although many variations of the originalMOCNESS design have been created in thepast two decades (Wiebe and Benfield,2003), none of these allow for simultaneouscollection of small zooplankton and larger,faster moving post-flexion ichthyoplankton.Collecting small zooplankton requires asmall mesh size while collecting larger, fastmoving ichthyoplankton calls for a large netopening with high filtration capacity (largermesh size). A large net with small mesh sizecould be used, but aside from the high levelof drag, sample biomass would be large andwould require a substantial amount of timeto sort for target species and quantify zoop-lankton. Net systems have been coupled withOptical Plankton Counters (Cass-Calay,2003; Mullin & Cass-Calay, 1997), butthese systems remain, thus far, limited tocounting particle size without providing ac-curate taxonomic information.

The system presented herein allows forthe sampling of both zooplankton prey andichthyoplankton predator fields synopticallyby employing a combination system of two


m2 net) simultaneously so that each depthbin sampled would have a prey and predatorfield as well as all the environmental data as-sociated with it. The motor control cablewhich sends a voltage pulse to the toggle re-lease system was spliced in two. The resultwas two parallel circuits going to each step-ping motor so that a single voltage pulsewould trigger the two toggle release systemssimultaneously and subsequently trigger thetwo nets at the same time. For our purpose,the system was equipped with two sets of nets:one set of five 1 m2 nets with 150 µm meshand one set of five 4 m2 nets with 800 µmmesh. Sensors incorporated in this samplinggear were pressure, temperature, conductiv-ity, fluorometry, downwelling light, volumefiltered, towing angle, and net response.

AssessmentThe system was deployed for the first

time offshore of Miami on 7 January 2003.As the net system was taller than the A-frameonboard the R/V F. G. Walton Smith, it waslaunched by sliding it off the edge of theship’s stern (Figure 2a-e). The frame wasfilmed via an underwater camera to assess

the behavior of the system underwater (Fig-ure 2f). The system was towed using a 10.16mm (0.400 inch) double-armored, threeconductor cable from the Rochester Cor-poration mounted on a Hawbolt IndustriesSPR-2840/S scientific winch. The systemwas sent to a target depth of 100 meters andsampled correctly and steadily with both nets

(1 m2 and 4 m2) opening and closing simul-taneously without affecting the overall framebalance. The flow counts were calibrated forthe 1 m2 nets and subsequently multipliedby four to obtain the volume filtered by the4 m2 nets. Subsequent to the initial testing,between January 2003 and April 2004, morethan 350 successful deployments have beenmade in sea conditions ranging from Beau-fort forces of one to six without any prob-lems with simultaneous net firings or stabil-ity. The tension on the tow cable ranged from1,000 kg to 2,750 kg depending on sea state.The mean net angle ranged from 46.07° to55.07° during the system deployment inconditions two to six on the Beaufort scale(Table 1). These data confirm the stabilityof the system in different sea conditions.

FIGURE 1Diagram of the coupled 1 m2 and 4 m2 Multiple Opening Closing Net and Environmental Sensing System.

FIGURE 2Launch sequence: since the system is taller than the A-frame onboard the R/V F.G. Walton Smith, the packagewas launched and retrieved by sliding it off the edge of the stern.

TABLE 1Mean net angle during ascent in different seaconditions.

MeanBeauford angle Standard Variance/

force (Degree) deviation mean

Force 2 51.22 2.86 0.16

Force 3 46.07 5.91 0.76

Force 4 51.18 5.51 0.59

Force 5 55.07 1.33 0.03

Force 6 54.90 1.78 0.06

a. b. c.

d. e. f.


ConclusionThis coupled MOCNESS responded to

the need of synoptic plankton sampling thatis useful in trophic and ichthyoplanktonstudies. The system is currently used in theStraits of Florida to assess the early life biol-ogy of billfishes and other economicallyimportant species. This system offers thepossibility of effectively and efficiently sam-pling a large size spectrum of zooplanktonand ichthyoplankton, while simultaneouslyreducing ship time requirements.

AcknowledgementsWe would like to thank the RSMAS

machine shop manager, Manuel Collazo, forhis expertise and skills, the captain and crewof the R/V F. G. Walton Smith, Erich Horganfor providing vital information aboutMOCNESS systems, as well as Claire Parisand Deanna Pinkard for reviewing the earlyversions of this manuscript.

ReferencesCass-Calay, S.L. 2003. The feeding ecology of

larval Pacific hake (Merluccius productus) in the

California Current region: an updated

approach using a combined OPC/MOCNESS

to estimate prey biovolume. Fish Oceanogr.

12(1):34-48.

Mullin, M.M. and Cass-Calay, S.L. 1997.

Vertical distributions of zooplankton and

larvae of the Pacific Hake (whiting), Merluccius

productus , in the California Current System.

Calif Coop Oceanic Fish Invest Rep. 38:127-136.

Tucker, G.H. 1951. Relation of fishes and

other organisms to the scattering of underwa-

ter sound. J Mar Res. 10:215-238.

Wiebe, P.H. and Benfield, M.C. 2003. From

the Hensen net towards 4-D biological

oceanography. Prog in Ocean. 56:7-136.

Wiebe, P.H., Morton A.W., Bradley A. M.,

Backus R.H., Craddock J.E., Barber V.,

Cowles T.J. and Flierl G.R. 1985. New

developments in the MOCNESS, an apparatus

for sampling zooplankton and micronekton.

Mar Biol. 87:313-323.

Wiebe, P.H., Burt K.H., Boyd S.H., and

Morton A.W. 1976. A multiple opening/

closing net and environmental sensing system

for sampling zooplankton. J Mar Res.

34(3):313-325.


A U T H O R SNadya VinogradovaSergey VinogradovAtmosphericand Environmental Research, Inc.

Dmitri NechaevVladimir KamenkovichThe University of Southern Mississippi

Alan F. BlumbergStevens Institute of Technology

Quamrul AhsanHonghai LiHydroQual, Inc

P A P E R

Evaluation of the Northern Gulf of Mexico LittoralInitiative Model Based on the Observed Temperatureand Salinity in the Mississippi Bight

A B S T R A C TTemperature and salinity measurements from the Northern Gulf of Mexico Littoral Ini-

tiative (NGLI) survey during August 30 - September 14, 2000 reveal a high level of tempo-ral and spatial variability in the Mississippi Bight. To support scientific studies using anumerical model, a three-dimensional hydrodynamic Estuarine and Coastal Ocean Model(ECOM) is implemented in the Mississippi Bight. The ECOM is run with realistic topogra-phy, stratification and meteorological forcing to hindcast circulation on a shallow and highlyvariable shelf of the Mississippi Bight. The results of the model are compared with obser-vation to evaluate the ECOM performance on different temporal scales. Based on the areaoceanography and data availability, three temporal scales are chosen for model/data com-parison: fine scale (less than an hour), diurnal, and large scale (a two-week period). Limi-tations of the ECOM application on each scale are discussed. The model is capable toreproduce observed water masses, describe spatial distribution of water properties, andsimulate areas with high horizontal gradient such as freshwater plumes. However, delayedresponse to meteorological forcing, overestimated mixing rates and uncertainties in com-putation of river discharges result in statistically significant bias in the simulations. Alongwith traditional linear correlations from all observational points and spectral analysis overthe diurnal cycle, a new technique of model validation is introduced. The technique is anew application of an existing variational interpolation method. Detailed description of themethod and numerical procedure allow one to apply this technique to any oceanographicdata with prescribed data variances for model/data comparison.

an important role in the estuary circulation.Mean depth of the Mississippi Sound is 3m (Kjerfve, 1983), which indicates that windis another mechanism that dominates in theestuary. In addition, multiple connectionsto the Northern Gulf of Mexico through anumber of passes between barrier islandsallow important interaction between the es-tuarine and the Gulf of Mexico waters. Thesefeatures present a unique circulation patternand control the long-term dynamics.

To describe complex features of the Mis-sissippi Bight circulation, an extensive set ofobservations is required. However, measure-ments used in the early studies were lim-ited. For example, Kelly (1991) studied cir-culation and hydrography of the MississippiBight using five current meter moorings andtwelve CTD stations, while the study ofBrooks (1984) was based on the observa-

tions from four moorings and eight CTDstations. Significant progress was achievedin 1999 – 2001, when a series of oceano-graphic surveys were conducted, collectinghigh-resolution data in the Mississippi Bightarea. More than a thousand CTD stationswere sampled over a two-year period, cover-ing the area of more than 30,000 km2

(Vinogradov et al., 2004). These surveyswere conducted in support of the NorthernGulf of Mexico Littoral Initiative (NGLI)program. The NGLI system was establishedas both an operational Navy product and aresearch tool used to benefit Gulf Coasteconomies and the marine environment(Asper et al., 2001). Besides observations,another constituent of the NGLI programis a modeling system. Prediction of thecoastal ocean circulation is one of the mostchallenging issues in numerical modeling.

TI N T R O D U C T I O N he Mississippi Bight shelf is an area of considerable interest to marine com-merce, human recreation, oil and gas explo-ration and development. In addition, com-mercial fishing activity is an integral part ofthe economy of the region. Geologically, theMississippi Bight shelf is wide due to com-parative tectonic stability; the coastal plainis broad and of low relief, allowing abun-dant and extensive estuaries to intrude in-land. As a typical Gulf of Mexico estuary,Mississippi Sound is a bar-built system witha flat topography (Schroeder and Wiseman,1999). It stretches for approximately 130 kmalong Louisiana, Mississippi and Alabamacoastlines and is separated from the North-ern Gulf of Mexico by sandy barrier islands(Figure 1) at a distance of about 15 km fromthe mainland (Kjerfve, 1983). MississippiSound receives large discharges of fresh wa-ter from the major rivers of Mississippi andMobile as well as many minor rivers andnumerous bayous. The freshwater influx datasuggests that buoyancy driven currents play


Circulation on the shelf is influenced by avariety of processes, such as winds, topogra-phy, complex coastline, shelf wave propaga-tion, storm surges and many others (Allenet al., 1995; Chen and Beardsley, 2002;Klink, 1995; Fong and Geyer, 2001). Inorder to ensure that a numerical model iscapable of reproducing and predicting acoastal circulation, thorough validation ef-forts are required. It is usually done throughcomparison of the model with the data,which leads to model improvement. TheEstuarine and Coastal Ocean Model(ECOM) is one of the NGLI models. It isdesigned by HydroQual, Inc. The ECOMis a three-dimensional, time-dependent,sigma-coordinate model that computes cir-culation and mixing patterns of the coastalocean. ECOM proved to be a useful toolfor investigating the mechanisms of the up-welling circulation along the Oregon Con-

tinental Shelf (Allen et al., 1995) as well aswind and tide forced processes in Chesa-peake Bay (Blumberg et al., 1990) andGeorges Bank (Chen et al., 1995). TheNGLI experiment provides a unique oppor-tunity for model/data comparison in theMississippi Bight.

The primary goal of this paper is to evalu-ate the capability of the ECOM to describehydrodynamics in the Mississippi Bight shelfon different temporal scales. Evaluation ofthe model on different scales is critical as itrepresents applicability of the model to dif-ferent oceanic processes. For example, themodel’s ability to reproduce processes on adaily scale is important for studying tidaldynamics, whereas in climatological andplanetary studies a large-scale model perfor-mance is crucial. The model skill assessmentis carried out through comparison of theECOM hydrodynamics with the observed

temperature and salinity. ECOM simula-tions correspond to one of the NGLIoceanographic surveys, collected duringAugust 30 - September 14, 2000. When thesimulated values are compared with tempo-rally and spatially matched data, the modelperformance on small scales can be evalu-ated. Analysis of time series stations (25-hours anchor stations) allows one to com-pare the model with the data on the dailyscale. The largest temporal scale, on whichthe ECOM performance is analyzed, is theperiod of the survey, i.e. two weeks. Unfor-tunately, the lack of simulations correspond-ing to other NGLI surveys does not allowone to consider larger temporal scales, suchas seasonal and inter-annual scales, neces-sary for a thorough model validation. How-ever, as mentioned earlier, for a highly vari-able shelf such as Mississippi Bight, atwo-week period is a sufficient time to es-tablish the circulation pattern. Consequently,a model/data comparison within this perioddoes provide the information about the abil-ity of the ECOM to identify major physicalprocesses on the Mississippi Bight.

The next section describes the oceano-graphic observations collected in the studyarea, followed by the description of themodel and the methods used to comparethe simulations with the measurements. Theresults and discussion provide informationon the advantages and limitation of theECOM implementation in the area.

Oceanographic ObservationsDuring August 30 - September 14, 2000,

a 15-day survey was conducted in support ofthe NGLI program. The fieldwork was car-ried out on the R/V Pelican. A total of 178full depth discrete conductivity – tempera-ture – depth (CTD) stations were employedas illustrated in Figure 1. The hydrographicdata went through a quality control proce-dure and oceanographic analysis (Vinogradovet al., 2004). The corresponding T-S diagram(Figure 2a) clearly demonstrates the existenceof four water masses. There are hot and verylow-salinity coastal waters, hot and low-sa-linity surface waters, warm and salty mid-water column (intermediate) and cool and

FIGURE 1Map of the Mississippi Bight region. White circles indicate the CTD sampling locations of the R/V Pelicanhydrographic survey during August 30 – September 14, 2000. Black dots represent the ECOM horizontal curvi-linear grid.


illustrates mixing between the shallow estua-rine and the deeper shelf waters. The split-ting of the T-S diagram here represents twodifferent mixed layers in the area. Both layersare characterized by hot temperature, ~29 °Cat a depth of about 20 – 25m (Figure 2b).However, their salinities are different. Onelayer is located closer to the coast. It has lowersalinity of about 32 ppt as a result of the closeproximity to the river inflow. There is a dif-ferent mixed layer further offshore. It is closeto the shelf break and it has a higher salinityof about 36 ppt. The water with temperature~29 °C and salinity 36 ppt is a typical watermass for the surface waters in the Gulf ofMexico, which as the observations show canbe found as far onshore as the shelf break area.

FIGURE 2(a) T-S diagram and (b) temperature and salinity vertical profiles from the CTD measurements collected asshown in Figure 1.

TABLE 1Hydrographic features of the four water massesobserved during August 30 - September 14, 2000.

Water Temperature Salinity Depthmass °C m

Coastal 30 15 – 26 < 20

Surface 30 26 – 34 < 20

Intermediate 22 – 28 36 – 37 20 – 100

Deep 15 – 20 35 > 100

Vinogradov et al. (2004) made the analy-sis of temporal and spatial variability of thehydrographic data in this region. They ex-amined temperature and salinity fields mea-sured over a period of two years. The oceano-graphic analysis revealed a high level ofspatial and temporal variability in the region,which has an important impact on the dis-tribution of the physical properties in thewater, such as heat, salt, oxygen, sound speedand others. The survey conducted duringAugust 30 – September 14, 2000 revealedthe smallest spatial and temporal variationsduring 1999 - 2000. Mean data variabilityover a two-week period was estimated as 1 °Cfor temperature and 2 ppt for salinity. It in-cludes the overall standard deviation of thedata from the mean field over the period ofthe survey. Compared to the winter months,when the variability in the data was about3 °C for temperature and 8 ppt for salinity,the variability during the summer is small(Vinogradov et al., 2004).

saline deep waters. The characteristics of thewater masses are given in Table 1. The watersin the mixed layer are well stirred, which isdemonstrated by the horizontal portion of

the T-S diagram (Figure 2a). The deep watershave the same salinity level, which corre-sponds to the vertical portion of the T-S dia-gram. The right corner of the T-S diagram


Numerical Model andExperiment Design

The ECOM is a three-dimensional, timedependent, sigma coordinate, free surfacemodel. The numerical model seeks a solu-tion of an initial-boundary value problemin a specified domain. Governing equationsare in the Eulerian frame of reference, in theflux-conservation form. The system consistsof the conservation equations for momen-tum, heat, salt and mass. Equation of statecompletes the set of seven equations for sevenquantities: temperature, salinity, density,pressure and three components of velocity.The model uses the hydrostatic andBoussinesq approximations. The small sub-grid processes, such as horizontal and verti-cal mixing of momentum and scalars, areparameterized using the turbulence closure.The vertical mixing coefficients are com-puted using the Mellor-Yamada level 2.5turbulence closure scheme (Mellor andYamada, 1982). The horizontal mixing ofthe momentum and scalars is representedby the Laplacian terms. The horizontaldiffusivity coefficients are mean-deforma-tion-rate-dependent (Smagorinsky, 1963).This implies that the eddy coefficients arerelated to the simulated flow scales, ratherthan being constant. For this implementa-tion, the background (constant) verticalmixing is 1 x 10-5 m2/s. The constant valueused in Smagorinsky’s formula for horizon-tal mixing is 1 x 10-1 (non-dimensional). Theratio of viscosity to diffusivity (Prandtl num-ber) is 1.0 both for horizontal and verticalmixing. The effect of rotation is introducedby the Coriolis parameter, which is com-puted using the beta-plane approximation.

The governing equations are formulatedin the local orthogonal curvilinear coordinatesin the horizontal and the bottom-followingsigma coordinate in the vertical. The hori-zontal curvilinear system allows one to resolvea complex geometry of the Mississippi Bightcoastline, featuring numerous bays, estuariesand bayous. The use of the sigma coordinate,which varies in proportion to depth, permitsone to resolve the bottom boundary layer. Ithas been shown that such a coordinate sys-tem suits to modeling of the shallow coastalocean better than the ordinary Cartesian co-

ordinate system (Gerder, 1993). The horizon-tal grid used in this study is shown on Figure1. The horizontal grid is non-uniform inspace, with the resolution varying from 3 kmto 100 m. The finest grid corresponds to theregions with the high gradients of water prop-erties, such as the passes between the barrierislands, ship channels and the MississippiRiver mouth. The 11 sigma levels in the ver-tical are evenly spaced. For a shallow regionsuch as Mississippi Sound, where maximumdepth is 10 m, the resolution exceeds 1 m inthe vertical. In the deepest areas of the do-main, close to the shelf break, the vertical reso-lution is about 7 – 10 m. A high-resolutionmodel grid contains 165 x 121 x 11 grid cells.

The ECOM equations are discretized onan Arakawa C-grid, and are solved explic-itly for the horizontal derivatives and im-plicitly for the vertical. The advection of saltand heat is handled using the Smolarkiewiczscheme (Smolarkiewicz and Grabowski,1990), which corrects numerical diffusionbetter than other advective schemes. Themodel incorporates a mode splitting tech-nique, solving the two-dimensional equa-tions for the fast (external) processes, andthe three-dimensional equations for theslower (internal) processes. For this study,the internal step is 60 s and the external stepis 6 s. The model is integrated from the stateof 00:00 UT, July 1, 2000 for 3 months,which covers the sampling period of August30 – September 14, 2000.

There are two types of the lateral bound-ary conditions used in this model configu-ration—coastal and open ocean boundaries(see Figure 1). At the coastal wall the nor-mal component of velocity and the normalgradients of temperature and salinity are zero.Along the open boundaries, the sea surfaceelevation, temperature and salinity fields arespecified. The open boundary conditions aretime-variable. The temporal increment isone hour for the surface elevation and twodays for the temperature and salinity fields.The elevation boundary conditions are de-rived from the Oregon State University tidalmodel (Blumberg et al., 2002). The tidalforcing at the open boundary is specifiedusing the inverse Reid and Bodine bound-ary condition, which allows longwave en-

ergy, such as tides, to enter and radiate outof the model domain. The time-varying tem-perature and salinity boundary conditionsare specified at six depths in the vertical. Thetemperature and salinity are derived usingthe Modular Ocean Data Assimilation Sys-tem, MODAS (Fox et al., 2001), providedby the Naval Oceanographic Office(NAVOCEANO).

The boundary conditions in the verticalare the conditions at the free surface and thebottom of the basin. The surface boundaryconditions are the net ocean heat flux, theevaporation-precipitation fresh water surfaceflux, and the wind stress. The surface condi-tions are computed from the meteorologi-cal parameters, such as wind speed and di-rection, air temperature and humidity,provided by the Coupled Ocean Atmo-spheric Mesoscale Prediction System,COAMPS (Hodur, 1997). The surface con-ditions are time-variable as well and have aone-hour increment. On the lower bound-ary, there is no flow normal to the bottomof the basin and the fluxes of heat and saltare zero. The bottom frictional stress is de-termined from the logarithmic law of thewall (HydroQual, 2002). The bottom fric-tion coefficient is set to 2.5 x 10-2; the bot-tom roughness is 3 x 10-3 m.

The freshwater discharge is specified atthe 26 grid cells, corresponding to the loca-tion of the Mississippi River, East and WestPearl River, Biloxi River, Wolf River, Eastand West Pascagoula River, Mobile River,and the other smaller rivers in the area. TheECOM freshwater sources, such as dischargeflow, temperature and salinity of the flow,are specified daily, using the USGS measure-ments from the river monitoring gauges.

Model Evaluation MethodsThe primary goal of this paper is to evalu-

ate the ECOM hydrodynamics by compari-son of simulated fields with ocean observa-tions. In particular, the performance of themodel on different temporal scales is of great-est interest. For a fine scale comparison, theECOM results are matched with the obser-vations in time and space. The model high-resolution horizontal grid allows matching the


station locations with high accuracy (Figure1). Simulated fields have one-hour temporalresolution. Consequently, in time, the simu-lated and observed data are matched within a30-minute interval. For example, 21:23 sam-pling time corresponds to the 21:00 modeltime, whereas 21:43 sampling time corre-sponds to 22:00 model time. Physical pro-cesses that have temporal scales smaller thana 30-minute interval are ignored in the cur-rent consideration. To match the model andthe data vertically, the observations are inter-polated onto the model depths using a cubicspline interpolation. The ECOM depths aredetermined using the computed sea surfaceelevation and the model bathymetry. It isworth noting that for certain stations therewas a difference between the model bottomtopography and the actual depth of the sta-tions. Ahsan et al. (2002) showed that themodel is extremely sensitive to the bathym-etry. Its contribution to variances in modelsalinity might be as high as 76% and as highas 88% in temperature. To exclude the possi-bility of high model errors due to the bottomtopography, the stations with large discrep-ancies between modeled and observedbathymetry are disregarded. About 10% ofmodel/data comparisons are discarded dueto bathymetric mismatches.

During this survey there were several sta-tions at which the data were collected everyhour over one day. The comparison of thetime-series measurements with the corre-sponding simulations provides the analysisof the model performance on a daily scale.The analysis of a time series is usually donein a frequency domain, using a spectralanalysis. The amplitudes, Xk, are computedusing the fast Fourier transform algorithm(Frigo and Johnson, 1998). To avoid thealiasing of the spectra, the Fourier coeffi-cients are computed for the frequencies lowerthan the Nyquist frequency,

. For the sampling interval

1 hour, the Nyquist frequency is

0.5 hr-1. The spectrum is computed as

(1)

where X*k is the complex conjugate of Xk ;

N = 26 is a number of the data points in atime series station; and k = 1. . .N/2, sincethere are only N/2 meaningful Fourier coef-ficients for a discrete time series of N datapoints (Teng, 2003).

As mentioned above, the largest tempo-ral scale of model/data comparison here is twoweeks. For a large-scale analysis, simulationsare temporally averaged over a two-week pe-riod, corresponding to the period of the sur-vey. At the same time, observations were madeonly once at each station, not during a two-week period. To obtain a mean state of theocean over two weeks having a single set ofmeasurements, a variational interpolationtechnique is designed. The basic premise ofthe variational interpolation is to determinean optimal estimation of an oceanographicfield approximating the data while exhibit-ing only small spatial variations. The varia-tional interpolation can be considered as anapplication of Gauss–Markov theorem(McIntosh, 1990). The theorem determinesthe optimal estimate of the field of interest,which is unbiased, is linear in the data, andhas the minimum variance, given prior theexpectation value and covariance of both thefield and the data. Specifically, an optimalestimation has to meet the following criteria:(1) the field is determined on a regular modelgrid rather than on a irregular observation net;(2) values of the field are consistent with theobservations in the data locations; (3) the fieldis smooth; (4) data variability is taken intoaccount; and (5) the field is dependent onthe bottom topography and the coastline ge-ometry. The last criterion is useful in the ar-eas of a large bottom gradient, such as shelfbreak and barrier islands. The cross-isobathicvariations across the shelf tend to be largercompared to the along-shelf variations. In thevicinity of the barrier islands, the circulationdiffers significantly on both sides of the is-land. The mathematical details and a numeri-cal procedure of the optimal field derivationare shown in Appendix A.

One of the advantages of the variationalinterpolation is that, in addition to the op-timal field, it allows one to compute theposterior error of the optimal estimator. Inother words, one can determine the result

reliability and see where the error of themethod is large. The limits depend on theprocess of interest. In this paper, an optimalfield estimates a biweekly mean state of theocean. Therefore, the interpolation errorincludes the data variability over a two-weekperiod. Based on that, the optimal field isconsidered reliable if its error does not ex-ceed the standard deviation of the data er-ror. The data error includes a mean variabil-ity of the measurements within the twoweeks and the measurement error. As men-tioned early, the mean variability of the datawas estimated as 1° C for the temperatureand 2 ppt for salinity during August 30 -September 14, 2000 (Vinogradov et al.,2004). The mathematical details of the er-ror derivations are given in Appendix B.

ResultsModel Performance on a FineTemporal Scale

Figure 3 compares simulated and ob-served temperature and salinity at the sur-face, mid-water column and the bottom.The two sets have a positive linear relation-ship with a high degree of linear interrela-tionship (R2 is 0.78 - 0.97). However, thereis a significant bias in model temperatureand salinity (see Table 2). The offset coeffi-cients are negative for all depths, implyingthat the simulations are fresher and colderfor ~80% - 90% of the sample points. Theslope coefficients are close to 1.0. The highervalues of the slope show that the rate ofchange in the simulations is generally highercompared to the observed rate of change.Figure 4 examines the regression residualsin order to check if the regression analysis isvalid. The residual values are determined asa difference between the linear regression fitand the ECOM values at the three depths.The regression analysis is valid when the er-rors are independent and are normally dis-tributed with the constant variances (Teng,2003). As shown in Figure 3, there is nocorrelation between the independent vari-able, i.e. observations, and the regressionresiduals. At all depths, the histograms ofthe residuals are close to the symmetric (nor-mal) distribution, slightly skewed right (posi-


tive skewness). The linear regression analy-sis is adequate here since there is no viola-tion of the general error assumption.

To examine the overall discrepanciesbetween the simulations and the data, thedistribution of the model error is computed(Figure 5). The model error is defined as thedifference between the matched simulationsand the measurements (Chu et al., 2001).The model temperature errors are close to aGaussian type. The mean temperature erroris close to zero. The values of the standarddeviation (STD) are 0.5 -1.2 °C. The slightlynegative mean values of the model tempera-ture errors show that the modelunderpredicts the temperature at all threelayers. The model salinity errors reveal a bi-modal distribution (non-Gaussian). Themean salinity error ranges from –2.6 ppt to–1.7 ppt and standard deviation varies from1.7 ppt to 3.7 ppt. The first mode of modelsalinity error is around zero. The secondmode is around –2 ppt, shifting the meanmodel salinity error to the left. The negativemodes and a high percentage (96% - 97%)of the negative errors are indications that themodel underestimated salinity.

Model Performance on aDaily Temporal Scale

The time series station, chosen for com-parison with the simulated time series, waslocated in the ship channel at the Mobile BayPass during September 4 – 5, 2000 (see mapin Figure 6). The depth at this location was9.0 m. Figure 5 a-c examines the diurnal vari-ability of the temperature and salinity at thesurface (0.5 m), mid-water column (4.5. m)and the bottom (9m). Surface waters (Fig-ure 6a) show a high degree of variability bothin the model results and the observed data.The rate of change in the simulations is fasterthan the observed rate of change, which wasalready seen from Figure 3a and Table 1 forthe surface water analysis. Bottom waters(Figure 6c) show a good model/data agree-ment, but with a negative temperature andsalinity offset. Mid-water model temperaturevariations (Figure 6b) seem to be negativelycorrelated with the data. The mid-watermodel salinity values are smaller and changeslower than the data. The same result of the

FIGURE 3Simulated vs. observed temperature and salinity (a) at the surface, (b) at the mid-water depth and (c) at the bottom.

TABLE 2Linear regression coefficients, computed for the simu-lated and observed temperature and salinity.

Water R2 Offset Slopemass Coefficient Coefficient

Salinity

Surface 0.97 -7.2 1.2

Mid-water 0.91 -11.9 1.3

Bottom 0.78 -13.8 1.9

Temperature

Surface 0.84 -4.1 1.1

Mid-water 0.86 -5.2 1.2

Bottom 0.90 -6.6 0.8

model underestimating salinity is demon-strated on Figure 3b and was mentioned ear-lier in the text.

Estimation of the surface observed andsimulated spectra is shown on Figure 6d. The

dominant power peak for the surface tem-perature is at ~0.04 hr-1, which is an inertialfrequency in the Mississippi Bight shelf (Keenand Allen, 2000). The corresponding periodis about 26 hours, which is very close to theperiod of the lunar diurnal tidal constituentO

1. The salinity spectra show a strong

subinertial signal at frequency ~0.08 hr-1,which corresponds to approximately a half-day period. The simulated spectra are veryconsistent with the observed ones. TheECOM shows the largest variability at thediurnal and semi-diurnal temporal scales,which is in agreement with the measurements.The spectra decrease with increasing fre-quency. For the temporal scales less than 6 -7 hours (frequencies higher than 0.2 hr-1), thesignal is very weak, showing a small variabil-ity both in the model and in the data.


Model Performance on a LargeTemporal Scale

The observations are variationally inter-polated according to the technique describedin Appendix A. Figures 7b and 8b show theoptimal estimations of the biweekly meansurface salinity and temperature fields, respec-tively. Note that, in order to avoid data ex-trapolation, the optimal fields are determinedonly on the part of the ECOM grid, enclosedby the data locations (see Figure 1).

The horizontal salinity pattern (Figure7b) reveals three distinctive regions. Thereis a high salinity offshore region (33 – 34ppt), a less saline coastal area (30 – 32 ppt),and the region of minimum salinity (22 –28 ppt) in the Mobile Bay. The relatively

fresh-water outflow, spreading through theMobile Bay Pass onto the shelf is clearly seenon Figure 7b. When the plume passes thebarrier island (Dauphin Island, AL), it entersthe shelf and propagates westward. The sur-face temperature (Figure 8b) is characterizedby the hot coastal waters (30 – 30.5 °C),slightly cooler offshore water (29 – 29.5 °C),and the region of the minimum tempera-ture west off Biloxi Bay (27 °C).

It is important to emphasize that the ob-tained estimations of the mean temperatureand salinity are optimal. The fields meet allfive criteria stated above (see Model Evalua-tion Methods section). The observations areinterpolated from the original data locationsonto the model fine resolution grid (condi-

tion 1). The values of the estimators are closeto the observations in the data locations (con-dition 2) (not shown here). Both tempera-ture and salinity estimators are smooth (con-dition 3) and are not spatially isotropic(condition 5). In particular, there is no corre-lation between the points on the different sidesof the islands, though the points are geo-graphically close. To examine the reliabilityof the estimators, the error analysis is per-formed (see Appendix B for the numericalprocedure). As stated earlier, the technique isconsidered to give a reliable solution, whenits error does not exceed the data error. Fig-ures 7c and 8c show the standard deviationsof the salinity and temperature interpolationerrors, respectively (hereafter, referred as esti-mator error). Examination of these graphsshows that the estimator error does not ex-ceed the prescribed data error, which is, asmentioned above, 1° C for temperature and2 ppt for salinity for this survey. Therefore,the interpolation results are reliable over thebasin shown in Figures 7 and 8.

The ECOM biweekly mean surface tem-perature and salinity are shown on Figure 7aand 8a, respectively. Both fields have zonal dis-tribution. The surface temperature graduallydecreases, whereas the surface salinity increasesoffshore. The observed freshwater plume inthe Mobile Bay Pass (Figure 7b) is clearly iden-tified in the simulations (Figure 7a). As previ-ously seen from the analysis of the surface layer(Figure 3a), the simulations change faster, com-pared to the observations. The simulated fresh-water plume is more pronounced comparedto the observed one, which implies a fasterpropagation rate in simulations. In addition,overall simulations are slightly fresher andcooler as compared to the data.

To obtain a quantitative estimation ofthe model/data consistency, the differencebetween the two is divided by the error esti-mates. The relation gives a quantitative evalu-ation of the model/data discrepancy in termsof the standard deviation of the variationalinterpolation error (estimation error). If thisratio is less than one, then these two fieldsare consistent within one standard deviationof the method. The normalized salinity andtemperature differences are presented in Fig-ure 7d and 8d. The ratio does not exceed

FIGURE 4Regression residual patterns and histograms of the temperature (right) and salinity (left) (a ) at the surface,(b) at the mid-water depth and (c) at the bottom.


one standard deviation almost over the en-tire area, which implies the simulation andthe data are consistent. The exceptions areseveral areas that are close to the coast. Spe-cifically, in the northern Mobile Bay, thesalinity ratio increases up to four standarddeviations and temperature ratio increasesup to two standard deviations. In addition,the temperature ratio grows in the west,where the minimum temperature was ob-served (Figure 8b).

DiscussionObserved temperature and salinity on

the Mississippi Bight shelf reveal spatial andtemporal variability during August 30 – Sep-tember 14, 2000. Different water types are

found near the coast and shelf break region,close to freshwater discharges and furtheroffshore. In addition, an irregular coastlinewith many bayous and island passes adds todata spatial variability altering water mix-ing. Temporal variation in data is due to di-urnal changes and shelf rapid response tometeorological forces such as wind. Thesemeasurements are used to validate one ofthe NGLI models, the ECOM. This typeof analysis is necessary to support numeri-cal studies in the Mississippi Bight and tojustify model results for operation use in thearea. Based on the region’s oceanography,three temporal scales are chosen to analyzemodel performance. These are fine (less thanan hour), diurnal, and large (two weeks)scales. As seen from observations, oceanic

processes on these scales are important inshelf circulation. In the Mississippi Bight astrong tidal signal is close to lunar diurnalconstituent, which is reflected in CTD datavariability. In addition, rotating winds in theMississippi Bight change their direction ap-proximately every seven to ten days (Keen,2002). Variations on the shelf are closelycorrelated with wind stress (Schroeder et al.,1987). Therefore, a two-week period is suf-ficient for circulation to develop in a shal-low wind-driven shelf such as the Missis-sippi Bight. It would be desirable to seelonger scales such as seasonal and inter-an-nual variations. However, only limited datawere available. Nevertheless, three temporalscales are dominant within seasonal changesin the area. Therefore, analysis on these par-ticular scales justifies model performanceduring a season in the shelf dynamics.

The ECOM skill assessment demon-strates the following strengths and weaknessesof the model. On a fine scale, about 80 –90% of the simulations are linearly relatedwith the observations. It includes shelf areaand nearshore region that are far enough fromriver inputs. The remaining 10 – 20% ofcomparisons show a non-linear relation.These points represent surface and mixedlayers, which are close to the coast with riverinflows. In these points both temperature andsalinity are underestimated by the model.There are two possible reasons for a non-lin-ear relationship between the data and thesimulation in these data points. The first rea-son is a difference between observed andsimulated mixing rates. The observed mix-ing of the estuarine and shelf waters, whichoccurs just below the surface, is captured onthe earlier stage as compared to the simu-lated one. In other words, the ECOM hori-zontal mixing is faster compared to the ob-served one. Simulated vertical mixing is alsofaster than the observed one. The faster simu-lated vertical mixing rates lead to smoothervertical gradients in the model profiles (notshown here), which result in the underesti-mated temperature and salinity. The secondpossible reason for the salinity bias is an un-certainty in specification of the river dischargeinformation. It includes the estimation of thedischarges in the un-gauged areas, the sensor

FIGURE 5Histograms of the model temperature (right) and salinity (left) errors (a) at the surface, (b) at the mid-waterdepth and (c) the bottom.


calibration and/or the specified inflow rates,which could be different from the actual rates.Therefore, on a fine temporal scale theECOM performs reasonably well in 80% ofthe data points except in the regions that areclose to the freshwater inflows and in themixed layer of the water column.

The next dominant scale in the area is thediurnal cycle. The strong observed diurnal and

semi-diurnal variability is well captured by theECOM. The simulated and observed signalsare in a good agreement, both having theirenergy peaks in the low frequencies (period of26 and 13 hours). Both observed and simu-lated signals are gradually decreasing towardthe higher frequencies (period less than 6hours). A slight non-synchronization betweenthe model and the data is likely due to differ-

ent response to meteorological conditions. Thenortheastern wind prevailed during Septem-ber 4 – 5, 2000, which is the day the timeseries data were collected. This resulted in themixing between the water in the Mobile BayPass (station location) and warm and freshcoastal waters. The corresponding event in themodel occurs with a 6 – 10 hours delay. Nev-ertheless, the general trend, observed in thisarea during a wind-induced mixing event, isreproduced by the ECOM.

The last temporal scale considered here isa two-week period. The observed integral large-scale horizontal patterns are seen in the tem-porally averaged simulations. Both qualitativeand quantitative analyses show a good model/data consistency within the estimator error.Similar to the observations, the ECOM tem-perature and salinity increase offshore. In ad-dition, the observed freshwater plume is alsofound in the simulations. When the simulatedplume comes out to the Gulf through theMobile Bay pass between the barrier islands, itis not destroyed by the horizontal mixing withthe ambient water. The simulated plume en-ters the shelf area and propagates southwestsimilar to the observed plume path. The abil-ity of the model to handle processes with highhorizontal gradients is very important in study-ing the dynamics of the fronts. However, thecold front passage that was observed duringCTD data collection is not captured by theECOM. This, as mentioned earlier, is a result ofthe delayed response of the model to the me-teorological forcing. In addition, there is an over-all negative bias in the simulated mean charac-teristics due to the reasons mentioned above.

ConclusionsA quantifiable summary of the ECOM

performance on different temporal scales isshown in Table 3. The main conclusion of thispaper is that the ECOM is found to be usefulto study coastal oceanography in the Missis-sippi Bight. It resolves general trends and dy-namical features in the area, including deter-mination of the main water masses andsimulation of their spatial distribution in thearea. In addition, areas of high horizontal gra-dients are well handled by the model, which iscrucial in modeling shelf and slope regions of

FIGURE 6Observed and simulated salinity (left) and temperature (right) as functions of time at the Mobile Bay Passstation (a) at the surface, (b) at the mid-water depth and (c) at the bottom. (d) The surface spectra computedbased on the simulations and measurements at the Mobile Bay Pass station.


the Mississippi Bight. However, the currentperformance of the ECOM is to be improvedin order to be used operationally. In particular,the model biases in hydrographic field estima-tion and the model’s delayed response to me-teorological forces are major concerns in thecurrent application of the ECOM.

Furthermore, the variational interpolationtechnique is reintroduced in this paper, whichfollows the approach described by McIntosh(1990). New application of an existing methodis proved to be useful and effective as a newtechnique of model validation. The techniqueprovides a reliable estimation of mean state ofthe ocean from observations. The methodcomputes both the optimal field and its error,so one can control the limits of the solutionacceptance. The numerical algorithm describedin this paper is applied to one CTD survey ofAugust 30 – September 14, 2000. Based onthe data, a bi-weekly mean state of the ocean isestimated with an error of the solution less thanthe prescribed error. It is worth noting that thisapproach is not limited to a particular surveyand can be applied to any data with prescribeddata variances. This technique was successfullyapplied to observations collected during fiveother NGLI surveys (Vinogradov andVinogradova, 2003). Prior to computation ofthe optimal fields, data variances for each sur-vey were estimated by Vinogradov et al. (2004).The obtained mean temperature and salinityfields could be used to study seasonal clima-tology and inter-annual variability in the area.It could also be used as reference fields in re-gional numerical models to improve forecastsystems in the Mississippi Bight.

AcknowledgementsThe authors wish to thank John Blaha

and Carl Szczechowski of NAVOCEANOfor their support and assistance in data col-lecting and preprocessing. We benefited fromvaluable comments and thoughts on the draftmanuscript by Mark Cobb (NRL) and dis-cussion with Chung-Chu Teng (NDBC). Weappreciate comments of two anonymous re-viewers, whose suggestions helped us to im-prove the manuscript. This work was fundedby CNMOC through N62306-01-D-BOO01-0002.

FIGURE 7Comparison of mean simulated and observed surface salinity. (a) biweekly mean model surface salinity; (b)biweekly mean surface salinity data, obtained using a variational interpolation method; (c) standard deviation ofthe variational interpolation error for surface salinity; and (d ) the difference between a biweekly simulated andobserved salinity fields divided by the error of the variational interpolation method.

TABLE 3Summary of the Model/Data comparison

Temporal Scale Model PerformanceStrength Weakness

Fine scales · 80% of comparisons show strong · 20% of comparisons show underestimated(less than positive linear correlation. temperature and salinity.an hour) · Surface layer and offshore areas · Mixed layer and areas that are close to

are resolved by the model in 90% freshwater discharge are not resolved byof comparisons the model in 60%-80% of comparisons.

Diurnal scale · Both inertial (0.04 hr-1) · Delayed response to meteorological and sub-inertial (0.08 hr-1) forcing (about 6 hours)signals are captured by the model

· General trend of response tometeorological forcing is similar

Large scales · Similar horizontal patterns are found · Model captures a rapid cold event with delay(two weeks) in the model and the data · Bias in the mean characteristics

· Model resolves areas of high gradients,such as freshwater plumes


Appendix A

Variational interpolation techniqueConsider the problem of estimating the field X (a mean temperature or salinity field over the observational period) from data

values D, which measure the field X with some error. The Bayesian Maximum Likelihood approach allows one to build the optimalfield X

OPT, which maximizes the conditional probability of the observed field X:

(A.1)

Maximization of the conditional probability (or minimization of the log-likelihood ratio, − log p(X D)) can be reduced to theminimization of the following cost function with respect to X:

(A.2)

FIGURE 8Comparison of a mean simulated and observed surface temperature. (a) biweekly mean model surface tem-perature; (b) biweekly mean surface temperature data, obtained using a variational interpolation method; (c)standard deviation of the variational interpolation error for surface temperature; and (d ) the difference betweenbiweekly simulated and observed temperature fields divided by the error of the variational interpolation method.


The term log p(D) in (A.2) is not included since it does not depend on X. Definition of the conditional probability density p(D X)requires the specification of the observational model: D = IX + ε, where I is the interpolation operator, which establishes the relations betweenD and X. The interpolation is required since these fields are represented on different grids. The errors in the data are introduced as ε. Underthe assumption that the errors in D are uncorrelated and the statistics is Gaussian, the probability density p(D X) (for a given X) can beexpressed in terms of the data error statistics:

(A.3)

In (A.3), WDis the inverse of the diagonal data error covariance matrix. For the Gaussian statistics of X, the cost function takes the

following form:

(A.4)

In (A.4), the second term represents − log p(X), which is a priori statistics of the observed field X with a covariance matrix Cprior

(X) W−1 andthe expectation value X

REF. Following the conventional approach (McIntosh, 1990), a priori correlations in the field X are approximated as

follows

(A.5)

In (A.5), WREF

, WSM

, WBOT

and are the diagonal weight matrices, and H is the bottom topography on the model grid. Finally, the optimalfield X

OPT is obtained as a solution of the minimization of the cost function in the following form

(A.6)

The cost function J(X)is a sum of the four terms in the right hand side of (A.6). The first term forces the algorithm to build the field that isclose to the observations in the data locations. The second and the third terms in (A.6) take into account the correlations in the observed field X.The weight matrix W

REF in the second term of (A.6) is a diagonal matrix of the inverse variances of the reference field X

REF. ∇2 is the approximation

of the Laplace operator. The correlations between the bottom topography and the observed field X are imposed by adding the fourth term into(A.6). Though the value of the forth term is small, it is useful in the areas of a large bottom gradient, such as ship channels or barrier islands.

Numerical ProcedureThe weights of the cost function (A.6) are estimated as follows

■ WREF

= ((D − XREF

)2)−1 = var(D)−1 , where var(D)−1 is a reciprocal of the data variances

■ WD = (10−2 ..... var(D))−1

■ WSM

= var(D)−1 .....d 4, where d is a characteristic distance between the data points

■ WBOT

= (var(∇Η)var(D))−1 .....d 2, where H is the ECOM bottom topography

The Conjugate Gradient (Fletcher and Reeves, 1964) descent method is used to minimize the cost function (A.6). The optimal estima-tion of the mean temperature and salinity fields are computed at the ECOM horizontal grid ( N

x = 165 × N

y grid points).

X


Appendix BVariational interpolation error

The conventional way to estimate the posterior error covariance of the optimal solution is to compute the inverse of the Hessian matrixassociated with the cost function:

(B.1)

Since the dimension of the problem is relatively large (dim(X) ∼ 106), we utilize a simplified approach to compute Cposterior

. This approachis based on the linearity of the variational interpolation procedure:

(B.2)

ND in (B.2) is a number of the data points, N

M = N

X × N

Y is a number of the grid points in the field X, L

1 and L

2 are linear operators. By

definition

(B.3)

The data covariance matrix, Χ(D) is defined as

(B.4)

Consequently,

(B.5)

where E is a unit matrix and σD is the standard deviation of the data error. As previously mentioned, σ

D was estimated as 1°C for

temperature, and 2 ppt for salinity for this survey (Vinogradov et al., 2004). The covariance matrix of the reference field is defined as

(B.6)

In (B.6), the standard deviation of the reference field error, σREF

, was estimated as 7 – 9 ° C for temperature and 7 – 10 ppt for salinity(Vinogradov et al., 2004).

To estimate the terms L1C(D)1/2 and L

2C(X

REF)1/2 in (B.3), the variational interpolation technique is applied to a special kind of data and

reference field. In order to estimate the term L1C(D)1/2, the technique is applied to the data, D* = σ

D and a zero reference field, X*

REF = 0.

To estimate the term L2C(X

REF)1/2, the technique is applied to a reference field, X* = σ

REF, and zero data, D* = 0. Since we are interested in the

error variance of the optimal field XOPT

, we discard the correlations in (B.3) and approximate var(XOPT

)as:

(B.7)


ReferencesAhsan, Q., A. F. Blumberg, H. Li, and J.

Blaha. 2002. The Calibration/validation of a

Mississippi Sound/Bight model. Proceedings of

the MTS/IEEE Oceans 2002, Conference &

Exhibition. 2:823-833.

Allen, J. S., P. A. Newberger, and J. Federiuk.

1995. Upwelling circulation on the Oregon

continental shelf. J Phys Oceanogr.

25:1843-1866.

Asper V., J. P. Blaha, C. Szczechowki, C.

Cumbeee, R. Williams, S. Lohrenz, D.

Redaljie, and A.-P. Diercks. 2001. The

Northern Gulf of Mexico Littoral Initiative

(NGLI): a collaborative modeling, monitoring,

and research effort. J Miss Acad Sci. 46(1):60.

Blumberg, A. F. and D. M. Goodrich. 1990.

Modeling of wind-Induced destratification in

Chesapeake Bay. Estuaries. 13:1236-1249.

Blumberg, A. F., Q. Ahsan, H. Li, and J.

Blaha. 2002. An operational forecast modeling

system for the Mississippi Sound/Bight.

Proceedings of the MTS/IEEE Oceans 2002,

Conference & Exhibition. 2:816-822.

Brooks, D. A. 1984. Current and hydro-

graphic variability in the northwestern Gulf of

Mexico. J Geophys Res. 89:8022-8032.

Chen, C. and R. C. Beardsley. 2002. Cross-

frontal water exchange on Georges Bank: some

results from an U.S. GLOBEC/Georges Bank

program model study. J Oceanogr. 58:403-420.

Chen, C., R.C. Beardsley, and R. Limeburner.

1995. A numerical study of stratified tidal

rectification over finite-amplitude banks. Part II:

Georges Bank, J Phys Oceanogr. 25:2111-2128.

Chu, P. C., S. H. Lu and Y. C. Chen. 2001.

Evaluation of the Princeton Ocean Model

using South China Sea Monsoon Experiment

(SCSMEX) data. J Atmos Ocean Tech.

18:1521-1539.

Fletcher, R. and C. M. Reeves. 1964. Function

minimization by conjugate gradients.

Computer Journal. 7:149-154.

Fong, D. A. and W. R. Geyer. 2001. Response

of a river plume during upwelling favorable

wind event. J Geophys Res. 106:1067-1084.

Fox, D.N., W.J. Teague and C.N. Barron.

2001. The Modular Ocean Data Assimilation

System (MODAS), J Atmos Ocean Tech.

19:240-252.

Frigo, M. and S. G. Johnson. 1998. FFTW:

An Adaptive Software Architecture for the

FFT. Proceedings of the International

Conference on Acoustics, Speech, and Signal

Processing. 3:1381-1384.

Gerdes, R. 1993. A primitive equation ocean

circulation model using a general vertical

coordinate transformation. 1. Description and

testing of the model. J Geophys Res.

98:14,683-14,701.

Hadivogel, D. B. and A. Beckmann. 1997.

Numerical modeling of the coastal ocean. In:

The Sea, K. H. Brink and A. R. Robinson,

eds., 10, 457 – 482. New York: Wiley.

Hodur, R. M. 1997. The Naval Research

Laboratory’s Coupled Ocean/Atmosphere

Mesoscale Prediction System (COAMPS).

Mon Wea Rev. 125:1414-1430.

HydroQual. 2002. A Primer for ECOMSED.

Version 1.3. 188 pp.

Keen, T. R., 2002. Waves and currents during

a winter cold front in the Mississippi Bight,

Gulf of Mexico: Implication for barrier island

erosions. J. Coastal Res., 18, 622 – 636.

Keen, T. R. and S. E. Allen. 2000. The

generation of internal waves on the continental

shelf by Hurricane Andrew. J Geophys Res.

105:26,203-26,224.

Kelly, F. J. 1991. Physical oceanography/water

mass characterization. Chapter 10 in:

Mississippi-Alabama Continental Shelf

Ecosystem Study: Data Summary and

Synthesis. Volume II, Technical Narrative. J.

M. Brooks, ed., Technical Rpt. OCS Study/

MMS 91-0063, U. S. Dept. of the Interior,

Minerals Management Service, New Orleans,

La., 10-1-10-151.

Kjerfve, B. 1983. Analysis and synthesis of

oceanic conditions in Mississippi Sound, April

Thru October 1980. Final Report to U.S.

Army Engineer District, Mobile, Alabama.

March 1983.

Klinck, J. M. 1995. Circulation near

submarine canyons: A modeling study. J

Geophys Res. 101:1211-1233.

McIntosh, P.C. 1990. Oceanic data interpola-

tion: objective analysis and splines. J Geophys

Res. 95:13, 529-13, 541.

Mellor, G. L. and T. Yamada. 1982. Development

of turbulence closure model for geophysical

fluid problems, Rev Geophys. 20:851-875.

Schroeder, W. W., S.P. Dinnel, W. J. Wisaman

Jr. and W. J. Merrell Jr. 1987. Circulation

patterns inferred from the mvement of

detached bouys in the eastern Gulf of Mexico.

Cont Shelf Res. 7:883-894.

Schroeder, W. W. and W. J. Wiseman Jr. 1999.

Geology and hydrodynamics of Gulf of Mexico

Estuaries. In: Biochemistry of Gulf of Mexico

Estuaries, T. S. Bianchi, J. R. Pennock, R. R.

Twilley, eds., 1 – 28. John Wiley & Sons, Inc.

Smagorinsky, J. S. 1963. General circulation

experiments with the primitive equations. I.

The basic experiment. Mon Weather Rev.

91:99-164.

Smolarkiewicz, P. K. and W. W. Grabowski.

1990. The multidimensional positive definite

advection transport algorithm. J Comput Phys.

86:355-375.

Teng, C. C. 2003. Oceanographic data

analysis. Notes for MAR 665, Stennis Space

Center, MS.

Vinogradov, S. and N. Vinogradova. 2003.

Northern Gulf of Mexico Temperature and

Salinity Variability and ECOM Assessment

Based on the Northern Gulf of Mexico

(NGLI) Conductivity-Temperature-Depth

(CTD) Data. Dept. of Mar. Sci. Final Report

2003-11., the Univ. of Southern Miss, Stennis

Space Center, MS

Vinogradov, S., N. Vinogradova, V.

Kamenkovich, and D. Nechaev. 2004.

Temperature and salinity variability in the

Mississippi Bight. MTS Journal, 38(1):52-60.


A U T H O RKent DaveyUniversity of Texas

P A P E R

Ship Component in Hull OptimizationA B S T R A C T

This document outlines an optimization to define the size of the components in thepower train of an electric ship, specifically one appropriate for an 80 MW Destroyer. Theobjective is to minimize the volume of the system, including the fuel. The size, numberand speed of the gas turbines, the electric generators, and the power electronics areconsidered as unknowns in the analysis. At the heart of the procedure is the powermission profile. The gas turbine is by far the most important component in terms ofinfluence on system volume. Integral to its selection is the specific fuel consumption as afunction of power and turbine size. The proposed procedure outlines a nested optimiza-tion to define both the best spread of turbines as well as the proper scheduling with loaddemand. Including fuel in the system volume is the key to meaningful component identi-fication. The optimized design has a system volume 603.5 m3 smaller than the base con-figuration, assuming both systems employ load scheduling among turbines. An optimizeddesign can save as much as 600 m3.

As will be shown shortly, the answer to allthree questions is “No”. What follows are thereasons the answer is no, and a methodologyfor selecting the optimal power train. Jiang etal. (2002) recognize the inefficiency problemspresented by gas turbines, and attempt to off-set the difficulty by supplementing their usewith diesel generators. This paper shows howto offset the difficulty with load scheduling andproper selection of the gas turbines a priori.

BackgroundMattick (1996) and Bishop (1997) have

documented the mission profiles for the vari-ous naval vessels depicted in Figure 2. Theaverage speed for a ship on general patrol is17.17 knots.

Future destroyers are estimated to requirea 72 MW peak propulsion power demandat 30 knots, and a hotel load of 8 MW.

The DDG-51, Arleigh-Burke Destroyerhas an average daily fuel consumption of98,883 kg, and an average mission profileduration of 14 days. A reasonable fuel con-sumption index for gas turbines is 202 kg/MW-Hr. The average power is therefore

The objective of this exercise is to choosethe power train configuration which mini-mizes the combined volume of fuel plus com-ponents for a 14 day mission subject to theconstraint that it have the capability of deliv-ering 80 MW peak. Marine diesel fuel has amass density ñ of 876 kg/m3. A 14 day mis-sion profile therefore requires approximately

FIGURE 1Ship component identification problem.

(1)

(2)

Poweravg

= = 20.4 MW98883

24 . 202

Volfuel

= = 1,580m314 . 98883876

TI N T R O D U C T I O N he objective of this exercise is to identify the components in the power trainof an all electric ship supplying ship powerat 60 Hz. This paper describes a method forquantifying the individual components inFigure 1. An 80 MW destroyer on a four-teen day mission is the focus. Among thequestions of interest are the following:1. Turbine size generally decreases with

speed. Are higher speed turbines favorablefor Naval ships?

2. Should gearboxes be employed to run theelectrical generators at higher speeds?The product of speed and volume areroughly constant for electrical generators.

3. Should both the turbine and the generatorbe driven at a common speed greater than3600 RPM, and a converter used toadjust the voltage and frequency?


Weight and volume are certainly linked.Ship designers must consider weight distri-bution of engine and electrical componentsand weapons to maintain ship stability. Inthis context the author recognizes the im-portance of weight as a separate issue fromvolume. For this paper, volume will be thefocus of the optimization. The techniqueoffered applies equally well to weight.

A typical power velocity profile for a4,000 ton navy frigate is shown in Figure 3.

The reader is cautioned not to make acommon mistake regarding the propellerpower. Force due to drag for a short bodytraveling in a medium with density ρ withcross-sectional area A, at velocity v is

CD is the drag coefficient based on the

hull shape and roughness. Power is the prod-uct of force and velocity. Multiplying by vdoes not account for the inefficiency of thepropeller itself. A percentage of the powerto the propeller goes to imparting radial ve-locity to the water, a component which isnot effective for imparting momentum. Italso does not account for the wetted area.

Longer ships have more viscous drag overthe percentage of wetted area. These twoadded complications are included in themeasured profile displayed in Figure 3,which is approximately proportional to ve-locity to the fourth power.

Combining the propeller power versusvelocity relationship with the published gen-eral patrol mission profile for an 80 MW de-stroyer, 72 MW of that being propulsion,yields the mission profile shown in Figure 4.The sum of the 29 ordinates is of course 100%.

Power PartitioningTo obtain a general idea of the size of

individual components, consider an 80MVA power train with four 20 MVA gasturbines. The fuel volume has been estimatedin equation (2). Estimates of the air ductingrequired for gas turbines indicate that 150m3 is a conservative estimate for the volumeof both air inlet and exhaust ducting if theturbines are placed on lower decks. The four20 MW gas turbines with skids will occupyapproximately 90 m3, based on the docu-mented specifications of four GE LM-2500’s. Experience at the Center forElectromechanics has shown that largepower converters require a little less than onecubic meter of volume per MW. The powerelectronics for a 22.5 MVA system is esti-mated to occupy 18.9 m3 and scale aspower0.7, yielding an anticipated volume of46 m3 for an 80 MVA system. A 3600 RPM20 MW generator will have a volume of

FIGURE 2Mission profile with speed.

FIGURE 3Typical power velocity profile for a navy frigate

(3)


approximately 2.1 m3, yielding a housing of8.4 m3 for an 80 MVA system. Figure 5 de-picts the relative volumes of the componentsgraphically.

Before discussing the optimization, anumber of recommendations can be drawnassuming from equation (2) that the mis-sion will consume 1600 m3 of fuel.1. Eliminate air ducting by placing the

turbines on the upper deck when allowedby weight distribution considerations.

2. Choose the turbine set to minimize thecombined fuel plus turbine volume.

3. Operate both the gas turbine andgenerator direct coupled at 3600 RPM.

There are five reasons for this recommendation:a. A 1 % change in generator efficiency

translates to 16 m3 of fuel for the basesystem, twice the volume of all thegenerators.

b. Eddy current losses grow as electricalfrequency squared, and hysteresis lossas frequency(1.7).

c. It eliminates both the volume of powerconverter electronics and the penalty

incurred from their use by way ofefficiency. The power converter has avolume of 46 m3 and a 97% efficiency,its use will incur a loss of another 48 m3

in fuel.d. Turbine bearing and windage loss grow

as speed squared. A survey of various

turbines shows that the loss in efficiencydoes not compensate the slightly smallerturbine volume. A 2% loss in efficiencyis commensurate with a 32 m3 loss in fuelvolume.

e. A large class of efficient turbines isavailable at this speed.

4. Avoid the use of gear boxes to ratio speeddifferences between the turbine and thegenerator. The expected minimal loss of3-4 % is equivalent to 48-64 m3 of fuelvolume.

5. Make the following changes to thegenerator:

a. Eliminate the exciter; the additional I2Rloss on the field winding is too high aprice to pay for the added field control.

b. Eliminate or replace the magnet shieldwith a thinner titanium bandi. The copper shield increases the air gap.ii. Shield losses are proportional toconductivity. Allowing the losses todissipate in the magnet is a better tradefor fuel volume.iii. The probability of a magnet lossduring short circuit is minor due tothe synchronous reactance.

c. Do not consider using any more than twopoles. A two pole generator minimizesthe ac/dc loss. The relatively small increasein stator back iron volume to carry theflux is worth the trade in efficiency.

d. Use small stator sub-conductors with fullslot transposition.

FIGURE 4Destroyer power mission profile.

FIGURE 5Volume distribution by component


The recommendations regarding the gen-erator are based on the objective of increas-ing efficiency. The volume of the generator isproportional to speed. The improvement ingenerator volume with increased speed is ac-companied by a decreased efficiency. Thedecreased efficiency has a much greater im-pact on the total fuel volume as compared tothe rather insignificant size of the generator.

Gas Turbine CharacterizationLarger turbines are more efficient than

smaller turbines if they are run at peak loadratings (Jiang et al., 2002). However, the largerturbines lose considerable efficiency at par-tial power settings. Operation of a large tur-bine at a percentage of its peak load ratingguarantees greater fuel expenditure over thatrequired with smaller turbines. This intuitivefact follows from the physics of bearing, wind-age, and thermal loss of the larger turbines.

It is necessary to formulate relationshipsto quantify the partial power performance ofany turbine. Let ξ

0 represent the specific fuel

consumption at the lowest power setting, andξ

2 represent the final specific fuel consump-

tion at rated power. The specific fuel con-sumption is closely represented by the equation

The exponential parameter m must bedetermined by minimizing the index

A task that can be accomplished by solving

.

This nonlinear least squares problem isdetermined numerically, and the results for7 different turbines are displayed in Table 1.The information on various turbines wascompiled primarily from the Gas TurbineHandbook, the Rolls Royce Marine Turbinewebsite, and Rocha (1998).

TABLE 1Specific fuel consumption results

Power(MW) î0 î 2 m

4.83 0.2821 0.2551 1.2455

5 0.344 0.29 1.5837

7.1 0.3234 0.2494 1.7871

13.5 0.4198 0.2268 2.5848

19.5 0.4 0.225 3.9354

20.142 0.5232 0.231 4.8428

50 0.5741 0.2 8.7273

Figure 6 shows the trend expected, i.e.that specific fuel consumption degradesrapidly in large turbines at small power.

Each of the factors in Table 1 can befitted with a bivariate cubic spline (deBoor, 1978) to the maximum power rat-ing of a turbine. Specifically x

0, x

2, and

m are determined as functions of themaximum power rating of the turbine.Figure 7 shows the comparison of thepredicted and measured specific fuelconsumption for seven different tur-bines using this approach.

The volume distribution used includesthe mounting skid and is tabulated inTable 2 and does not include exhaust andin-take ducting volume.

TABLE 2Turbine volume – power dependence

Power (MW) Volume (m3)

5 4.83

14.9 13.77

20 23.4

27 37.1

50 65

Gas Turbine OptimizationAssuming the mission load dictated in

Figure 4, there are three issues that must beidentified:1. How many turbine–generator sets should

be employed?2. What are the ratings of the generators?3. What is the appropriate partial power

distribution among the generators forloads less than 80 MW?

FIGURE 6Exponential dependence on turbine power in setting sfc characterization.

(4)

(5)

(6)


Redundancy requirements demand that that there be more than one turbine generator.The Navy’s directive to reduce man hours in maintenance argues for fewer machines. Thissecondary factor introduces issues outside the scope of this paper. A compromise of four tur-bines will be assumed in this analysis.

The turbine sizes displayed in Figure 7 range between 4.8 and 50 MW. Let x be a randomvector of length three such that 0<x(k)<1. Maintain the requirement that the four turbineshave a combined 80 MW capacity. The vector x(3), where 0<x<1, fully describes a fourturbine set. If the turbines range between P

min=5 MW and P

max=50 MW, with P

system=80

MW, the rating R of each turbine is

One objective of the ensuing optimization is to determine the best vector x to minimizethe combined fuel plus turbine volume.

Yet another optimization must be performed within the one characterizing turbine rat-ings. This inner optimization determines the appropriate load sharing among turbines, atask referred to as scheduling. At each of the partial power demands P in Figure 4, considerassigning a vector y with length four to represent the fractional use of each turbine, a fractionmultiplying the rating R of each turbine,

If ρ is the density of the fuel, and t the time of the mission

The 29 values for η are taken from Figure 4.Each of the quantities m, x

0, and x

2 are com-

puted using the bivariate spline developedfrom Figure 7. The logic of the code is there-fore a nested optimization:

Optimize the vector x(3), where 0<x<1, tominimize combined fuel and turbine volume

Optimize y(4) for every powerdemand over the complete missionEnd;

Sum the fuel volume for each power demandEnd;

Both optimizations are performed usinga trust region variable metric method. Spe-cifically, the SQP method closely approxi-mates Newton’s method for unconstrainedoptimization (Powell, 1983 and Schittowski,1985). At each iteration, an approximationof the Hessian is made using a quasi-New-ton updating method (Fletcher, 1963), andthe problem is approximated locally as aquadratic function.

Load SchedulingThe importance of the inner load sharing

algorithm cannot be underestimated. Perhapsthe easiest way to think through the problem isto compare two simpler cases each with four20 MW turbines. In the first case, consider run-ning these four turbines balanced, all activatedand sharing the load equally. In the second caseconsider sharing the load by minimizing totalspecific fuel consumption. The balanced casehas a mission fuel consumption of 2,792 m3,while the optimized case consumes only 1,852m3, a difference of 940 m3! Table 3 shows howthe load should be shared among these fourequal turbines as a function of demand power.As might be expected intuitively, the optimiza-tion tries to run the turbines as far out in themore efficient regime as possible. That meansloading up one turbine completely before us-ing the second. When the demand exceeds therating of one turbine, the algorithm attemptsto form an equal share among two turbines.Note the continued trend and the equal distri-bution among four turbines as soon as the de-mand exceeds the rating of three turbines. Thesystem power demand is displayed in the 5th

FIGURE 7Predicted versus measured specific fuel consumptions using equation.

(7)

(8)


column. The 6th and 7th columns show the spe-cific fuel consumption for these two scenarios.A graphical representation of the load schedul-ing is depicted in Figure 8.

The problem becomes more challeng-ing when all the generators are different.Smaller turbines offer more flexibility inmatching the loads.

TABLE 3Optimized load sharing for a four 20 MW turbine system.

% Turbine % Turbine % Turbine %Turbine Power Demand Burn Rate Burn Rate1 2 3 4 (MW) (kg/kW-Hr) balanced (kg/kW-Hr)

40.0 0.0 0.0 0.0 8.0 2.36 3.94

40.9 0.0 0.0 0.0 8.2 2.39 4.02

41.8 0.0 0.0 0.0 8.4 2.41 4.12

42.8 0.0 0.0 0.0 8.6 2.44 4.21

43.5 0.0 0.0 0.0 8.7 2.46 4.28

43.9 0.0 0.0 0.0 8.8 2.48 4.32

44.4 0.0 0.0 0.0 8.9 2.49 4.37

45.5 0.0 0.0 0.0 9.1 2.52 4.47

46.6 0.0 0.0 0.0 9.3 2.56 4.59

47.9 0.0 0.0 0.0 9.6 2.60 4.72

51.1 0.0 0.0 0.0 10.2 2.70 5.03

57.3 0.0 0.0 0.0 11.5 2.90 5.64

64.4 0.0 0.0 0.0 12.9 3.15 6.14

70.0 0.0 0.0 0.0 14.0 3.36 6.42

75.5 0.0 0.0 0.0 15.1 3.57 6.67

82.8 0.0 0.0 0.0 16.6 3.86 6.98

91.0 0.0 0.0 0.0 18.2 4.21 7.30

99.6 0.0 0.0 0.0 19.9 4.57 7.60

100.0 8.8 0.0 0.0 21.8 5.45 7.91

59.8 59.8 0.0 0.0 23.9 5.98 8.25

66.6 66.6 0.0 0.0 26.6 6.47 8.66

75.4 75.4 0.0 0.0 30.2 7.14 9.17

86.5 86.5 0.0 0.0 34.6 8.03 9.82

100.0 99.9 0.0 0.0 40.0 9.16 10.65

77.0 77.1 77.0 0.0 46.2 10.90 11.68

88.8 0.0 88.8 88.7 53.3 12.33 12.93

76.4 76.5 76.5 76.5 61.2 14.44 14.44

87.6 87.6 87.6 87.6 70.0 16.24 16.24

100.0 100.0 100.0 100.0 80.0 18.35 18.35

TABLE 4Optimized Turbine configuration

Combinedfuel and

Turbine 1 Turbine 2 Turbine 3 Turbine 4 Fuel Volume turbine(MW) (MW) (MW) (MW) (m3) volume (m3)

Base 20 20 20 20 1851.6 1945.2

Optimized 9.24 9.78 10.99 50 1248.1 1334.2


FIGURE 9Optimized power train for an 80 MW Destroyer.

FIGURE 8Load scheduling among four equal turbines.

Optimization ResultsThe nested optimization determines the

optimized turbine shown in Table 4. Thisconfiguration saves 603.5 m3 fuel over thefour 20 MW base configuration, each em-ploying proper load scheduling.

Generator DesignThe directive for the generator is to maxi-

mize efficiency, essentially to the exclusionof everything else. Based on equation (2),and a DDX system using approximately1600 m3 of fuel, improving the efficiencyby 1% saves 16 m3 on fuel. This objective iscompatible with a synchronous generatorcommensurate with the following:1. No exciter2. No magnet shield3. 60 Hz output4. Two poles5. Laminations on the rotor (4130 Steel)6. Very thin fully transposed stator conductors

to minimize ac/dc loss.The governing rule is that nothing

should be considered that compromises gen-erator efficiency. The equivalent of two gen-erator sets is gained for every 1% gain ingenerator efficiency.

ConclusionsComponent selection on an electric ship

must be performed on a system level. If vol-ume is the figure of merit, the size of the fueltank will dwarf other considerations and makeit clear why high speed drive trains, gear boxes,and power electronics must be eliminated.Turbines should be located on the upper decksto eliminate air duct volume. Among the spe-cial considerations given to the generator arethat it should be two pole, 3600 RPM, per-manent magnet, without shield, and withfully transposed, thin wall, stator conductors.The turbines must be carefully selected com-mensurate with the expected mission profileto minimize fuel and turbine volume. Loadscheduling should be employed during amission to properly allocate sharing amongturbines. Figure 9 shows the optimized powertrain for an 80 MW Destroyer. The most sig-nificant message for the Navy is that they have


an alternative allowing for fuel tanks with 600m3 less volume.

The conclusions are based on net vol-ume optimization. Reliability and mainte-nance considerations provide a counter per-spective. If the larger turbine fails, only37.5% of the total required power is avail-able from the remaining three turbines. Re-liability considerations would suggest thatthe analysis be repeated with an additionalconstraint, e.g., that the system maintain75% of full capability with the loss of anyone turbine. Maintenance considerationsadd yet another constraint, e.g., that tool-ing and spare parts be required for no morethan three different turbines. Both requireadditional analysis with added constraints.

AcknowledgmentsThe author wishes to acknowledge the

helpful support of Mark Pichot and JoeyZierer, both at the Center forElectromechanics, in providing much use-ful data on multiple turbines.

ReferencesBishop, G.N., Shelley, N.A. and Edmonds,

M.J.S. 1997. Electric propulsion for surface

fighting ships. Eleventh Ship Control Systems

Symposium, South Hampton, UK.

DDG-51 Air Duct Drawings, Northrop

Grumman Ship Systems Inc. Ingalls Operations

Division, Cage Code 34293, Drwg. No. D-

BM-AIOO-26, (drawings were obtained over

Internet from Web site that is no longer active).

de Boor, C. 1978. A Practical Guide to

Splines. Springer-Verlag.

Fletcher, R. and Powell, M. J. D. 1963. A

rapidly convergent descent method for

minimization. Computer Journal, (6):163-168.

GAO/NSIAD–98-1. 1998. Underwater

Replenishment Report.

Gas Turbine World Handbook, 2003 ed.

Pequot Publishing, Southport CT, 2003 Edition.

Jiang, C., Forstell, B., Lavis, D., and Ritter, O.

2002. Ship Hull and Machinery Optimization

using Physics Based Design Software, Marine

Tech. 39(2):109-117.

Mattick, D.R. 1996. The electric warship.

INEC 96 3rd International Naval Engineering

Conference, Paper 5.

Powell, M. J. D. 1983, Variable Metric

Methods for constrained optimization. In:

Mathematical Programming : The State of the

Art, eds. A. Bachem, M. Grotschel and B.

Korte. Springer Verlag, pp. 288-311.

Rocha, G., Etheridge, C. J. 1998. Evolution of

the Solar Turbines Titan 130 Industrial Gas

Turbine. Am. Society of Mechanical Engineers,

paper 98-GT-590.

Rolls Royce Marine Gas Turbine Web site,

http://www.rolls-royce.com/marine/product/

gasturbines/default.jsp.

Schittowski, K. 1985. NLQPL: A Fortran-

Subroutine Solving Constrained Nonlinear

Programming Problems, Annals of Operations

Research, (5):485-500.


A U T H O RLinwood H. PendletonProgram in Environmental Scienceand EngineeringDepartment of EnvironmentalHealth SciencesUniversity of California, Los Angeles

P A P E R

Understanding the Potential Economic Impactsof Sinking Ships for SCUBA Recreation

A B S T R A C TShips, planes, and other large structures are finding their way to the bottom of the

sea along coasts in North America, Europe, Australia, and elsewhere. More and more,coastal communities and even not-for-profit organizations (e.g. the San Diego OceansFoundation and Artificial Reef Society of British Columbia) are actively promoting andfinancing “ships to reefs” projects as a means of providing new destinations for recre-ational SCUBA diving tourists.

Creating a “ships to reef” site can be costly. The cost to prepare a ship for reefingcan range from $46,000 to $2 million, depending on the size of the vessel (Hess et al.,2001). The benefits, however, can be equally large or larger. In order to get a better ideaof the potential economic value of ships to reefs, I review the literature on the value ofrecreational diving to artificial reefs in the United States. Using data from the literature,I estimate that potential net present value of expenditures associated with the recentlyplaced Yukon ship to reef site in Southern California could be on the order of $46 mil-lion and the potential net present non-market value of the sunken ship could be as highas $13 million. These estimates are within an order of magnitude of estimates based ona preliminary survey of divers at the Yukon.

The scale and pace of sinking ships tocreate artificial reefs, especially reefs designedfor recreational diving, is increasing rapidly.In Florida, over 380 vessels have been sunkto create artificial reefs. In 2004 the U.S.S.Spiegel, a 510-foot naval vessel, was sunk inthe Florida Keys National Marine Sanctu-ary. To date, over 700 vessels serve as artifi-cial reefs in the waters off the continentalU.S. coastline. The majority of these shipsare found off the coast of Florida (380), NewJersey (129), South Carolina (100), and NewYork (65) (http://njscuba.net/reefs/index.html accessed 9.18.2004). Other stateslag far behind in the creation of artificial reefstructures. For instance, while steps havebeen made to increase the use of artificialreefs in California, the state has only ten shipscurrently in place as artificial reefs intendedfor recreational diving.

While the attention paid to artificial reefdevelopment has increased dramatically inthe past decade, artificial reefs are not amodern development. Two thousand yearsago, the Greek geographer Strabo recorded

that the Persian Kingdoms built reefs acrossthe Tigris River (Hess et al., 2001). In theUnited States, artificial reefs have beenaround for over 150 years; as long ago as1830 log huts were sunk off the coast ofSouth Carolina to improve fishing (Hess etal., 2001). What differentiates modern arti-ficial reefs from past reef making is the scaleand cost of artificial reefs and the potentialeconomic benefits that could be producedby the strategic placement and marketingof artificial reefs.

Creating a ship to reef site can be costly.The cost to prepare a ship for reefing canrange from $46,000 to $2 million, depend-ing on the size of the vessel (Hess et al.,2001). These costs represent direct outlaysby cities, counties, states or not-for-profitorganizations and are considered an invest-ment that is expected to produce economicreturns. For governments, especially localgovernments, ships to reefs are intended asnew recreational revenue sources that willstimulate tourism, increase local expendi-tures, and support new tax revenues. For

S I N T R O D U C T I O N hips, planes, and other large structures are finding their way to the bottom ofthe sea along coasts in North America, Eu-rope, Australia and elsewhere. While manypurists see the scuttling of ships and planesin coastal waters as something akin to dump-ing, more and more coastal communities areturning to these structures as a means of pro-tecting shoreline, creating habitat for fish andsea life, and providing new destinations forrecreational fishing and SCUBA diving tour-ists (Baine, 2001). In many cases (e.g. thesinking of the Yukon off the San Diegocoast), the goal of “ships to reefs” is exclu-sively to create new destinations for non-consumptive SCUBA diving; in some casesships to reefs are even designated as “no fish-ing zones”. Despite the success of these re-gional organizations in raising funds to sup-port ships to reefs, it is never certain thatany one ship to reef site represents a goodeconomic investment—especially when theintended goal of a “ship to reef” is limitedto SCUBA recreation alone.


diving groups and not-for-profits, the returnsfrom these investments need not be so obvi-ous. Acting on behalf of the diving public,these groups seek to create new recreationalresources that increase the quantity and qual-ity of dive opportunities for their members.The benefits these users derive may not beapparent in the market and include the valueof these ship-based dive experiences beyondwhat divers spend in the market. These lattervalues, known as non-market values, can besignificant, but often are difficult to measure.

To better understand the potential eco-nomic benefits of artificial reefs generally, anumber of studies have been undertaken toestimate both the market and non-marketvalues of artificial reefs. In the late 1970s,studies began to quantify the economic ben-efit of recreational fishing and diving on ar-tificial reefs (Daniel, 1976). Over time, theaccuracy and comprehensiveness of thesestudies have grown to provide a more com-plete picture of the potential economic ben-efit of artificial reefs.

Ships to reefs potentially could supporta number of diverse uses and values includ-ing shoreline protection and fishery enhance-ment or concentration. Nevertheless, shipsto reefs projects are increasingly undertakenwith the primary purpose of supporting rec-reational, non-consumptive diving. I focuson this limited use of ships to reefs as a con-servative estimate of their value. (Further, Ilimit the scope of the research to diving inorder to keep the demonstration of valuesstraightforward and clear.) Recreational div-ing is a rapidly growing industry and increas-ingly artificial reefs are being prepared, sunk,and maintained for the express use of recre-ational diving. Leeworthy et al. (2005) esti-mate that 2.86 million people over the ageof sixteen years participated in SCUBA div-ing activities in 2000. Of nineteen settings/activities for which participation was esti-mated, SCUBA diving was estimated to bethe fastest growing recreational activity inthe United States.

In the paper that follows, I review theliterature to develop a better understandingof the potential economic value of ships toreefs for recreational SCUBA diving. Increas-ingly, studies from both the peer reviewed

and gray literature are easily available on theWeb. The National Ocean Economics Pro-gram provides a literature portal that includesa searchable bibliographic database of ma-rine non-market valuation studies(www.oceaneconomics.org). Similarly, theNational Oceanic and Atmospheric Admin-istration (NOAA) has a Web site that con-tains many technical reports on the eco-nomic valuation of marine resources(www.marineeconomics.noaa.gov). Econo-mists can use data from these studies andon-site data regarding environmental andsocio-economic conditions to estimate theeconomic value of marine resources that havenot yet been valued rigorously. Less formally,these studies can be used by non-economiststo better understand the potential range ofvalues that may be associated with a marineresource or policy.

In this paper, I assess the state of the artin the quantification of the recreational val-ues of artificial reefs that may provide recre-ational experiences that are similar to shipsto reefs. I provide an overview of the litera-ture, describe the kinds of estimated valuesin the literature, and provide a non-techni-cal demonstration of how these values couldbe used to gain a better knowledge of thepotential economic returns of sinking shipsfor SCUBA tourism.

II. The Economic Value ofArtificial Reefs for SCUBARecreation

Artificial reefs yield economic benefitsthrough the enhancement of shoreline pro-tection, fishery resources, and recreationalfishing and diving opportunities. The val-ues of these benefits are difficult to quantifybecause they involve both market and non-market values. The market impact of a reefresource usually is assessed by examininghow much money artificial reef users con-tribute to the local economy by spendingmoney to participate in activities on the reef(such as recreational fishing and diving).Commonly, the focus of market-based stud-ies is on gross expenditures with fewer stud-ies focusing on profits or taxes. While gross

expenditures do not represent net benefitsto the economy, gross expenditures do cap-ture the magnitude of importance that arti-ficial reefs may have in the overall localeconomy. Further, gross expenditures rep-resent the base upon which tax revenues canbe generated. The promise of increased taxrevenues may lead local, state, and even fed-eral agencies (e.g. National Marine Sanctu-aries) to approve the sinking of ships andthe creation of other artificial reef structures.The lure of increased expenditures on divecharters and hotel stays can encourage localbusinesses to support such endeavors.

The non-market value of recreationaldiving is more difficult to determine. Non-market values represent the value reef usersplace on a reef, beyond what they have topay to use the reef. Non-market values areoften associated with outdoor recreationalresources, including dive sites, and have beenshown to generate substantial economicvalue beyond the expenditures generated bythese resources (see Cesar, 2000 andPendleton, 1995). These non-market valuesrepresent a true net economic value of reefsto divers; these values capture the increasein economic well-being that divers enjoy asa result of access to reefs. At a minimum,funds raised directly from divers to supportthe creation of artificial reefs reflect a lowerbound for these non-market values. Thesefunds are only a lower bound, however, be-cause most artificial reefs, including ships toreefs, are open access public resources; manyreef users will be able to “free ride” on thecreation of ships to reefs.

In the literature, two primary methodsare used to estimate the non-market valueof artificial reefs. Travel cost methods are usedto estimate a demand curve for recreationaldiving to artificial reefs by modeling the in-fluence of travel cost and travel time on thefrequency of visitation by divers. Travel costmethods use real diver behavior to estimatethe consumer surplus of recreational diving(the value divers place on a reef visit beyondwhat they have to pay), but the method canonly estimate the value of current uses bynon-resident divers. When travel cost meth-ods are inappropriate, authors have usedcontingent methods to estimate values for


artificial reef maintenance or abundance.Specifically, several authors use contingentvaluation methods to ask divers to place avalue on their current recreational use of a)existing artificial reefs and/or b) proposednew artificial reefs.

Below I summarize studies that provideestimates of both market values (expendi-tures) and non-market values associated withrecreational uses of artificial reefs. Most ofthe comparable studies focus on sunkenships or oil rigs. It is important for the readerto note that the methods for finding thesemarket and non-market values often differbetween studies. In the following I providethese estimates (all converted to US$ in2004, all figures greater than $10 arerounded to the nearest dollar) with briefexplanations of the basic methods. Further,when possible, I break down the value esti-mates based on the value per visitor per day.By doing so, I hope the reader will be ableto better compare these results across stud-ies and also understand how these values maycompare to the values that would be gener-ated by future artificial “ships to reefs” valu-ation studies.

The Market Value ofRecreational Diving AtArtificial Reefs

Gross expenditures by divers generate netrevenues for local firms and businesses andalso have substantial secondary impacts.Expenditures by divers support jobs andwages for dive charter captains and crews,employees at local hotels and eateries, andnumerous other ancillary services.

While much of the literature focuses onthe economic value of recreational anglingand diving combined, many of these stud-ies also provide data on the independentvalue of artificial reefs for recreational div-ing. (Of course, many recreational diversmay also spearfish. We do not attempt todifferentiate between non-consumptive andconsumptive recreational diving.) Two stud-ies estimate the expenditures associated withrecreational SCUBA diving at oil rigs. Hiett& Milon (2002) surveyed divers that wentdiving or fishing within 300 feet of offshore

oil and gas structures in the Gulf of Mexico;the authors calculate that the average perperson-day expenditures at artificial reefs inAlabama, Mississippi, and Louisiana was$119, and total annual spending for the threestates combined was over $7.4 million. Fol-lowing a similar approach, McGinnis et al.(2001) calculate the average per person-dayexpenditures of divers visiting decommis-sioned oil rigs in California to be $64, witha total annual spending of $10,700 for allrig diving in the state.

Expenditures by divers visiting artificialreefs are similar to divers visiting oil rigs (seeTable 1). Hess et al. (2001) provide grossrevenue estimates for a variety of artificial reefsites made from sunken ships. The authorsfind that these reef sites, located around theworld, generate an average of $3.4 millionannually per reef site. Ditton et al. (2001)and Ditton and Baker (1999) find that non-resident divers who visited an artificial reefon at least one dive trip each year spent justover $193 per person-day on their last trip toa dive site in coastal Texas waters; residentsspent over $184 per person-day. Brock (1994)surveyed a dive-tour operator in Hawaii whoconducted trips exclusively on a surplus yardoiler and calculated the total gross annual in-come generated by these trips to be $494,840.Bell et al. (1998) also provide a breakdownof expenditures per person-day for divers vis-

iting artificial reefs in Northwest Florida. Theauthors find that divers spend $50 to approxi-mately $90 per person-day (for residents andnon-residents respectively), a value that lieswithin the range of the other studies; together,resident and non-resident divers visiting arti-ficial reefs spend more than $14 million an-nually in Northwest Florida. Johns et al. findeven higher levels of expenditures by SCUBAdivers and snorkelers visiting artificial reefs inSoutheast Florida. The authors estimate theper person-day expenditures of $61 to $204for residents and non-residents respectively.

The Non-Market Value ofRecreational Diving AtArtificial Reefs

Artificial reefs, including sunken ships,can generate substantial non-market valuesfor recreational divers (Table 2). Roberts etal. (1985) use contingent valuation meth-ods to estimate the mean annual per divernon-market value of oil rig diving in the Gulfof Mexico to be $339, with a total annualvalue ranging from $905,216 to $1,264,640for all sites. Other studies provide estimatesof per person-day non-market values. Bellet al. (1998) use both travel cost and con-tingent valuation methods (specificallyTurnbull and Dichotomous Choice analy-ses) to estimate per person-day non-market

TABLE 1Market Value (Expenditure) Estimates for Diving at Artificial Reefs

Author Location Habitat Type Market ValuePer Person-Day($2004, figuresare rounded)

Hiett & Milon (2002) Gulf of Mexico Oil and Gas Structures $119

McGinnis et al. (2001) Southern California Platform Grace (Oil Rig) $64

Ditton and Baker (1999) Texas Various types of artificial reefs $185 for resident$194 for non-residents

Ditton et al. (2001)

Bell et al. (1998) North West Florida Ships, reef balls, and other $50 for residentsprivate and public artificial reefs $90 for visitors

Johns et al. (2003) South East Florida Ships, reef balls, and other $61 for residentsprivate and public artificial reefs $204 for visitors

Wilhelmsson et al. (1998) Eliat, Israel Navy Ship $28

Brock (1994) Waikiki Surplus yard oiler $26-$60


values. The authors find that the value forvisitors may be as high as $11/ person-dayfor non-residents and $4.30/ person-day forresidents. Ditton and Baker (1999) estimatethe non-market value of diving in Texaswaters, for divers that visited at least one ar-tificial reef in the past year, to be between$45 and $75 per person-day for non-resi-dents. The values estimated by Ditton andBaker, however, are not exclusively for arti-ficial reef divers.

Johns et al. estimate the value of main-taining artificial reefs and creating new arti-ficial reefs; the authors conclude that thenon-market use value per person-day formaintaining existing artificial reefs ($3.41-$14 for residents and non-residents respec-tively) was generally higher than for creat-ing new artificial reefs ($0.80 - $5.61 forresidents and non-residents respectively); thefinding suggests that there are decliningmarginal returns to increasing the supply ofreefs in an area in which reefs (artificial ornatural) already were abundant. Milon(1989) also estimates the economic value ofnew artificial reefs, what the author calls

“option” values.” Milon finds that estimatesfor the option value of new artificial reefsrange from $4.48 to $128 per visitor peryear, depending on the method used. Inthese cases, reef diving opportunities are notas scarce as in other locations (e.g. SouthernCalifornia or the Mid-Atlantic UnitedStates). Where reef diving opportunities arescarce, it is likely that the non-market valueof new artificial reefs will be relatively higherinitially, but the value of additional artificialreefs should be expected to decline (a com-mon tenet of economics known as declin-ing marginal returns).

At least two studies find that artificialreefs are not perfect recreational substitutesfor natural reefs when both types of reefsexist together. Johns, et al. found a prefer-ence among boaters, fishers, and divers fornatural reefs; the per person-day use valuefor natural reefs averaged $14 compared tothe value for artificial reefs that averaged$9.18. In addition to the higher willingnessto pay for natural reefs, the Johns et al. studyalso shows that in most counties in Florida,the percent of dives conducted on natural

reefs was much higher than that of divesconducted on artificial reefs. In an unpub-lished manuscript (personal communica-tion), Ditton also finds that artificial reefsare not as highly valued as natural reefs;Ditton estimates the per trip value for artifi-cial reefs is $76 lower than that of naturalreefs ($114 and $190 respectively).

III. ConclusionOur base of knowledge regarding the

economic value of dive recreation at ship-based artificial reefs is still limited. In thepublished literature, only a handful of stud-ies examine the economic impacts of ship-based artificial reefs and most of those stud-ies focus on ship-based artificial reefs incoastal Florida. Clearly, there is a need toknow more about the economic impacts ofthe more than 300 ship-based artificial reefsin place around North America, but out-side of Florida. The potential economic valueof a ship-based artificial reef depends bothon the value of a reef to the individual diver(which is a function of diver interest, the

TABLE 2Non-Market Value Estimates for Diving at Artificial Reefs

Author Method Location Habitat Type Market Value Per Person-Day($2004, figures are rounded)

STUDIES OF DIVING ON ARTIFICIAL REEFS

Ditton and Baker (1999) Contingent Valuation: Texas Various types of artificial reefs 1. $752. $45Ditton et al. (2001) 1.dichotom-ous choice

2. open-ended

Bell et al. (1998) Travel Cost North West Florida Ships, reef balls, and other $11structures

Contigent Valuation Residents: $3.50 - 4.30Visitors: $6.30-7.70

Roberts et al. (1985) Contingent Valuation Gulf of Mexico Petroleum Structures ($339 annually per diver)

Johns, et al. (2003) Contingent Valuation Southeast Florida Ships, reef balls, and other Residents:(dichotomous choice) private and public artificial reefs $3.40 (to maintain existing artificial reefs)

$0.80 (new artificial reefs)Visitors:$14 (to maintain existing artificial reefs)$5.60 (new artificial reefs)

STUDIES OF DIVING AND FISHING ON ARTIFICIAL REEFS

Milon (1988) Contingent Valuation Florida Network of 7 different reefs $29.04 to $42.77 per yearfrom various materials

Milon (1989) Contingent Valuation Florida Ships and steel debris $4.48 to 127.56 per year


quality of the artificial reef, and substitutedive sites) and the total number of diversthat are expected to use a ship to reef site.Individual value, individual expenditures,and the total numbers of visitors will varyfrom region to region.

In 2000, The San Diego Oceans Foun-dation (SDOF), paid $238,000 to acquirethe 366 foot, Canadian Destroyer EscortYukon. The Foundation also paid an addi-tional $97,000 to prepare the vessel and$100,000 for towing, mooring, and sink-ing. Can the values taken from the litera-ture give us an idea of the potential economicbenefit of this “ship to reef” project?

The San Diego Oceans Foundation es-timates that 10,000 divers made 26,700 dayvisits to the Yukon “ship to reef” site betweenAugust 2002 and August 2003; roughly6,000 of these diver trips were made by out-of-town visitors accounting for 15,600 per-son-days (Pendleton, 2005). Conservativeestimates from diving in Florida (Bell et al.,1998 and Johns et al., 2003) suggest thatdivers to the Yukon might spend on the or-der of $50 per dive day for residents and$200 for non-residents. A non-random sur-vey of 814 divers to the Yukon revealed thatexpenditures associated with trips to thewreck were approximately $95 for dive-re-lated expenditures with out-of-town diversspending an additional $580 on food andlodging. The literature provides estimates ofpotential expenditures that are well withinan order of magnitude of what are likely tobe the actual expenditures by divers.

A similar transfer of values from the lit-erature can also be used to estimate the po-tential non-market value of visits to thesunken Yukon. Bell et al. (1998) and Johns etal. (2003) find only modest non-market val-ues for artificial reef diving (ranging fromapproximately $3/person-day for residents to$13/person-day for non-residents) while otherauthors find much more substantial valuesfor diving on rigs (on the order of $50 perperson-day). A travel cost study by Pendleton(2005), using a non-random sample of 4,256diver day trips to the Yukon over 3 years, esti-mated that the non-market value of diving atthe Yukon was on the order of $110/person-day. Estimates from the literature are clearly a

conservative estimate of the potential non-market value, but they still provide a guidefor considering the potential non-market eco-nomic value of the site to divers. The non-market value of artificial reefs to local diversmay explain the increasing role of diver-based“not-for-profit” organizations in the creationof new artificial reefs.

The value of sinking ships to create reefsdepends on the degree to which the newartificial reef site generates increased diverexpenditures and non-market values. Thevalues from the literature only indicate thepotential value of trips to a new reef site likethat which would be created by a ship toreef project. For the Yukon, the analyst wouldneed to know how the provision of the Yukonas a recreational destination for divers effec-tively changed the level of expenditures andtotal non-market values enjoyed by thesedivers. If we consider all of these trips as anupper-bound estimate of the incrementaleffect of the Yukon, we can use the literatureto estimate the maximum value of the Yukon.The literature suggests that the potentialmagnitude of annual expenditures by alldivers to the Yukon could be $3.5 million($3 million by non-residents and over $0.5million for residents; note that this figure isalmost exactly equal to the average estimatedexpenditures per site from Hess et al. [2001])for a Net Present Value of $46 million at adiscount rate of 4% over twenty years—roughly one hundred times the costs of buy-ing, preparing, and sinking the ship. Ofcourse, only a fraction of these expendituresrepresents true economic benefits (net rev-enues). Nevertheless, the values from the lit-erature suggest that the fiscal impact of theYukon could well exceed the costs of creat-ing the new dive site.

A similar analysis could be conductedto estimate non-market value of the Yukon.The literature suggests that the annual valueof non-market benefits of the Yukon are likelyto be between $80,000 and $1.3 million fora Net Present Value of between $1 millionand $13 million dollars. Again, the non-market values from the literature provideevidence that the economic returns fromcreating a new reef site in Southern Califor-nia could justify the creation of the site.

Original valuation studies and technicalbenefits transfer analyses are beyond themeans of many organizations that might liketo finance and promote ships to reefs. Thispaper demonstrates that estimates of mar-ket and non-market values, taken from theliterature, may provide a reasonable approxi-mation of the potential economic benefitsfrom creating new ships to reefs, artificialreefs, or other marine and coastal projects.

AcknowledgmentsFunding for this research was provided by

a grant from the San Diego Oceans Founda-tion. The author wishes to thank numerousreviewers for their excellent comments. Vernon(Bob) Leeworthy provided invaluable help inunderstanding values from studies in Florida.

ReferencesBaine, M. 2001. Artificial reefs: a review of

their design, application, management and

performance. Ocean Coast Manage. 44:241-259.

Bell, F., Bonn, M., and Leeworthy, V. 1998.

Economic Impact and Importance of Artificial

Reefs in Northwest Florida. NOAA Paper

Contract Number MR235.

Brock, R. 1994. Beyond Fisheries Enhance-

ment: Artificial Reefs and Ecotourism.

B Mar Sci. 55(2-3):1181-1188

Cesar, H. S.J. 2000. Collected Essays on the

Economics of Coral Reefs, pp. 250, CORDIO,

Kalmar University, Kalmar, Sweden.

Daniel, D.L., 1976. Empirical and theoretical

Observations on the Potential Economic

Benefits and Costs Associated with Mississippi-

Alabama Liberty Ship Reef Program.

Hattiesburg: Bureau of Business Research,

University of Southern Mississippi.

Ditton, R.B. and T.L. Baker. 1999. Demo-

graphics, Attitudes, Management Preferences,

and Economic Impacts of Sport Divers using

Artificial Reefs in Offshore Texas Waters,

Report prepared for the Texas Parks and

Wildlife Department through a research

contract with Texas A&M University


Ditton, R. B., C.E. Thailing, R. Riechers and

H. Osburn. 2001. The Economic impacts of

sport divers using Artificial Reefs in Texas

Offshore Waters. Proceedings of the Annual Gulf

and Caribbean Fisheries Institute 54:349- 360.

Ditton, R. personal communication. Valuing

Recreational SCUBA Diving Use of Natural

and Artificial Reef Habitats. Abstract only.

Hess, R., Rushworth, D., Hynes, M., Peters, J.

Disposal Options for Ships National Defense

Research Institute RAND 2001. 59-80.

Hiett, R. and Milon, J.W. 2002. Economic

Impact of Recreational fishing and Diving

Associated with Offshore Oil and Gas

Structures in the Gulf of Mexico. DOI

Minerals Management Service Document

MMS Study 2002-010.

Johns, G., Leeworthy, V., Bell, F. and Bonn, M.

2003. Socioeconomic Study of Reefs in Southeast

Florida, Final Report for Broward County,

Florida, Fort Lauderdale, October 19, 2001.

Leeworthy, V.R., Bowker, J.M., Hospital,

J.D., and Stone. E.A. 2005. Projected

Participation in Marine Recreation: 2005 &

2010. U.S. Department of Commerce,

National Oceanic and Atmospheric Adminis-

tration, National Ocean Service, Special

Projects, Silver Spring, MD. March 2005.

McGinnis, M. Fernandez, L. and Pomeroy, C.

2001. The Politics, Economics, and Ecology of

Decommissioning Offshore Oil and Gas

Structures. DOI Minerals Management Service

Document. MMS Publication 2001-006.

Milon, J.W. 1988. The Economic Benefits of

Artificial Reefs: An Analysis of the Dade

County, Florida Reef System. Gainesville, Fla.:

Sea Grant Extension Program, University of

Florida, 1988. Report / Florida Sea Grant

College; no. 90

Milon, J.W. 1989. Contingent valuation

experiments for strategic behavior. J Environ

Econ Manag. 17:293-308.

Pendleton, L. 1995. “Valuing Coral Reef

Protection.” Ocean Coast Manage. 26:119-131

Pendleton, L. 2005. Towards A Better

Understanding of the Economic Value of Ship

to Reef Scuba Diving in Southern California.

San Diego Oceans Foundation.

Roberts, K. Thompson, M., and Pawlyk, P.

1985. Contingent Valuation of Recreational

Diving at Petroleum Rigs, Gulf of Mexico.

Transactions of the American Fisheries Society

114: 214-219.

Wilhelmsson, D., Ohman, M.C., Stahl, H.,

Shlesinger, Y. 1988. Artificial Reefs and Dive

Tourism in Eilat, Israel. AMBIO. 27(8):764-766.


A U T H O R ST.S. SidhuS. PrakashR.D. AgrawalMetallurgical and Materials EngineeringDepartment, Indian Institute ofTechnology Roorkee

P A P E R

State of the Art of HVOF Coating Investigations—A Review

A B S T R A C TCorrosion, erosion and abrasion, or combinations of these mechanisms, are the main

cause of degradation of materials used in marine, aircraft, waste incinerators, power gen-eration, chemical, and paper and pulp industries. One possible way to address these prob-lems is by applying a thin layer of wear and corrosion resistant coatings. Due to thecontinuously rising cost of materials as well as increased material requirements, coatingtechniques have been given more importance in recent times. Among the different coat-ings techniques, high velocity oxy-fuel (HVOF) spraying process is a new and rapidlydeveloping technology, which can yield high density coatings with porosity less than 1%,having high hardness and adhesion values, and good erosion, corrosion and wear resis-tance properties. The very high kinetic energy of the powder particles in the HVOF pro-cess results in the deposition of high quality coatings. It is possible to obtain a coatingthickness of more than 1.5 mm with careful control of cooling to reduce residual stresses.The purpose of this paper is to review the physical, mechanical, erosion-corrosion andwear properties of the HVOF coatings and effects of deposition parameters of the processon the properties of the coatings.

ing to the substrate, seawater may permeatethem and reach the interface between the coat-ing and the substrate. When a conductive so-lution contacts different conductive materi-als, it forms a galvanic cell. A combination ofnoble coating and less-noble substrate accel-erates substrate corrosion more than a baresubstrate with the same surface area.

The high velocity oxy-fuel (HVOF)process belongs to a family of thermalspraying technologies being used to en-hance the surface properties of base mate-rials. HVOF coatings have comparativelyless porosity as compared to plasma spraycoatings. As porosities of the coatings playa significant role in the corrosion resistanceof thermal spray coatings, HVOF coatingsare being studied extensively for their cor-rosion resistant properties. Moreover, thecomposition of HVOF coatings is nearlythe same as that of spraying powder.

The HVOF sprayed coatings have foundwide application in marine, aircraft, auto-motive and other industries. For reclaiminga wide range of petrochemical-process com-ponents such as storage vessels, heat exchang-ers, pipe end fittings and valves, which are

subjected to severe erosive, wear and corro-sive conditions, Amoco Oil Company rou-tinely employs the HVOF process by ap-plying AISI 316 L and Hastalloy C-276coatings (Moskowitz, 1992).

HVOF thermal spraying is a techniquewhereby powder material is melted and pro-pelled at high velocity, with the use of oxygenand fuel gas mixtures, towards a surface.Propylene, propane, hydrogen, acetylene,methane, ethylene, crylene, SPRAL 29 kero-sene, MAPP (methyleacetylene-propadiene-stabilised gas), LPG etc. are used as com-bustion fuels. The HVOF system consistsof a spray gun, powder feed unit, flow meterunit, and an air and gas supply unit. Thepowder feed unit comprises a hopper assem-bly, air vibrator, feed rate meter and controlcabinet. The desired powder is fed from thepowder feed unit by means of a carrier gasto the gun, where combustion occurs. Theamount of powder required for depositionmay be regulated using the powder feed-ratemeter. In the combustion zone, the powdermaterial enters the flame, where it becomesmolten or semi-molten, depending on themelting temperature and the feed rate of the

SI N T R O D U C T I O N tructural steels used in marine applications, energy conversion and utili-zation systems, and in chemical and petro-leum industries are often required to have along service life because environmental regu-lations and labor costs of repairs are expectedto become increasingly severe and highhereafter. In order to obtain a long lifetime,erosion-corrosion and abrasion of these ma-terials are the main problems to be solved.

In the marine environment, structuralsteels are subject to severe corrosion damagedue to the abundant presence of sea saltsand water. Up to now, cathodic protection,thick anticorrosion paint and cladding, havebeen mostly used for corrosion protectionfor marine corrosion. However, it is ques-tionable whether these methods can providelong service life, over 100 years, without anymaintenance.

One possible way to attack these prob-lems is the use of thin anti-wear and anti-cor-rosion coatings. However, the coatings to beused in a marine environment require an im-permeable nature above all. If such coatingshave even a small amount of pores connect-


material. The flame temperature for theHVOF process is around 30000 C (Sobolevet al., 2004). The molten or semi-moltenparticles are then propelled out of the gunnozzle at supersonic velocities towards thetarget/substrate, where the material is depos-ited. Powder particles, typically in the range10-63 µm, attain velocities of 300-800 ms-

1 at the substrate to be coated (Kowalsky etal., 1991; Irving et al., 1993; Knight et al.,1994; Smith & Knight, 1995; Herman etal., 2000). The basic scheme of the HVOFspray system is shown in Figure 1 using theDiamond Jet gun as an example (Stokes &Looney, 2001).

Due to high velocity and high impact ofthe sprayed powder particles, the coatingsproduced by HVOF spraying process are lessporous and have higher bond strength thanproduced by other methods such as plasmaspraying, flame spraying, and electric arcspraying (Roa et al., 1986; Crawmer et al.,1992; Jarosinski et al., 1993; Provot et al.,1993). Particle speed, flame temperature andspray atmosphere are the main parameterswhich differentiate the various spraying tech-niques. Coating porosity, bond strength andoxide content are typical properties influ-enced by the coating procedure. Table 1shows the characteristics of the WC-Co coat-ings sprayed by different spray techniques

FIGURE 1Schematic cross-section of HVOF gun (Stokes & Looney, 2001)

FIGURE 2Characteristics of HVOF and standard plasma process coatings (Helali & Hashmi, 1992)

TABLE 1Thermal spraying processes

Deposition Heat Source Propellant Typical Typical Average Spray Coating Relative BondTechnique Temperature(°C) Particle Velocity Rate (kg h-1) Porosity Strength

(m s-1) (%by Volume)

Flame Spraying Oxyacetylene/ Air 3000 30-120 2-6 10-20 FairOxyhydrogen

Plasma Spraying Plasma Arc Inert Gas 16000 120-600 4-9 2-5 Very Good toExcellent

Low Pressure Plasma Arc Inert Gas 16000 Up to 900 - <5 ExcellentPlasma Spraying

Detonation Gun Oxygen/ Acetylene/ Detonation 4500 800 0.5 0.1-1 ExcellentSpraying Nitrogen Gas Shock Waves

Detonation

High Velocity Fuel Gases CombustionJet 3000 800 2-4 0.1-2 ExcellentOxy-fuel (HVOF)


(Sobolev et al., 2004). In comparison withair or vacuum plasma spraying, the HVOFhas the advantage of being a continuous pro-cess. Figure 2 indicates the characteristics ofHVOF coating compared with those pro-duced using the standard plasma sprayingprocess (Helali & Hashmi, 1992).

Quality of the coatings depends signifi-cantly on the velocity and temperature ofthe powder particles impinging onto thesubstrate surface, which in turn is associatedwith the gas pressure developed in the com-bustion chamber. In the HVOF spray sys-tems of the first and second generations(Continuous Detonation Spraying, TopGun, Jet-Kote and Diamond Jet) combus-tion occurs at pressures in the range of 3-5bar and the flame attains a supersonic ve-locity in the process of expansion at thenozzle exit. These systems produce compa-rable particle velocities with the standardspray parameters and the same fuel gases andpowders. For instance, during spraying ofthe WC-17%Co powder with the particlesize distribution of -45+10 µm using pro-pane the particle velocities are about 450 m/s (Kreye, 1997).

HVOF systems of the third generation(Diamond Jet Hybrid 2600 and 2700, JP-5000, OSU Carbide jet, and TOP Gun K)

operate at higher combustion pressures inthe range of 6-10 bar. These systems permithigher particle velocities and higher sprayrates. For examples, in the case of the WC-Co powder the velocities are about 600-650m/s and the spray rates increase up to 10kg/h and in JP-5000 system even up to 18kg/h without any deterioration of the coat-ing quality (Kreye, 1997).

Stokes & Looney (2001) have modi-fied the HVOF spraying gun by incorpo-rating a traverse unit to traverse the spraygun back and forth, thermocouple to mea-sure the spraying temperature of thesprayed surface and a carbon dioxide cool-ing system to cool the sprayed region, toreduce and control the spraying tempera-ture. These additional features reduce theresidual stresses caused by interruption ofthe spraying process for controlling thespraying temperature to a set value. Fig-ure 3 shows the end-view schematic of aHVOF gun, traverse unit and carbon di-oxide cooling system and Figure 4 showsa schematic of a carbon dioxide nozzle lay-out and distances.

In the HVOF spraying process, somein-flight powder particles get oxidized.This problem can be minimized by shield-ing the in-flight particles from the atmo-sphere by inert gas. Recently, Kawakita etal. (2003) have conducted studies on theHVOF spraying process, in which an in-ert gas shroud system is attached. In thissystem, a pipe is attached to one end ofthe barrel of a commercial HVOF gun andinert gas is injected from both the ends ofthis pipe. They termed this mechanism asthe ‘gas shroud mechanism’ or the ‘shroudmechanism’. Nitrogen gas is injected into

FIGURE 3End-view schematic of HVOF gun, traverse unit and carbon dioxide cooling system (Stokes & Looney, 2001)

FIGURE 4Schematic of carbon dioxide nozzle layout and distances used (Stokes & Looney, 2001)


the shroud at the flow rates of 1.5–2.5 m3

min-1 from upstream and at 0.3 m3 min-1

from downstream. This attachment en-abled in-flight spray particles to be accel-erated over 750 ms-1 and suppressed oxi-dation significantly. The coating ofHastelloyC nickel base alloy by HVOFspraying with the gas shroud attachmenthas zero through-porosity and 0.2 mass%of oxygen content. The laboratory corro-sion tests showed that the on-shroudHastelloyC coating is comparable to thebulk material of HastelloyC in terms ofcorrosion resistance.

Vacuum plasma spraying (VPS) is stilla frequently used thermal spray processfor deposition of corrosion and oxidationresistant materials and is recognized forproviding reasonably density and relativelyless oxidization under thermal and oxida-tion load. However, it is a very expensiveprocess since equipment costs are greaterthan $2 million US. Furthermore, thechamber process requires time-consumingevacuation and flooding cycles, which in-hibits its efficiency. Additionally, an on-line process control by optical diagnosticmeans such as in-flight particle or sub-strate pyrometry is difficult in the vacuumchamber. In contrast, HVOF spray sys-tems are operated in the atmosphere, in-vestment costs are roughly a tenth com-pared to VPS, and process monitoring iseasier (Irons & Zanchuk, 1993; Yamasakiet al., 1995; Meyer, 1995).

The electrochemical behaviour ofHVOF sprayed WC-12%Co coatings ap-plied on low alloyed Cr-Mo steel has beenstudied in artificial seawater using ZeroResistance Ammetry techniques. The elec-trolyte artificial seawater used in this studywas prepared according to ASTM D-1141(Collazo et al., 1999). Similarly HVOFsprayed stainless steel coatings have beensubjected to the corrosion test in 3.4%NaCl + saturated Ca(OH)

2 solution. The

corrosion performance of the coatings hasbeen evaluated using linear polarization,AC impedance, and salt spray techniques(Gu et al., 1998). These studies concludedthat the HVOF sprayed coatings have highcorrosion resistance.

2. Physical and MechanicalProperties of the Coatings

Coatings used in marine structure mustbe strong, hard and adherent. Immersioncoatings must have good impact and abra-sion resistance and must be able to flex wellenough to maintain contact with the steelsubstrate when it is bent.

The high kinetic energy of the particlesin the HVOF process leads to the formationof dense and hard coatings due to the defor-mation of the particles in a plastic state ratherthan a molten state. As a result, oxidation ofspray metal during flight and flattening isrelatively less, since oxidation can occur onlyby a relatively slow diffusion mechanism. Inspite of the plastic state, the high kinetic en-ergy of the particles still allows flattening bydeformation and leads to dense and pore-free coating with low oxygen content. Thischaracteristic of the HVOF process is of highimportance for spraying mechanically alloyedmaterial (Provot et al., 1993; Nestler et al.,1994; Voggenreiter, 1996; Lugscheider et al.,1998; Zhao et al., 2001; Zhao &Lugscheider, 2002).

Detailed microstructural examination ofHVOF sprayed powders shows that coat-ings exhibit characteristic splatlike, layeredmorphologies due to the deposition andresolidification of molten or semi-moltenpowder particles. During HVOF spraying,powder particles are generally comprised ofthree separate zones; fully melted regions,partially melted zones, and an unmeltedcore. However, the relative proportionformed in an individual powder depends onits particle size, trajectory through the gun,the gas dynamics (velocity/temperature) ofthe thermal spray gun and the type of gunemployed (Dent et al., 2001; Kong et al.,2003; Zhang et al., 2003).

HVOF sprayed carbide dispersed Ni-basedalloy (Cr

3C

2- NiCr) can have hardness of 1150

Hv and adhesion strength of 200 MPa. Fur-ther, by using the smaller primary powder sizeand NiCr of 20 mass %, adhesion strength ofCr

3C

2- NiCr coating can be improved to 250

MPa. In the case of NiCr< 20 mass%, due tothe high carbide rate, the strength of the coat-ing declines. On the other hand, if NiCr> 20mass%, the softer particles could derive a de-

crease of adhesion strength. Further, in case ofusing smaller particles, the relatively extensivesurface area, causing effective kinetic momen-tum and heat transfer from the gas flame tothe particles attributes superior acceleration andheat, which result in better adhesion strengthof the coatings (Hamatani et al., 2002).Through porosity of HVOF sprayedHastelloyC coating has been chemically de-tected using ICP atomic emission spectrom-etry of dissolved substances permeating viaconnecting pores in such coatings. It is foundthat through porosity depends on coatingthickness and on the sprayed-particles stack-ing structure. Coating with zero through po-rosity has been prepared under a higher com-bustion pressure (0.86 MPa) than the standard(0.68 MPa) and was approximately 400 µmthick (Kawakita et al., 2003a). HVOF sprayedAl

2O

3-dispersion-strengthened NiCr powder

results in dense coating with homogenousmicrostructure. The microstructure of pow-der is retained in the coating after spraying.However, the coating showed lower hardnessthan the powder after spraying. Wear resistanceof the coatings is found to be dependent onthe properties of powders. Homogeneous andhigher volume fraction of Al

2O

3 powder (50

Vol. %) produced more wear resistant coat-ings (Zhao et al., 2004).

Addition of CeO2 and Cr in the HVOF

thermal sprayed NiAl intermetallic-basedcoatings improves the wetting and bondstrength of the coatings to the substrate,which decreases the tendency of brittle peel-ing during thermal spraying. These coatingslayers have higher hardness, improved elas-tic modulus with less cracks and pores ascompared with pure NiAl coatings. TheNiAl base intermetallic alloy coatings exhib-ited excellent carburization resistance at hightemperature. This may result from the dif-fusion barrier role of the NiAl coatings, asafter carburizing, oxide films such as Al

2O

3

and rare earth compound, CeAlO3, are

formed in the intermetallic-based alloy coat-ings. These oxides may obstruct the inwarddiffusion of carbon during high-temperaturecarburization, resulting in low carbon con-centrations in the NiAl coatings, and pre-vent the formation of carbides in the sub-strate (Wang & Chen, 2004).


3. Wear Resistance PropertiesHigh velocity oxy-fuel thermal spraying

is one of the most versatile processes of depo-sition of coating materials to enhance wearperformances. It is a process with almost nolimitation of materials and has the ability todeposit coatings on a great variety of shapesand sizes, with thicknesses ranging from sev-eral micrometers to tens of millimeters.

HVOF sprayed carbides based cermetscoatings are widely used against wear andcorrosion in gas and oil industries. Theirwear resistance is three to five times that ofelectroplated chromium, and their manu-facturing costs are low. These coatings canbe used as a possible replacement for hardchromium plating in gas turbine shaft re-pair. HVOF sprayed Cr

3C

2–NiCr and WC–

Co coatings exhibit high hardness with ahigh volume fraction of carbides being pre-served during the spraying process. Weartests of HVOF sprayed Cr

3C

2–25NiCr and

WC–12Co coatings have been carried outwithout lubrication and under extreme load-ing conditions. The hardness (Figure 5) andwear resistance (Figure 6) of the WC-Cocoatings are better and porosity (Figure 7) isless as compared with Cr

3C

2-NiCr coatings.

Micrographs reveal that on the HVOFsprayed Cr

3C

2–25NiCr coatings there is evi-

dence of particle pull-out or scratchingwhich supports the wear by abrasion. Dam-aged surfaces contain craters whose rate anddimension are more significant than thosecaused in WC–Co deposits (Sahraoui et al.,2003). Considering the economical and eco-logical requirements, HVOF sprayedTribaloy©-400, Cr

3C

2-25%NiCr, WC-

12%Co coatings can possibly replace elec-trodeposited hard chromium (EHC) in a gasturbine shaft repair. The friction coefficientsof HVOF coatings are found to be close tothat of chromium electroplated deposits(Sahraoui et al., 2004).

Chromium carbide/nickel chromiumcoating can be deposited by various thermalspraying processes i.e. detonation gun pro-cess (Tucker et al., 1998), plasma spray pro-cess (Hwang & Seong, 1995), and a varietyof HVOF processes (Li et al., 1998). Amongthese thermal spray processes, the HVOFprocess has a relatively lower temperature

FIGURE 6Evolution of the weight loss of HVOF coated samples vs. the applied loads (Sahraoui et al., 2003)

FIGURE 5SEM observations of worn region of samples coated with: (a) Cr3C2–25NiCr; (b) WC–Co (Sahraoui et al., 2003)

FIGURE 7Optical micrographs of the as-sprayed coatings: (a) Cr3C2–25NiCr; (b) WC–Co (Sahraoui et al., 2003)


and supersonic gas jet for the deposition ofheat sensitive materials in the atmosphere.Supersonic gas jet which ultimately formsdiamond shock waves, is due to the nozzlewhich is of convergent divergent type. Atthe exit of the gun, pressure exceeds atmo-spheric pressure, so the gas jet expands witha corresponding increase of the Mach num-ber above 1 and the so-called diamondsshock are formed. These benefits enableHVOF to be a promising and most popularprocess for the preparation of wear resistantcermet coatings.

Sudaprasert et al.(2003) have studied thesliding wear behavior of HVOF sprayedWC-Co coatings deposited using both gas-fuelled (HVOGF) and liquid-fuelled(HVOLF) systems and reported that with adense powder feedstock, the HVOGF de-posited coating was superior to the HVOLFdeposited coating, as the HVOLF sprayedcoating was associated with a mechanicaldamage to the WC-Co powder particles asthey impact with the substrate resulting incarbide cracking, and a reduction in the in-tegrity of the bond between the carbide par-ticles and the matrix phase.

Triobological study of NiCrBSi coatingshows that HVOF sprayed coatings had ahigher value of micro hardness and wear re-sistance than the fused process as well as theplasma spraying process. Plasma sprayedNiCrBSi had the worst sliding wear resis-tance (Miguel et al., 2003).

4. Erosion and CorrosionBehavior of the Coatings

Marine corrosion includes the deterio-ration of structures and vessels immersed inseawater, the corrosion of machinery andpiping systems that use seawater for coolingand other industrial purposes, and corrosionin the marine atmosphere. Erosion corro-sion is acceleration in the rate of corrosionattack in metal due to the relative motion ofa corrosive fluid and a metal surface.

The corrosion resistance of the HVOFsprayed HastelloyC deposit was found to becomparatively high under the seawater envi-ronment. Its corrosion rate was estimated tobe in the order of 10 µm year-1 from the re-

sult of the electrochemical AC impedancemeasurement. The primary corrosion reac-tion of the deposit was uniform formation ofthe oxide or the hydroxide on its whole sur-face. When pores existed between the sprayedparticles of the deposits, such places were sub-ject to the predominant corrosion reactionand the corrosion rate there was considerablyfaster than that in the normal sprayed parts.As the result, the local corrosion seemed totake place there (Kawakita et al., 2003).

Erosion corrosion tests of HVOF NiAl-40Al

2O

3 intermetallic-ceramic coating us-

ing bed ash and fly ash retrieved from anoperating boiler as erodent materials revealedthat the coating exhibited excellent thermalshock resistance and high erosion resistance,especially at a steep impact angle and hightemperature (Wang & Lee, 2000).The cor-rosion behavior of HVOF-sprayed Inconel625 coatings showed that the coatings pro-duced with the liquid-fueled gun exhibitedreduced interconnected porosity and in-creased corrosion resistance compared withdeposits obtained from the gas-fueled sys-tem (Zhang et al., 2003).

Wang (1996) conducted erosion tests ona proprietary HVOF Cr

3C

2-NiCr cermet

coating (DenSys DS-200) at a temperatureof 4500C and impact angle300 and 900Cunder generally oxidizing conditions usingbed ash and fly ash retrieved from over 60CFBC units in North America and Europe.They observed that HVOF Cr

3C

2-NiCr

coatings showed excellent erosion-corrosionbehavior as compared with 1018 steel, A213-T22 steel, and other thermal sprayed coat-ings tested under both shallow and steepangles. The high erosion-corrosion resistanceof HVOF Cr

3C

2-NiCr coating is attributed

to its high compactness, fine grain size struc-ture, and a homogeneous distribution of theskeletal network of hard carbide within aductile, corrosive-resistant metal binder.

Corrosion and wear behavior of HVOFsprayed nano-powder Cr

3C

2-25%NiCr coat-

ings on AISI 1045 steel substrate displayed amarkedly smaller weight loss value with re-spect to hard chromium and HVOF sprayed45µm grain sizes coatings (Figure 8). Thisbehaviour can be related to the lower surfaceroughness and to the better distribution of

carbides in the metal matrix and also to thelowest porosity of the coating. Micrographsof the morphology of the coatings surfacesare shown in Figure 9. The finer microstruc-ture of the coating obtained using nano-sizedpowder is well evident, whereas some porescan be observed in the coating obtained us-ing standard powders of grain sizes 45 µm.The research involves substantial benefits forthe environment, as the proposed HVOFtechnique can replace some highly pollutingsurface treatment techniques, such as chro-mium-plating, with a perfectly clean process(Fedrizzi et al., 2004).

In an electrochemical test, HVOFsprayed NiCrBSi coatings exhibited an ex-cellent corrosion resistance in alkali solu-tions, as the surface can form protective filmand keep in a self-passivation condition. Thecorrosion current of the coating in sour so-lutions is greater than that in 3.5% NaCl.Corrosion of HVOF NiCrBSi coating firstoccurred around the particles that had notmelted during spraying and the defects suchas pores, inclusions and microcracks, thenfollowed by the development along the pathsformed by pores, microcracks and lamellarstructure, resulting in exfoliation or laminarpeeling off. Adjusting the thermal sprayingparameters to reduce the electrochemicalunevenness or sealing the pores can improvethe corrosion resistance of the coating [Zhaoet al., 2004a; Zhao et al., 2005).

The effects of inclusion and porositieson the corrosion behaviour of HVOFsprayed NiCrBSi coating has been reportedby Zhao et al. The results revealed that if theinclusion were big enough, local corrosionwould occur. The effects of porosities on theearly corrosion of the coating are not seri-ous unless there are penetrating porosities.However, porosities can do harm to the per-sistent corrosion resistance of the coating andthe presence of porosities may weaken thecohesive strength within the coating (Zhaoet al., 2004a).

Investigation of HVOF sprayed Ni-basealloy coating (N160JH and Delelo) re-vealed that corrosion weight loss rate ofN160 JH and Delelo 50 coating was 1/30and 1/16 that of substrate, respectively, anderosion corrosion resistance of N 160 JH


and Delelo 50 coating was 15 and 5 timeshigher than that of substrate, respectively(Anfeng et al., 2003).

For some years, the hydraulic pistons inneed of repair are also coated with the highpressure-high velocity oxygen fuel (HP-HVOF) process. The coatings used aremainly chromium carbide in a nickel chro-mium metal matrix. The applied thicknessof the coating, as sprayed, is usually about150 µm thick and may reduce to around100 µm after polishing with diamond grind-ing to remove the surface oxide layer. Thecoatings obtained in this way show a muchlonger life combined with a better corrosionresistance than the usual galvanic chromium-plating (Barbezat et al., 1993; Russo &Dorfmann, 1995; Zimmermann & Kreye,1996; Stein et al., 1999).

5. Effects of DepositionParameters on thePerformance of Coatings

Spraying distance, fuel/oxygen ratio andpowder feed rate exert a significant influ-ence on the porosity and corrosion resistanceof the coating. The higher the total gas flowrate, lower powder feed rate and shorter thespray distance, the higher the particle veloc-ity and temperature. The particle velocity ismore sensitive to the spray parameters thanparticle temperature. In general, the coat-ing hardness increased with increasing theparticle temperature and velocity and coat-ing porosity decreased (Lugscheider et al.,1998; Gil & Statia, 2002; Zhao et al.,2004b). With a high temperature, powderparticles will be more in a molten state be-fore striking the substrate surface. Thus, theyflow more easily and may fill the voidsformed across the boundaries of splats ef-fectively. However, in the case of tungstencarbide coating, hardness decreases with anincrease in temperature.

Significant melting of spray particlesdoes not contribute to the increase in theadhesion of HVOF metallic coatings. Onthe other hand, the deposition of partiallymelted large particles contributes to the sub-stantial improvement of adhesive strengthof HVOF coatings and yields an adhesive

FIGURE 8Volume loss of the HVOF coatings under different tribo-corrosion conditions (Fedrizzi et al., 2004).

FIGURE 9(a) Top view of the surface morphology of the HVOF sprayed Cr3C2-NiCr standard coating (b) Top view of thesurface morphology of the HVOF sprayed nano-sized Cr3C2-NiCr coating (Fedrizzi et al. 2004).


strength of over 76MPa, double the coatingdeposited with completely molten particles.These coatings adhere to the substrate by themechanism of mechanical interlocking on thepeaks of valleys of substrate, which is rough-ened by grit blasting. Semi-molten particlesmay have better mechanical interlocking withthe substrate than the molten particles. Spray-ing distance is an important parameter in theHVOF process and it affects the velocities ofspraying particles and therefore the porosityof the coating. The higher the velocity ofspraying particles, the less the porosity (Gil& Statia, 1999; Li et al., 2000). Sobolev et al.(1994) have studied the effect of increase inspraying distance on velocities of Al

2O

3, WC-

12%Co and pure Ni particles and results arereported in Figure 10. It is observed from thegraphs that with an increase in spraying dis-tance, the powder particle velocities increase,attain maximum values, and then decrease inthe direction of the substrate. They furtherobserved that the maximum velocity of theparticle decreases and its axial position is dis-placed in the substrate direction as the par-ticle diameter increases.

Hearley et al., while studying the effectof spray parameters on the properties ofHVOF NiAl intermetallic coatings, foundthat gas mixing ratio and powder size werecritical in determining coating properties.The best quality coatings can be obtainedwith an 80% stoichiometry gas ratio, aspherical inert gas atomized powder with anarrow particle size range between 15-45 µmand a small percentage of particles > 50 µm.With this optimized gas mixing ratio andpowder size, it is possible to deposit a highquality NIAl coating with porosity levels of2 vol%, low oxygen content (0.93 wt%),high Young’s Modulus (281 Gpa) and hard-ness (420 Hv) (Hearley et al., 1999; Hearleyet al., 2000).

Lih et al (2000) have studied the affectsof HVOF process parameters such as oxy-gen flow rate, fuel gas flow rate, powder car-rier gas flow rate, powder feed rate, gun bar-rel length, stand-off distance substratesurface speed as given in Table 2 and Table3, on the coating quality and their resultsare reported in Table 4. Particle speed and

FIGURE 10Variation in particle velocity with spraying distance

TABLE 2Spraying parameters used for design of experiments (DOE)

Label Parameters Unit Level1 Level 2

OF O2 flow rate l/min 300 450

FF C3H6 flow rate l/min 55 75

CF N2 carrier gas flow rate l/min 20 35

PF Powder feed rate g/min 25 50

GB Gun barrel length inch 4 5

SD Stand-off distance mm 200 300

SS Substrate surface speed m/min 63 126

TABLE 3DOE matrix for coating deposition

Trial no. OF FF CF PR GB SD SS

Run-1 1 1 1 1 1 1 1

Run-2 1 1 1 2 2 2 2

Run-3 1 2 2 1 1 2 2

Run-4 1 2 2 2 2 1 1

Run-5 2 1 2 1 2 1 2

Run-6 2 1 2 2 1 2 1

Run-7 2 2 1 1 2 2 1

Run-8 2 2 1 2 1 1 2

Run-9 1 2 1 2 1 1 1

Run-A 2 2 2 2 2 1 1

Run-10 2 2 2 2 1 1 1


temperature data have been generated fromthe average of detected good particles within30 s accumulation measuring. The value inparentheses under temperature data is thetotal detected good particles of each mea-surement. It is observed that coatings de-posited by higher kinetic energy and ad-equate surface temperature of moltenparticles are dense and hard.

The oxidation of powder during HVOFspraying results in the formation of metaloxides on the splat boundaries. The pres-ence of these oxides in the HVOF coatingswill degrade the resistance of the coatings tocorrosion and also affect mechanical prop-erties. To analyse the affect of the oxidationduring HVOF coatings, Feng et al. preparedHVOF NiCrAlY coatings with oxygen con-tents ranging from 3 to 21 at.% introducedduring spraying. Isothermal oxidation testsconducted at 10000 C up to 1000h onHVOF sprayed NiCrAlY coatings with vari-ous oxygen contents introduced during thethermal spraying process showed that a denseoxide scale consisting mainly of µ-Al

2O

3

formed on the coating with the lowest oxy-gen content (3 at.%). A duplex oxide scalewith an µ-Al

2O

3 sub-layer and a Ni(Al,

Cr)2O

4/Cr

2O

3 upper layer formed on the

coating with medium oxygen content (11at.%). Porous Cr

2O

3/NiCr

2O

4 oxide scale

formed on the coatings with the highest

oxygen content (21 at.%). These resultsshow that the oxide scale formation on thecoatings can be affected significantly by thedegree of oxidation that occurs in the coat-ings during the HVOF spraying process.Low oxygen (3 at.%) content in the coatingis beneficial to the formation of a protectiveα-alumina scale (Feng et al., 2004).

6. Effects of Pre- andPost-treatments of Coatings

Residual stresses are developed in HVOFsprayed coating and in the coated material.The developed stresses have been measuredwith the hole-drilling strain-gauge method.The measured strains are found to be nega-tive, indicating that the residual stresses aretensile stresses as per ASTM standards. Pre-heating of the specimen surface reduces theresidual tensile stresses. Residual stresses canbe reduced by selecting a coating materialwith matching properties to the substratesurface, and macro roughening of the sub-strate surface. Pre-heat the surface slightly,usually with a single pass of the torch with-out powder flowing, to remove any adsorbedgases or condensate and to cause some ex-pansion of the part. The part surface tem-perature should be pre-heated to 79–93°C.The pre-heating should be done carefully toavoid contamination of the surface. For

many applications, the maximum allowedtemperatures are related to the componentshape and the specimen material. When anyrestrictions are present, it is recommendedthat the maximum working temperatureshould be about 150°C. However, the pre-heating temperature may not be increasedbeyond 200°C because there may be achange in the structure of the steel beyondthis temperature (Hashmi et al., 1998).

Kinos et al. (1994) had reported thatcorrosion resistance of the HVOF sprayedcoating can not be improved by post treat-ments; however, shrouding of the in-flightparticle with inert gas seemed to reduce theamount of oxide in coating. Lee et al. (2000)while studying the corrosion properties ofHVOF sprayed Ni-Cr-W-Mo-B coating,observed that annealing of the coatings in avacuum furnace of 10-7 MPa at 550, 750and 950°C for 2 h after HVOF sprayingimproves the corrosion resistance due to in-creased microstructural and chemical homo-geneity, such as the reduction of porosity,densification and reduction of the eutecticphase. Uusitalo et al. (2002) also supportedthe results of Lee et al. (2000) and reportedthat laser remelting of HVOF sprayed Ni-50Cr, Ni-57Cr, Fe

3Al, Ni-21Cr-9Mo coat-

ings did not suffer from any corrosion dam-age, whereas as-sprayed coating waspenetrated by corrosive species. Laser remelt-ing efficiently removed the interconnectednetwork of voids and oxides at splat bound-aries of the HVOF coating.

Guilemany et al. (2001) have studied theinfluence of thermal treatments on the elec-trochemical corrosion resistance of HVOFCr

3C

2-NiCr coatings. Firstly, a few layers of

coatings were sprayed followed by thermaltreatment with a gun and finally the remain-ing layers were sprayed. Results reveal thatthe thermal treatment with the gun duringthe spraying process increases the protectionthat the coatings offer against the pass of elec-trolytes to reach the substrate. Further, a longtime spent between spraying, thermal treat-ment and spraying the rest of the coatinglayers again results in a poor corrosion resis-tance in comparison with the faster process.

Guilemany et al. (2002) have further re-ported that the sliding wear resistance of the

TABLE 4Properties of HVOF sprayed CrC/20NiCr coatings

Coating Run-1 Run-2 Run-3 Run-4 Run-5 Run-6 Run-7Properties

Particle 1590 1674 1580 1693 1710 1746 1668temperature (0C) (984) (591) (1407) (684) (612) (614) (1286)

Particle 475 385 453 525 609 432 481Speed (m/s)

Deposition rate 9.03 7.34 3.87 20.86 3.9 14.37 7.75(µm/pass)

Porosity 0.85 0.55 0.99 0.49 o.51 0.80 0.44Content (%)

Roughness 6.71 4.67 4.6 8.3 5.86 4.09 3.83(µm, Ra)

Microhardness 7.47 7.51 7.66 8.43 8.08 7.56 8.03(GPa)

Tensile bond 79.6 99.4 87.7 90.8 86.6 85.2 93.4Strength (MPa)


Cr3C

2–NiCr coating can be improved by an

adequate heat treatment (1 h at 1033 K) inan inert atmosphere. The formation of ox-ides, when heat treatment is carried out in anoxidizing atmosphere, increases the hardnessvalues. But these oxides have prejudicial ef-fect on the wear properties of the coatingswhen compared with coatings after similarheat treatments in an inert atmosphere. Smalland well distributed Cr

3C

2 produced by heat

treatment at 1033 K (inert atmosphere)greatly enhances the wear resistance of thecoating (Guilemany et al., 2002).

7. Repair, Maintenance andOther Applications

The HVOF thermal sprayed processcould be successfully used for forming thefreestanding solid and industrially relevantcomponents with various thicknesses. Thisoutcome is promising for numerous appli-cations of the HVOF process, in terms ofmanufacturing. The HVOF thermal spray-ing process has also been successfully usedto repair stainless steel and D2 tool steelsubstrate with different depth of damageto a built-up thickness of up to 5.5 mm.Sprayed material had good adherence to thesubstrate under various types of aggressivemachining processes (Tan et al., 1999;Stokes & Looney, 2001).

Traditionally, HVOF spraying tech-niques are predominantly being used aswear, corrosion and oxidation resistant bar-riers, resulting in increased lifetimes as com-pared with the uncoated substrate compo-nents (Sampath & McCune, 2000).However, as the technology has advanced,through new deposition techniques andimproved tool design, the range of materi-als that can be effectively deposited by thisprocess has increased to include low melt-ing point ceramics such as alumina andmaterials for biomedical applications suchas bioceramic coatings for dental implants(Haman et al., 1995). Now, HVOF coat-ings are increasingly used in many areassuch as petrochemical and offshore indus-tries, automotive components and generalengineering applications including print-ing, textiles and mining (Sturgeon et al.,

1994). HVOF technique can be used tospray-form thermocouples, humidity sen-sors, strain gages, and sensor arrays(Fasching et al., 1995).

Dent et al. (2001a) successfully depositedBaTiO

3 by HVOF for use as prototype di-

electric layers for applications as meso-scaleconformal circuits. Dielectric constants of upto 115 have been achieved in 150 µm thicklayers of the HVOF deposited material.

8. Concluding Remarks1. During the HVOF spraying process, hy-personic gas velocities of about 1800 ms-1

and a combustion temperature of about30000 C are generated. These rapidly ex-panding gases accelerate the powder particlesto velocities up to 800 ms-1, which allowdevelopment of dense coatings with very lowporosity usually below 1%, considerably lessthan plasma sprayed coatings (2-3%).2. Coatings produced by the HVOF pro-cess have increased thickness capability, highhardness values and less effect on the envi-ronment (reduced decarbonization, oxida-tion and loss of key elements by vaporiza-tion) during the spray process.3. HVOF-sprayed MCrAlY coatings arealso replacing some low-pressure plasma-sprayed coatings for high temperature oxi-dation/hot corrosion and TBC bond-coatapplications for repair and restoration ofexisting components.4. A good amount of work has been doneto evaluate the performance of HVOFsprayed coatings for their corrosion-erosionand wear protection properties. However,formation of oxides during HVOF spray-ing may affect the performance of the coat-ings in corrosive environments. High tem-perature oxidation and hot corrosionbehaviour of HVOF coatings need to beinvestigated in detail to explore the possibil-ity of applying these coatings in high tem-perature aggressive environments.5. Contradictory results are published onthe post treatments of the HVOF sprayedcoatings and more detailed study is requiredto evaluate the effects of post heat treatmentof the coatings on their physical, mechani-cal and chemical characteristics.

9. ReferencesAnfeng, Z., Yuyue, W., Jiandong, X.,

Chanqjiu, Li., 2003. Erosion-corrosion

characteristic and electrochemical behavior of

high velocity oxy-fuel sprayed coatings and two

types of steels, Journal of Xi’an Jiaotong

University, v 37, n 11, p 1150-1153, 1158.

Barbezat, G., Nicoll A. R. and Sickinger, A.

1993. Abrasion, erosion and scuffing resistance

of carbide and oxide ceramic thermal sprayed

coatings for different applications, Wear 162:529.

Collazo, A., Novo, X.R. and Perez, C. 1999.

Corrosion behaviour of cermet coatings in

artificial seawater, Electrochim, Acta 44:4289.

Crawmer, D.C., Krebsback, J.D. and Riggs,

W.L. 1992. Coating development of HVOF

process using design of experiments, Proceed-

ings of the 13th International Thermal

Spraying Conference, Florida, pp. 127–136

Dent, A.H., Horlock, A.J., McCartney, D.G.

and Harris, S.J. 2001. Microstructural

characterisation of a Ni-Cr-B-C based alloy

coating produced by high velocity oxy-fuel

thermal spraying, Surf Coat Tech. 139:244-250.

Dent, A.H., Patel, H., Gutleber, J., Tormay,

E., Sampath, S. and Herman, H. 2001a. High

velocity oxy-fuel and plasma deposition of

BaTiO3 and (Ba,Sr)TiO

3, Materials Science

and Engineering B: Solid-State Materials for

Advanced Technology, v 87, n 1, Oct 24,

2001, p 23-30.

Fasching, M., Prinz, F.B. and Weiss, L.E.

1995. ‘Smart’ coatings: a technical note. JTST

4(2):133-136.

Fedrizzi, L., Rossi, S., Cristel, R. and Bonora,

P. L. 2004. Corrosion and wear behaviour of

HVOF cermet coatings used to replace hard

chromium, Electochimica Acta 49: 2803-2814.

Feng, T.. Ajdelsztajn, L. and Schoenung, J. M.

2004. Characterization of oxide scales formed

on HVOF NiCrAlY coatings with various

oxygen contents introduced during thermal

spraying, Scripta Materialia. 51:25-29.

Gil, L. and Statia, M.H. 1999. Microstructure

and properties of HVOF thermal sprayed

NiWCrBSi coatings. Surf. Coat Tech. 120-

121, 423-429.


Gil, L. and Statia, M.H. 2002. Influence of

HVOF parameters on the corrosion resistance

of NiWCrBSi coatings. Thin Solid Film.

420-421, 446-454.

Gu, P., Arsenault, B., Beaudoin, J.J., Legoux,

J.G., Harvey, B. and Fournier, J. 1998.

Polarization resistance of stainless steel-coated

rebars. Cem Conc Res. 28:321.

Guilemany, J.M., Fernandez, J. and Delgado,

J. 2001. Influence of thermal treatments in the

corrosion behaviour of HVOF Cr3C

2-NiCr

coatings, In: Thermal Spray 2001: New

Surfaces for a New Millennium, C.C. Berndt,

K.A. Khor, and E.F. Lugscheider (eds.).

Materials Park, Ohio: ASM International.

pp. 1165-1169.

Guilemany, J.M., Miguel, J.M., Vizcaýno, S.,

Lorenzana, C., Delgado, J. and Sanchez, J.

2002. Role of heat treatments in the improve-

ment of the sliding wear properties of Cr3C

2 –

NiCr coatings. Surf Coat Tech.157:07 –213.

Haman, J.D., Lucas, L.C. and Crawmer, D.E.

1995. High-velocity oxyfuel thermal spray

coatings for biomedical applications. JTST,

4(2):179.

Hamatani, H., Ichiyama,Y. and Kobayashi, J.

2002. Mechanical and thermal properties of

HVOF sprayed Ni based alloys with carbide,

Sci Technol Adv Mat. 3(4):319-326.

Hashmi, M.S.J., Pappalettere, C. and Ventola,

F. 1998. Residual stresses in structures coated

by a high velocity oxy-fuel technique. JMPT

75:81-86.

Hearley, J.A., Little, J.A. and Sturgeon, A.J.

1999. The erosion behaviour of NiAl

intermetallic coatings produced by high

velocity oxy-fuel thermal spraying. Wear

233-235, 328-333.

Hearley, J.A., Little, J.A. and Sturgeon, A.J.

2000. The effect of spray parameters on the

properties of high velocity oxy-fuel NiAl

intermetallic coatings, Surf Coat Tech.

123:210-218.

Helali, M.M. and Hashmi, M.S.J. 1992. A

Comparative study of plasma spraying and

High Velocity Oxy-Fuel (HVOF) Thermal

Spraying, Proceeding of the 10th Conference

of the Irish Manufacturing Committee (IMC

10), Galway, Ireland, pp 377–387

Herman, H., Sampath, S. and McCune, R.

2000. Thermal spray: current status and future

trends. MRS Bull 25 (7):17-25.

Hwang, S.Y. and Seong, B.G. 1995. In:

Ohmori (ed.), proc. 14th ITSC, Koke, Japan,

pp. 59-63.

Irons, G. and Zanchuk.,V. 1993. Comparison

of MCrAlY coatings sprayed by HVOF and

low pressure processes. C.C. Berndt (ed.),

proc. NTSC, Anaheim, California, 191-197.

Irving, B., Knight, R. and Smith, R.W. 1993.

The HVOF process: the hottest process in

thermal spray technology, Welding J. 72:25-30.

Jarosinski, W.J., Gruninger, M.F. and Londry,

C.H.1993. Characterisation of tungsten

carbide cobalt powder and HVOF coatings,

Proceedings of the Fifth National Thermal

Spray Conference, California, pp. 153–158.

Kawakita, J., Kuroda, S., Fukushima,T. and

Kodama, T. 2003. Development of dense

corrosion resistant coatings by an improved

HVOF spraying process, Surf Coat Tech.

4:281-289.

Kawakita, J., Kuroda, S., and Kodama, T.

2003a. Evaluation of through-porosity of

HVOF sprayed coating. Surf Coat Tech.

166:17-23.

Knight, R., Smith, R.W., Xiao, Z. and

Hoffman, T. T. 1994. Particle velocity

measurements in HVOF and APS Systems.

Proc. 7th National Thermal Spray Conf.

[NTSC ’94], ASM Int’lÆ, Boston, MA.

pp. 331-336.

Knios,T., Siitonen, P., Kettunen, P. and

Laurila, U.J. 1994. Proc. NTPC, 7:537.

Kong, G., Zhang, D., Brown, P.D.,

McCartney, D.G. and Harris, S.J. 2003.

Microstructural characterisation of HVOF

thermally sprayed Stellite 6, Mater Sci Tech.

19:1003-1011.

Kowalsky, K.W., Marantz, D.R., Smith, M.F.

and Oberkampf, W.L. 1991. In: T.F. Bernecki

(ed.), Thermal Spray Research and Applica-

tions, ASM International, Ohio. pp. 587-592.

Kreye, H. 1997. Proceedings of 4th Confer-

ence HVOF Spraying, German Thermal Spray

Society, HÖllriegelskreuth, pp. 13-21.

Lee, C.H. and Min, K.O. 2000. Effects of heat

treatment on the microstructure and properties

of HVOF-sprayed Ni–Cr–W–Mo–B alloy

coatings, Surf Coat Tech. 132:49-57.

Li, C.J., Sonoya, K., Ji, G.C. and Wang,Y.Y.

1998. Effect of powder type on the relation-

ship between spray parameters and properties

of HVOF sprayed Cr3C

2-NiCr coatings, In:

C Coddet (ed.) Proc. 15th ITSC, Nice, France,

pp 287-292.

Li, C-J., Wang,.Y-Y and Ohmori, A. 2000.

Effect of melting state of spray particles on the

adhesion strength of HVOF of Ni- based alloy

coating, proceed. ITSC, Montreal,QUE,

United States, May 8-11, pp. 791-96.

Lih, W.H., Yang, S.H., Su C.Y., Huang, S.C.,

Hsu, I.C. and Leu, M.S. 2000. Effects of

process parameters on molten particle speed

and surface temperature and the properties of

HVOF CrC/NiCr coatings, Surf Coat Tech,

133-134, 54-60.

Lugscheider, E., Herbst, C. and Zhao, L.

1998. Parameter studies on high-velocity oxy-

fuel spraying of MCrAlY coatings. Surf Coat

Tech. 108-109, 16-23.

Meyer, P.J. 1995. In: A. Ohmori (ed), Proc.

ITSC, Kobe, Japan, pp 217-222.

Miguel, J. M., Guilemany, J. M. and Vizcaino,

S. 2003. Tribological study of NiCrBSi coating

obtained by different processes, Triobology

International, 36:181-87.

Moskowitz, L. N. 1992. Application of

HVOF Thermal Spraying to Solve Corrosion

Problems in the Petroleum Industry, Proceed-

ings of 13th International Thermal Spray

Conference, Florida, USA. 611-618


Nestler, M.C., Hartman, S. and Höhle, H-M,

1994. In: H. Kreye et al. (eds.) Conference

Proc., Kollequium Hochgeschwindig-

keitsflammspritzen, Erding, Germany, pp. 93-104.

Provot, X., Burlet, H., Vardayoulias, M.,

Jeandin, M. and Richard, C. 1993. Compara-

tive studies of microstructures, residual stress

distributions and wear properties for HVOF

and APS WC-Co coatings of Ti6Al

4V, Proc. of

NTSC, Anaheim, USA, pp. 159-166.

Roa, K.V. et al. 1986. Properties and

characterisation of coatings made using jet kote

thermal spray technique, Proceedings of the

11th International Thermal Spray Conference,

Montreal, Canada, pp. 873–882

Russo, L. and Dorfmann, M., 1995. Thermal

Spraying: Current Status and Future Trends.

High Temperature Society of Japan, p. 681.

Sahraoui, T., Fenineche, N.-E., Montavon G.

and Coddet C., 2003. Structure and wear

behaviour of HVOF sprayed Cr3C2–NiCr and

WC–Co coatings. Mater Design. 24: 309-313.

Sahraoui,T., Guessasma, S., Fenineche, N. E.,

Montavon, G. and Coddet, C., 2004. Friction

and wear behaviour prediction of HVOF

coatings and electroplated hard chromium using

neural computation, Mater Lett. 58:654-610.

Sampath, S. and McCune, R., 2000, Thermal-

Spray Processing of Materials, MRS Bulletin,

25(7):12.

Smith, R.W. and Knight, R. 1995. Thermal

spraying I: powder consolidation - from

coating to forming, JOM. 47:32-39.

Sobolev, V.V., Guilemany, J.M. and Nutting, J.

2004, HVOF Spraying, B0655, Maney,

IOM3 p.5.

Sobolev, V.V., Guilemany, J.M., Garmier, J.C.

and Calero, J.A. 1994. Modelling of particle

movement and thermal behaviour during high

velocity oxy-fuel spraying, Surf Coat Tech.

63:181-187.

Stein, K.J., Schorr, B.S. and Marder, A. R.

1999. Erosion of thermal spray MCr–Cr3C2

cermet coatings. Wear 224:153.

Stokes, J. and Looney. L. 2001. HVOF system

definition to maximize the thickness of formed

components, Surf Coat Tech. 148:18-24.

Sturgeon, A.J., et al. 1994. High velocity oxy-

fuel spraying, T I Met Finish. 72:139-140.

Sudaprasert, T., Shipway, P.H. and McCartney,

D.G. 2003, Sliding wear behaviour of HVOF

sprayed WC–Co coatings deposited with both

gas-fuelled and liquid-fuelled systems, Wear,

255:943-949

Tan, J.C., Looney, L. and Hashmi, M.S.J.

1999. Component repair using HVOF

thermal spraying, JMPT. 92-93, 203-208.

Tucker, R.C., Ashari, A.A.. et al. 1998. The

structure property relationship of erosion

resistant thermal spray coatings. Proc. 15th

ITSC, Nice, France, pp. 259-62.

Uusitalo, M.A., Uuoristo, P.M.J. and Mantula,

T.A. 2002. High temperature corrosion of

coatings and boiler steels in reducing chlorine-

containing atmosphere, Surf Coat Tech.

16:275-285.

Voggenreiter, H. 1996. Ph.D., thesis. Hertbert

Verlag, muenchen, Germany.

Wang, B. 1996. Erosion-corrosion of thermal

sprayed coatings in FBC boilers. Wear. 199:24-32.

Wang, B. and Lee, S.W. 2000. Erosion–

corrosion behaviour of HVOF NiAl–Al2O3

intermetallic-ceramic coating. Wear. 239:83-90.

Wang, Y. and Chen, W. 2004. Microstruc-

tures, properties and high-temperature

carburization resistances of HVOF thermal

sprayed NiAl intermetallic-based alloy

coatings. Surf Coat Tech. 183:18-28.

Yamasaki, R., Takeuchi, J., et al. 1995. In: A.

Ohmori (ed.), Proc ITSC, Kobe, Japan, pp.

863-868.

Zhang, D., Harris, S. J. and McCartney, D. G.

2003. Microstructure formation and corrosion

behaviour in HVOF-sprayed Inconel 625

coatings. Mater Sci Eng. A344:45-56.

Zhao, L., Zwick, J. and Lugscheider, E. 2004.

HVOF spraying of Al2O

3-dispersion-strengthened

NiCr powders. Surf Coat Tech. 182:72-77.

Zhao, L. and Lugscheider, E. 2002. High

velocity oxy-fuel spraying of a NiCoCrAlY and

an intermetallic NiAl—TaCr alloy. Surf Coat

Tech. 149:230-35.

Zhao, L., Lugscheider, E., Fischer, A. and

Reimann, A. 2001. Thermal spraying of high

nitrogen duplex austenitic–ferritic steel. Surf

Coat Tech. 141:208-215.

Zhao, L., Maurer, M., Fischer, F., Dicks, R.

and Lugscheider, E. 2004. Influence of spray

parameters on the particle in-flight properties

and the properties of HVOF coating of WC-

CoCr. Wear, 257:1-2, 41-46.

Zhao, W.-M., Wang, Y., Han, T., Wu, K.-Y. and

Xue, J. 2004. Electrochemical evaluation of

corrosion resistance of NiCrBSi coatings deposited

by HVOF. Surf Coat Tech. 183:118-125.

Zhao, W.-M., Wang, Y., Dong, L.-X., Wu, K.-

Y., and Xue, J. 2005. Corrosion mechanism of

NiCrBSi coatings deposited by HVOF. Surf

Coat Tech. 190:293-298.

Zimmermann, S. and Kreye, H., 1996.

Chromium carbide coatings produced with

various HVOF spray systems, Proceedings of

the 9th National Thermal Spray Conference,

ASM International, Materials Park, Ohio. p. 147.


A U T H O R SSteven A. RubergNOAA-Great Lakes EnvironmentalResearch Laboratory

Dwight F. ColemanInstitute for Exploration and University ofRhode Island

Thomas H. JohengenGuy A. MeadowsHans W. Van SumerenUniversity of Michigan

Gregory A. LangNOAA-Great Lakes EnvironmentalResearch Laboratory

Bopaiah A. BiddandaGrand Valley State University


Groundwater Plume Mapping in aSubmerged Sinkhole in Lake Huron

A B S T R A C TA multidisciplinary exploratory project team from the Institute for Exploration,

the Great Lakes Environmental Research Laboratory, Grand Valley State University,and the University of Michigan located and explored a submerged sinkhole in LakeHuron during September 2003. A CTD system and an ultra-short baseline (USBL)acoustic navigational tracking system integrated with an open frame remotely oper-ated vehicle (ROV) provided high-resolution depth, temperature, and conductivitymaps of the sinkhole and plume. Samples were also peristaltically pumped to thesurface from a depth of 92 meters within and outside of the sinkhole plume. A 1-2m thick cloudy layer with a strong hydrogen sulfide odor characterized the watermass close to the plume. Relative to ambient lake water, water samples collectedwithin this layer were characterized by slightly higher (4-7.5 oC) temperatures, veryhigh levels of chloride and conductivity (10-fold) as well as extremely high concen-trations of organic matter (up to 400 mg C/L), sulfate, and phosphorus. Our obser-vations demonstrated the occurrence of unique biogeochemical conditions at thissubmerged sinkhole environment.

sinkhole vents, producing a visible cloudylayer above the lake bottom (Figure 1), werea serendipitous discovery made during a2002 remotely operated vehicle (ROV) sur-vey of the sinkholes. Recharge areas of fresh-water replenishment for the Silurian-Devo-nian aquifers have been documented on landin the Lake Huron basin; these areas are typi-cally sinkholes (Figure 2). In this report, wediscuss the mapping of the Isolated Sink-hole located approximately 10 miles fromshore at a depth of 93 m in the north cen-tral region of the Thunder Bay National Ma-rine Sanctuary during September 2003.

Survey MethodsAll survey operations were conducted on

the 80-foot R/V Laurentian. The Laurentianis jointly operated by NOAA’s Great LakesEnvironmental Research Laboratory and theUniversity of Michigan. A Seabird SBE-19conductivity, temperature, and depth(CTD) recorder and LinkQuest TrackLink1500HA USBL acoustic navigational sys-

tem were integrated with the University ofMichigan Hydrodynamics Laboratory’sBenthos open-frame ROV (Figure 3). Aperistaltic pump was used to pump water tothe deck of the research vessel to collect wa-ter samples for laboratory analyses. Thepump tubing inlet was attached to the ma-nipulator arm and routed through the SBE19 conductivity cell.

InstrumentationThe SBE-19 sensors included an inter-

nal-field glass conductivity cell with plati-num electrodes (range: 0-7 S/m, accuracy:+/- 0.001 S/m), thermistor temperature sen-sor (range: -5 to +35 C, accuracy: +/- 0.01C), and a mechanical strain gauge pressuresensor. The sensor system was attached tothe lower ROV frame, as shown in the right-hand image in Figure 3, in order to collectdata from just above the lake bottom.

Positioning SystemThe LinkQuest TrackLink 1500HA

operates at 31– 43.2 kHz with a 120-150

TI N T R O D U C T I O N he Laurentian Great Lakes were formed about 10,000-12,000 years beforepresent (ybp), and presently contain approxi-mately 19% of the Earth’s surface liquidfreshwater (Beeton, 1984). The Lake Hu-ron Basin is mostly covered with a layer ofglacial till, sand, silt and clay. Underlyingthese sediments are aquifers formed withinPaleozoic (Silurian-Devonian) bedrock.These bedrock aquifers were laid down whenthe shallow seas still spread widely over thecontinental areas approximately 350- 430million ybp. The Silurian-Devonian aqui-fer consists of carbonate, shale, and sand-stone matrix with some evaporite beds, andhas fresh and saline water, which can con-tain varying amounts of sulfates, chloridesand iron. Dissolution of the Silurian-Devo-nian evaporites has produced the major karstfeatures (Olcott, 1992) such as the sinkholesdiscovered during the 2001 acoustic surveyexpedition (Coleman, 2002) conducted bythe Thunder Bay National Marine Sanctu-ary and the Institute for Exploration. The


degree beamwidth and has a slant range ac-curacy of 0.2 m and positioning accuracy of0.25 degrees. The surface unit was fixed-mounted to a mast attached to the vessel.The surface transceiver interrogates the trans-ducer located on the ROV, receives the trans-ducer uplink, and resolves position based onrange and angle of incidence on the multi-element acoustic array. The positioning sys-tem software operates on a desktop com-

puter and uses multiple inputs including thedifferential GPS signal (< 1m accuracy), aKVH fluxgate compass (+/- 0.5 degree ac-curacy) heading signal, and the output ofthe acoustic transceiver (slant range and bear-ing) to resolve underwater vehicle position.All signals from the positioning softwarewere fed into HYPACK, Inc.’s HYPACKsystem, the primary navigational softwareused during this research cruise.

ROVThe University of Michigan’s Re-

motely Operated Vehicle for Educationand Research (M-ROVER) was the un-derwater experimental platform utilizedfor this investigation. M-ROVER is aBenthos Open Frame Sea Rover specifi-cally designed to accommodate a widevariety of instrument packages includingthe Seabird CTD, selective sampling gear,and the LinkQuest Tracklink positioningsystem used in this sortie (Figure 3). Flex-ible tubing (3/8 in. I.D.) was attached tothe ROV tether and run from the peri-staltic pump on the deck of the R/VLaurentian to the M-ROVER located atthe experimental site. The flexible tubewas routed through the vehicle frame andinto the CTD conductivity cell. Using theROV’s three function articulated arm toposition the sampling tube, real time, se-lective sampling was conducted at severallocations throughout the sinkhole. TheSeabird CTD data was transmitted to thesurface through spare conductor wires inthe ROV tether. The M-ROVER isequipped with several autopilots that al-low precise movement of the vehiclethrough the experimental arena. The ROVpropulsion system includes four horizon-tal thrusters and two vertical thrusters ina vertran configuration. The thruster gaincan be precisely controlled to diminishbottom disturbance. Precise navigationthrough the experimental area was accom-plished using M-ROVER’s 675 kHz, highresolution, color, scanning, imaging sonar.Detailed video observations were con-ducted with the ROV’s onboard colorvideo camera and all images were digitallyrecorded.

Chemistry MethodsWater samples from the deep region

of the sinkhole were collected by pump-ing the samples to the surface via a peri-staltic pump directly into acid cleanbottles. Water samples from the ambientwater column (lake water) were collecteddirectly above the sinkhole, at depths of 5m and 25 m, using teflon-coated Niskinbottles. All samples were immediately pro-

FIGURE 2Regions of Karst in Alpena County and in Lake Huron

FIGURE 1Vertical gradient of 1-2m thickness resulting from sinkhole venting into ambient Lake Huron waters (photofrom 2001 survey by ROV Little Hercules).


cessed. For total phosphorous (total P),50 ml of sample was measured into anacid-cleaned 70 ml Pyrex test-tube andthen refrigerated until analyzed. Eachsample was processed in duplicate with theaverage value recorded in Table 1. Forchloride, water was filtered through a 0.2uM Nylon filter and dispensed into apolypropylene test-tube, sealed tightly andstored in a refrigerator until analyzed. To-tal P and chloride concentrations were de-termined using standard automatic colo-rimetric procedures on an Auto AnalyzerII as detailed in Davis and Simmons(1979). Total P was determined follow-ing digestion in an autoclave after addi-tion of potassium persulfate (5% finalconc.) (Menzel and Corwin, 1965). Sul-fate concentrations were determined byion chromatography (APHA, 1992).

Samples for particulate organic carbon(POC) were processed through pre-com-busted (4h at 450oC) Whatman GF/F fil-ters and frozen until analysis. Prior toanalysis filters were thawed and soaked in1.0N HCl and then dried at 80oC for 24h.Samples for dissolved organic carbon(DOC) were taken from the filtrate of thePOC sample, after discarding the first 50ml as a rinse, and poured into acid-cleanedand pre-combusted glass test-tubes andthen frozen until analysis (Cotner et al.,2000). POC concentrations were deter-mined on a Perkin-Elmer Model 2400 el-emental analyzer. DOC concentrationswere determined by high temperaturecombustion on a Shimadzu TOC 5000carbon analyzer.

Survey Results andObservationsVisible Vertical Gradient(the cloudy layer)

An examination of in situ measure-ments using ROV mounted instrumentsand shipboard laboratory analyses of physi-cochemical conditions within the nearbottom cloudy layer (Figure 1), and itscomparison to lake water properties pre-vailing at a comparable depth in the lakeaway from the venting water showed thatthe venting water was characterized by dis-tinctly unique properties (Table 1). Vent-ing water was warmer than ambient lakewater at depth only by a few degrees Cel-sius, but was characterized by 10-foldhigher concentrations of chloride, 100-foldhigher concentrations of sulfate, and 1000-fold higher concentrations of total P. Highchloride concentrations are characteristicof the Silurian-Devonian aquifer (Olcott,1992), and help explain the measured highconductivity. The high sulfate ion concen-trations in the venting water could serveas substrate for sulfate reducers in this en-vironment. Indeed samples from thecloudy layer smelled strongly of hydrogensulfide (H

2S) when brought to the sur-

face—suggesting anaerobic pathways ofcarbon transformations may be occurringin this sinkhole ecosystem. Furthermore,the significantly higher temperature pre-vailing in the venting water is likely to en-hance microbial metabolism (Biddandaand Cotner, 2002), and thereby expeditebiogeochemical cycling of bioactive ele-ments in this near bottom environment.

Phosphorus (P) concentrations in the vent-ing water were 1000-fold higher than the am-bient lake water. Because P is commonly thelimiting nutrient for primary productivity inall of the Laurentian Great Lakes (Schindler,1977; Fahnenstiel et al., 1998), the input of Pinto Lake Huron from sinkhole discharges maybe of significant local importance in terms ofwater quality as well as overall productivity.However, such P entering the lake via sub-merged sinkholes at aphotic depths (as in thepresent study at 93 m), will have to becomemixed into the surface sunlight layer over daysto months before it can influence autotrophicprimary production by phytoplankton. A thor-ough quantification of such nutrient fluxesfrom sinkhole discharges to the Lake Huron Pbudget and its ecological consequences withinthe Great Lakes opens the possibilities for fu-ture investigations.

Coincident with the presence of high con-centrations of dissolved nutrients, the ventingwater was characterized by 5-fold higher con-centrations of DOC and 400-fold higher con-centrations of POC, relative to ambient lakewater. Both DOC and POC are utilized byheterotrophic bacteria, to fuel the microbialfood web in aquatic environments (Cotner andBiddanda, 2002). Such high abundance oforganic matter prevailing in the sinkhole plumeappears to be the result of localized but intensebacterial chemosynthesis and heterotrophicproduction processes occurring at this sub-merged sinkhole ecosystem within the LakeHuron basin (Biddanda et al., in preparation).The combination of high abundance of inor-ganic nutrients and organic matter providesthe context for both aerobic and anaerobic bio-

FIGURE 3M-ROVER with tracking system visible on the starboard stern corner in the image at left and the CTD attachedto the ROV frame in the image at right

TABLE 1A comparison of ambient Lake Huron water and sink-hole vent water (see text for explanations).

Parameter Lake Vent(Units) Water Water

Conductivity (µS/cm) 140 1700

Temperature (oC) 3.5 7.0

Chloride (mg/L) 13 175

Sulfate (mg/L) 16 1457

Total P (mg/L) 0.004 3.230

DOC (mg/L) 2.5 9.8

POC (mg/L) 0.9 405


geochemical processes to occur within the sink-hole environment, which may explain the vis-ibly cloudy nature of the venting plume.

Depth, Temperature, andConductivity Maps

Temporally referenced CTD and acous-tic positioning data, in the Universal Trans-verse Mercator (UTM) format converted toa local reference in meters, were merged tocreate figures 4, 5, and 6. IDL (ResearchSystems, Inc.) was used to plot contouredvisualizations. The vertical and horizontalaxes of each plot are in meters. The tem-perature and conductivity data were filteredand depth-averaged providing observationsat about 1 meter above the floor of the sink-hole. Median filtering, a non-linear tech-nique that applies a sliding window to thedata sequence, was used to eliminate anyspurious noise in the data.

The dimensions of the sinkhole from thedepth map (Figure 4) are approximately 55meters by 40 meters. The depth range from thedeepest point in the sinkhole to the surround-ing lake depth is approximately 3 meters. The

0 20 40 60 80East-West (Meters)

0

20

40

60

80

Nor

th-S

outh

(M

eter

s)

3.5 7.5Temp

4.0

4.0

4.04.5

4.5

5.0

5.0

5.5

5.5


0

20

40

60

80

Nor

th-S

outh

(M

eter

s)

90.5 93.8Depth

91

91

9191

91

91

91

92

9292

92

93

93

93


0

20

40

60

80

Nor

th-S

outh

(M

eter

s)

0 1825Cond20

0

200

200

200

600

600

600

1000

1000

FIGURE 4Depth map of sinkhole vent plume

FIGURE 5Temperature map of sinkhole vent plume

FIGURE 6Conductivity map of sinkhole vent plume


depth map indicates the existence of two deepersections of the sinkhole; one on the northeastrim and a second area separated by a low ridge(approximately 1 meter) about 20 meters tothe south. Both areas appear to be sources ofplume water as observed from the temperatureand conductivity maps (Figures 5 and 6). Cur-rents could contribute to other areas of increasedtemperature and high conductivity that do notappear to coincide with increased depth. Thetemperature map covers a range from 3.9oC to7.5oC. Conductivity map values range from122.6 uS/cm to 1821.2 uS/cm. The high con-ductivity of the plume is attributed to the highlevels of Chloride and Sulfate.

The temperature of the plume water ap-pears to come into thermal equilibrium withthe larger mass of surrounding lake water morequickly than mixing can dilute the highly con-ductive plume waters. Normal temperaturesin the Great Lakes at these depths are typically4oC. The chemistry results (Table 1) and highconductivity levels measured during this pre-liminary investigation appear to indicate thatthe source of the sinkhole plume is the Sil-urian-Devonian aquifer.

Future WorkComplex linkages existing between sur-

face water and groundwater driven by hy-drologic and climatic conditions may influ-ence both the flow rate and composition ofgroundwater venting in karst sinkholes(Gibert et al., 1994). Future plans include acontinued exploration and survey of near-shore and deeper-water sinkholes compar-ing chemistry and microbial parameters.Further investigations will be done to deter-mine the age of groundwater in various sink-holes throughout this karstic system. Thedeployment of instrumentation in the Iso-lated Sinkhole measuring flow, light, dis-solved gasses (e.g., oxygen, hydrogen sulfide),temperature and other parameters will pro-vide insight into seasonal/annual variability.

AcknowledgmentsThe authors acknowledge the important

contributions of Jeff Gray, Manager, Thun-der Bay National Marine Sanctuary, to thesuccess of this preliminary investigation. Wewould like to thank the crew of the R/VLaurentian – Andrew Yagiela, GlennTomkins, Denis Greenwald, and DennisDonahue. This work was funded by NOAA’sNational Marine Sanctuary program and theOffice of Ocean Exploration; and is GreatLakes Environmental Research Laboratorycontribution number 1355.

ReferencesAPHA. 1992. Standard methods for the

examination of water and wastewater. 18th

edition, American Public Health Association.

Beeton, A. M. 1984. The world’s great lakes.

J. Great lakes Res . 10: 106-113.

Coleman, D. F. 2002. Underwater Archaeology

in Thunder Bay National Marine Sanctuary,

Lake Huron. Mar Technol Soc J. 36(3):33-44.

Biddanda, B. A., and J. B. Cotner. 2002. Love

handles in aquatic ecosystems: the role of

dissolved organic carbon drawdown, resuspended

sediments, and terrigenous inputs in the carbon

balance of Lake Michigan. Ecosystems 5:431-445.

Biddanda, B. A., D. Coleman, T. Johengen, S.

Ruberg, G. Meadows, H. vanSumeren, R.

Rediske, and S. Kendall. Exploration of a

submerged sinkhole ecosystem in Lake Huron.

In preparation.

Cotner, J. B., and Biddanda, B. A. 2002. Small

players, large role: Microbial influence on

biogeochemical processes in pelagic aquatic

ecosystems. Ecosystems. 5:105-121.

Cotner, J. B., T. H. Johengen and B. A.

Biddanda. 2000. Intense winter heterotrophic

production stimulated by benthic

resuspension. Limnol Oceanogr. 45:1672-1676.

Davis, C.O. and M.S. Simmons. 1979. Water

Chemistry and Phytoplankton Field and

Laboratory Procedures. Special Report No. 70.

Great Lakes Research Division, University of

Michigan, Ann Arbor, MI.

Fahnenstiel, G. L., A. Krause, M.

McCormick, H. Carrick and C.Schelske.

1998. The structure of planktonic food web in

the St. Lawrence Great Lakes. J Great Lakes

Res. 24:531-554.

Gibert, J., D. Danielpol and J. Stanford. 1994.

Groundwater Ecology. New York: Academic

Press. p. 571.

Menzel, D.W. and N. Corwin. 1965. The

measurement of total phosphorus liberated in

seawater based on the liberation of organically

bound fractions by persulfate oxidation.

Limnol Oceanogr. 10:280-281.

Olcott, P. G. 1992. Ground Water Atlas of the

United States, Iowa, Michigan, Minnesota,

Wisconsin, U.S. Geological Survey HA730-J.

Schindler, D. W. 1977. Evaluation of

phosphorus limitation in lakes. Science.

195:260-262.


A U T H O R SSungHyun NamGuebuem KimKyung-Ryul KimKuh KimSchool of Earth and EnvironmentalSciences/Research Institute of Oceanography,Seoul National University

Lawrence Oh ChengKi-Wan KimHyong OssiOTRONIX Co., Ltd.

Young-Gyu KimAgency for Defense Development,Jinhae, Korea

P A P E R

Application of Real-time Monitoring Buoy Systemsfor Physical and Biogeochemical Parameters in theCoastal Ocean around the Korean Peninsula

A B S T R A C TWe introduce technological achievements while developing real-time ocean moni-

toring buoy systems in the key coastal regions around the Korean peninsula, andhighlight their potential contribution to oceanographic studies in the region. Majorachievements are an integration of physical and biogeochemical sensors, real-timeand two-way communication, sustainable maintenance with stable power supply andmooring design, and the two-way control of sensor and sampling strategies with highsampling rates (as often as every minute). The time-series data from two buoy sys-tems deployed in the key coastal regions are given as examples to show their poten-tial use in studying oceanographic issues, such as major current variations along theeast coast of Korea, wind-driven episodic events including typhoon passages, andfrequent changes due to internal wave passages. The real-time and high-frequencymonitoring of biogeochemical properties of seawater together with physical param-eters could be used for numerous oceanographic studies in the coastal region, i.e.,air-sea gas exchange, harmful dinoflagellate bloom, interaction between physical andbiogeochemical processes.

Recent enabling technologies applied tomonitoring buoys are compared for typicalsystems around the world’s (mostly coastal)oceans (Table 1), though most of them arestill in the stage of being developed or up-graded. The MAREL buoy and Smart buoyare equipped with various new sensors tomeasure biogeochemical properties as wellas physical properties of coastal water belowthe surface at a high frequency in continu-ous and autonomous mode, as operated inthe Iroise Sea, Brest, France and the UKshelf-seas, respectively. Continuous cleaningof the sensor, preventing it from biologicalfouling (Blain et al., 2004), and redundancyof key parameters for data reliability (Millset al., 2002) are the outstanding features ofthese buoys. Such environmental monitor-ing, for biogeochemical parameters as wellas physical parameters, is also available alongwith newly developed controllers e.g. theOASIS (Ocean Acquisition System for In-terdisciplinary Science) controller used inMonterey Bay, U.S. (Chavez et al., 1997).

Chavez et al. (1997) illustrated the capabili-ties of the OASIS controller including flex-ible 28 sensors and two-way telemetry thatallow users to communicate directly withvarious instruments and controllers. Thetwo-way communication is widely appliedto monitoring buoy systems using satelliteor cellular telemetry, e.g. MAREL buoy,OASIS moorings, TABS (Texas AutomatedBuoy System)/SEMB (Surface EnvironmentalMonitoring Buoy), etc. (Table 1).

Individual systems, however, have certainlimitations. Still, some monitoring buoy sys-tems, such as MAREL buoy, Smart buoy,ATLAS buoy etc., are not equipped with waveand current sensors to measure these vital pa-rameters for the physical environment. Onthe other hand, though the monitoring buoysused in the COMPS (Coastal Ocean Moni-toring and Prediction System) program andTABS/SEMB are monitoring water columncurrent and wave parameters as well as me-teorological parameters, they can not provideany biogeochemical information by them-

TINTRODUCTION here has long been recognized a need for lengthy and continuous observationand real-time monitoring of the environ-ment to characterize and understand climateand ocean variability. The coastal area in par-ticular, because of its high spatio-temporalvariability in the environment due to vari-ous factors such as shallow water depth, pres-ence of the coast, enhancement of tidalmotion, terrestrial influences, etc., has beenthe subject of considerable efforts directedtowards the development of a monitoringbuoy in the world’s coastal oceans (Maloneet al., 2000; Chavez et al., 1997). Moreover,recent technical advances in telemetry, powersupply, data capacity, and mooring stability,and development of new sensors have stimu-lated an increasing demand for such moni-toring buoys to observe various coastal areasmore effectively (Chavez et al., 1997) at ahigher temporal resolution and with morefacile communications than before, i.e. real-time and two-way (Frye et al., 1991).


Recent enabling technologies applied toESROB and NMB, and also to other suchocean monitoring buoys in the coastalocean, are not widely documented exceptin a few articles (Table 1) despite the factthat they have a significant role in provid-ing valuable time-series data for marinemeteorological, physical and biogeochemi-cal process studies. In this paper, therefore,we introduce technological achievementswhile developing such monitoring buoysystems in the key coastal regions aroundthe Korean peninsula, and highlight theirpotential contribution to environmentalstudies in the coastal region. Generalprogress in technology and its applicationto scientific purposes with possible utiliza-tion are described in sections 2 and 3, re-spectively. Then, in section 4, a brief sum-mary and conclusion are presented,together with a view of possible future de-velopments of buoy monitoring in thecoastal ocean.

TABLE 1Comparison of enabling technologies applied to typical ocean monitoring buoy systems.

Buoy Documentation Enabling technologies

(Program) Body hull Sensors & Power supply Telemetry EtcName parameters

ESROB & NMB This study 2.4 m diameter Met., physical, Solar panel Cellular CDMA, 1 min. samplingdiscus biogeochemical (25W, 12V) Two-way possible

MAREL buoy Blain et al. (2004) 4m diameter Physio-chemical Solar panel Cellular GSM-France, Sensor calibrationdiscus (30W, 24V) Two-way every three months

Smart buoy Mills et al. (2002) Met., physio- Satellite, One way ESM controllerbiogeochemical Redundancy for

key parameters

OASIS moorings Chavez et al. ~2m diameter Met., physical and Solar panel Cellular, Two-way OASIS controller(1997) donut biochemical (10W, 12V)

(COMPS)USF Weisberg et al. Met., physical GOES satellite,(2002) One-way

TABS/SEMB Guinasso, et al. 0.79 m diameter Met., physical Solar panel Satellite/Cellular,(2001) (10W, 12V) Two-way

(NDBC) Moored http://seaboard. Various (12 m discus Met. physical Solar panel ARGOS satellite,buoy) ndbc.noaa.gov/ – 1.5 m colos) (one-way)

Coastal buoy SYSTEA Brochure Physio- Solar panel (20W) GSM wireless, Nutrient Probebiogeochemical Two way Analyzer, SMS

CMBS METOCEAN Flexible Flexible, One- or Flexible controllerBrochure Two-way

ATLAS Milburn et al. 2.3 m diameter Met., T/S Battery pack ARGOS satellite,(TAO/TRITON) (1996) (84 D-cell alkaline) One-way

selves (Weisberg et al., 2002; Norman et al.,2001). Moreover, data from many monitor-ing buoy systems are still transferred in one-way communication using an ARGOS satel-lite or UHF/VHF radio telemetry.Cost-effectiveness is another critical limita-tion for some monitoring buoy systems.

To address these limitations, we have de-veloped (in partnership with other agencies inKorea) new real-time monitoring buoy systems(Kim et al., 1999; Nam et al., 2003) in the keycoastal areas around the Korean peninsula (Fig-ure 1), applying enabling technologies for en-vironmental studies. These include a stablepower supply in order to equip more sensors,two-way communication and its utilization formore effective environmental monitoring, sus-tainable maintenance (biological fouling, warn-ing message for breaking away from position,etc.), and sensor and sampling strategies (high-frequency sampling for short-period internalwave monitoring, calibration and assessmentof various biogeochemical sensors for examin-

ing environmental issues such as harmful al-gae bloom). Among the above buoy systems(Figure 1), two of them (buoys S and J) areidentified as the East Sea Real-time Ocean Buoy(ESROB) and Narodo Monitoring Buoy(NMB). The ESROB is located about 9 kmoff the port Donghae, in the middle of the eastcoast of Korea, at a depth of 130 m. The area isknown for high spatial and temporal variabil-ity in both the current and the water proper-ties, mainly due to the convergence of coldwater and warm water (Kim et al., 1983),which are mostly carried by the North KoreanCold Current (NKCC) and the East KoreanWarm Current (EKWC), respectively. TheNMB operates in the region near the southcoast of Korea at a depth of 10-15 m (depend-ing on tide) where the Red Tide frequentlyoccurs. Besides the ESROB and NMB, thereare several other monitoring buoys oper-ated in the key areas around the Koreanpeninsula as shown in Figure 1 (‘B’, ‘C’,‘D’, ‘H’, ‘I’, ‘V’, and ‘W’).


2. Instrumentation2.1. Buoy Specifications

The ESROB and the NMB are 2.4 m indiameter (1.8 m in the early stages forESROB) and 5.67 m (4.1 m above sea level)in height. They are divided into three parts:the ‘air’ part above sea level, the ‘seawater’part below it, and the ‘inside’ part inside thehull (Figure 2). In the ‘air part’, four solarpanels for power supply, navigation light(beacon) and radar reflector for safety, andantennae for DGPS and telemetry, are

placed with various meteorological sensors.All the physical and biogeochemical oceano-graphic sensors (Figure 2, Table 2) are at-tached in the ‘seawater’ part, together withinductive cables for CTD (Conductivity-Temperature-Depth) data transfer, a chainand anchor for stable mooring, and an acous-tic releaser for safe recovery. The main con-troller, which gathers all the data and sup-plies the system power, is placed in the ‘insidepart’ with specific sensors for checking theinternal status of the buoy.

2.2. Installation ofEnvironmental Sensors

Every 10 minutes, meteorological sen-sors measure wind speed, gusts (maximumwind speed for 10 minutes) and direction,air temperature and pressure and relative

humidity. Detailed information, e.g., manu-facturer, measured item, accuracy and otherspecifications of each sensor, is listed in Table2. Surface waves are parameterized every 10minutes from 1,024 accelerometer data read-ings measured at 2 Hz, using both the zero-up cross method1 and the Fast Fourier trans-form (FFT) method. The wave parametersinclude maximum and significant waveheights, significant and FFT wave periods,etc. The sensor for the surface waves is placedinside the hull below an electronic canister(Figure 2).

Three-dimensional subsurface currentsat 26 vertical levels are measured every 10or 1 minute(s) by mounting a 300 kHzADCP (acoustic Doppler current profiler)below the bottom frame, in looking downform (Figure 2) with a relatively low powerconsumption of 1916 mAh/day (Table 2).It is a commercial Monitor-type WorkhorseADCP, manufactured by RD Instrumentsand offers 1) extreme accuracy (Table 2) andreliability, 2) versatility, 3) high data resolu-tion and minimal power consumption and4) a four-beam solution (http://www.adcp.com/). The east-west, north-south, vertical, and error currents are ac-quired at all the depth levels at each timeinterval (1 or 10 minutes flexible), with echointensities of four beams and ADCP atti-tudes, i.e. pitch, heading, and roll, for as-sessing the data quality. Besides these data,water temperature measured at the ADCPnear the surface is used as an indicator ofsea-surface temperature.

Physical properties (temperature andsalinity) of seawater at several depth levelsare measured every 10 or 1 minute(s), byattaching SBE37 CTDs (Figure 2) to theinductive modem (mooring) line in the ‘sea-water’ part. The accuracies of conductivityand temperature are 0.0003 S/m and 0.002,respectively (Table 2), which are sufficientto apply this sensor for oceanographic pur-poses in this shallow water region. Data mea-sured from the CTD are transferred to themain controller through the inductive mo-dem cable (mooring line). Sensor calibra-tion should be performed (typically everysix months) together with a battery changefor long-term operation.

Footnote1In this method, maximum wave height and sig-

nificant wave height are determined by the maxi-

mum and one-third or one-fourth values of the

wave heights in order of magnitude, respectively.

FIGURE 1Positions of real-time coastal ocean monitoring buoy systems around the Korean peninsula. The ‘buoy S’ and‘buoy J’ denote ESROB (East Sea Real-time Ocean Buoy) and NMB (Narodo Monitoring Buoy), respectively.Similar monitoring buoys are also operated at positions ‘B’, ‘C’, ‘D’, ‘H’, ‘I’, ‘V’, and ‘W’.


TABLE 2Typical specifications of the devices the ESROB and NMB equip.

Device Manufacturer Measured Accuracy Sample Current Output Sensor OutputItems Rate (mA h/day) (byte/day) Location Type

300kHz RD 3D Current, echo intensity, ±0.5% of 1 min.or 1916 585 Near surface RS232w/h ADCP Instrument ADCP heading, pitch, roll, the velocity 10 min.selectable (26 bins)

and temperature

CTD (SBE37) SBE conductivity, temperature, 0.0003S/m, 1 min.or 45 167/set 5 - 124 m RS485pressure 0.002°C, 0.15% 10 min.selectable

Wave sensor OTRONIX wave height, wave period ±1% 10 min. 1040 4320 0M Inside RS232

Met- Wind RM-Young wind speed, wind gust, ±0.3m/s, ±3° 10 min. 16 2016 2M Top voltagesuite wind direction Air Vaisala air temperature, air ±1%°C, ±3hPa, 10 min. 1.6 2592 2M Top voltage

pressure, relative humidity ±1%RH

Compass NAVICO buoy heading ±3° 10 min. 2 576 0M Inside RS232

DGPS JRC latitude, longitude 1.5 m (average) 10 min. 140 6912 2M Top RS232

Controller OTRONIX telemetry, battery, internal 1320 6768 0M Inside RS232temperature, internalhumidity

EcoLAB Enviro Tech. nutrient 1-3 % 3 hours 8.8 1 m

METS CAPSUMS methane 10 min. 64 1 m RS232

Glowtracka bioluminescence 10 min. 16 1 m RS232

Fluorometer SEAPOINTs chlorophyll a 10 min. 6 1 m

We used a submersible multi-channelanalyzer (Eco-LAB) for nutrient measure-ment (Figure 2, Table 2). Eco-LAB uses aconventional colorimetric method employ-ing a fluorometric detector, which is a well-established wet chemistry technique. Eco-LAB takes samples by withdrawing thesyringe plunger at the inlet port and a reac-tion is achieved by moving the valve andadding reagents to a sample by retractingthe syringe plunger at each appropriate port.Samples and reagents can then be mixed,color develops as the reaction progresses andthe sample is injected into the detector formeasurement. For nitrate measurement, anactivated cadmium column is used for re-duction. All these procedures are automati-cally operated by Eco-LAB every 3 hours(Table 2). In the laboratory, a calibrationcheck should be periodically performed. Adetailed chemical procedure can be obtainedfrom the Eco-LAB manufacturer.

Chlorophyll a is measured using a Fluo-rometer manufactured by Seapoint (Figure 2,Table 2). This fluorometer is a high-perfor-mance and low-power (6 mAh/day) instrumentfor in situ measurements of chlorophyll a. This

FIGURE 2Design images of the real-time ocean monitoring buoy (deployed) and various types of sensors equipped in thebuoy system. Below are the time lines of ESROB and NMB operations with the instrument attachments num-bered from (1) to (4).


sensor is operated with a pump and usesmodulated blue LED lamps and a blue exci-tation filter to excite chlorophyll a. The fluo-rescent light emitted by the chlorophyll a passesthrough a red emission filter and is detectedby a silicon photodiode.

Glowtracka, which is a kind of precisionfluorometer especially stimulating biolumines-cent organisms, is utilized for dinoflagellatedetection. Some dinoflagellates themselves ex-hibit a large range of luminous output, de-pendent on their species. Thus, Glowtrackamonitors the visible emissions from materialsuspended in a liquid using a photodetectoras it flows down a pipe. We used one sub-mersible pump to feed water through boththe Fluorometer and Glowtracka.

Methane was measured by using anunderwater methane sensor (METS),which is available from CAPSUM Tech-nologies (Figure 2, Table 2). The principleof measurement is that the hydrocarbonmolecules diffuse out of the liquid througha special silicone membrane into the de-tector chamber. The adsorption of hydro-carbons on the active layer leads to elec-tron exchange with oxygen and thus tomodification of the resistance, which istransduced electronically into a voltage.The membrane is made up of silicon, witha thickness of 10 mm. The detection limitis 20 nmol/L, with a reaction time of 3 to30 minutes.

2.3. Real-time Monitoring ofBuoy Status and Maintenance

In addition to meteorological andoceanographic data, the buoy also incorpo-rates operational status data for mainte-nance, on a real-time and two-way basis.We can call to the buoy system as requiredto check the real-time status of the systembattery level, communication antenna level,internal temperature and humidity insidethe hull (separately from the air tempera-ture and humidity), the buoy position inlatitude and longitude and motion in termsof speed and direction (calculated from theposition changes during 10 minutes), andthe attitude of the buoy hull (heading, pitch,and roll). If necessary, we can provide im-mediate actions for the buoy system re-

motely with a personal cellular phone us-ing two-way telemetry or by going directlyusing a small boat. Real-time monitoringof these status data is important for sustain-able buoy operation under severe or unex-pected environmental conditions or humanactivity. A warning message is sent to userswhen the buoy goes outside a specific user-defined area by checking the buoy statusevery ten minutes.

Figure 3 shows a time series of (a) buoyspeed (in knots), (b) internal temperature,(c) battery, (d) pitch and roll, and (e) head-ing of the buoy for May 2004. The speed,pitch, and roll of the buoy are typicallywithin 1 knot and ±5° except for periodsof strong wind and wave conditions. Formost periods of deployment, the telemetryantenna and battery are maintained at highlevels, independently of the wind, wave,and solar conditions. The heading of thebuoy rotates mostly in a clockwise direc-tion at a frequency near the local inertialperiod of 19.6 hours. To prevent the moor-ing line from twisting due to the rotationof the buoy hull, several swivels are used inbuoy mooring.

Currently we cannot monitor the bio-fouling status of sensors in real-time butdivers manually clean the sensors. Normallysome instruments such as ADCP transduc-ers were painted with anti-fouling chemi-cals and equipped with a specific housing toprevent bio-fouling. Due to the sensor cali-bration drift, meteorological sensors are cali-brated periodically in the OTRONIX labo-ratory and SBE 37 CTDs are replaced everysix months.

2.4. Power Supply andSystem Power Consumption

We have improved the power supply sys-tem of the buoy since the first deploymentof the system. The original 4 solar panels of15 W were replaced by 25 W units with anincrease of battery capacity to 135 Ah for anefficiency of 75 %. Currently the total powerconsumption of the buoy system is about6.4 Ah/day, so that the buoy can operatewithout a power supply for 21 days (135/6.4). Here, the total power consumption ofthe buoy system is calculated by summingthe power consumptions of all the individualdevices both for working and idle modes.

FIGURE 3Time series of (a) buoy speed (in knots), (b) internal temperature, (c) battery, (d) pitch and roll, and (e) headingof the buoy for May 2004. Here, the buoy headings are in degrees clockwise from the north. All the data aretransferred to users in a real-time and two-way sense to check the buoy status and send commands.


For example, the power consumption of theCDMA (Code Division Multiple Access)controller is 0.804 Ah/day, which is a sum-mation of the power consumption duringdata receive mode (0.12 Ah/day=3 min./ 60min. × 0.1 A × 24 h/day) and that duringidle mode (0.684 Ah/day=57 min. / 60 min.× 0.03 A × 24 h/day). Increased power sup-ply capacity and decreased power consump-tion, both improved from the initial deploy-ment of the buoy system in 1999, are vitalfactors for uninterrupted buoy operation.

2.5. Data Flow and CellularPhone Communication

The main controller gathers all the databefore sending to the laboratory. Meteoro-logical and ADCP data are sent to the maincontroller through a ZENO (Figure 4). TheCTD, wave, and biogeochemistry data donot go through the ZENO but directly tothe controller as shown in the data flowchart. The data collected in the main con-troller are transmitted to the database serverat OTRONIX co. via the telecom networkusing cellular phone (CDMA) communi-cation and then transferred to the data server

in the laboratory via FTP. The data storedin the database server are again transferredto the Web server and provided to publicusers in real-time through the Web sites(http://www.otronix.com, or http://eastsea.snu.ac.kr) or via FTP.

The CDMA communication has manyadvantages, e.g. fast data transmission, highcost-effectiveness, low power consumption,and messaging service with a standard cel-lular phone, when compared to other telem-etry techniques such as Orbcomm,Globalstar, and Inmarsat. The maximumtransmission rate of the CDMA system is153.6 kbps which is considerably faster thanOrbcomm (4.8 kbps), Globalstar (9.6 kbps),and Inmarsat-C (4.8 kbps). Moreover, thetime for linking data from the terminal unitin the buoy system to the Ground Earth Sta-tion is only a few seconds without data lossin the case of the CDMA system while ittakes 3-5 minutes (often 30 minutes) for theOrbcomm system, and 5-10 minutes for theInmarsat-C system. The Inmarsat-C systemusually causes data loss for data exceeding abuffer size limit of a few thousand bytes. Ap-plying the CDMA system, the data from

the buoy are transmitted at a rate of 19200bps (not a maximum rate) and transferredinto the OTRONIX server in 10 secondsregardless of the operational modes.

Besides the data rate, the CDMA sys-tem has high cost-effectiveness comparedto other telemetry systems. For example,typical data charges to transmit the samesize of data (80 bytes) for a month are about$20, $100, and $150 for the CDMA,Orbcomm, and Inmarsat-C systems, re-spectively. Terminal units are generallymore economical for CDMA system thanfor the other systems.

The power consumption of the termi-nal units for the CDMA system (BellwaveBSM-850) and for the Orbcomm system(Stellar ST-2500) are only 1.2 W (100 mA,12 V) in data receive mode while those forthe Globalstar system (GSP-1620) and forthe Inmarsat-C system (Trimble TNL-7001) are 2.4 and 12 W in data receivemode with a maximum of 5.4 and 110 W,respectively.

Moreover, the CDMA system pro-vides the short message service (SMS)that enables users to communicate theirown messages in a real-time and two-waysense. This SMS is applied to control thebuoy monitoring in a two-way process.For example, if a user wants to sampleADCP and CTD data at 1-minute in-tervals for a specific period, he/she canchange the sampling time interval from10 minutes (default sampling interval) to1 minute for that period very conve-niently by sending a specified short com-mand to the buoy system using a per-sonal cellular phone without any specifictools. In addition to the change of sam-pling time interval, various actions suchas specifying the monitoring area (thebuoy sends a warning ‘SOS’ alarm mes-sage when it breaks away from this area),switching on and off sensors, checkingstatus etc. are available by sending simpleSMS commands to the buoy. All com-mands are received by the buoy withinseconds and the buoy sends responsemessages automatically for the user toknow if the commands worked correctlyin the buoy system.

FIGURE 4Flow chart for monitoring buoy data. The data is gathered into the main controller and transmitted to the dataprovider (OTRONIX) through the telecom company. The client (SNU) as well as the data provider can control thesampling mode, specific sensor on/off, etc., with a standard cellular phone through the SMS service of thetelecom company.


3. Application toOceanographic Studies

The ESROB and NMB are operatingto monitor physical and biogeochemicalparameters in the key coastal region aroundthe Korean peninsula, though the data ob-tained from the buoys are also used for vari-ous other applications, beyond their mainoperational purpose. In this section we high-light the potential of such time-series dataand effective monitoring strategies to obtainsuitable data for scientific purposes.

3.1. Utilization of Meteorologicaland Physical Parameters in the‘S’ Region

The influence of major current systems,EKWC and NKCC, along the east coastof Korea was noticed from the ADCP dataof ESROB. They showed dominant south-ward or southeastward currents at all depthlevels (especially at lower depth levels)throughout the entire observational periodfrom 1999 to 2004. The physical proper-ties of seawater below the thermocline forthe period of southward or southeastwardcurrents are mostly similar to those of theNorth Korean Cold Water (NKCW), basedon Kim and Kim’s (1983) definition (tem-perature of 1-6 °C and salinity below34.00). In spite of the predominance ofNKCC in the coastal region (related to theoffshore movement of the EKWC path) formost of the observational period, north-ward or northwestward currents sometimesoccurred near the surface, as in March,April and July 2001 and February to May,August, and December 2003. The sea sur-face temperature images in those periodsindicate that the EKWC passed furthernorthward close to the coast. Figure 5 is astick diagram of (low-pass filtered with half-power at 30 hours) currents at 5 depth lev-els (10, 30, 50, 70, and 90 m), observedfrom ESROB in 2001. Northward ornorthwestward currents near the surface inApril and July are noticed in contrast tosouth or southeastward currents at all thedepth levels in May and June, 2001. Mea-surements of water column current andphysical properties of seawater are of par-ticular importance for addressing the two

current system in the ‘S’ region. Long-termmonitoring of the current and physicalproperties of seawater in the region can con-tribute to the local climate relevant to thecurrent systems.

Wind-driven currents are frequentlyobserved from ESROB, and the most strik-ing example can be shown during the pas-sage of typhoon ‘Maemi’ in September,2003. It recorded a maximum wind gustof 25 m/s (10-minute average speed of 20m/s) and a minimum air pressure of 980hPa when the eye of ‘Maemi’ passed by nearUljin, Korea at 03:00 on September 13,2003 (Figure 6a). Along with the wind di-rections of 0-90° (northeasterly) and 270-360° (northwesterly) before and after thepassage of the eye respectively, this indi-cates strong northerly winds at that time.Also noticeable are currents near the sur-face, of which the speeds reached up toabout 0.1 m/s in the southward directionat 13:00 (10 hours after the passage of‘Maemi’) (Figure 6b). The mixed layer(characterized by high temperature and lowsalinity) thickness, which was accompaniedby a strong southward current, graduallyincreased from 20 m to 40 m over 10 hours.A simple two-layer model for the response

to an impulsive alongshore wind over auniformly sloping bottom developed byCsanady (1984) showed reasonable esti-mates of alongshore and offshore currentsand interface displacement in comparisonto data from the area under the conditionsof the passage of ‘Maemi’ for 10 hours(Nam et al., 2004a). This model result alsoconfirmed the downwelling type responseto the southeastward wind stress(downwelling favorable wind conditions)in the region, reproducing a strongbaroclinic jet at the upper layer in the along-shore direction.

The passage of a typhoon also causesproblems in buoy operations. Since 1999,we lost the buoy mooring systems sev-eral times due to the extreme environ-ments during the passage of the typhoons(Figure 6a) named Bart (1999), Saomi(2000), and Rusa (2002). According toa witness from a local fishery, many float-ing materials including big trees movedsouthward very fast in the coastal areaduring the typhoon periods. The ESEOBalso measures the strong southward cur-rents of about 1.0 m/s (abnormally largevalues in the ‘S’ region) with extremewind and wave conditions during the

FIGURE 5Stick diagram of (low-pass filtered) current data obtained from the ESROB. Shown are the data at depths of 10,30, 50, 70, and 90 m from April 1 to September 17, 2001. Strong current variability in the ‘S’ region is due to theeffect of the major current systems - EKWC and NKCC - as well as various coastal processes. Monitoring ofcurrents at all the depth levels in the region is particularly important for addressing the two current systems withthe simultaneous monitoring of water properties.


‘Maemi’ period. One possible explana-tion of the ESROB lost in September of1999, 2000, 2002 is that the hull ofESROB was damaged by floating objects,high wind and waves, etc. and submergedlosing its buoyancy (Figure 6c). When itloses its buoyancy lower than 500 kg(usually larger than 3000 kg) it goesdown below the surface for a maximumcurrent of 1.5 m/s as shown in Figure 6c.

In the ‘S’ region, physical oceanographicparameters are highly variable due to the twocurrent systems (NKCC and EKWC) andtheir relative influence in the coastal area,making a characteristic front near the region.In addition to the current system, the vari-

ous coastal processes, i.e. wind-driven cur-rents, internal waves, etc. can cause com-plex variability of the coastal water as ob-served from ESROB in the region.Moreover, the mountainous landmass on theeastern part of the Korean peninsula mayaffect the peculiar variability of meteorologi-cal parameters in the region, which is causedby the Asian monsoon, the passages of me-soscale weather systems, including typhoons,and diurnal sea/land breezes, etc. (Nam etal., 2004b). Likewise, the real-time moni-toring data for meteorological and physicalparameters in the ‘S’ region has particularsignificance for marine meteorological andphysical oceanographic studies.

3.2. Utilization of Meteorological,Physical and BiogeochemicalParameters in the ‘J’ Region

The methane data obtained in the ‘J’ re-gion using the NMB were utilized to exam-ine the air-sea gas exchange study (Hahm etal., 2005). Careful comparison among thetime-series data of water level, column cur-rent, surface wind, and methane, suggestedthat the sea-to-air transfer of methane is en-hanced greatly during the spring tide due tothe increased gas transfer velocity and verti-cal methane transport from the bottom wa-ter to the surface layer. They suggest that thesea-to-air transfer of gases is controlled notonly by episodic wind events but also by regu-lar tidal turbulence in the coastal ocean. Thisstudy shows one good example for the po-tential usage of continuous monitoring dataof meteorological, physical, and biogeochemi-cal parameters in the coastal ocean. By simul-taneously monitoring both physical and bio-geochemical parameters in the region, somecauses for the biogeochemical changes can beexamined by comparing them in terms ofphysical parameters. Moreover, the interac-tions between physical and biogeochemicalprocesses can be investigated by analyzing thetime series data of both parameters.

Monitoring of the biogeochemical pa-rameters is of particular importance in the‘J’ region due to the red-tide outbreak. It isknown that harmful dinoflagellate bloomshave occurred every year in the ‘J’ regionsince the first blooms were recorded in 1982,while the cause has remained poorly under-stood (Kim et al., 2005; Lee et al., 2005).Because the NMB provides time-series datafor dinoflagellates, NO

3, NO

2, Si(OH)

4,

PO4 and chlorophyll a, as well as methane

(Figure 7), careful comparison of dissolvedinorganic nitrogen and phosphorus maydeduce new possible causes for harmful di-noflagellate blooms as suggested by Kim etal. (2005) and also by Lee et al. (2005), alongwith supplementary data obtained by con-ducting adaptive sampling through a shipsurvey in the region.

Our laboratory data analysis for nutrientsobtained from time-series shipboard samplingsupport the view that the nutrient data pro-duced with the sensor (Eco-LAB) were reli-

FIGURE 6(a) Path of the eye of typhoons which passed by the region of ESROB in September 1999 (Bart), 2000 (Saomai),2002 (Rusa) and 2003 (Maemi), (b) time-series data of wind direction (in degrees clockwise from the north)and speed, air pressure, wave height, and east-west and north-south current measured from the ESROB duringthe passage of Maemi 2003 (Nam et al., 2004a), (c) mooring designs under the strong surface currents whichseem to have happened during the passage of the typhoon in the region. Four cases are compared in (c) wherethe maximum currents are 0.8, 1.5, 1.5, and 1.5 m/s and surface buoyancies are 500, 500, 200, and 100 kgfrom left to right.


able, suggesting minor problems with datadrifting and calibration during the measure-ment period. However, we raise two majorconcerns in utilizing this sensor for a longerperiod: firstly the reagents and standardscan go bad in seawater at its normal tem-perature, and secondly a significantbiofouling effect can damage the sensor.Thus, much more effort toward the im-provement of this sensor is necessary be-fore it can be used on widespread basis. Allother sensors were run without any suchproblems during the measurement period,promising easy use of them in real-timemonitoring of biogeochemical parametersin coastal seawater.

3.3. Utilization of Two-way ControlTechnology and Strategy forEffective Sampling

Novel aspects of the buoy system arethe two-way telemetry and the high-fre-quency sampling (as often as everyminute), which enable us to monitor thecharacteristics of phenomena having pe-riods of a few, or a few tens of minutes,i.e. short-period (near-buoyancy fre-quency) internal waves. In the ‘S’ region,highly nonlinear internal solitons wereobserved using data-logging instrumentsfor water temperature and currents(moored at a depth of about 100 m forthree days) during the field experiment in

May 1999, for the first time in this region(Kim et al., 2001). They have vertical dis-placements of up to 26 m and periods ofminutes or a few tens of minutes. In or-der to monitor such short-period internalwave characteristics continuously since2003, we regularly reduced the samplingtime interval to 1 minute for several daysusing the two-way telemetry. By analyz-ing CTD and ADCP data during each sev-eral-day period of 1 minute sampling, wecharacterized local internal wave proper-ties in the ‘S’ region (Nam et al., 2004c).In this coastal region, we observed thepassage of near-inertial internal waves,semi-diurnal internal tides as well as near-buoyancy internal solitons. High-fre-quency sampling of current and watertemperature in the ‘S’ region is able toprovide valuable time series data for in-ternal wave studies in the coastal regionunder various stratifications.

The ‘SOS’ warning messages are anotherexample of two-way control. On December18, 2003, ESROB was separated from themooring line due to an unexpected accident(abrasion of a stainless steel shackle connectedto the mooring line) and drifted outside thespecified monitoring area (in the ‘S’ region).As soon as the accident happened, it sent‘SOS’ messages to us every 10 minutes as itmoved toward the open sea (northeastward).Contingency plans were immediately imple-mented to recover the bouy by monitoringthe position, speed and direction of the bouyevery 10 minutes with individual cellularphones in real-time. As a result of real-timetracking, it was successfully found and recov-ered by a nearby navy vessel about 100 kmaway from its original position, as it movedvery fast (~1m/s) due to the influence ofEKWC offshore.

Two-way control could also be usedfor an effective sampling strategy of sev-eral biogeochemical properties. Figure 7shows abrupt changes in dinoflagellate at20:30 13 August and 01:00 14 August,and PO

4 at 06:00 14 August in 2003. At

those times, no significant changes are no-ticeable in other biogeochemical param-eters. Comparing several parameters withon- and off-specific sensors with a cellu-

FIGURE 7Time-series data of (a) Chlorophyll a, Dinoflagellate, (b) Methane, (c) NO3, NO2, (d) PO4, and Si(OH)4 measuredfrom the NMB in August, 2003. Abrupt changes in Dinoflagellate at 20:30 13 August and 01:00 14 August, andPO4 at 06:00 14 August were recorded in 2003, owing to this high sampling rate of biogeochemical sensors.


lar phone we can conduct various test ex-periments without directly going to thesea when the NMB starts to record sig-nificant changes in some specific param-eters relevant to the harmful dinoflagel-late bloom. In addition, comparisons withthe physical parameters, i.e., local wind,tidal current, air-sea heat flux, etc. couldhelp to explain various biogeochemicalprocesses in the coastal region.

4. Summary and PotentialFuture Developments

Owing to recent technical advances intelemetry, power supply, data capacity, andmooring stability and development of newsensors, real-time ocean buoys provide thepossibility of monitoring various coastalareas more effectively at a higher tempo-ral resolution (sampling every minute)with more facile (two-way) communica-tion than before. General progress in suchtechnology provides buoy operations witha stable power supply, the use of two-waycommunication for effective environmen-tal monitoring, sustainable maintenance,sensor and sampling strategy, and calibra-tion/assessment of biogeochemical sen-sors. The continuous time-series data ob-tained from the buoy systems in the keycoastal regions could track major currents(EKWC and NKCC) along the east coastof Korea, wind-driven episodic events (in-cluding typhoon passage), passage of in-ternal waves of near-inertial and higher fre-quencies, and sea-surface wind and wavevariability near the coast. They could alsobe used for air-sea gas exchange and harm-ful dinoflagellate bloom studies in thecoastal region.

Development of real-time ocean moni-toring buoys and their application in thecoastal ocean around the Korean penin-sula provide implications for future direc-tions of buoy monitoring. In order to in-crease payload capacity allowingadditional sensors to be incorporated,power supply and data storage need to beenhanced to a maximum. With regards totwo-way telemetry, communication utiliz-ing larger quantities of data, commands,

and responses needs to be pursued in thefuture. Other efforts toward the improve-ment of mooring design would open thepossibility of integrating additional sen-sors for the seafloor and the earth below,as suggested by Detrick et al. (2000), com-bining the advantages of a subsurfacemooring with the capacities of a surfacemooring.

AcknowledgementsWe are grateful to many individuals

who participated in the early stage ofESROB development, and offer specialthanks to Mike Kelly, Shaun Bolger, Dong-Joo Joung and Sang-Uk Lee for English cor-rections and academic and technologicalhelp. This work was in part supported bythe Ministry of Science and Technology,and by the Ministry of the Environment,Korea, through the National ResearchLaboratory (NRL) program (2000-N-NL-01-C-012) and the Echo-technopia 21program (121-041-033). This work is alsosupported by the Agency for DefenseDevelopment, Korea through the Under-water Acoustics Research Center(UD970022AD). The first author waspartly supported by the Ministry of Edu-cation, Korea through the BK21 project.

ReferencesBlain, S., J. Guillou, P. Treguer, P. Woerther,

L. Delauney, E. Follenfant, O. Gontier, M.

Hamon, B. Leilde, A. Masson, C. Tartu, and

R. Vuillenmin. 2004. High frequency

monitoring of the coastal marine environment

using the MAREL buoy. J Environ Monitor.

6(6):569-575.

Chavez, F. P., J. T. Pennington, R. Herlien, H.

Jannasch, G. Thurmond, and G. E. Friederich.

1997. Mooring and drifters for real-time

interdisciplinary oceanography. J Atmos Ocean

Tech. 14(5):1199-1211.

Detrick, R. S., J. A. Collins, and D. E. Frye.

2000. Seafloor to surface to satellite to shore:

Moored buoys offer potential for continuous,

real-time observations anywhere in the ocean.

Oceanus 42(1):12-13.

Frye, D. E., W. B. Owens, and J. R. Valdes.

1991. Ocean data telemetry: New methods for

real-time ocean observation. Oceanus 34:46-52.

Guinasso, N. L. Jr., J. Yip, R. O. Reid,

L. C. Bender, III, M. K. Howard, L. L. Lee III,

J. N. Walpert, D. A. Brooks, R. D. Hetland,

and R. D. Martin. 2001. Observing and

forecasting coastal currents: Texas Automated

Buoy System (TABS). In: OCEANS 2001

MTS/IEEE Proceedings: 1318-1322,

Washington DC: Marine Technology Society.

Hahm, D. S., G. Kim, Y. W. Lee, S. H. Nam,

K. R. Kim, and K. Kim. 2005. Tidal influence

on the sea-to-air transfer of CH4 in the coastal

ocean. Tellus B (in revision).

Kim, G., D. W. Hwang, D. J. Joung, and Y.

W. Lee. 2005. Tracing excess nitrogen causing

the outbreak of dinoflagellate red tides in the

South Sea of Korea using radium isotopes.

Marine Chemistry (under review).

Kim, K., Y. G. Kim, K.-W. Kim, and H. Ossi.

1999. Real-time ocean buoy off the east coast

of Korea. In: Proceedings of International

Symposium on the progress in coastal

engineering and oceanography, 1-6, Seoul, Korea.

Lee, Y. W., G. Kim, D. J. Joung, K. R. Kim,

and K. Kim. 2005. The real-time monitoring

of red-tide outbreak under conditions of

depleted inorganic nutrients. Marine Ecology

Progress Series. (under review).

Malone, T. C., and M. Cole. 2000. Toward a

global scale coastal ocean observing system.

Oceanography 13:7-11.

Milburn, H., P. D. McLain, and C. Meinig.

1996. ATLAS buoy-Reengineered for the next

decade. In: OCEANS 1996 MTS/IEEE

Proceedings, 1318-1322, Washington DC:

Marine Technology Society.


Mills, D. K., R. W. P. M. Laane, J. M. Rees,

M. Rutgers van der Loeff, J. M. Suylen, D. J.

Pearce, D. B. Sivyer, C. Heins, K. Platt, and

M. Rawlinson. 2002. SmartBuoy: A marine

environmental monitoring buoy with a difference.

In: 3rd EuroGOOS Conference Proceedings.

Nam, S. H., J.-Y. Yun, and K. Kim. 2004a.

Observation on the coastal ocean response to

typhoon MAEMI at the East Sea Real-time

Ocean Buoy. The Sea: Journal of Korean

Society of Oceanography 9(3):111-118.

Nam, S. H., Y. H. Kim, K.-A. Park, and K.

Kim. 2004b. Spatio-temporal variability in sea

surface wind stress near and off the east coast

of Korea. Acta Oceanol Sin. 24(1):107-114.

Nam, S. H., B. W. Han, Y. H. Kim, J.-Y. Yun,

K. Kim, S.-I. Kim, and Y.-G. Kim. 2004c.

Internal wave characteristics and their effects

on acoustic transmission in the coastal ocean

near Donghae city. In: Proceedings of 5th

seminar on naval weapon system development:

127-134, Agency for Defense Development,

Jinhae, Korea. (in Korean).

Nam, S. H., K.-W. Kim, H. R. Kim, C. B.

Cho, S. J. Lyu, Y. G. Kim, and K. Kim. 2003.

Development of ESROB (East Sea Real-time

Ocean Buoy), In: Scientific Report series from

2002 MONITOR workshop, North Pacific

Marine Science Organization, Sidney, BC,

Canada.

Weisberg, R., R. Cole, R. He, J. Donovan, M.

Luther, C. Merz, J. Walsh, and V. Subramanian.

2002. A coastal ocean observing system and

modeling program for the West Florida Shelf.

In: OCEANS 2002 MTS/IEEE Proceedings,

Washington DC: Marine Technology Society.


A U T H O RDavid J. CseppAuke Bay LaboratoryAlaska Fisheries Science CenterNational Marine Fisheries Service

P A P E R

ROV Operation from a Small BoatA B S T R A C T

Remotely operated vehicles (ROVs) are useful tools for aquatic research, but areunderutilized because of the high operational costs and limited versatility of older mod-els. Scientifically useful ROVs were originally very expensive to purchase or lease, andcostly to operate with limited models available. With an increase in the number of modelsavailable and lower operating costs, the use of ROVs in science is steadily increasing.However, scientists may be reluctant to use this technology because of past budgetaryand operational constraints, or are unaware of the lower-priced models that are now avail-able. There are a number of ROVs that are relatively inexpensive and can be operatedwithout specialized teams and platforms. We describe how to greatly reduce ROV opera-tional costs and increase versatility by operating the ROV with two to three people from asmall boat without hydraulics and in limited space. This combination can be safely usedin a variety of weather conditions. For this purpose, any recreationally used personal boatis considered a small boat. We used a Phantom XTL ROV1 and boats (without cabins)ranging from 16 to 24 feet long; both were chosen for low price and light weight, and hadadequate features and design for safe and reliable nearshore scientific research.

tations of depth, current and wind speeds andsea states that will keep surveys nearshore inless intense weather and sea conditions.

Beginning in the mid-1990s, scientists atthe NOAA Fisheries Auke Bay Laboratoryneeded to survey a plethora of nearshore habi-tats throughout southeastern Alaska to docu-ment biota from water depths of 250 ft toshore. Research objectives included determin-ing the distribution, habitat, and behavior ofrockfishes (Sebastes spp.), assessing the avail-ability of nearshore prey to sea lions at twosoutheastern Alaska haulouts, and document-ing nekton found in numerous nearshorehabitats for Essential Fish Habitat (EFH)studies. The surveys needed to be conductedat frequent intervals throughout the year forseveral years in remote bays, coves, and ca-nals close to shore in shallow (<250 ft depth)rocky or vegetated areas. Most of these areascould not be surveyed using a larger boat or alarger ROV.

Traditional and modern sampling tech-niques using fishing, drop cameras, andhydroacoustics addressed only some of thesurvey needs. Fishing methods such as trawl-ing or seining were labor intensive, expen-sive, destructive, and did not allow for study-ing ethology or habitats on a fine scale. Dropcameras were useful for studying behavior at

a variety of depths and at a fine scale, but didnot have enough control. Hydroacoustics didnot give a fine enough habitat scale, madefish identification difficult, and is unable tosurvey the benthic biota.

With the scientific use of ROVs on therise (Auster, 1992, 1997) and their appar-ent usefulness and cost effectiveness for ouraquatic research, we decided to purchase asmall ROV for about $60,000. The ROVwe decided on was the Phantom XTL1 fromDeep Ocean Engineering; this unit is 42 inlong, 18 in high, 21 in wide, and weighs100 lbs. The ROV has 41 lbs of thrust, aspeed of 2 knots, a maximum operatingdepth of 1000 ft, two 150-watt halogen ad-justable lights, two lasers 10 cm apart, and ahigh resolution, color, video camera withzoom and tilt. The Phantom XTL1 can beeasily carried, deployed, and retrieved by oneto two people without using any hydrau-lics. This ROV is not as small as theMiniRover MK II1 or Phantom S21

(Norcross and Mueter, 1999) which weigh65 lbs, and not as large as the PhantomHD21 (Fox et al., 2000) or Phantom DS41

(Hardin et al., 1992) which weigh 220 lbs.All of these ROVs have been proven as use-ful scientific machines. The smaller ma-chines were a little under-powered for our

Footnotes1 Reference to trade names does not imply

endorsement by the National Marine Fisheries

Service, NOAA.

R I N T R O D U C T I O N emotely operated vehicles (ROVs) have proven to be useful aquatic re-search tools. Most ROVs used for scientificresearch are large, such as ROPOS1 (Shepardand Juniper, 1997) and ISE Hysub ATP 401

used by MBARI (Robison, 1992). These largeand powerful machines are expensive to leaseand more expensive to operate. They requiretrained operators and special platforms thatuse cranes and hydraulics to deploy and re-trieve the ROV (Fox et al., 2000). SmallerROVs have been found to be effective researchtools, are 40% cheaper to lease, use a chartervessel that is 70% cheaper (Hardin et al.,1992), and are simpler to deploy, operate, andretrieve. Use of a small ROV has been provenadvantageous (Norcross and Mueter, 1999)by giving the scientist more insight into small-scale habitat use (Auster, 1991), spatial dis-tribution (Parker, 1994), and ethology(Spanier, 1994). Small ROVs can be oper-ated with quicker response time and in moreremote locations (Donaldson and Trusting,1997). Small ROVs also allow close side-by-side interaction between the ROV pilot andscientist, who is offered better interaction withthe environment for more accurate and effi-cient scientific study (Robison, 1992). Op-erational costs for small ROVs could remainhigh due to the continued use of large char-ter vessels with specialized teams and plat-forms. Use of a small boat < 24 feet longwould eliminate this cost. The drawbacks tousing small ROVs in small boats are the limi-


needs and the larger machines were too heavyto be safely and efficiently operated from asmall boat.Our remote survey locations,design, budget, and schedule made using acommercially chartered vessel to operate ourROV unfeasible. Most scientific ROV sur-veys are conducted from large charter ves-sels (Fox et al., 2000; Adams et al., 1995;Norcross and Mueter, 1999; Auster et al.,1989; Badd and Scheifele, 1992) offshore,in deep water, using specialized teams andplatforms. There is nothing published ona scientific team operating an ROV from asmall vessel, nearshore, in more shallowwater. To determine whether an ROV canbe safely operated from a small boat for sci-entific research, we gathered the equipmentneeded and developed a suitable design.

To operate an ROV from a small boatrequires a variety of water sensitive electron-ics, which all need to be operable in ma-rine, rainy, and windy conditions. Themore sensitive electronics and control con-sole should ideally be integrated into aworkstation, and the boat laid out for thework station, power source, ROV, opera-tors and umbilical.

We designed a workstation that couldbe positioned for safe and reliable operationof the ROV in southeastern Alaska’s insidewaters, in a variety of weather conditions andsea states. These waters are protected andhave less intense sea states than the openocean, but can still produce conditions dan-gerous for boat operations. Weather condi-tions range from continuous light rain tosporadic, heavy downpours. Saltwater sprayfrom winds and boat wash can also be aproblem. A boat had to be chosen for theareas we need to survey and the money wehad to spend. The survey areas werenearshore from 250 ft deep to shore, in bays,coves and channels throughout southeast-ern Alaska. Boat transit is limited by windsup to 15 knots and sea states with 3-4 ftwaves. The ROV will need to be operatedin <10 knot winds and no more than 2 ftseas with < 2 knots of current. The stationdesign features used for these weather andsea conditions are shelf lips, padding, andtie downs to keep the equipment in the sta-tion. Station and boat covers, and station

positioning are used to keep the electronicsprotected from spray or blowing rain.

The boat needed to be light enough sothe ROV could pull it (with a little motorassistance if needed) along a transect in re-mote, rocky, or shallow areas from 250 ftdepth to shore. A larger boat with a forwardor rear cabin offers more protection for theROV equipment, but we found such vesselstoo heavy, expensive, and not maneuverableenough for our research. Deploying andoperating the ROV from the bow of the boatand having the boat light enough for theROV to pull was the easiest way of control-ling the ROV and boat throughout our ex-tended transects in our survey locations.

Our surveys were located throughoutsoutheastern Alaska, which required the boatto be transported by trailer, towed by anynon-commercial truck or SUV, wheneverthe survey area was near the road system.Southeastern Alaska has very few roads; allbut two towns throughout the area are notconnected by road to any other town. There-fore, use of a trailer to transport our boatwas limited. Whenever a survey area is noton the road system or within small boattravel a charter vessel is needed to transportthe boat. A small, light boat will help keepcharter cost low (<$1,500 p/day) whenevera charter is needed, and will allow us to usea larger variety of charter vessels, making iteasier to find an appropriate vessel.

After much deliberation on which equip-ment, design, and layout to use, we put to-gether an ROV and boat combination, hadthe boat and workstation custom built, andscheduled it for extended field testing. Wepurchased the boat, motors, ROV, and allaccessories, and designed and built every-thing needed for sustained research forroughly $75,000. We started field testing in1998 and have made numerous successfulresearch dives to date. We have completedapproximately 300 ROV dives from a 16-ftlong, 4-ft wide flat-bottom skiff, and 100dives from an 18-ft long, 8-ft wide, semi-flat-bottom skiff. During these ROV dives,we conducted video surveys along transectsthroughout southeastern Alaska to notehabitat types and fish assemblages for EFH,sea lion prey base, and rockfish studies.

Transects were perpendicular to the shore-line starting from 250 ft depth offshore tothe water’s edge, and took place in a varietyof weather conditions. We had few prob-lems and collected a large amount of scien-tific data, including video, position, conduc-tivity, temperature and depth (CTD), andbiological data. Video data was collected withtemperature, depth, and position on screen,with occasional acoustic tracking for real-time ROV positioning. This ROV modelhas the capability to collect samples or posi-tion and collect data loggers with the addi-tion of manipulator and grabber-cutter arms.

Our small boat and ROV combinationproved invaluable in collecting a variety ofimportant and useful data. We used thiscombination to effectively collect detaileddata on nearshore habitat use, spatial andseasonal distribution of forage fishes over athree year period in Frederick Sound andLynn Canal in Southeast Alaska (in prep).We dependably used the ROV to qualita-tively verify fishes that were jigged andseined around two sea lion haulouts,(Brothers and Benjamin Islands) in south-eastern Alaska over a three year period. Theverification was important because itshowed us that our jigging and seiningmethods accurately sampled the area. Thevideo data collected by the ROV also addedimportant habitat data to our species com-position, allowing us to associate specificspecies with habitat type (in prep). TheROV was also used to qualitatively verifythe fish assemblages found on the outeredges of eelgrass and kelp beds, outside thereach of our beach seines (in prep). Thevideo data collected by this small boat andROV combination allowed us to collectdetailed information on the distribution,habitat and ethology of Rockfishes, Sebastesspp., in nearshore waters of southeasternAlaska (Johnson et. al., 2003). The datacollected by this combination of small boatand ROV made the $75,000 investmentmore than worth it, because the data col-lected added valuable data points and veri-fications, which made our research morecomplete, accurate and in turn useful andcould not have been collected using anyother method.


Workstation Designand Construction

A key to small boat ROV operation is todesign a workstation that fits the equipment,protects it from moisture, and can be posi-tioned in the boat for safe operations. Thefirst step is to put together all the equipmentneeded to operate the ROV. Equipment mustbe small in order to keep the workstation assmall as possible. The equipment will beworking on the water, possibly in marine en-vironments and in the rain, and in condi-tions unfriendly to the electronics needed forROV operation. The equipment needed tooperate an ROV includes a monitor, videorecorder, control console, audio mixer, video/PC interface, portable 2000-watt generator,transformer, and umbilical. Everything butthe ROV, generator, umbilical, and trans-former needs to be in the workstation for pro-tected and comfortable operation (Figure 1).

The dimensions of the station we usedare 16 in deep, 44 in wide, and 24 in tall,with the laptop computer area and trackingbox included. The station could be madeeven narrower if the laptop computer areaand tracking junction box were eliminated.The laptop area is 14 in deep, 16 in wide,and 19 in tall and the tracking box is 6 indeep, 16 in wide, and 16 in tall. The di-mensions of the station with only the essen-

tial ROV operational components could beas small as 16 in deep, 24 in tall, and 24 inwide. The back and bottom are made of onecontinuous piece of ½-inch plywood for acomplete water seal and structural integrity.The 24-in tall walls of the center box arealso made of ½-inch plywood for addedstrength with ¼-inch plywood used for allother walls and tops. Additional 1 × 1 inchsupports are added in the appropriate cor-ners for strength; all the seams are sealed withsilicone and the entire box inside and out issealed with 4-6 coats of marine-grade poly-urethane varnish. A surge strip is securelyfixed inside the box and plugged into thegenerator to supply power to all the elec-tronics. All the electrical lines that leave andenter the workstation are fed through one2.5-in hole located in the bottom back cor-ner of the station. For shock protection andto help keep the equipment from movingaround, thick neoprene is glued to all sur-faces where electronics will be positioned.Wood lips were installed on the shelf, roof,and bottom to help keep the electronics inplace and to add strength. Plastic could beused in place of the plywood and dimen-sional lumber. However, plastic is heavier,not as rigid, more difficult to find, and moreexpensive. A rigid ½-inch ABS plastic tablewas designed with a small footprint and legs

that are out of the way for permanent in-stallation along the side of the boat. The tablehad holes drilled in it that matched holes inthe bottom of the workstation so the work-station could be bolted to it and easily re-moved when not in use.

Whenever the electronics are not in useand at all times during travel the worksta-tion needs to be covered (Figure 2). Coversneed to be designed to completely protectthe sensitive equipment from rain and spray,stay out of the way when not in use, andeasily be put back in place when needed.We accomplished this by making the coversoverlap the box’s edges, padding the cover’sedges, and securing the cover in place bysnapping it to the box. This keeps rain andspray out no matter what the water’s angleof attack is. We also designed the covers sothey can be easily unsnapped, rolled up outof the way, and kept close at hand. We didthis by keeping the covers connected to thetop of the station, which allowed us to rollthe covers up and stow them on the top ofthe box (Figure 1). With the covers still con-nected and rolled up on the top of the boxwe can easily and quickly cover up the sta-tion when needed.

Construction of the workstation dependson the skills and materials available, and thesize of the station depends on the equipmentneeded and its size. Equipment should beas small as possible; we use a 9-in monitorand small Sony1 digital recorder to savespace. Materials used for construction couldbe anything that keeps the contents dry andis structurally sound; plywood, aluminum,or plastic work fine. We chose plywood be-cause it is light and easy to work with, widelyavailable, inexpensive, and sharp edges canbe easily eliminated. Plywood is easy to re-pair and modify, and most work can be donewith a limited amount of specialized toolsand skills. Aluminum and plastic are bettersuited for water but are heavier, harder towork, and more expensive.

The workstation needs to be firmly at-tached to the boat, and electronics need tobe secured in the station with padding (e.g.,thick neoprene) for shock absorption, andall shelf edges need lips to keep items in place(Velcro can be useful). The main consider-

FIGURE 1ROV workstation designed for ergonomic, safe, and protected operations.


ations for the design of the station are to keepthe electronics dry and protected from shock,and to have an ergonomic layout. The sta-tion and equipment will need to be removedfrom the boat for proper storage when not inuse. The only problem we found with ourcustom workstation was that the electronicswere more difficult and confusing to set upand take down than in the manufactured rackstation. We solved this problem by color cod-ing all lines and connections. Logistically,having the equipment separate from the sta-tion makes it easy for one person to moveand pack the electronics and station. Elec-tronics can be packed in waterproof cases fortransportation, making for excellent foul-weather protection, easy storage and logistics.With proper design and construction of theworkstation and cover, water and shock dam-age can be successfully eliminated from thesensitive ROV electronics. Our workstationhas been used for years in a variety of weatherconditions throughout southeastern Alaska’sinside waters, without any water damage byrain or saltwater spray. It has been operatedand allowed to sit in the rain for days withoutany water leaking into the station. We havetraveled many miles in a variety of sea stateswithout any shock damage sustained by anyelectrical device.

Manufacturers of ROVs usually offer arack-mount workstation designed for every-thing needed for ROV operation except thecontrol console. These stations are water re-

sistant and have shock absorbers built in.They make setup and takedown fairly easyand protect the equipment fairly well. How-ever, they are heavy, bulky, and are designedvertically and deep; a horizontal and shal-low design would lay out better in a boat.The console needs to be close to and prefer-ably in front of the monitor during opera-tion, so a table with the proper dimensionsis needed to position the control console forcomfortable operation (Figure 3).

The dimensions of the manufactured sta-tion are 24 in wide, 28 in tall, and 42 in deep,including the console, which adds 14 in tothe depth. The 42-in depth is what hinderswhere this station can be located in the boat.These stations are too heavy for one personto move, and too deep for the best position-ing in a boat. They cannot be closed onceconnected (without modification), makingweather protection impossible when travel-ing. In order to close the box while connected,it is necessary to cut a 3-inch hole in the bot-tom of the box to feed the lines through. Thisallows closure of the box while traveling inorder to protect the instruments from rain

FIGURE 2Covered workstation for protection from the elements.

FIGURE 3Manufactured rack-mount workstation set up for operation; note the position of the workstation and console.


and spray; however, a separate cover is neededto protect the console (Figure 4).

The workstation in Figure 4 is more dif-ficult to cover for travel and cannot be leftout in the rain even when closed because waterwill leak through the aluminum seams. Rack-mount stations are functional, but a customstation is necessary for the workstation to bestfit the available space of a small boat and forthe optimal mix of ergonomic layout andprotection from rain and spray. However, forresearchers who do not have the means todesign and built a workstation, this rack-mount system would be a workable alterna-tive. If the station and console are laid out,covered, and secured in the boat properly,equipment should be protected well enoughfor years of reliable service.

Boat LayoutThe ROV skiff needs to be laid out for

safe, comfortable, and efficient boat opera-tions, generator use, and ROV operation andmust offer adequate protection from the el-ements for personnel and equipment. Nec-essary items include the workstation, space

for the ROV and umbilical operators, gen-erator, transformer, ROV, and umbilical. Theworkstation along with the ROV operatorneeds to be positioned where neither is inthe way (visual or physical) of the boatoperator(s). The workstation and ROVoperator’s seat have to be positioned highenough for ergonomic operation. The work-station needs to be positioned within theboat so that water from rain and spray willnot inundate it. Positioning the station fac-ing forward or aft would allow spray to moreeasily reach the electronics during operation.Orienting the station so it faces port or star-board with the solid back facing out bestprotects electronics from spray.

The generator needs to be positioned ina place where the noise and fumes do notaffect the operators, and needs a cover thatwill protect it from water, keep it in place,protect it from damage and insulate the op-erators from noise. The generator coverserves four roles; to protect the generatorfrom water and physical damage, to lowerthe noise levels, to create a step to enter andleave the boat, and to create a flat work area.A four-stroke generator was chosen for light,

quiet, and efficient operations. This size gen-erator doesn’t produce enough noise tohinder communication much, but long ex-posure intervals to motor noise necessitatesgetting the quietest generator available andinsulating it if possible.

The power line(s) coming from the gen-erator need to be protected from damageand not cause accidents. This was done bylaying the line(s) tight to the floor, securingit in electrical conduit tight on the floor, andpositioning it using the shortest distancefrom the transformer and workstation to thegenerator. This kept all power lines protectedfrom physical damage and eliminated anytripping accidents.

The transformer needs to be out of theway and protected from water (Figure 5). Wefound a metal ammo box works well for trans-port, storage, and protection of the trans-former while in use. No modification of thebox is needed; all that is needed is properpositioning of the transformer cord so thatthe top can be laid as low as possible. Thenposition the box parallel with the side of theboat, out of the way of boat traffic, with theopen end of the top facing away from spray.Heat is produced by the transformer so thebox needs to be metal and clean of all flam-mable residues inside and out.

The boat needs to be wide enough andlong enough for boat, ROV, and other scien-tific operations. Our first ROV skiff was anopen 4-ft wide, 16-ft long flat bottom boatthat was a bit small and uncomfortable. Itoffered questionable protection for the elec-tronics on rainy and windy days, and was toocluttered for ergonomic and safe operations.However, with a well-designed workstation,generator cover, and boat layout it was suc-cessfully used for four years on nearly 300dives without any personal injuries and withminimal equipment damage. Our currentROV platform was designed much better forsafe and comfortable operations. We chose asemi-flat bottom 18-ft long, 8-ft wide boatand added a canvas top, a four-cycle genera-tor, and a more ergonomic workstation (Fig-ure 6). The boat’s canvas top (Bimini) coversthe top and sides of the boat and protects theelectronics when in use, and makes the boatmore comfortable for the operators on rainy

FIGURE 4Manufactured workstation completely covered, note the additional blue cover needed for the console.


days and during travel. The boat canopy, workstation, and laptop computer covers weremade and positioned to protect the electron-ics during operation and travel.

The considerations for our second boatwere to choose a design that allows 3-5people to comfortably and safely travel inwinds up to 15 knots, in 3-4 ft seas and incurrents up to 10 knots. The boat had toenable us to conduct a variety of researchprocedures in rainy and windy conditions,with winds <10 knots, in seas up to 2 ft andcurrents <2 knots. The vessel needed to beshort and light enough to trailer easily andload on any charter vessel with a boom anddeck space. A semi-flat displacement hulldesign was chosen for speed, beach seining,and for shallow and rocky areas. A length of18 ft was chosen to keep the boat short andlight while still being long enough for op-erations. A width of 8 ft was chosen for boatstability while at sea and to give us enoughspace for unobstructed operations. Gunwaleheight and interior design were limited bythe available models that would be suitablefor all our varied operations and projects (i.e.,beach seining). There were two consider-ations for gunwale height, the gunwale hadto be high enough to handle 3-4 ft seas yet

low enough for the ROV to be pulled out ofthe water by two people.

A very sparse interior was chosen in or-der to give us an “empty slate” so we couldlay out the seats and equipment to best suitour needs. We chose a manufacturer thatcould provide us with the a stern bench,steering and control console in front of thestern bench, next to the gunwale, and a raisedplatform on the bow. We now neededenough power to get this vessel and all our

gear up on step well, and push us along at aspeed >20 mph, while not being too heavyor expensive. We chose a 50-hp, two-strokeoutboard for our main power and a 6-hpoutboard as a backup.

Once the boat was in hand we were ableto set up the generator, workstation and seatsto allow for unobstructed movement aroundthe boat at all times. Seat layout was veryimportant for ergonomic travel, vessel bal-ance and research operations. Comfortableseats with backs were chosen and positionedwith one seat on each side at the stern forpiloting the boat and computer operations.The ROV operator’s seat was positioned infront of the two stern seats at the correctheight and distance to the workstation inorder to minimize back strain while keep-ing the boat balanced (Figure 5).

Safe operation of the boat was the pri-mary concern during the design, construc-tion and operation of this research platform.Our design and operations allowed for un-hindered views for all people, especially forthe captain while under way and nothingwas positioned on deck tha would cause anytripping accidents. All U.S. Coast Guardregulations and requirements were followedto the letter, with float plans distributed andthe proper floatation, EPIRBs, communi-cation, and signaling devices on board. Thetotal cost for the boat, two motors, all safetyequipment, Bimini top, and trailer wasaround $10,000.

FIGURE 6Second-generation ROV skiff; notice the open space on the side in front of the boat canopy, and the space on thebow for the umbilical operator.

FIGURE 5Unobstructed boat layout; notice how the workstation, generator, transformer, seats, and power lines arepositioned in the boat.


An important boat layout considerationis the location for deploying, retrieving, andoperating the ROV and umbilical. We foundthe bow of the boat to be the best locationto control the ROV and boat. To operatethe ROV from the bow, a flat raised plat-form is needed at the bow and a fairly flatopen space is needed behind this platformfor the umbilical operator. Many boat manu-facturers design just such a platform in theirbows. Umbilical operations involve one per-son feeding out and coiling up the umbili-cal while deploying and retrieving the ROV.The raised bow platform and flat area tostand make bow work more ergonomicwhile coiling and uncoiling the umbilical.With too low a platform or uneven footing,back, leg, and foot problems arise. Ampleclear gunwale space (2-3 ft) is also neededfor one to two people to retrieve the 100-plus lb ROV from the water (Figure 6).

A dive door would make retrieval easierbut the boat would require more elaborateconstruction and engineering in order for adoor to be installed. This would add con-siderable weight and cost to the boat andlimit its design. Small hydraulic arms areavailable that can be easily modified for ROVretrieval; however, we did not find this nec-essary. We found deploying and operatingthe ROV from the bow of the boat was easi-est and offered the most efficient use of theROV and boat. Operating the ROV fromthe bow gave better control of the boat andROV during surveys, allowing for greaterROV maneuverability and distance traveled.

Having a boat light enough for the ROVto tow increases the survey range and allowsthe ROV operator to maneuver withoutfighting the boat. This also frees up the boatoperator to perform scientific duties such asdata collection and data entry. Only in stron-ger currents (1-2 knots) and on windy dayswas the boat’s power needed to position theboat so it wasn’t fighting with the ROV.Careful control of the amount of umbilicalout during operation also helps the ROVfrom fighting with currents and the boat.Too much umbilical out creates drag andincreases chances for the umbilical to hangup, and too little umbilical out forces theROV to fight the boat.

The ROV is easy to deploy and retrievewith the ROV and umbilical sitting on theraised platform in the bow. To deploy theROV, it is positioned on top of the gunwaleand pushed into the water. The umbilical isthen allowed to freely follow the ROV, withcare taken to avoid any kinks. To retrievethe ROV, the umbilical handler coils theumbilical as the ROV comes in. When theROV is at the side of the boat, two peoplegrab the crash frame and pull the ROV overthe gunwale and onto the coiled umbilical,which is on the flat raised area on the bow.Conditions are unsafe for deploying theROV when sea states make it too difficultto stand, making it unsafe to be by the edgeof the boat, or when winds or currents aremoving the boat faster than 2-3 knots. Com-mon sense and some sea experience in a smallboat are needed to properly evaluate whetherconditions are safe.

Design TestingWith proper boat layout, workstation

design, and construction we were able to de-pendably conduct several years worth of strin-gent scientific video surveys using a smallROV from a small boat in one of the wettestenvironments in North America. Our ROVresearch used two generations of boats andworkstations. The first-generation boat wasa 16-ft flat bottom, which was as small andbasic as one can get and still operate an ROV.The first-generation workstation was madeof questionable materials but was designedwell for ROV operations and measured 32in wide, 18 in tall and 16 in deep. The sta-tion and table were of one piece with pipesfor legs, which were bolted to the boat. Theboat was barely big enough (16 ft) for threepeople and the seats were not very ergonomic,but the workstation design and boat layoutallowed safe and dependable operation. Theworkstation and boat were custom designedand built by a local craftsman with no out-side specialty contractors.

Once the ROV proved useful, our de-sign proved dependable and safe, and thefunds became available, a second-generationvessel was purchased and a redesigned work-station was built. This second station and

boat were redesigned for safer and more er-gonomic operations and once again the boatwas put together by a local craftsman. Thelarger 18-ft boat allowed us to better sepa-rate operators from generator noise andfumes. We covered the boat with a canopyand set up the boat for more comfortableoperations. The second-generation worksta-tion was still made of plywood but this timeit had no legs and was bolted to an ABS plas-tic table that was permanently mounted tothe boat. The location of the monitor wasraised and centered for less neck and handstrain, and laptop computer and tracking sta-tions were added. The front cover was cus-tom made by a local canvas maker for morecomplete and comfortable coverage of theoperator and station. A quieter and lightergenerator was purchased and the boat wasbetter wired for its use. The front umbilicalwork deck was higher and the foot deck flat-ter for a more ergonomic layout for feet, legs,and back. This boat was far more comfort-able for work and travel, considering that oursurvey locations and scheduling dictated con-siderable travel from site to site.

This second-generation ROV worksta-tion and boat has proven to be well designedand constructed for ROV operations afteralmost 100 dives and three years of use. OurROV has been used on a number of projectsover the years (see Johnson et al., 2003).Other projects the ROV was used for in-clude video surveying biota found from adepth of 250 feet to shore around sea lionhaulouts, in eelgrass and kelp, and in othersub-tidal habitats. The ROV has also beenused for less scientific projects such as sur-veying species in areas of interest for localtowns, surveying areas before and after con-struction, and assisting local authorities withsearch and rescue operations. The boat withall the equipment needed was easy to loadon charter vessels as small as 42 ft that costroughly $1000 per day. It was also easy totrailer and tow with any SUV. When we hadto trailer the boat at speeds greater than 45mph, the Bimini top needed to be loweredand secured.

During our years of operation we had fewproblems with the ROV, boat, or electronics.Problems that occurred included digital re-


corder operation and a leaking ROV housing.The digital recorder was more sensitive tomoisture from condensation and needed to beremoved from the workstation at the end ofthe day. The leaking problem with the ROVhousing was caused by poor annual mainte-nance; ROVs usually require annual mainte-nance of the dynamic seals, and certain partsneed to be kept in place. If the annual mainte-nance is not done, if the end caps were removedtoo often, or if the purge plug was left out inmoist or humid conditions when the ROVwas not in use, leaking problems could occur.If the ROV gets proper maintenance and ischecked properly, there should be few prob-lems with the ROV or with electronics. Withour designs, operational and maintenance pro-cedures, we were able to collect roughly 200hours of video data, make 538 beach seine setsto collect shallow species identifications, andjig for 100 hours to collect species identifica-tions from waters ≤250 ft deep with little tono equipment or personnel problems.

ConclusionOperating small ROVs from small boats

can allow more scientists to take advantageof this useful technology. Not having to char-ter a large commercial boat and not need-ing hydraulics allows for inexpensive andsimple ROV operation. Buying an ROVeliminates the lease cost and scheduling con-straints. Operating the ROV from a smallboat eliminates the charter vessel lease andits limitations. Operating the ROV directlygives the scientist the best interaction andcontrol over the ROV which enhances theresearch experience and the quality of datacollected. Remotely operated vehicles cannow be purchased and operated at a reason-able cost by marine scientists and other re-searchers working in aquatic environments.With the right workstation and layout, re-searchers can expect years of dependable,safe, and efficient service from a combina-tion of a small ROV and a small boat. Theonly drawbacks are the limitations of depth,current and wind speeds, and sea states thatcan be handled by a small ROV and smallboat, which will keep surveys nearshore inless intense weather and sea conditions.

Remotely operated vehicles are not dif-ficult to operate; with a little practice andaverage hand-to-eye coordination, anyonecan handle one like a professional. MostROV manufacturers offer training classes for2-4 days that are included with the purchase,and teach basic construction, maintenance,repair, and operation. The hardest thing tooperate on an ROV is the video interface,which puts information on the screenthroughout the video survey. Once this isset up and the basic commands are learned,easy commands can be used to make theappropriate screen informational changes.Small ROVs are fairly simple devices thattypically have a hull, lights, thrusters, cam-era, manipulating arm(s), power circuitboard(s), and control/data board(s). The twoboards, electric thruster motors, and all wir-ing connections are sensitive to water, butwith the right maintenance, problems canbe kept to a minimum. Anyone who is tech-nically inclined and fairly good with toolscan take most of the ROV apart, includinglights, thrusters, power circuit board, andall the seals (dynamic and static). This isimportant in the event that the ROV getsflooded with saltwater and must be takenapart, rinsed with fresh water, dried, and putback together.

In-house maintenance includes proce-dures that are simple enough and need to bedone enough times a year to make it finan-cially and chronologically impractical to sendthe ROV back to the manufacturer. Themanufacturer’s course teaches what must bedone to maintain the ROV. With a little prac-tice and some technical skill, these procedureswill become second nature. The main thingthat reduces maintenance is to remove capsand plugs only when necessary, and to useonly a little silicon grease on seals. The greasekeeps the seals from freezing in place, mak-ing them difficult to remove. By keeping thecaps and plugs in place, these parts seal bet-ter and seal damage is minimized. The mainitems that must be maintained once or twicea year, depending on use, are ALL the dy-namic seals. Dynamic seals are the rubberO-rings that are used to seal up moving partssuch as thruster shafts and control arms (i.e.,camera tilt). These are exposed to friction and

thus are prone to wear. Working with ROVsdoes require a little technical prowess; other-wise annual service can be scheduled withthe manufacturer.

A small ROV can be inexpensively, safely,dependably, and comfortably operated froma small boat if the boat and workstation arelaid out well, making it possible to collect datathat could be invaluable and quite possiblyimpossible to collect using any other method.

AcknowledgmentsI thank M. Murphy, S. Johnson, J.

Thedinga, C. Lunsford, J. Fujioka, and thecrew of NOAA ship John N. Cobb, who as-sisted with the ROV operations. M. Sigler,S. Johnson, and P. Rigby reviewed an earlierdraft of the manuscript and N. Muirheadedited the final version.

ReferencesAdams, P. B., Butler, J.L., Baxter, C.H., Laidig,

T.E., Dahlin, K.A., Wakefield, W.W. 1995.

Population estimates of Pacific Groundfishes

From Video Transects and Swept-area Trawls.

Fish Bull. 93:446-455.

Auster, P. 1997. ROV Technologies and

Utilization by the Science Community. Mar

Technol Soc J. 31(3):72-76.

Auster, P.J. 1992. Introduction: ROVs for

Science. Mar Technol Soc J. 26(4):3.

Auster, P.J., Malatesta, R.J., LaRossa, S.C.,

Cooper, R.A. and Stewart, L.L. 1991.

Microhabitat Utilization by the Megafaunal

Assemblage at a Low Relief Outer Continental

Shelf Site–Middle Atlantic Bight USA. J

Northwest Atl Fish Sci.. 11:59-69.

Auster, P.J., Stewart, L.L., Sprunk, H. 1989.

Scientific Imaging with ROVs: Tools and

Techniques. Mar Technol Soc J. 23(3):16-20.

Babb, I.G., Scheifele, P. 1992. Commentary:

Marine Science Education Using Low-Cost

ROV Technology. Mar Technol Soc J.

26(4):73-78.


Donaldson, P. and Tusting, R. 1997. Water,

Sediment, and Organism Sampling Systems

Designed for Small ROVs. Mar Technol Soc J.

31(3):5-10.

Fox, D., Amend, M., Merems, A., Appy, M.

2000. Coastal Zone Management Section 309

Grant: 2000 Nearshore Rocky Reef Assess-

ment: Final Report for 2000 Grant. Contract

No. 01-01, Oregon Dept. of Fish and Game

Marine Program.

Hardin, D.D., Graves, D., Imamura, E. 1992.

Investigating Seafloor Disturbances with a

Small ROV. Mar Technol Soc J. 26(4):40-45.

Johnson, S.W., Murphy, M. L. and Csepp, D.

J. 2003. Distribution, Habitat, and Behavior

of Rockfishes, Sebastes spp., in Nearshore

Waters of Southeastern Alaska: Observations

from a Remotely Operated Vehicle. Envir Biol

Fish. 66:259-270.

Norcross, B. L. and Mueter, F. J. 1999. The

Use of an ROV in the Study of Juvenile

Flatfish. Fish Res. 39:241-251.

Parker, R.O. Jr., Chester, A.J., Nelson, R.S.

1994. A Video Transect Method for Estimat-

ing Reef Fish Abundance, Composition, and

Habitat Utilization at Gray’s Reef National

Marine Sanctuary Georgia. Fish. Bull.

92:787-799.

Robison, B.H. 1992. Midwater Research

Methods with MBARI’s ROV. Mar Technol

Soc J. 26(4):32-39.

Shepard, K. and Juniper, S.K. 1997. ROPOS:

Creating a Scientific Tool from an Industrial

ROV. Mar Technol Soc J. 31(3):48-54.

Spanier, E., Cobb, J.S., Clancy, M. 1994.

Impacts of Remotely Operated Vehicles

(ROVs) on the Behavior of Marine Animals:

an Example Using American Lobster”, Mar

Ecol Prog Ser. 104:257-266.


A U T H O R SMichaela DommisseSchool of Fisheries and Ocean Sciences,University of Alaska, Fairbanks andSchool of Geography and EnvironmentalScience, Monash University

Dan UrbanAlaska Department of Fish and Game

Bruce FinneySchool of Fisheries and Ocean Sciences,University of Alaska, Fairbanks

Susan HillsSchool of Fisheries and Ocean Sciences,University of Alaska, Fairbanks


Potential Depth Biasing Using the Biosonics VBTSeabed Classification Software

A B S T R A C TThe use of digital echosounders with post-processing commercial seabed classifica-

tion software is becoming increasingly popular to create high-resolution resource mapsof marine habitats over large scales. Here, we examine the Biosonics Visual Bottom Typ-ing (VBT) seabed classification software (version 1.91). Although the VBT software usespotentially robust seabed classification parameters and has many useful features, a majordrawback of the software is that it does not normalize echoes to a reference depth (typi-cally the average survey depth). Depth normalizing adjusts for a change in echo lengthwith depth. Without it, the VBT classification parameters that are calculated from theenergy integral of the returned echo envelope over a fixed sampling window can be depthbiased. The degree of misclassification from depth biasing will depend on the particularechosounder specifications (beamwidth (at -3dB) and pulse length) and characteristicsof the survey area (depth variation and bottom acoustic diversity). Depth normalizationapplied before classification parameters are calculated is a very simple solution to poten-tial depth-related misclassification and should be incorporated into the VBT software as amatter of priority to ensure its reliability and broad user application.

tems (RoxAnn, designed and manufacturedby Marine Microsystems Ltd., AberdeenScotland; QTCView and QTC Impact, de-signed and manufactured by Quester Tan-gent Corporation, Sidney BC, Canada ) andit uses both reliable (E1 and E2) and novel(e.g. fractal dimensions) acoustic classifica-

tion parameters (see Table 1). Further, as theVBT software analyses digital data it allowsa lot of flexibility in data processing. How-ever, a major drawback of the VBT softwareis that current versions do not normalize re-turning echoes to a reference depth (J.Burcszynski, Biosonics Inc, pers. comm.).

TABLE 1A description of VBT classification parameters

Parameter Description Sediment character

E0 The energy integral of the above-surface echo Above sediment biomasswith a sampling window of one pulse lengthbefore the start of the bottom pick

E1’ The integral of the energy of the first part of Hardnessthe first bottom echo, with a sampling windowof one pulse length from when the start of theecho is picked.

E1 The integral of the energy of the second part Roughnessof the first bottom echo, starting after the E1’sampling window for a sampling window ofthree pulse lengths

E2 The integral of the energy of the second bottom Hardnessecho with a sampling window of three pulse lengths

Sediment thickness(Sed) An estimate of the thickness of the sediment layer Height of vegetation(above sediment echo)

Fractal Dimension(FD) A measure of irregularity of the first echo Roughness

RI N T R O D U C T I O N ecent legislation worldwide is increasing pressure on fisheries managers toidentify and protect fish habitats for effec-tive and sustainable management of dwin-dling fish stocks (Conover et al., 2000). Tra-ditional methods of seabed habitat mappingsuch as grab sampling, video imagery anddiving are expensive and time consumingand limit habitat maps to narrow spatialscales (meters) (Davies et al., 1997;Greenstreet et al., 1997). Amongst emerg-ing techniques is the use of acoustic tech-nology to remotely sense the seabed overlarger spatial scales (kilometers) (Davies etal., 1997; Greenstreet et al., 1997). Coupledwith Geographic Information Systems,acoustic data may be used to create high-resolution resource maps over unprec-edented scales (Kloser et al., 2001).

A system that is relatively new to themarket (1995) is the Visual Bottom Typing(VBT) system (Biosonics, Seattle USA), asoftware product that analyses digitalechosounder data. The system is attractiveto users because it is much cheaper thanother commercial seabed classification sys-


This can result in depth-relatedmisclassification by the VBT software sincethe echoes shape and apparent energy char-acteristics are affected, and energy integralsare used to differentiate between sedimenttypes. Here, we investigate how bottomdepth may be affecting VBT classificationand suggest modifications to the softwarethat would increase its reliability for users.

The VBT SoftwareThe VBT Seabed Classifier software is de-

signed to process digital data collected from adigital echosounder. Signal processing involvessampling the echo envelope at a constant rate(the sampling rate) for a given number ofsamples (the sampling window) to calculatethe time integral of the squared amplitude ofdifferent parts of the digitized echo envelope(i.e. energies). A Time Varied Gain (TVG)adjustment (-20 Log [Range]) is made beforeclassification because the energy of a return-ing echo changes significantly with waterdepth due to spreading and absorption losses(Medwin and Clay, 1998). Spreading loss istaken as inversely proportional to range. Ab-sorption can also be allowed for.

VBT classification parameters are based onthe shape and energy characteristics of the ad-justed echo envelope (Table 1). Four param-eters are based on the energy characteristics ofthe echo: E0, E1’, E1, E2 (Table 1). Other echoderived parameters are Sediment thickness layer(Sed) and Fractal Dimension (FD) (Table 1).The FD is calculated for the echo envelope us-ing the box dimension method as per£ubniewski and Stepnowski (1997) and detailedin the Biosonics manual (Biosonics Inc, 1998).Too few details are provided on the Sed thick-ness parameter to determine how it is calculatedbut “thickness” suggests it may be the distance(time) from the start of a minor echo due tovegetation until the start of the first bottom echoproper. For classification, the VBT software com-pares the classification parameters from differ-ent sediments against pre-determined types (su-pervised classification). Alternatively, users canclassify bottom types without any pre-deter-mined habitat types using the software fuzzymean clustering module or in an external analysis(unsupervised classification).

Adjusting Echo Lengthto Depth

Although the VBT software makes aTVG adjustment, no adjustment is madefor the effect of bottom depth on echo length(J. Burczysnki, Biosonics Inc. pers. comm.).The length of the echo is a function of thepulse length, spreading time of the wavefrontacross the bottom and the travel time in thesediment. The spreading time of thewavefront can have the most effect on echolength as the survey area changes because itis proportional to depth. This can be illus-trated in Figure 1. After the leading edge ofthe ping hits the seafloor, i.e. after the echostarts to form, the time taken to ensonify tosome given angle θ increases with depth (D)(Figure 1), and the echo signal lengthens asdepth increases (Figure 2) (Clarke andHamilton, 1999). However the VBT soft-ware calculates VBT classification param-

eters with a user fixed sampling window thatdoes not take into account the change inecho length with depth. For example, therecommended sampling window of the E1classification parameter (the tail of the firstecho return) is three times the pulse length.However, the tail of the echo may extendwell beyond the sampling window in deeperwaters where the spreading time of thewavefront is greater thus omitting some ofthe energy integral (Figure 2). In shallowerwaters, the tail of the first echo may beshorter than the sampling window and in-clude too much of the energy integral forcorrect classification (Figure 2). The sameargument can be made for the other VBTclassification parameters (E1’, E2 and FD).We cannot comment on the effect of depthon the E0 and Sed parameters because toolittle information is provided on how theyare calculated.

FIGURE 1A source ensonifying the same sediment type at a reference depth (D0), deeper and shallow depth (D1 and D2)to a specific angle (θ or half beamwidth). The time taken to ensonify to an angle θ after the Leading Edge (LE) ofthe ping first hits the bottom (LE(t0)) increases with depth i.e. from LE(t0) to LE(t1) where time (t1) > time (t0).Adapted from Clarke and Hamilton (1999.)


As is evident from Figure 1, the degreeto which echo length (and misclassification)is affected is a function of the particularuser’s echosounder specifications (pulselength etc, particularly beamwidth) anddepth of the survey area. Users collectingdata from shallow areas of high acousticdiversity, or over a constant survey depthwill still collect meaningful results. How-ever, users cannot rely on data where thesurvey depth has varied significantly.

In other seabed classification systems,to allow for the effects of a fixed samplingrate, echo length is adjusted by normaliz-ing echoes to a reference depth (Hamilton,2001) or equivalent (e.g. Kloser et al.,2001). Usually the reference depth is takenas the mean or geometric mean depth ofthe survey area (QTC Impact™ manual,1999; Hamilton, 2001). The length of thereturning echo is then either lengthened or

shortened along the time axis, in relationto the reference depth, and classificationparameters are calculated from the normal-ized echo. This allows echoes from the ac-tual seafloor depth and the reference seaf-loor depth to maintain the same time/anglerelationships (Caughey et al. [1994] citedin Clarke and Hamilton [1999]). Clarkeand Hamilton (1999) have already mod-eled the effect of increasing depth on echolength (see their Figure 4) and demon-strated the need for depth normalizing.Their results provide excellent in-principlesupport for the potential depth-relatedmisclassification by the VBT software notedin our paper. They provide simple algo-rithms to correct for depth biasing. Thetime adjustment to transfer an echo re-ceived from a depth (D) to a referencedepth (D0) simply involves multiplicationby the factor D/D0.

ConclusionsAt the time of writing the VBT software

will have been on the market for almost adecade and has users worldwide. The use ofa digital based platform and the incorpora-tion of many useful features make it a po-tentially valuable tool for seabed classifica-tion. However without depth normalization,the current VBT software is unnecessarilyrestricted in its use to shallow areas or con-stant survey depths. We were prompted towrite this note because a very simple solutionto potential depth biasing would be for themanufacturer to depth normalize echoes inroutine signal processing before the classifi-cation parameters are calculated. This wouldexpand the operational limits of what is oth-erwise a promising seabed classification tool.For current users the outlook is also good.The advantage to collecting digital data isthat it can be reprocessed once the softwareincorporates depth normalizing and userswill not have surveyed for nothing.

AcknowledgmentsHelpful comments on the manuscript

by Drs. Jason Beringer, Les Hamilton andan anonymous reviewer are greatly appreci-ated. The study was funded by grants fromthe North Pacific Marine Research Initia-tive and the West Coast and Polar RegionsUndersea Research Center, Alaska undergrant no. NA 86RG0050 (project no. RR/99-02), and from the University of Alaskawith funds appropriated by the state.

ReferencesBiosonics Incorporated. 1998. Guide to using

VBT – seabed Classifier. Biosonics Incorpo-

rated: Seattle, USA. 91 pp.

Burczynski, J. Bottom Classification. 14pp.

See http://www.biosonicsinc.com for a

downloadable document (“Bottom

Classification6.doc”).

FIGURE 2A hypothetical echo envelope from the same bottom at different depths using a fixed sampling window. Indeeper waters (D2) the echo length is dilated in relation to a reference depth (D2 >D0), while in shallower waterthe echo length is compressed in relation to the reference depth (D1< D0). Modified from Clarke and Hamilton(1999). Shaded areas are examples of the energy (area) integrals for E1 used by VBT.


Caughey, D.A., Prager. B. and Klymak. J.

1994. Sea bottom classification from echo

sounding data. Contractors report 94-56,

prepared by the Defense Research Establish-

ment Pacific, Canada. Document number

SC93-019-FR-001. Quester Tangent

Corporation, Marine Technology Center:

British Columbia, Canada. 35 pp.

Clarke, P.A. and Hamilton L.J. 1999. The

ABCS Program for the Analysis of Echo

Sounder Returns for Acoustic Bottom

Classification. Report DSTO-GD-0215.

Defence Science and Technology Organiza-

tion, Aeronautical and Maritime Research

Laboratory: Fishermen’s Bend, Australia. 58pp.

See http://www.dsto.defence.gov.au/ publications

for a downloadable pdf.

Conover, D., Travis. J. and Coleman. F. 2000.

Essential Fish habitat and marine reserves: An

introduction to the second symposium in

fisheries technology. Bull Mar Sci. 66(3):527-534.

Davies, J., Foster-Smith. R. and Sotheran. I. S.

1997. Marine Biological mapping for

environment management using acoustic

ground discrimination systems and geographic

information systems. J Soc Undw Tech.

22(4):167-172.

Greenstreet, S. P. R., Tuck. I. D., Grewar. G.

N., Armstrong. E., Reid. D. G. and Wright. P.

J. 1997. An assessment of the acoustic survey

technique, RoxAnn, as a means of mapping

seabed habitat. ICES J Mar Sci. 54(5):939-959.

Hamilton, L.J. 2001. Acoustic seabed

classification systems. Report DSTO-TN-

0401. Defence Science and Technology

Organization, Aeronautical and Maritime

Research Laboratory: Fishermen’s Bend,

Australia. 65 pp. See http://

www.dsto.defence.gov.au/ publications for a

downloadable pdf.

Kloser, R.J., Bax. N., Ryan. T., Williams. A.

and Barker. B.A. 2001. Remote sensing of

seabed types in the Australian South East

Fishery, development and application of

normal incident acoustic techniques and

associated ‘“ground truthing”. Mar and Freshw

Res 52(4):475- 490.

Lubniewski, Z and Stepnowski. A. 1997. Sea

bottom typing using fractal dimensions. In:

International symposium on hydroacoustics

and ultrasonics. Gdansk-Jurata, Poland. 6 pp.

Medwin, H. and Clay. C. 1998. Fundamentals

of acoustic oceanography. Academic

Press:USA.

QTC Impact™ manual, 1999. Acoustic

seabed classification user guide version 2.00.

Quester Tangent Corporation: British

Columbia, Canada. 108 pp.


A U T H O R SGuido CalcagnoF. Di FeliceM. FelliF. PereiraINSEAN, Italian National Institute forNaval Architecture Studies and Testing

P A P E R

A Stereo-PIV Investigation of a Propeller’s Wakebehind a Ship Model in a Large Free-surface Tunnel

A B S T R A C TAn experimental investigation of a five blade MAU propeller wake behind a Series 60

Cb=0.6 ship model has been performed using Stereo Particle Image Velocimetry (Stereo-PIV) in a large free-surface cavitation tunnel. The investigation of the wake and its evolu-tion during propeller revolution and at different longitudinal stations has pointed out thecapability of Stereo-PIV in resolving the complex flow field with great accuracy. As a firststep of this investigation, a systematic measurement analysis has been done to find thebest angle between the two cameras. The blade viscous wake, developing from the bladesurface boundary layers; the trailing vortex sheets, due to the radial gradient of the boundcirculation; and the velocity fluctuation distributions are identified and discussed. Thecomplex interaction between the hull wake and propeller is described through the evolu-tion of the mean velocity components and the vorticity fields. In the near field the effectsof turbulent diffusion and viscous dissipation, which cause a rapid space-broadening ofthe velocity gradients in the trailing edge wake, are also examined. Comparison with LDVmeasurements shows a substantial agreement between the two techniques.

to difficulties in unsteady flows or havingto keep the facility working characteristicsconstant for long periods of time.This is the case of the experimental inves-

tigation of a propeller wake in a non-uniforminflow. This analysis requires a sufficientlydense grid into the whole measurement plane,for each different propeller angle, in order toresolve the flow structures during the propel-ler revolution (Felli et al., 2000; Esposito etal., 2000; Di Felice et al., 2000).

From this point of view, considering thatthe Particle Image Velocimetry (PIV) tech-nique allows the instantaneous measure-ment of the velocity at a plane, it offers manyadvantages over single point techniques: theexperimental analysis can be done quicklyand easily by acquiring images at each an-gular position of the blade, drastically re-ducing the testing time. Over the past de-cade the PIV technique has experiencedconsiderable progress and is today consid-ered to be a powerful whole field measure-ment technique, continuously broadeningits range of applications. The growth of thePIV technique due to the improvement ofthe hardware components is clear: high-en-ergy and nanosecond pulse duration lasers,

high-resolution and low-noise CCD cam-eras, fast frame grabbers, as well as fastercomputers and large data storage hard disksare among the major factors that have raisedthe capabilities of the measurement ap-proach. Recent literature demonstrates theapplicability of the technique to the navalfield, in particular in towing tank applica-tions (Guj et al., 2001) and in the case ofthe propeller flow (Cotroni et al., 2000; DiFelice et al., 2000) even if investigations havebeen limited only to two velocity compo-nents in the measurement plane (planar PIVor 2C-PIV).

However, it is apparent that this infor-mation is not always sufficient to character-ize the flow field especially for propeller flowswhere the presence of a strong three-dimen-sional flow structure with strong velocitygradients requires knowledge of all the ve-locity components.

Stereoscopic PIV, using two cameras view-ing the flow from two perspectives, is theobvious extension of planar-PIV for measur-ing all three velocity components in a plane.Two components of velocity nominally per-pendicular to the camera optical axis are mea-sured from each camera viewpoint. The pair

II N T R O D U C T I O N n the last 20 years the application of advanced optical measurement tech-niques like laser doppler velocimetry (LDV)has provided a deep insight in the complexflow field, such as propeller flow. Most ofthe actual knowledge on propeller flow, es-pecially regarding the characteristics of tur-bulence, derives from the application of suchmeasurement technique. (Min, 1978;Kobayashi, 1981; Cenedese et al., 1985;Jessup, 1989; Chesnack et al., 1998, Stellaet al., 2000). LDV, which is today routinelyused by the main research organizations inthe world and in different fields, will makeit possible in the future to gain allow also inthe future to obtain highly accurate and valu-able information on the mean and fluctuat-ing velocity field in complex flows to be usedfor physical modeling and to validate 2Dand 3D computer codes. However, like anyexperimental technique, beside its undeni-able advantages, LDV has some limitations:■ it can hardly give an idea of the spatial

characteristics of large coherent structuresthat are generally encountered in complexand separated flows, because of its singlepoint measurement nature,

■ it can induce significant errors on theintensity of unsteady vortical structures,due to its fixed location and time averagingnature,

■ it needs long periods of operation of thefacility to get a whole velocity fieldbecause of its point measurement nature,increasing the testing costs and leading


of two-dimensional velocity vectors for a pointin the flow are then combined to yield a three-dimensional velocity vector. By combiningthe vector fields from the two cameras, thethree-dimensional velocity field for the planein the fluid is computed.

In this present work Stereo-PIV is ap-plied in the analysis of a ship model withpropeller in the large INSEAN CirculatingWater Channel. The experiment is a partialreplica of the Toda et al. (1990) experiments:measurements are performed in several cross-planes behind a Series 60 C

b=0.6 ship model

with a 5 blades MAU propeller. The Series60 model was selected for the experimentsto complement the many previous studieswith this geometry.

Results of the wake survey are discussed,pointing out the measurement techniquecapabilities in resolving the wake structuresand outlining the major problems in apply-ing stereo-PIV in a large facility.

Experimental Set-upMeasurements were carried out at the

INSEAN Circulating Water Channel, a freesurface cavitation channel with a 10mlength, 3.6m width and 2.25 m depth testsection which has a capability of 5.2 m/smaximum flow speed. More informationregarding the facility’s capabilities can befound at http://www.insean.it

The ship-model was a Series 60 Cb=0.6,6.096 m length, conforming to the standardoffsets with some minor modification in thestern geometry to allow the propeller instal-lation. The propeller was a five blades MAUpropeller, with the following features: diam-eter D=221.9 mm, pitch-diameter ratio P/D

07=1.031, expanded area-disk area ratio A

e/

A0=0.74. Tests have been carried out at the

propeller angular velocity of 6.7 rps with thetunnel water velocity of 1.22 m/s, corre-sponding to a Froude number Fr = 0.16. Inthe following presentation of the results anddiscussion, a Cartesian coordinate system isadopted in which x,y,z axes are in the direc-tion of the uniform flow, respectively thestarboard side of the hull and the upward.Measurements were performed in threecross-planes orthogonal to the shaft and lo-

cated downstream of the propeller disk re-spectively at x/L.= 0.9997, 1.0000, 1.0187(Figure 1) where L is the model length. In areference frame with the propeller center, theabove measurement planes are located re-spectively at x/D= 0.59, 0.76 and 1.8. Inthe first plane, measurements also withoutthe propeller were performed to have roughinformation on the propeller inflow. Thelocations of the last two measurement planesare similar to the Toda et al. (1993) experi-ment, while the first one is very close to thepropeller and was selected to test the stereo-PIV capability because of the stronger ve-locity gradients expected in such a region.

The experimental set-up is shown in Fig-ure 1. The light sheet, generated by a two-head Nd-YaG laser, was delivered to themeasurement plane by means of underwa-ter optics. The repetition rate of the lasersheet was 10 Hz, with an energy output of200 mJ per pulse. A rotary 3600-pulse/revo-lution encoder supplied the actual propellerposition as an electrical trigger signal to thesynchronizer. This in turn provided the trig-ger signals to the two flash lamps and twoQ-switches of the double-head laser, as well

as to both cameras. Therefore, the imageacquisition was synchronized with the pro-peller angle. During the test campaign 129acquisitions at a given propeller angle wereperformed in order to obtain the mean ve-locity field. Propeller angles from 0° to 69°have been considered with a step of 3°.

Stereo-PIV System Set-upStereoscopic PIV uses two cameras to

view a flow field from two different perspec-tives so that, as for the vision of living crea-tures, an estimate of the out-of-plane veloc-ity component can be calculated. Eachcamera measures displacement nominallyperpendicular to its optical axes. Combin-ing the pair of the two-dimensional velocityvectors coming from the two cameras, forthe same physical point, it’s possible to com-pute the three-dimensional velocity vector.Therefore, repeating it for each point, it’spossible to obtain the whole three-dimen-sional velocity field for the laser sheet plane.

In the present experiment the angulardisplacement method was adopted for theoptical configuration of the stereo-PIV sys-tem. The stereo-PIV system consisted of twocameras with 1280x1024 pixels image sizeand a 12 bits gray level on each pixel. Thefirst one (left camera) was located 2 m down-stream from the ship model in an underwa-ter housing and the second one (right cam-era) outside the tunnel test section lookingat the measurement plane through the tun-nel access windows. To reduce the strongdiffraction and aberration due to the thickglass window and to the water/glass/air in-terfaces, a water-filled prism has been placedin front of the right camera.

The adopted configuration is not symmet-ric and has never before been presented in pre-vious stereo-PIV measurements. The main rea-sons for this choice are the following:■ symmetrical configuration—two camera

looking at the measurement planethrough the windows on the oppositesides of the test section—which guaranteesthe maximum accuracy (Weesterweeland Van Oord, 1999) because of the largeangle between the cameras, was notpractical due to the size of the facility,

FIGURE 1Experimental set-up and location of the measurementplanes


the limited length of the camera cablesand the need to control remotely thefocus and Scheimpflug angle of the farthercamera located in the opposite side of thetest section in respect to the control room.

■ with the adopted configuration theunderwater camera was directly measuringthe cross-flow components V and W,allowing the maximum accuracy in mea-suring such components with no needfor a remote control of the Scheimpflugangle which requires more complexityand a larger underwater case to host allthe necessary equipment. In this way, allthe errors related to the stereo reconstruc-tion are only confined to the evaluationof the longitudinal component.

■ the adopted configuration, with an anglebetween the right and left camera rangingfrom 36° to 40 ° depending on the mea-surement plane, allows a maximumerror around 5% in the evaluation of theout-of-plane component. Such errorrepresents the best achievable resultamong all possible optical configurationshaving the cameras on the same side ofthe test section. This result has beenassessed on a test bench by measuring aknown displacement of a referenceobject placed on a translation stage.

Image Analysis and StereoReconstruction

As a first step towards the determinationof the three velocity components at the mea-surement plane, the images were processedto obtain the vector fields viewed by the leftand right cameras. The acquired images werethen analyzed using an algorithm in whichthe window offset correlation method hasbeen implemented (Westerweel, 1997). Fur-thermore a recursive processing method wasused by implementing a hierarchical ap-proach in which the sampling grid was con-tinually refined and also the size of the in-terrogation windows was reduced during theiterations. During the iteration the interro-gation window was also weighted by usinga Gaussian function that was stretched inthe direction of the window offset to fur-ther improve the signal-to-noise ratio of the

correlation function (Di Florio et al., 2001).In the last iteration the windows were alsooverlapped to obtain a better reconstructionof the whole flow field especially in the re-gions with strong gradients. This procedurehas the added capability of applying inter-rogation windows in a smaller size than theparticle image displacement increasing boththe dynamic range and the spatial resolu-tion. To eliminate the remaining spuriousvectors, each data set was subjected to a vali-dation procedure to detect and replace spu-rious displacement vectors (Di Felice et al.,2000). For the results presented in the fol-lowing sections, a final window size of 32 x32 pixels, with 75% overlap between twoadjacent windows, has been adopted as thebest compromise in terms of spurious vec-tor reduction and spatial resolution. Thiswindow size was equivalent to 7 x 7 mm2 inreal space.

In the stereo reconstruction, the proce-dure described by Soloff et al. (1997) is used.The camera views are calibrated using a spe-cial target providing a mesh of 20 x 20 dotsin two planes. This calibration was requiredto determine the transformation functionneeded to reconstruct the 3 velocity compo-nents from the two separate planar PIV mea-surements. Besides the geometrical correctionof the perspective, this non-linear transfor-mation also took into account the optical dis-tortions introduced by the presence of mul-tiple interfaces (air, glass and water).

In the adopted configuration, the mea-surement area was defined by the overlap-ping region between separate views and hada dimension of 250 mm x 200 mm thatwas sufficient for investigating the wholepropeller disc.

The analysis of the acquired images pre-sented some difficulties due to the fact thatboth the left and right camera were imagingthe rotating propeller in the background.The propeller, even if painted black withcare, was scattering the light diffused by theparticles, especially for the plane x/L=.9997closer to the propeller. The scattered lightfrom the propeller was masking the particlesand locking the velocity at the propellerspeed especially in the regions in proximityto the hub and at the blade edges. This fact

leads to the erroneous evaluation of the 3Cvelocity field in the region extending alonga horizontal radius from the hub as shownin Figure 2a.

To overcome such a problem the imageshave been pre-processed: a mean image hasbeen calculated by using all the acquiredimages at a given angle and this referenceimage has been subtracted to the actual im-age before the analysis (Figure 2). This pro-cedure allows the elimination of the back-ground propeller and drastically improvesthe final result.

The preprocessing procedure was usedonly in the analysis of the measurementplane closer to the propeller. For the planesat x/L = 1.000 and 1.018 the preprocessing

FIGURE 2Image preprocessing: a) original image; b) meanimage over 129 acquisitions; c) final image with pro-peller removed in the background


of the images was not deemed necessarybecause of the larger distance between thepropeller and the measurement planes as wellas the small depth of field of the cameraobjectives that reduced the occurrence of theabove problem.

Measurements UncertaintyThe uncertainty of the velocity measure-

ments by each single camera is mainly dueto the error on the particle displacementevaluation. This is normally considered lessthan 0.1 pixel for the present image analysisalgorithm (Raffel et al., 1998), which isequivalent to approximately 4 cm/s in termsof velocity. This error is essentially presentin the measurement of the instantaneousflow field, in particular in the evaluation ofthe cross flow components which are directlymeasured by the underwater camera as ex-plained before.

The error in the measurement of the lon-gitudinal component is mainly related to thestereo reconstruction. In the present case,the configuration chosen for the experimenthas been assessed on a test bench and com-pared with a symmetrical standard arrange-ment. The angle between the camera look-ing normally to the measurement plane (i.e.the underwater camera) and the cameraplaced outside the observation section hasbeen varied from 10 to 70°. A target dis-playing a typical PIV image as in figure 2awas used and put on a flat plate, which wasthen moved along the normal axis by 1 mm.

The errors, expressed by the measure-ment standard deviation, were calculatedacross the complete measurement plane. Thefirst graph in Figure 3 shows measurementuncertainty from the three-dimensional re-construction on U, V and W as a functionof θ using the asymmetrical configuration,and corresponding values for the symmetri-cal configuration with θ =52°.

The results shown pointed out errorsunder 2.5% for the in-plane components Vand W, with values fairly constant across thewhole range of angles. The errors on thenormal component U were found to reacha minimum lower than 3% between 30 and60°, while the errors for all angles outsidethis range are more important, as one would

expect from a stereoscopic reconstruction.For indication, the error for the standardsymmetrical configuration is also displayedfor an angle of 104°, i.e. double the angleused in the experiment described here. Theerrors reported for this configuration arenaturally inferior for all the componentscompared to the respective errors registeredwith the asymmetrical setup.

However, the previous numbers couldbe underestimated in the facility where ob-jectives with a longer focal length are usedand with the additional window aberrations.The errors due to light reflections from thehub and from the blade edges were impor-tant in flow field regions mapped in the rightor left camera in proximity of these reflec-tion spots. The moving propeller in the back-ground of the measurement plane is anothersource of error. Even if pre-processing theimages mostly removes the propeller image

and reduces this effect, the correlation peakis still locked at the propeller velocity, in theregions where there is a lack of particle traces.In the post-processing phase, the validationprocedure is very effective to detect suchspurious vectors due to a large differencebetween the flow and the blade velocity es-pecially at the tip of the blade. Detected er-roneous vectors are eliminated and replacedby interpolation. Nevertheless, spurious vec-tors might also be validated biasing the sta-tistics especially in proximity of the hubwhere the flow and propeller velocity becomecloser. This effect is relevant especially forsecond order statistics.

Finally, the accuracy of the mean veloc-ity field definitely depends on the numberof acquired samples and on the shape of thevelocity probability distribution function.The probability density function in the tipvortex core and in the blade wake markedly

FIGURE 3Measurement uncertainty results


differs from a Gaussian distribution. Fur-thermore, a lack of data and fewer samplesavailable for statistics have been observed inthese regions. However by using the t-Stu-dent distribution (for which the confidenceinterval at 95%, is ±1.96*rms /Ö(N-1), withN=129), it is possible to estimate the uncer-tainty on a velocity component which isabout 1/6th of the measured velocity rms.

The second graph in Figure 3 representsthe uncertainty error as a function of the dis-placement along the in-plane axis correspond-ing to the V component. The target wasmoved in steps corresponding to 0.1 pixel.The plot shows two distinct trends, respec-tively for displacements under and above 0.5pixel. This behavior is documented elsewherein the literature, both experimentally andtheoretically (see Westerweel, 1997).

ResultsInstantaneous Flow Field

An example of an instantaneous flowfield obtained in the first measurement planefor the revolution angle q=0° is shown infigure 4a and 4b. For graphical reasons thevectors have been skipped of a factor four.

From the left field and the right field it’spossible to evaluate the 3D velocity fieldshown in figure 4c. The reconstructed flowfield shows many holes due to the fact thatthe number of spurious vectors is approxi-mately the union of the right field and leftspurious vectors. During the mean valuescalculation, performed using only validatedvectors, the holes in the reconstructed fieldare recovered as shown in figure 4d wherethe 3D field is shown after averaging over129 acquisitions.

Hull Wake and Propeller InflowBefore going into the details of the pro-

peller wake, and to better understand thebehavior of flow field during propeller rota-tion, in figure 5 the mean fields at x/L=0.9997measured without the propeller are shown.Mean field has been obtained averaging over500 acquisition. Although in the present casemeasurements of the propeller inflow werenot practical due to the limitation in the op-tical access with the Stereo-PIV, such infor-mation, even if obtained downstream of thepropeller plane, can give an idea of the nomi-nal wake and of the propeller inflow.

The mean velocity field is characterizedby a sharp and strong axial velocity defectdue to the diminishing cross-section of thehull at the stern. The cross flow generally isdirected upwards and towards the hull cen-ter-plane where it tends to roll up generat-ing a diffused circulation well resolved bythe vector field and vorticity field.

Figure 5 shows also the axial wake pro-files comparison with the results from Todaet al. (1990), performed at the same Froudenumber. The differences among the velocityprofiles is rather apparent with a wider thick-ness of the velocity defect in Toda et al. (1990).The difference can be related to the differ-ence in Reynolds number, that in the presentexperiment is almost double, as well as to thedifferent characteristics of the facilities.

Near Propeller Wake EvolutionIn the following, the evolution of the wake,

with the propeller installed, during the revolu-tion at x/L=.9997 and only for q = 0° alongthe longitudinal cross-sections will be discussed.

FIGURE 4Left and right camera instantaneous field with 3-D stereo reconstruction and mean field over 129 acquisitions

FIGURE 5Ship model wake without propeller at x/L =.9997.Comparison with Toda et al. (1990)


In figure 6 the evolution of the longitudi-nal component U for two propeller angles 0°and 45°, corresponding to a blade passage, isshown. The main characteristics of a propellerwake working with non uniform inflow dueto the hull wake can be pointed out:■ the propeller wake loses the axial-

symmetric morphology typical of anisolated propeller and the largest long-itudinal velocity is achieved in the lowerpart of the propeller disc and at radialpositions r/R=0.6-0.7.

■ the thin wake released by each blade canbe recognized in the longitudinal com-ponent plot even if the wake thicknessis approximately of the same order ofmagnitude of the spatial resolution ofthe PIV measurement leading tounderestimation of the velocity defectsdue to the smoothing effect

■ trace of the tip vortex are apparent inthe velocity map and are pointed out bythe strong velocity gradients at a differentangular position at about r/R= 0.9Tip vortices and blade wake are more

evident in the V and W contour plot andthis also demonstrates the importance of

the strong circulation generated by thehub vortex that characterizes the propel-ler slipstream.

The passage of the blade in the narrowwake of the hull at 12 o’clock generatesstrong three-dimensional effects. In theaxial velocity contour plot, when the bladeis in the range of 18°< q < 36° two parallelstripes of maximum and minimum veloc-ity near the blade tip can be noticed. Thisphenomenon points out the presence of astrong vortex structure having the axis par-allel to the measurement plane. The strongthree-dimensional effects due to the bladepassage in the sharp hull wake can also benoticed in the vorticity evolution shown infigure 7. From the previous figure the fol-lowing considerations can be also made:■ the trailing vorticity, shed from the blade

trailing edge, is well identified andconsists of two layers of opposite signwhich overlap at about r/R =0.7, in theblade section of maximum loading

■ The link of the tip vortex with the bladetrailing vorticity is more apparent withrespect of the mean velocity field wherethis information is almost confused or lost

■ the vorticity field provides informationon the radial distribution of the bladeloading and points out differencesamong the five blades due to the differentinflow through the blades.The strong variation of the propeller in-

flow produces important shape and inten-sity modifications of the wake released bythe blade during the revolution, as shownin figure 8, where the location of the wake,identified by the vorticity field, is reportedfor the different blades for the propeller angleq = 0°.

The strong modification of the wake’sshape points out the importance of consid-ering the angular variation of the propellerinflow which most of the time is ignored insome potential flow calculation by integrat-ing the nominal wake along the circumfer-ence at a given radius.

Figure 9 shows the evolution of the totalturbulence intensity distribution

where u’, v’, w’ are the standard deviations ofthe velocity components.

Even if the confidence of the statisticalestimator is limited, due to the fact that ithas been evaluated using only 129 samples,some important features of the wake can bepointed out:■ the maximum values of turbulence

intensity are achieved in the tip and hubvortex cores

■ the trace of the hull wake, representedby a vertical stripe of turbulence, can stillbe noticed in the measurement plane

■ the passage of the blade in the hull wakeproduces a strong turbulence generationdue to the interaction of the hull wakewith the blade tip vortex

■ the intense spikes in the turbulencelevel distribution located at the bladetrailing edge, near the hub, at differentpropeller angles are due to some spuriousvectors locked at the blade velocity.Such an effect is due to the backgroundmotion of the propeller and is notcompletely removed by the image pre-processing and by the vector post-pro-cessing validation as explained previously

FIGURE 6Longitudinal velocity U (m/s) at two different angles(0°, 45°) for plane x/L=0.999

FIGURE 7Vorticity at two different angles (0°, 36°) for planex/L=0.999


Longitudinal Wake EvolutionThe description of the wake longitudi-

nal evolution will be limited to the longitu-dinal velocity component U and the vor-ticity shown in figure 10.

The longitudinal evolution of the vor-ticity shows the following features on threedifferent transversal planes:■ very strong dissipation; The wake blade

is broadened very quickly from the firstto the second measurement plane. In thelast one the hub and the tip vortex arestill present, even if very attenuated, whilethe blade wake has almost disappeared

■ very strong distortion of the wake structuredue to a different pitch of the wake shedfrom the trailing edge and the pitch ofthe tip vortices can be noticed from thefirst to the second plane .

■ diffusion of the wake due to viscosity;this is most visible for the hub wake andthe tip vorticesThe evolution along the longitudinal axisof the axial velocity component U shows:

■ wake contraction between the first twoplane (they are very close together and

also close to the propeller plane)■ strong dissipation of the axial velocity

component between the first two planesand the last one which is furtherdown-stream.

■ stronger radial gradients of the axialvelocity component for the first twoplanes compared to the last one wherethe radial distribution of the axial velocityis smoother.

Comparison between PIVand LDV Techniques

Analysis of the average axial velocity,vorticity, and cross-flows shows a substan-tial agreement between the two techniques.The main error sources can be ascribed tothe residual light reflection, to the differ-ent spatial resolution of the LDV and PIVprobe and measurement meshes, and to thedifferent number of samples in the statisti-cal analysis. In the stereo-PIV measure-ments, the errors due to light reflectionsfrom the hub and from the blade edges wereimportant in flow field regions mapped in

the right or left camera in proximity of thereflection spots. The moving propeller inthe background of the measurement planeis another source of error. Even if the pre-processing of images mostly removes thepropeller image and reduces this effect, insome regions the correlation peak is stilllocked at the propeller velocity, as shownby the spurious vectors behind the hub andin correspondence of the lower starboardside of the measurement plane. In the post-processing phase, the validation procedureis very effective to detect these spurious vec-tors due to the large difference between theflow and the blade velocity, especially at thetip of the blade. Detected erroneous vec-tors are eliminated and replaced by inter-polation. Nevertheless, spurious vectorsmight also be validated biasing the statis-tics as, for instance, in proximity of the hubwhere the flow and propeller velocity aresimilar. This effect is relevant especially forthe second order statistics as confirmed bythe intense spikes in the iso-contours of thePIV data in correspondence of the bladetrailing edge at approximately 12 o’clock.

FIGURE 8Shape of trailing vorticity wake for each blade for propeller angle q= 0°

FIGURE 9Total turbulence at two angles (0°, 36°) for planex/L=0.99


The stereoscopic PIV measurements pro-vide a better accuracy than LDV in thereconstruction of the blade wake. This canbe explained as a consequence of the mea-surement grid used for the LDV cam-paign, which has a mesh size bigger thanthe wake thickness and hence leads toaliasing effects in the blade wake resolu-tion. This is not true for the stereo-PIVmeasurements where the use of 32 x 32pixels interrogation windows with 75%overlap generates a grid approximately half

of the wake thickness. This allows, withsufficient accuracy, resolution of the wakestructures released from the blade trailingedge. The effects of the different spatialresolution in the velocity field reconstruc-tion is apparent in the velocity profiles:the relative low resolution of the LDVmesh introduces a low pass spatial filterthat smoothes the velocity peaks and thevelocity gradients, thus overestimating theturbulent wake thickness. The differentintensity in the turbulence levels of theaxial component can be explained as aconsequence of the stereoscopic recon-struction effect which introduces, for thechosen configuration, an error for the out-of-plane component around 10%. A dif-ferent behavior and a lower deviationwould be noticed for the turbulence lev-els of the cross flow components measuredusing only the underwater camera in astandard 2D PIV configuration.

ConclusionsThe analysis of a propeller wake behind

a Series 60 ship model, in a large waterchannel, has been performed by using ste-reo-PIV. Both instantaneous and averagedvelocity fields are achieved, the latter afterphase sampling averaging over the sameangular position of the propeller blade. Theexperimental results, in terms of velocityand vorticity fields, reveal some of the dif-ferent contributions to the complex pro-peller flow field of a propeller working be-hind a ship:1. The viscous part of the wake generatedby the boundary layers on the blade’s surfaces.2. The potential part of the wake derivingfrom the vortex sheet at the blade’s trailingedge.3. The varying loading condition of theblade during the revolution which causes astrong wake deformation4. The complex and three-dimensionalbehavior of the tip vortex when passingthrough the narrow hull wake.5. The rapid broadening of the propellerwake in the downstream evolution which isfaded and smoothed by the turbulent diffu-sion and viscous dissipation.

FIGURE 10Longitudinal Wake Evolution of vorticity at 0° propellerangle for planes X/L=0.999, 1.00, 1.018

From the experimental setup point ofview and with regards to the future imple-mentation in standard ship model testingprocedures, the stereo-PIV has shown a num-ber of advantages in comparison to the wellassessed LDV technique. In particular, con-sidering the limited time usually given tothese tests combined with management andtechnical difficulties typical of a large testingfacility, the PIV technique can provide re-sults within a short period. Instead, the LDVtechnique requires up to three or four timesmore testing time to obtain the same infor-mation, which consequently translates intoadditional costs of facility occupancy. Themeasurement time is drastically reduced withthe stereo-PIV method, where the plane ofmeasurement is mapped instantaneously andprovides all three velocity components in onesingle step, while the LDV technique requiresa long scanning of the interrogation domain.In this sense, the PIV approach offers thefreedom of extending the wake survey to alarger number of areas of interest, with verylimited setup changes.

The major drawback of the PIV tech-nique is a reduced accuracy with respect tothe LDV technique as well as the hugeamount of information generated both atthe measurement and at the processing time(about 200 Gb in the case of the presentedresults): one must address the critical prob-lem of storing, managing and processing thisinformation without compromising the testscosts by extended data processing time.

AcknowledgementsThe authors are grateful to the INSEAN

Circulating Water Channel personnel andto Mr. Tiziano Costa who supported the PIVmeasurements.


ReferencesCalcagno, G., Di Felice, F., Felli, M, Pereira, F.

2002. Propeller wake analysis behind a ship by

stereo PIV. Proc. of the 24th Symposium on

Naval Hydrodynamics, Fukuoka, Japan. 8-13

July 2002.

Cenedese, A., Accardo, L., Milone, R. 1985.

Phase sampling techniques in the analysis of a

propeller wake. Proc. of the International

Conference on Laser Anemometry Advances

and Application, Manchester UK.

Chesnack C., Jessup S. 1998. Experimental

characterization of propeller tip flow. 22th

Symposium on Naval Hydrodynamics,

Washington D.C.

Cotroni A., Di Felice, F., Romano, G. P.,

Elefante M. 1999. Propeller Tip vortex analysis

by means PIV. 3rd International Workshop on

PIV, Santa Barbara, CA.

Di Felice, F., Felli, M., Ingenito, G. 2000.

Propeller wake analysis in non uniform inflow

by LDV. Proceedings of the Propeller and

Shafting Symposium, Virginia Beach. VA

Esposito, P., Salvatore F., Di Felice F., Ingenito

G., Caprino G. 2000). Experimental and

numerical investigation of the unsteady flow

around a propeller. 23rd Symposium on Naval

Hydrodynamics, Bassin D’Essais Des Carènes,

France.

Felli, M., Di Felice, F., Romano, G.P. 2000.

Installed Propeller wake analysis by LDV:

phase sampling technique. Proceedings of the

9th International Symposium on Flow

Visualisation, Edinburgh.

Jessup, S.D. 1989. An experimental investiga-

tion of viscous aspects of propeller blade flow.

Ph.D. Thesis. The Catholic University of

America, Washington D.C.

Keane, R.D., Adrian R.J. 1992. Theory of

cross-correlation analysis of PIV images. Appl

Sci Res. 49:191-215.

Keane, R.D., Adrian, R.J. 1990. Optimization

of Particle Image Velocimeteres. Part 1: Double

pulse system. Meas Sci Technol. 1:1202-1215.

Kobayashi, S. 1981. Experimental methods for

the prediction of the effects of viscosity on

propeller performance. Dept. of Ocean

Engineering, Rep. 81-7, MIT.

Kobayashi, S. 1982. Propeller wake survey by

laser Doppler velocimeter. Proc. of the

International Symposium on the Application

of Laser-Doppler Anemometry to Fluid

Mechanics, Lisbon.

Meyers, J.F. 1991. Generation of Particles and

Seeding. Von Kárman Institute for Fluid

Dynamics, Lecture Series 1991-05, Laser

Velocimetry, Brussels.

Min, K.S. 1978. Numerical and experimental

methods for prediction of field point velocities

around propeller blades. Dept. of Ocean

Engineering, Report no. 78-12, MIT.

Raffel, M., Willert, C., Kompenhans, J. 1998.

Particle Image Velocimetry. Springer, ISBN 3-

540-63683-8.

Stella, A., Guj, G., Di Felice, F., Elefante, M.

1998. Propeller wake evolution analysis by

LDV. Proc. of the 22nd Symposium on Naval

Hydrodynamics, Washington, D.C.

Stella A., Guj G., Di Felice F., Elefante M.

2000. Experimental investigation of propeller

wake evolution by means of LDV and flow

visualization. J Ship Res. Sept. 2000:155-169.

Toda, Y., Stern, F., Tanaka, I., Patel, V.C.

1990. Mean Flow measurements in the

boundary layer and wake of a series 60 C

b=0.6 model ship with and without propeller.

J Ship Res. 34(4):225- 252.

Wang, M.H. 1985. Hub effects in propeller

design and analysis. Dept. of Ocean Engineering,

Report. 85-14, MIT.

Westerweel, J. 1994. Efficient detection of

spurious vectors in particle image velocimetry

data. Exp Fluids. 16:236-247.

Westerweel, J. 1997. Fundamentals of Digital

Particle Image Velocimetry. Meas Sci Technol.

8:1379-1392.


A U T H O RMark GustafsonUniversity of Rhode Island


Shipboard Scientific Dive Van—Meeting theStandard for UNOLS Portable Scientific Vans

17-year history the van was used primarilyon board URI’s research vessel Endeavor andWHOI’s research vessel Oceanus. While onboard these two vessels the van was used bynumerous divers from various institutions:Yale University, the University of Connecti-cut, Northeastern University, Brown Uni-versity, The New England Aquarium, Or-egon Coast Aquarium, Oregon StateUniversity, the University of Maryland, theU.S. Geological Society, and the Universityof New Hampshire.

However, the operation of URI’s dive vanwas quickly coming to a close with the on-set of new standards for scientific vans. Lo-gistic integration and containerizationwithin the maritime industry helped reshapeUNOLS’s standards for scientific vans usedon board the academic fleet. It became ap-parent that the older van would not meetthe new standard and therefore a replace-ment for the van with a state-of-the-art div-ing system was initiated.

UNOLS StandardizesVan Specifications

The incentive in standardizing portablescientific vans is to allow for a common de-sign where the container, or scientific van,can easily be interchangeable throughout theUNOLS fleet. In addition to these special-ized laboratories, the standardized van wouldhave a minimal mobilization time based onthe 20-foot ISO shipping container and canbe easily and economically transported by aland carrier (M. Hawkins, 2001, pers.comm.) and lifted on board vessels using aship’s crane.

Meeting the StandardThe proposed diving system that was to

be incorporated within the van was designedto be simple, logical and user friendly. Thecentral component to the van is its compres-sor, manufactured by Undersea BreathingSystems, and uses a semi-permeable mem-brane to produce oxygen-enriched air, or tocoin a phrase from the SCUBA community,Enriched Air Nitrox (EAN). EAN is pro-duced when the ratio of oxygen to nitrogenis greater than 21 percent. There are a num-ber of advantages to using Nitrox: 1) it helpsin preventing decompression sickness; 2) itlessens post-dive fatigue; 3) it can shortensurface intervals between dives; and 4) it canincrease the no-decompression limits allow-ing for longer bottom times. The theorybehind the membrane system is instead ofadding pure oxygen into the SCUBA cylin-der and then mixing it with air, the mem-brane system removes nitrogen from the airthrough a series of fibrous membranes us-ing a low-pressure compressor, then the en-riched oxygenated air is pressurized with asecond compressor in order to fill SCUBAcylinders to the desired output pressure, aprocess known as de-nitrogenation. Simplyput, the oxygen molecules pass through themembrane easier than the nitrogen mol-ecules, allowing for different concentrationsof oxygen as the final product. The mem-brane system can produce a Nitrox mix withoxygen concentrations varying from 22 per-cent to 40 percent at a maximum pressureof 3600 psi. The operator simply dials-inthe appropriate mix and the compressor doesthe work. The excess nitrogen that is re-moved from the air is then purged to the

OI N T R O D U C T I O N ver the years it has been well established that the scientific diveris an important and effective tool in marineresearch. It can be further established thatscientific diving within the euphotic zoneof deep oceanic waters from a UNOLSvessel is a common technique used to col-lect data during a research cruise. Theamount of diving that is done from a re-search vessel can vary in intensity from 2to 3 people diving per day to as many as10 to 15 people diving per day. The equip-ment used to support these efforts are con-siderable, from the ship’s crew thatlaunches, operates, and recovers the diveboats, to the instrumentation the scien-tists bring on board to support their div-ing operations. The area(s) of operationcan extend to hundreds of miles fromshore, therefore critical to the success of acruise that supports scientific diving is thedive locker that houses the compressor andsupporting equipment. Since mostUNOLS vessels do not have permanentdive lockers on board, the ability of hav-ing an enclosed portable scientific van de-signed solely to support diving operationsand made available to the UNOLS fleetcan make an enormous difference in theproductivity of the cruise.

HistoryThe University of Rhode Island’s Re-

search Dive Program has provided a portabledive van to scientists since 1985, as a sharedfacility through the New England Consor-tium of Oceanographic Research. Over its


outside through a fitting in the van’s bulk-head. The key advantage to the membranesystem is that large amounts of pure oxygendo not have to be stored and then mixed withair to achieve the desired Nitrox mix. Elimi-nating the need to store large volumes of high-pressure oxygen can significantly increase thesafety to those on board the vessel.

The high-pressure end of the compressordelivers 9 cubic feet per minute of Grade EAir or Nitrox at a maximum pressure of 3600psi. The diving system can fill four SCUBAcylinders directly from the compressor, orfrom a high-pressure bank of four 350 cubicfoot cylinders. The high-pressure componentsfor this two-gas system are completely iso-lated from each other, including two separatebanks: one for air the other for Nitrox. Thefill manifold has four whips with a DIN toyoke adapters. The compressor is positionedalong the centerline of the van on an alumi-num forklift frame six inches above the van’sdeck and can be easily removed through a setof double aluminum doors.

The van, manufactured by Sonic Enclo-sures is constructed of aluminum allowingfor weight and maintenance considerations.It measures 20 feet long, 8 feet wide and 8-1/2 feet in height, and uses ISO specifica-tions for securing the van to the deck of aship. The gross weight of the van is approxi-mately 11,000 pounds, including the com-pressor and high-pressure banks. The elec-trical system (now standard to UNOLSScientific Vans) has two electrical services, aship supply of 460/240 volts 3 phase, and ashore supply of 120/208 volts 3 phase. Ei-ther service can provide power to the vanand its components.

Some unique features include a Polymercoating that waterproofs the sub floor withdrains in each corner. Above the waterproofcoating is a fiberglass deck with an anti-slipsurface. There is an interior exhaust ventila-tion system with a thermostat control thatcan circulate 1,300 cfm of air with a manualdamper and a removable weather plate onthe exterior side. Spare parts for the com-pressor, and SCUBA equipment along withtools are stored in a workbench with a hardmaple surface, a 4-inch vise and steel cabi-net drawers. Vertically mounted on the in-

terior walls are stainless steel channelmounts, 24-inches on center to secureSCUBA cylinders, high-pressure bank cyl-inders and scientific gear.

ConclusionThe development of standardized scien-

tific vans as a shared facility within theUNOLS fleet has simplified the logistics andeconomics of transporting these specializedlaboratories between institutions and re-search vessels. Meanwhile, marine scientistscontinue to make advances in scientific andtechnical diving. The use of Nitrox is justone example, where it extends bottom time,allows for shorter surface intervals and in-creases one’s safety margin. However, thesuccess of any cruise depends upon thepeople and equipment that is on board. Thisis especially true when the science involvesSCUBA diving. With this new generationof scientific vans on the horizon, having onethat is portable, easily mobilized, and withNitrox capabilities can only enhance thedivers’ ability to safely work underwater.


A U T H O R SPeter C. ChuMichael CorneliusNaval Ocean Analysis and PredictionLaboratoryDepartment of OceanographyNaval Postgraduate School

Mel WagstaffNaval Oceanographic OfficeStennis Space Center


Effect of Suspended Sediment onAcoustic Detection Using Reverberation

A B S T R A C TSonar operates by ensonifying a broad swath of the seabed using a line array of

acoustic projectors with acoustic backscattering from the ensonified sediment. The sus-pended sediment layer affects the sonar imagery through the volume scattering strength.Understanding the acoustic characteristics of the suspended sediment layer can aid theNavy in detecting sea mines with sonar imagery. In this study, the Navy’s Comprehen-sive Acoustic Simulation System is used to investigate such an effect. A range of criti-cal values of volume scattering strength for buried object detection is found throughrepeated model simulations.

The suspended sediment layer occupiesthe lower water column. Presence of the sus-pended sediments creates a volume scatter-ing layer which affects acoustic detection.Understanding the acoustic effects of thesuspended sediment layer leads to the de-velopment of acoustic sensors with capabil-ity to scan the seafloor and to detect ord-nance such as sea mines.

2. Comprehensive AcousticSimulation System

Sonar equations provide guidelines forsystem design (Urick, 1983). The govern-ing equation for beam patterns with domi-nating volume reverberation is given by

SL – TLi – TL

r + TS - RL = SNR

(1)

where SL is the source level; TLi is the

transmission loss of the incident wave;TL

r is the transmission loss of the re-

flected echo; TS is target strength; RL isthe reverberation level of sediments; SNRis the signal-to-noise ratio of the sonardata. The transmission losses of the inci-dent and reflected waves, TL

i and TL

r,

account for spherical loss (such as spheri-cal spreading), acoustic attenuation, andboundary loss.

Volume reverberation depends on thephysical properties of the water column.With the presence of a suspended sedimentlayer, large quantities of sediment remain inthe water column and significantly affect theacoustic transmission in the water. Thedenser the suspended sediment layer is, theharder it would be for sonar to penetratethrough the water. Moreover, suspendedbottom sediment layer increases the densityof the lower water column and in turnchanges the sound velocity profile and pre-vents acoustic energy from reaching a pos-sible buried object.

Reverberation can be used to represent theeffects of a suspended sediment layer on acous-tic detection. The differential form of the re-verberation can be obtained from integrationover the ensonified area (Keenan, 2000),

d(RL) = SL + 10 log (SA) + SS – TLt - TL

f

+ BPt + BP

r (2)

where SA is the scattering area; SS is the scat-tering strength per unit area; TL

t is the trans-

mission loss to scatterer; TLf is the transmis-

sion loss from scatterer; BPt is the beam

pattern to scatterer; and BPf is the beam pat-

tern from scatterer. The most importantdesign criterion for detecting mine-like ob-jects is to maximize the SNR, the target echoto scattering noise ratio in decibels.

AI N T R O D U C T I O N coustic detection of undersea objects is difficult due to the uncertain en-vironment (Chu et al., 2002, 2004) andeven more difficult when the objects areburied in the seabed. First, sediments gen-erate high backscattering noise due to het-erogeneous scatters within the sedimentsclouding the object.

Second, the acoustic wave attenuationin sediments is much higher than in water.Acoustic shadows make the buried targetsabsent in the sonar images due to diffrac-tion around the target, transmission throughthe target and relatively high acoustic noisedue to backscattering from sediments sur-rounding the target. Classification of bur-ied targets is also more difficult since thereare no shadows, and the images do not con-tain much information about target shapesince scattering from oblique target surfacesis not detectable.

Acoustic images of buried targets pri-marily consist of echoes from the target sur-faces that are normal to the incident acous-tic ray path. Target surfaces with an obliqueaspect to the incident ray path will back-scatter much less energy at the lower oper-ating frequencies of sub-bottom profilerssince the acoustic wavelength is muchlonger than the surface roughness of mosttargets of interest.


The Comprehensive Acoustic Simula-tion System (CASS) is the Navy’s standardmodel for acoustic and sonar analysis. It in-corporates the Gaussian Ray Bundle(GRAB) eigenray modes to predict range-dependent acoustic propagation in the 150Hz to 100 kHz frequency band (Keenan etal., 1996; Keenan and Weinberg, 2001).

CASS contains several equations forsound speed conversions. The current OAMLapproved Sound Speed Algorithm (Chen andMillero, 1977; Millero and Li, 1994) has beenincorporated into CASS. The Chen-Millero-Li equations compute sound speed based ondepth, temperature, and salinity. The Chen-Millero-Li, Wilson (1969), and Leroy (1969)equations are all very close in the salinity range30 ppt to 40 ppt. For lower salinities theChen-Millero-Li equation should be used.Near shore off the coast of Louisiana may havesalinity variability especially near the Missis-sippi Delta. Thus, the Chen-Millero-Li equa-tions are used in this study.

CASS simulates the sonar performancereasonably well in the littoral zone with givenenvironmental input data, such as bottomtype, sound speed profile and wind speedand accurate tilt angle of the sound source(Chu et al., 2002, 2005). CASS successfullymodeled torpedo acoustic performance inshallow water exercises off the coast of South-ern California and Cape Cod. Recently,CASS was used to simulate mine warfaresystems performance in the fleet (Keenan etal., 1996), and for AN/SQQ-32 mine hunt-ing detection and classification sonar.

CASS calculates the reverberation innested do loops, seven deep. Reverberation isa function of time and the inner loop col-lected all the reverberation contributions overthe user-requested sampling times. There aretwo loops on eigenray paths, one for the pathsconnecting the transmitter to the scatteringcell and the other for the paths connectingthe scattering cell to the receiver. Since thereverberation is calculated in the time domainand there may be contributions in the sametime bin from different ranges, the next loopincrements the range. CASS combines thefive possible eigenray paths at each range stepand decides if the ray paths contribute to thereverberation time bin (Keenan, 2000).

Test rays are sorted into families of com-parable numbers of turning points andboundary interactions. Ray properties arethen power averaged for each ray family toproduce a representative eigenray of thatfamily. Target echo level and reverberationlevel are computed separately, and sub-tracted to get the signal-noise ratio in theabsence of additive ambient noise—noiselevel is typically power summed with thereverberation level for total interference. Adetection threshold is applied to computeSE, and then the peak signal is used to de-termine SNR (Figure 1).

3. Environmental andAcoustic Parameters

A nearshore location with silty clay bot-tom on the Louisiana shelf is selected.Horizontal extension of the area is 50 m × 60m. Total depth including water and sedi-ment is around 100 m. A hydrographicsurvey was conducted in the area(Cornelius, 2004). The sound speed in-creases a little from 1520 m s-1 near theocean surface to 35 m depth, and reducesdrastically to 1510.5 m s-1 at 55 m depth.Below 55 m depth, the sound speed de-creases slightly with depth and reaches 1510m s-1 at 100 m.

Suppose the sonar is towed at a depth of30.4 m with a source level of 240 dB. The wa-ter depth in the area varies from 77 to 95 m.The extent of the image is approximately 60 min the y-direction and 50 m in the x-direction.The grain size index for a silty clay bottom is 8from the Naval Oceanographic Office’s stan-dard. This parameter is related to specificgeoacoustic parameters of bottom reflection(APL-UW, 1994). The bottom reflection ef-fects are modeled using the Rayleigh model.

The water column is assumed to be rela-tively clear above 77 m depth with a vol-ume scattering strength of -95 dB. Thesource of the water column scatteringstrength is extracted from a volume scatter-ing strength database (CNMOC, 2004). Asuspended sediment layer is present below77 m depth characterized by a different scat-tering strength (-65 dB). The source level is240 dB. The sonar frequency is 100 kHz.The pulse length is 0.001 seconds. The timeincrement for modeling should not exceedone half of the pulse length to achieve properresolution of each time step. Since the totaldistance traveled from the sonar to the endof the image is approximately 50 m, the to-tal reverberation time is only 0.12 seconds.The maximum number of bottom and sur-face reflections is set at 30 to allow interfer-ence with reflected eigenrays.

FIGURE 1Steps taken to create a total reverberation image from CASS, compared to the side scan sonar image.


Test rays are sorted into families of com-parable numbers of turning points and bound-ary interactions. Target echo level and rever-beration level are typically computed separately,and then subtracted to get the signal-noise ra-tio. In the absence of additive ambient noise,the signal-noise ratio is typically power summedwith the reverberation level to calculate totalinterference. A detection threshold is appliedto compute SE, and then the peak signal isused to determine SNR (Figure 1).

4. Mine-Like ObjectLet a hollow mine-like steel object (8 m

× 5 m × 2 m) be placed near the center ofthe area with a height of 2 m. For the object,the grain size index is changed to -9 (clay)and the target strength is set as -35 dB. Thebathymetry is also changed to represent theexistence of the object. The water depth iskept the same (87 m) in the vicinity of theobject, and changed into 85 m over the object(Figure 2).

CASS is integrated with the sonar param-eters, sound speed profile (SSP), bottom type,bathymetry, and the scattering characteristics.The model output of the seafloor reverbera-tion is used to represent the model generatedsonar imagery (MGSI). An increase of thevolume scattering strength in the lower watercolumn reflects the presence of the suspendedsediment layer. Clearly, the object is visible inthe reverberation imagery. Since MGSI (Fig-ure 3) is a replica of the sonar image, the ef-fect of suspended sediment on acoustic de-tection may be studied using MGSI.

5. Effect of SuspendedSediment

Suspended sediment increases the vol-ume scattering strength. To simulate its ef-fect, the volume scattering strength is in-creased by an increment of 5 dB from thevalue of -65 dB below 78 m depth whilekeeping the volume scattering strength con-stant (-95 dB) above 78 m depth, and CASSis integrated with increasing the value of thevolume scattering strength to investigate theeffect of the suspended sediment on theobject detection.

FIGURE 2Bathymetry with mine-like object.

FIGURE 3Bottom reverberation with mine-like object inserted. Here, the horizontal-axis is cross track indices representedby time and the vertical-axis is along track indices represented by distance.


The following procedure is used to de-termine the criterion of the volume scatter-ing strength for the object absent fromMGSI. If the simulated mine-like object isstill visible in MGSI after increasing the vol-ume scattering, the simulated suspendedsediment layer is not strong enough to rep-resent a layer that would prevent a mine-like object from sonar detection. So the vol-ume scattering strength is increased further.This procedure is performed until the mine-like object is no longer visible. During thesimulation, the water column is assumed tobe relatively clear above 77 m depth, withan initial suspended sediment layer charac-terized by a slightly stronger scatteringstrength below 78 m.

Sediment in the water column wouldincrease the volume scattering and ulti-mately the volume reverberation. ForMGSI with an object, a critical value ofthe volume scattering strength can be de-termined below 78 m depth to render themine-like object undetectable. Volume at-tenuation and changes in the sound veloc-ity profile will also have an effect, but theyare not addressed in this work.

As the volume scattering strength of thesediment layer (below 78 m depth) in-creases to a value of -30 dB, the object be-comes nearly undetectable. The CASSmodeling continues with the increase of asmaller increment of 1 dB. The mine-likeobject is completely obscured (Figure 4) bythe suspended sediment layer at a value of-22 dB (below 78 m depth), which is takenas the threshold.

This threshold (-22 dB) is large com-pared to existing observational data. For ex-ample, Kringel et al. (2003) measure theacoustic volume scattering strength in WestSound, Orcas Island, Washington, using thebenthic acoustic monitoring system. It is abottom-mounted, radially scanning sonardesigned to record high-frequency scatter-ing from the seafloor. The acoustic volumescattering strength is measured from 22 mof water column with 0.5 m vertical and 2min temporal resolution. Their measure-ment shows that the volume backscatteringstrength varies between -75 dB to -25 dB.This indicates that the threshold (-22 dB) is

in the high end, which implies that highsuspended sediment density is needed tocompletely block the mine-like object fromacoustic detection.

6. Conclusions(1) CASS is used to investigate the effect ofsuspended sediment on detecting a mine-like object in the silty clay bottom at theLouisiana shelf with water depth around 100m. Hydrographic and meteorological sur-veys were conducted. The environmentaldata (wind and SSP) are taken as the inputinto CASS to simulate the sonar image witha mine-like object present through its rever-beration characteristics.(2) A threshold of volume scattering strength(-22 dB) for the mine-like object detectionis found through repeated model simula-tions. When the volume scattering strengthincreases to the threshold, the mine-likeobject is acoustically undetectable. It is noted

that this threshold is very large and valid onlyfor this case and that the purpose of thisstudy is to show the methodology rather thanto provide an accurate threshold.(3) While CASS is useful, several short-falls remain in this study. First, the pro-cess by which the environment and theobject are modeled is cumbersome. Sec-ond, the appropriate volume scatteringstrength for the buried object should begiven. A thorough study is suggested onthe relationship between the suspendedsediment layer density and type (e.g., sand,silt or clay), particle density in the layer,associated volume scattering strength andattenuation, and changes in the soundspeed profile.

AcknowledgmentsThe Office of Naval Research, Naval

Oceanographic Office, and the Naval Post-graduate School supported this study.

FIGURE 4Reverberation plot of bottom with mine object inserted. The horizontal axis is cross track indices represented bytime and the vertical axis is along track indices represented by distance with mine object inserted and sus-pended sediment layer with -22 dB volume scattering strength.


ReferencesAPL-UW. 1994. High-Frequency Ocean

Environmental Acoustics Models Handbook.

APL-UW TR 9407, Applied Physics Laboratory,

University of Washington, Seattle, 210 p.

Chen, C. T. and Millero, F. J. 1977. Speed of

Sound in Seawater at High Pressures. J Acoust

Soc Am. 62(5):1129-1135.

Chu, P.C., C.J. Cintron, S.D. Haeger, D.

Schneider, R.E. Keenan and D.N. Fox. 2002.

Yellow Sea acoustic uncertainty caused by

hydrographic data errors. In: Impact of Littoral

Environment Variability on Acoustic Prediction

and Sonar Performance, eds. N.G. Pace and F.

B. Jensen. pp. 563-570. Boston: Kluwar

Academic Publishers.

Chu, P.C., M.D. Perry, E.L. Gottshall, and

D.S. Cwalina. 2004. Satellite data assimilation

for improvement of Naval undersea capability.

Mar Technol Soc J. 38(1):12-23.

Chu, P.C. and N.A. Vares. 2005. Variability in

shallow sea acoustic detection due to environ-

mental uncertainty, U.S. Navy Journal of

Undersea Acoustics, in press.

Commander Naval Meteorology and

Oceanography Command (CNMOC). 2004.

Oceanography and Meteorology Master

Library, Stennis Space Center, MS.

Cornelius, M. 2004. Effects of a Suspended

Sediment Layer on Acoustic Imagery. MS

Thesis in Meteorology and Physical Oceanog-

raphy, Naval Postgraduate School, Monterey,

CA, 48 pp.

Keenan, R. E., H. Weinberg, F.E. Aidala.

1996. Modeling Cape Cod site C and site D

torpedo reverberation data with CASS. Naval

Undersea Warfare Center Division, Newport,

RI, NUWC-NPT Technical Report 10,590,

11 July 1996.

Keenan, R. E., 2000. An introduction to

GRAB eigenrays and CASS reverberation and

signal excess. Proceedings on IEEE/MTS

Oceans 2000, 6 pages (in CD Rom). 11-14

September 2000, Providence Rhode Island.

Keenan, R. E. and H. Weinberg. 2001.

Gaussian ray bundle (GRAB) model shallow

water acoustic workshop implementation. J

Comput Acoust. 9:133-148.

Kringel, K., P.A. Jumars and D.V. Holiday.

2003. A shallow scattering layer: High-

resolution acoustic analysis of nocturnal

vertical migration from the seabed. Limnol

Oceanogr. 48(3):1223-1234.

Leroy, C.C. 1969. Development of Simple

Equations for Accurate and More Realistic

Calculation of the Sound Speed in Sea Water.

J Acoust Soc Am. 46(1):216-226.

Millero, F.J. and Li, X. 1994. Comments on

“On Equations for the Speed of Sound in

Seawater”. J Acoust Soc Am. 95(5):2757-2759.

Naval Oceanographic Office Systems

Integration Division. 1999a. Software Design

Document for the Gaussian Ray Bundle

(GRAB) Eigenray Propagation Model.

OAML-SDD-74. Stennis Space Cneter, MS.

Naval Oceanographic Office Systems

Integration Division. 1999b. Software

Requirements Specification for the Gaussian

Ray Bundle (GRAB) Eigenray Propagation

Model. OAML-SRS-74. Stennis Space

Center, MS.

Urick, R.J. 1983. Principles of Underwater

Sound, McGraw-Hill, New York, 423 pp.

Weinberg, H., and R. E. Keenan. 1996.

Gaussian ray bundles for modeling high-

frequency propagation loss under shallow

water condition. J Acoust SocAm.

100(3):1421-1431.

Wilson, W.D. 1960. Equation for the speed of

sound in sea water. J Acoust Soc Am.

32:1357-1367.


A U T H O R SC.A. BaransM.D. ArendtMarine Resources Research InstituteSouth Carolina Department of NaturalResources, Marine Resources Division

T. MooreSkidaway Oceanographic InstituteUniversity of Georgia

D. SchmidtLJT & Associates, Inc.

P A P E R

Remote Video Revisited: A Visual Technique forConducting Long-term Monitoring of Reef Fisheson the Continental Shelf

A B S T R A C TFisheries observation data collected at similar times and frequencies as hydrographic

data could contribute significantly to the development of models for predicting responsesof fish assemblages to changing environmental conditions. Traditional collections of fish-ery observation data rarely occur with sufficient temporal replication for modeling; how-ever, remotely collected underwater video data represents a promising remedy. In August1999, an experimental underwater fish habitat and video data collection system wereestablished on the middle continental shelf off the coast of Georgia. Short (10 s) videodata files were collected hourly during daylight and transmitted 72 km to shore by micro-wave once daily. An extraordinarily large and valuable dataset was generated despite tech-nical problems, which precluded data collection during more than half of the days withinthe 1999-2002 study period. Evaluation of 5,590 usable video files from 429 observationdays resulted in documentation of presence, relative abundance and behavior for at least50 species, including several highly migratory pelagic species for which little scientificdata exists. Future efforts of this ongoing study will attempt to mitigate technical prob-lems discovered during 1999-2002. Expansion of automated visual sampling to numer-ous index stations along the continental shelf has the potential of greatly complementinginfrequent and expensive sampling cruises to collect ecological and behavioral data.

SCUBA (Brock, 1954; Hobson, 1972;Parker et al., 1994), submersibles (Parker andRoss, 1986), or remotely operated camerasystems (Powles and Barans, 1980;Baamstedt et al., 2003; Stevens, 2003), thereare few physical limits, other than water clar-ity, to where photographic observations canbe recorded. Unfortunately, none of theseinnovative visual techniques have providedlong-term, continuous observations, whichcould be used in developing physical-bio-logical models to predict relationships be-tween fish abundance and environmentalfactors, such as those reported by Runge etal. (2004). Furthermore, moderate sea statesare necessary to use such techniques; thus,sampling schedules and the temporal conti-nuity of data can be greatly compromisedby weather.

In contrast to point-in-time visual sam-pling, visual observations at fixed locations(“Eulerian” approach) using remote, semi-

permanently deployed data collection sys-tems enable sampling over extended peri-ods (due to reduced weather-dependence)under more ‘natural’ conditions. Semi-per-manently deployed systems become part ofthe landscape with time; thus, attraction toor avoidance of recently introduced remotecontrolled camera systems by many species(Parker et al., 1994) is reduced. Similarly,biased fish behavior due to SCUBA diverpresence and “bubble curtains” (partiallymitigated with re-breather systems, Lobel,2001) is also not a factor with semi-perma-nent systems.

An early attempt to use a remote videodata collection system occurred off of NorthBimini Island in the early 1970’s (Smith andTyler, 1973). The camera system, located in17 m of water with a 10 m field of view, waswired to and controlled from a shore-basedlaboratory 1.6 km away (Steinberg et al.,1965; Myrberg, 1972; Smith and Tyler,

DI N T R O D U C T I O N irect observation in the marine environment provides several sam-pling advantages with respect to samplinginvolving traditional sampling gears. Dur-ing daylight, direct observation generallydocuments more species (and individuals)of management interest than other samplinggears (Grace et al., 2000; Cappo et al.,2004). Direct observations result in mini-mal damage to habitats and species loss,making this approach especially valuable insensitive areas difficult or otherwise impos-sible to sample (Barnett and Pankhurst,1996; Westera et al., 2003). Direct obser-vation is one of the only methods that en-ables documentation of species interactions,which may subsequently affect catch com-position due to changes in distributionalhabits or altered ability to collect species.Given these advantages, recording directobservations through photography has beenpursued by both terrestrial (Sanderson andTrolle, 2005) and marine (Stryke) scientistsfor more than a century.

Underwater photography in the marineenvironment has largely been limited topoint-in-time efforts. Through the use of


1973). Technological advancements such asradio-signal control and improved data col-lection and storage abilities have made suchstudies more feasible; however, most effortscontinue to be conducted close to shore andfocus on general public (CWC; Collins)rather than scientific (Nelson) interests. Todate, long-term efforts to apply this fixedstation approach for conducting research atremote locations in the marine environmenthave not been reported.

In August 1999, an automated videosystem was established to monitor fish as-semblages at a small artificial reef and, even-tually, correlate fisheries video data withhydrographic information as part of theSouth Atlantic Bight Synoptic OffshoreObservational Network (SABSOON). Inthis paper we describe the remote video sys-tem and discuss: the mechanics of data col-lection for this revisited, remote fisheriesapproach; problems encountered with thecurrent configuration and potential rem-edies; and examples of the types of scientificdata that can be collected and analyzed us-ing temporally intensive video data. Ex-panded use of offshore video monitoring forlong-term monitoring of areas where under-water visibility is fairly good can comple-ment traditional fisheries data and providebehavioral information on temporal andspatial scales previously unattainable.

MethodsResearch Site

In May 1999, a research site consistingof concrete fish habitats was established onthe mid-continental shelf approximately 72km off the Georgia coast in approximately25 m of water to evaluate a small-scale no-take fishing area for stock enhancement pur-poses. To reduce the probability of fishingactivity, the exact location of the site was notdisclosed to the public. Similar to many ar-tificial reefs created in this region, the re-search site was established on an otherwisesandy bottom separating low relief, “livebottom” sponge/coral or “hard-bottom”habitats, located sporadically throughout thecontinental shelf of the South Atlantic Bight(Struhsaker, 1965).

Fish habitats were arranged in a circulardistribution to influence the frequency ofpositive fish observations by a centrally lo-cated camera system. Twelve large pyramid-shaped concrete fish habitats (Artificial ReefsInc.), each measuring 2.5 m wide at the baseby 1.5 m tall, were arranged in six clustersof two structures each and evenly spacedalong the perimeter of a ‘circle’ approxi-mately 15 m in diameter. Each of the threesides of each habitat had three triangular-shaped openings, each measuring about 0.1m (b) x 0.2 m (h), which were penetrable bymost reef fishes.

Data Collection and StandardizationCollection of visual observation data

began in August 1999. Video data files werecollected by six underwater cameras posi-tioned on the top of one pyramid unit (Fig-ure 1) at the center of the reef circle, roughly7 m from the outer-edge of the reef-unit clus-ters. Small black and white security cameras(Supercircuts PC-23C) with low light capa-bilities (< 0.04 lux), wide-angle lenses (8mm, 12° angle of view), and relatively lowresolution (460 lines) collected short (10 s)video data files hourly during daylight.

The camera housing contained a micro-processor and several basic sensors. The Intel8031 microprocessor with on-board SCSIand video controllers and PCI BUS framegrabber with JPEG compression (SensorayCo.) provided the means to multiplex ana-log video signals (NTSC RS-170) from eachcamera to a single coaxial cable running be-tween the pressure housing and the videocapture computer. The capture computeroperated Windows NT 4.0, with an Intel-Pentium II single board (Advantech ModelSBC 8259) and was remotely located on anabove-water platform. The microprocessorinterfaced with a video titler and sensors,while the embedded computer acted as a webserver; thus file transfer and system param-eter updates were possible at anytime via “PCANYWHERE” software. A console-con-trolled video frame grabber then convertedanalog video signals to digital images.

Data logging constraints and broadbandwidth transmission resulted in the se-lection of 10 s video data files (treated asindependent point-counts) for this study.Ten-second video data files with 15-20frames/s (full motion = 30 frames/s) resultedin compressed files of about 450 KB, which

FIGURE 1In situ image of the camera housing, showing two of the six camera windows. Note the establishment of sessileinvertebrates (sponges, bryozoans) on and fish near the camera housing support structure.


were a compromise between data storagespace and minimum resolution needed forspecies identifications. Power limitationsprohibited the use of strobes to monitor theresearch site at night, when species assem-blages and behavior can be vastly different(Stark and Davis, 1966; Wenner, 1983).

Data TransmissionDigital image data from the bottom-

mounted camera system was transmitted byan established microwave linkage (U.S. Navy,Tactical Air Combat Training System) toshore (Skidaway Institute of Oceanography,SkIO) in Savannah, GA, and finally via theInternet to the Marine Resources ResearchInstitute (MRRI) of the South Carolina De-partment of Natural Resources (SCDNR) inCharleston, SC. The transmission system wasessentially a transparent wireless local areanetwork (LAN) using a 10/100-Base-TEthernet topology running a TPC/IP proto-col, with the offshore computer appearing asnode on the SkIO network.

Video data files were transferred usingFTP protocols. The LAN interface, bothoffshore and at SkIO, was an Ethernet porton a Remote LAN Bridge (RLB) line cardinstalled in a Coastcom T1 Multiplexer. TheRLB mapped the Ethernet data packets intoa DS1 data stream, or frame. A DS1 framecomprised 24 Time Division Multiplexed(TDM) 8-bit data slots, or DS0’s. Data weremapped to 23 DS0’s, with one DS0 beingreserved for control and monitoring func-tions. A DS1 frame was transmitted at a rateof 8000 frames per second, resulting in abandwidth of 8000 x 23 x 8 bits, or 1.472Mega bits per second (Mbps).

The DS1 frames from the T1 Mux werefurther multiplexed into the Navy’s datastream and transmitted via a terrestrial mi-crowave system to an onshore site atShellman Bluff, GA. The data stream wassubsequently transported via a series of mi-crowave rely links to the Navy’s facility atMCAS Beaufort. At one of these rely sitesnear Savannah, GA, the SABSOON DS1was de-multiplexed from the Navy’s datastream and interfaced to a T1 landline viaa Spannet CSU. The T1 line was a copperbased four wire dedicated data line leased

from AT&T. After being routed througha myriad number of AT&T interconnec-tions and devices, the leased line termi-nated at SkIO and into a Coastcom T1Multiplexer where, in an identical (but re-versed) process as on the offshore platform,the TDM data were converted back intoEthernet data packets.

Data Management and AnalysisVideo data files were manually reviewed

and a database of video metadata and anno-tated content data for each recorded file wascreated and stored in MS Access. A search-able database of video files was made acces-sible to the public on the Internet at: http://fishwatch.dnr.sc.gov/

Video metadata was used to documentvisibility and sample quality. “Good” vis-ibility was defined as the ability to see (1)to and beyond the closest fish habitat and(2) the insides of the structure crevices,which corresponded to at least 10 m ofhorizontal visibility. “Fair” visibility wasdefined as the ability to see the extent ofthe closest habitat and corresponded toapproximately 3-10 m of horizontal visibil-ity. “Poor” visibility was defined as the in-ability to see the full extent of the closesthabitat and an inability to differentiate de-tails of the structure (i.e., crevices) and cor-responded to approximately <3 m of hori-zontal visibility. Files that could not beopened or contained a solid-color imagewere classified as “corrupt”.

Video content data consisted of themetadata collection number, species identi-fication to the lowest possible taxon andqualitative codes denoting (1) confidence ofproper identification, (2) species-specific

relative abundance estimates and (3) species-specific behavioral observations.

Hourly meteorological and hydrographicobservations collected within 20 km (Station41008, “Gray’s Reef”) of the research sitewere obtained online from the National DataBuoy Center (NDBC). Using the relationaldatabase, video metadata were matched withwave heights (m) based on the day and hourof collection. Descriptive statistics and cor-relations (MS Excel) were used to comparerelationships between the percent of videometadata files with respect to four visibilityconditions and six wave height intervals: <0.5m; 0.5-0.99 m; 1.0-1.49 m; 1.5-1.99 m; 2.0-2.49 m; and >2.5 m.

Detailed analyses of species behavioraland biological-physical relationships werebeyond the scope of the present descriptions.Examinations of trends in temporal distri-bution are the subject of a companion pa-per; however, examples of some unique find-ings are included here to demonstrate thetypes of data available for analyses.

ResultsAutomated video monitoring resulted in

collection of an enormous database of vi-sual observations over an unprecedentedamount of time (Table 1) despite some op-erational difficulties. At least one usable videofile (n=5,590 total) was recorded daily on429 of 1,216 d (35% of study) between 27August 1999 and 24 December 2002. Fur-thermore, although video monitoring wasonly conducted for approximately 10 s(0.3%) of every hour, 77% (n=4,306) ofthese files contained at least one ‘species’observation.

TABLE 1Comparison of sampling efforts by this study and other remote camera investigations.


Video files represented 15.5 h of obser-vations, which were disproportionately col-lected during 1999 (0.6 h), 2000 (8.5 h) and2002 (6.4 h). Annual differences in samplingeffort were attributed to variable numbers ofcameras available for sampling. Electricalproblems with computers resulting in filecorruptions or incapacitated transmissionability were implicated in 23 incidences to-taling 153 d (13% of study) in which no us-able video files were available. Malfunction

or maintenance of the microwave transmis-sion system resulted in 634 d (52% of study)for which no data could be collected between19 November 2000 and 14 August 2002.

Fifty-one unique animal ‘species’ (mostidentified to genus and species) were observed,of which 35 were positively identified (Table2). Using the classifications of Smith and Tyler(1973), 12 species could be considered benthicresidents, three as benthic cryptics, 15 as mid-water visitors, four as supra-benthic nomads,

and one species, the common loon (Gaviaimmer) was unique enough to treat separately.Nearly half of all species with less certain iden-tifications were mid-water visitors (Table 2).

The accumulated number of positivelyidentified species observed over time in-creased little after the first few months. Forty-nine percent (n=17) of positively identifiedspecies were first observed during 0.6 h ofsampling between August and December1999. Substantially increased sampling ef-fort (4.5 h) during the first six months of2000 produced observations of 13 additionalspecies, including several highly migratoryspecies such as cobia (Rachycentroncanadum), loggerhead sea turtle (Carettacaretta), common loon (G. immer) and theroughtail stingray (Dasyatis centroura).Monitoring during August-December 2000(4.0 h) and August-December 2002 (6.4 h)only resulted in documentation of five ad-ditional species, all but one of which werenot observed at the research site until 2002.

Wave heights were correlated with vis-ibility categories (n=8,523 observations; Fig-ure 2). A strong positive correlation(R2=0.95) was observed for the relationshipbetween “poor” visibility and wave heightintervals (y=11.731x+8.6895). Similarly, amoderate negative correlation (R2=0.80) wasobserved for the relationship between “fair”visibility and wave height intervals (y=-6.1776x+38.528). A strong negative corre-lation (R2=0.87) was noted for “corrupt” filesand wave height intervals (y=-5.1265x+49.905).

Although “poor” visibility and camerafouling (Figure 3) commonly reduced thefield of view for recording observations, fishviewing was rarely completely lost. Ninety-seven percent of “good” visibility files(n=171) contained at least one species ob-servation, similar to “fair” visibility files(90%; n=2,060) but quite different from“poor” visibility files (66%; n=2,075). Astrong positive correlation existed betweenthe percent of files with at least one ‘species’observation among the three visibility cat-egories (R2=0.90; y=-15.443x + 115.44).Mean number of ‘species’ per file was slightlygreater for “good” (1.82 species/file) visibil-ity than “fair” (1.62 species/file) visibility,

TABLE 2List of species and species groups (Smith and Tyler, 1973) identified from remote video sampling at the re-search site between 1999 and 2002. B=Benthic resident; MV=Mid-water visitor; CB=Cryptic benthic;SN=Suprabenthic nomad; A = Avian, not observed by Smith and Tyler, 1973.


but was nearly twice as great as the meannumber of species observed for “poor” (0.98species/file) visibility. Minimum (0) andmaximum (5-6) numbers of ‘species’ per filewere similar for all three visibilities.

Observations of interactions between andamong species (i.e., feeding habits, co-occur-rence, etc.), particularly during the late fall andwinter were likely unattainable using any othermeans of data collection. As many as three in-dividual ocean sunfish (Mola mola, Figure 4)were observed at a time in 23 video files, pre-dominantly collected between January andMarch. Courtship displays and the presenceof pre-spawning “gray-head” male black seabass (Centropristis striata, Figure 5) were com-mon during the late fall and early winter in allyears. Although juvenile black sea bass werenot observed, juveniles of what appeared to bevermilion snapper (Rhomboplites aurorubens)were exceedingly abundant in the late fall ofall years. Eighty-five percent (n=18 of 21 files)of sandbar shark (Carcharinus plumbeus) ob-servations were associated with large schoolsof predatory teleosts including blue runner(Caranx crysos), greater amberjack (Serioladumerili) and/or little tunny (Euthynnusalletteratus, Figure 6), with all but two sandbarshark observations occurring between Augustand November in all years.

DiscussionIntensive temporal sampling, combined

with the visual nature of the data, resulted inthe creation of a large quantitative and quali-tative dataset for detailed examination of spe-cies distributional patterns (examined in acompanion paper), that probably could not

have been collected without the use of a re-mote video camera system. Infrequent point-in-time methods, whether visual in nature(Gledhill et al., 1996) or involving removalof individuals can provide good estimates ofdifferences in species relative abundance inspace; however, they rarely provide adequateshort-term temporal replication for assessinglong-term trends given documented relation-ships between natural physical cycles, such astides and photoperiod, on short-term distri-butional patterns of many fishes(Wirjoatmodjo and Pitcher, 1984; Klimleyet al., 1988; Meyer et al., 2000; Arendt et al.,2001b; Hartill et al., 2003). Hourly samplingduring daylight also afforded greater prob-ability of observing cyclic or stochastic eventsunder natural conditions than traditionalpoint-in-time sampling, particularly duringperiods of high seas, when non-remote sam-pling is often not possible. The data collectedin this study should enhance the ability offisheries managers to make informed deci-sions, particularly once data are incorporatedinto ecosystem or biological-physical mod-els. Although many benefits resulted from theuse of the remotely operated video system,these benefits have not yet reached their fullpotential due to several disadvantages.

FIGURE 2Descriptive relationships between visibility and wave height.

FIGURE 3Temporal distribution of video files (%) with “poor” visibility and camera fouling between 1999 and 2002.


plexity offshore, dependence on solar andwind power sources, associated battery stor-age cells, and sophisticated electronics oftencontributed to relatively long interruptionsin data acquisition. This unfortunate realityshould improve with the development ofmore reliable remote power sources andmarine electronics, or other designs for re-mote collection of visual data offshore.

An alternative to the microwave videosystem used in this study may be to use au-tonomous video logging systems, whichwould require the same diver support forrecovery and basic maintenance of camerawindows as the present system, but withoutcable replacement and transmission prob-lems. Data loggers could be recovered fromshallow shelf depths by divers or by ship-board scientific crews during routine main-tenance of oceanographic instruments asso-ciated with existing offshore mooringsystems. A network of such video systems atregional index stations along latitudinal anddepth gradients on the continental shelfwould greatly expand spatial data coverageof reef fish assemblages in areas where real-time transmission to shore is unreliable orotherwise not feasible.

Deployment durations for data loggervideo systems are presently limited by win-dow fouling, power supply and data stor-age. Data memory chips are rapidly increas-ing in capacity, while battery packs arebecoming smaller and more powerful. Cop-per hoods shielding camera windows mayhelp reduce settlement of fouling organisms.In collaboration with researchers in NorthCarolina, South Carolina, and Georgia, re-mote video data loggers will be evaluated atseveral locations in mid-shelf waters of theSouth Atlantic Bight beginning as early assummer 2005.

As electronic improvements enhance theability to collect and store large video datafiles using either microwave-based or auto-mated data logging systems, additional em-phasis must be placed on increasing the effi-ciency of processing and managing volumesof video data, so that their near real-timevalue is not reduced or lost. Presently, anno-tation of video file content is labor-inten-sive, although data management in a rela-

FIGURE 5In situ image of a pre-spawning “gray-head” male black sea bass (C. striata) near a camera.

FIGURE 4In situ image of the three individual ocean sunfish, Mola mola, at the research site.

The primary disadvantage of the presentremote camera system was unreliability ofthe multi-faceted data collection and trans-mission system, which resulted in short- andlong-term interruptions in sampling. Mostdisruptions to data collection were caused,primarily, by electronic (power source andtransmission) problems. Furthermore, elec-trical problems could not be immediatelyaddressed because of the remote, offshore

location of the system; thus, maintenance andrepair of the at-sea system was much moredifficult and complicated than if the systemwere shore-based due to weather dependence,transportation requirements (i.e., vessels orhelicopter support), opportunistic transpor-tation availability, and expense. Althoughmicrowave transmission has the advantageof near real-time data transfer and substan-tial transmission distance, its technical com-


tional database format enables expeditiousanalyses. Automated annotation methods,using artificial intelligence attention selec-tion algorithms, have been developed for usewith video data files collected by remotelyoperated vehicle and submersible videoprojects (Edgington et al., 2003) at theMonterey Bay Aquarium Research Institute.Successful application of this technology tothe short video data files collected in thecurrent study could result in significant timesavings and increased efficiency in conduct-ing the critical process of annotating largequantities of fisheries video data; thus, au-tomated annotation would increase the over-all effectiveness of the video sampling meth-odology, including incorporation of datainto future fisheries models.

The video camera system described inthis paper documented a diversity of speciesover several years while only sampling 0.3%of every hour at best. Benthic reef fishes, thefocus group for this study, were readily iden-tifiable even under difficult viewing condi-tions due to characteristic swimming pat-terns and other behavioral cues. Mid-water

visitors were common throughout the study,yet many of these species do not appear, orappear in greatly reduced numbers, in col-lections made using trawls, traps and long-line sets on the mid-continental shelf in theSoutheastern U.S. (Powles and Barans, 1980;Wenner, 1983; Collins, 1990) and elsewhere(Grace et al., 2000; Cappo et al., 2004).Many mid-water visitors were large teleostsor elasmobranches of management interest,supporting the assertion of Cappo et al.(2004) that intense video monitoring canplay an important role in the study of thesespecies, as well as large suprabenthic nomadssuch as sea turtles and stingrays. Similar toSmith and Tyler (1973), our camera systemwas not ideal for studying small cryptic spe-cies; however, remote camera systems witha modified field of view have been used forthis purpose (Nelson).

Failure to observe a species did not nec-essarily imply species absence. Inability toobserve species, particularly benthic reef spe-cies that remained at the research site butwere not seen, could have resulted from (1)poor visibility, (2) high concentrations of

mid-water visitors restricting the field of viewand/or (3) shifts in horizontal and verticalpositions within the visible habitat. Whileresearchers cannot control environmentalconditions or species dynamics, the prob-ability of observing resident reef fishes at leastonce per day was greatly increased withmultiple cameras and sampling events. De-termination of more detailed temporal dis-tribution and habitat utilization patterns forindividual fishes, and insight into whethernon-observation resulted from absence orsome other factor, could be accomplishedusing automated acoustic telemetry systems(Arendt et al., 2001a; Hartill et al., 2003;Egli and Babcock, 2004). Joint use of auto-mated telemetry and video monitoringmight assist in developing video data ‘cor-rection factors’ for estimates of species rela-tive abundance; similar correction factorshave been developed for aerial surveys usingtelemetry data for bluefin tuna (Brill et al.,2002) and loggerhead sea turtles (Keinathet al., 1996). The combination of visual andacoustic methods could also provide furtherdata on species interactions, which influencehabitat carrying capacity, and therefore theeffectiveness of small “no fishing” areas.

Visual data represent some of the high-est quality fisheries information availablebecause they allow for positive species iden-tifications, particularly for species whichmight not otherwise be collected using tra-ditional sampling gears. Visual data also con-tribute to understanding behavioral re-sponses to hydrographic conditions andcomplement other remote data collectionmethods such as passive and active acous-tics. Pending mitigation of technical prob-lems, remote visual sampling will provide ameans for collecting data at rates compa-rable with oceanographic measurements, arequirement for correlating visual/biologi-cal and physical data sets. Technological ad-vances since the work of Smith and Tyler(1973) suggest continued improvements insystem reliability, making their widespreaduse more likely in the near future. Given themerits of remote collection of fisheries videodata, this approach should be expanded spa-tially to include a wider range of samplingin both depth and latitude.

FIGURE 6In situ image depicting the predatory associations among sandbar sharks (C. plumbeus), little tunny (E. alluterus)and greater amberjack (S. dumerili, not visible) during a feeding event on forage species, predominantly scad(Decapterus sp.).


AcknowledgementsAddressing the continuing challenges of

developing a reliable video system in a veryinhospitable marine environment requiredthe technical support of numerous electronicand computer “wizards,” like T. McKissick,as well as many others to whom we are verygrateful. The office of NOAA’s Undersea Re-search Center at UNC Wilmington pursu-ant to grant NA03OAR4300088 contributedmuch of the technical dive assistance; specialthanks to G. Taylor, J. Stryon, T. Potts, S.Hall and M. Bailey for their help in this re-gard. Additional dive support for this projectwas provided by SCDNR divers (S. Meisterand P. Powers); the diving staff of the GraysReef National Marine Sanctuary (R. Bohne,G. McFall); and the USN EOD Mobile Unit12. Numerous vessel days aboard the R/V’sBluefin, Ferguson, Palmetto and Sea Otter wererequired to conduct maintenance at the re-search site, and we thank all of the captainsand crews for their significant contributions.C. Metz (SkIO) and T. Snoots (SCDNR)provided invaluable IT assistance throughoutthis study. This research was initially fundedin part by the National Oceanographic Part-nership Program as part of the SABSOONprogram with continued support through theSoutheast Atlantic Coastal Ocean ObservingSystem (SEA-COOS). SEA-COOS is a col-laborative, regional program sponsored by theOffice of Naval Research under Award No.N00014-02-1-0972 and managed by theUniversity of North Carolina-Office of thePresident. This is contribution number 563of the South Carolina Department of Natu-ral Resources, Marine Resources Division.

ReferencesArendt, M.D., J.A. Lucy and T.A. Munroe.

2001a. Seasonal occurrence and site utilization

patterns of adult tautog, Tautoga onitis (Labridae)

at manmade and natural structures in lower

Chesapeake Bay. Fish Bull. 99(4):519-527.

Arendt, M.D., J.A. Lucy and D.A. Evans.

2001b. Diel and seasonal activity patterns of

adult tautog, Tautoga onitis, in lower Chesapeake

Bay, inferred from ultrasonic telemetry.

Environ Biol Fishes. 62:379-391.

Baamstedt, U., S. Kaartvedt and M. Youngbluth.

2003. An evaluation of acoustic and video

methods to estimate the abundance and

vertical distribution of jellyfish. J Plankton Res.

25(11):1307-1318.

Barnett, C.W. and N.W. Pankhurst. 1996.

Effect of density on the reproductive behavior

of the territorial male demoiselle Chromis

dispilus (Pisces: Pomacentriade). Environ Biol

Fishes. 46:343-349.

Brill, R., M. Lutcavage, G. Metzger, J.

Stallings, P. Bushnell, M. Arendt, J. Lucy, C.

Watson, and D. Foley. 2002. Horizontal and

vertical movements of juvenile bluefin tuna

(Thunnus thynnus) in the western North

Atlantic, determined using ultrasonic

telemetry. Fish Bull. 100(2):155-167.

Brock, V.F. 1954. A preliminary report on a

method of estimating reef fish populations.

J Wildl Manage. 18:299-308.

Cappo, M., P. Speare and G. De’ath. 2004.

Comparison of baited remote underwater

video stations (BRUVS) and prawn (shrimp)

trawls for assessment of fish biodiversity in

inter-reefal areas of the Great Barrier Reef

Marine Park. J Exp Mar Biol Ecol. 302:123-152.

Caribbean Web Cam (CWC), LLC. “Bonaire

Web Cams.” Available at:

http://www.bonairewebcams.com/

BonaireReefCam.php? Accessed 22 Mar 2005.

Collins, M.R. 1990. A comparison of three

fish trap designs. Fish Res. 9:325-332.

Collins, S. “Cams and Reports”. Available at

http://www.surfline.com/home/index.cfm

Accessed 22 March 2005.

Edgington, D.R., D. Walther, K.A. Salamy,

M. Risi, R.E. Sherlock and C. Koch. 2003.

Detection of visual events in underwater video

using a neuromorphic saliency-based attention

system. EOS Transactions, AGU 84(46), Fall

Meet. Suppl., Abstract H11F-0912.

Egli, D.P. and R.C. Babcock. 2004. Ultrasonic

tracking reveals multiple behavioural

modes of snapper (Pagrus auratus) in a

temperate no-take marine reserve. ICES

J Mar Sci. 61:1137-1143.

Gledhill, C.T., J. Lyczkowski-Shultz, K.

Rademacher, E. Kargard, G. Crist and M.A.

Grace.1996. Evaluation of video and acoustic

index methods for assessing reef-fish populations.

ICES Journal of Marine Science. 53:483-485

Grace, M., M. Bahnick and L. Jones. 2000. A

preliminary study of the marine biota at

Navassa Island, Caribbean Sea. Mar Fish Rev.

62(2):43-48.

Hartill, B.W., M.A. Morrison, M.D. Smith, J.

Boubée and D.M. Parsons. 2003. Diurnal

and tidal movements of snapper (Pagrus

auratus, Sparidae) in an estuarine environment.

Mar Freshwater Res. 54:931-940.

Hobson, E.S. 1972. Activity of Hawaiian reef

fishes during the evening and morning

transitions between daylight and darkness.

Fish Bull. 70:715-740.

Keinath, J.A., J.A. Musick and D.E. Barnard.

1996. Abundance and distribution of sea

turtles off North Carolina. Final report for the

Minerals Management Service (Contract

14-35-0001-30590) Atlantic OCS Region.

OCS Study MMS 95-0024. Prepared by the

Virginia Institute of Marine Science, College of

William and Mary. Gloucester Point, VA, 156 p.

Klimley, A.P., S.B. Butler, D.R. Nelson and

A.T. Stull. 1988. Diel movements of scalloped

hammerhead sharks, Sphyrna lewini Griffith

and Smith, to and from a seamount in the

Gulf of California. J Fish Biol. 33:751-761.

Lobel, P.S. 2001. Fish bioacoustics and

behavior: Passive acoustic detection and the

application of a closed-circuit rebreather for

field study. Mar Technol Soc J. 35(2):19-28.

Meyer, C.G., K.H. Holland, B.M. Wetherbee

and C.G. Lowe. 2000. Movement patterns,

habitat utilization, home range size and site

fidelity of whitesaddle goatfish, Parupeneus

porhyreus, in a marine reserve. Environ Biol

Fishes. 59:235-242.

Myrberg, A.A. 1972. Using sound to influence

behavior of free-ranging marine animals.

Pp. 435-468 In: Winn, H.E. and B.L. Olla

(eds.). Behavior of Marine Animals: Current

Perspectives in Research, Vol. 2, Vertebrates.

New York: Plenum Press.


National Data Buoy Center (NDBC),

National Weather Service (NWS), National

Oceanograhic and Atmospheric Administration

(NOAA). “Historical Data and Climate Summaries.”

Available at: http://www.ndbc.noaa.gov/

station_page.php?station=41008

Accessed 08 April 2005.

Nelson, R. “Usage of unique video surveillance

techniques to study Hawaiian shrimp gobies

and their snapping shrimp partners.” Available

at: http://www2.hawaii.edu/~robertne/

research.htm. Accessed 22 March 2005.

Parker, R.O., Jr. and S.W. Ross. 1986. Observing

reef fishes from submersibles off North Carolina.

Northeast Gulf Science. 8:31-49.

Parker, R.O., Jr., A.J. Chester and R.S. Nelson.

1994. A video transect method for estimating

reef fish abundance, composition, and habitat

utilization at Gray’s Reef National Marine

Sanctuary, Georgia. Fish Bull. 92:787-799.

Powles, H. and Barans, C.A. 1980. Ground-

fish monitoring in sponge-coral areas off the

Southeastern United States. Mar Fish Rev.

42(5):21-35.

Runge, J.A., P.J.S. Franks, W.C. Gentleman,

B.A. Megrey, K.A. Rose, F.E. Werner and

B. Zakardjian. 2004. Chapter 13: Diagnosis

and prediction of variability in secondary

production and fish recruitment processes:

development in physical-biological modeling.

In: The Sea, Vol. 13, eds. A.R. Robinson and

K.H. Brink. pp. 413-473. New York: John

Wiley & Sons.

Sanderson, J.G., and M. Trolle, 2005.

Monitoring elusive mammals: unattended

cameras reveal some of the world’s wildest

places. Am Sci. 93:148-155.

Smith, C.L. and J.C. Tyler. 1973. Population

ecology of a Bahamian suprrabenthic shore

fish assemblage. Am Mus Novit. 2528:1-37.

Stark, W.A., II and W.P. Davis. 1966. Night

habits of fishes of Alligator Reef, Florida.

Ichthyological Aquarium Journal, 38:313-356.

Steinberg, J.C., W.C. Cummings, B.D.Bray

and J.Y. MacBain (Spires). 1965. Further

bio-acoustic studies off the west coast of North

Bimini, Bahamas. Bull Mar Sci. 15(4):942-963.

Stevens, B.G. 2003. Timing of aggregation

and larval release by Tanner crabs, Chionoedi

bairdi, in relation to tidal current patterns. Fish

Res. 65(1-3): 201-216.

Stryke, D. “Diving History: Picture This!”

Available at:

http://www.enekton.com/divehistory3.html.

Accessed 22 March 2005.

Struhsaker, P. 1969. Demersal fish resources:

Composition, distribution and commercial

potential of the continental shelf stocks off

southeastern United States. U.S. Fish and

Wildlife Service, Fisheries Industrial Research,

4:261-300.

Wenner, C.A. 1983. Species associations and

day-night variability of trawl-caught fishes

from the inshore sponge-coral habitat, South

Atlantic Bight. Fish Bull. 81(3):537-552.

Westera, M., P. Lavery and G. Hyndes. 2003.

Differences in recreationally targested

fishes between protected and fished areas of a

coral marine park. Journal of Experimental

Biology and Marine Ecology, 294:145-168.

Wirjoatmodjo, S. and T.J. Pitcher. 1984.

Flounders follow the tides to feed: evidence

from ultrasonic tracking in an estuary. Estuar

Coast Shelf S. 19:231-241.


By Seelye MartinCambridge University Press,September 2004426 pp., $75.00

Reviewed by Gary M. MineartMitretek Systems

B O O K R E V I E W

An Introduction to Ocean Remote Sensing

W sensors—such as lidar and coastal high-fre-quency radar—are not addressed. For mostreaders, their absence is not likely to be agreat distraction given the thorough treat-ment of space-based applications and theuniversal applicability of most of the funda-mental theory and measurement techniques.

The book is structured with a de factoadoption of measurable oceanographic pa-rameters (ocean color, sea surface tempera-ture) to pose the visible and infrared appli-cations and instrument technologies (radar,scatterometer, altimeter) to organize thoseof the microwave. While this presents orga-nizational challenges and might seem a lesselegant approach than a layout organizedentirely by parameter, technology, or someother attribute, Martin’s choice in one senseenhances the readability of the book by pre-senting the information in a manner that iscommonly accepted and recognizable. Iloved the early encounter with the listingsof symbols, abbreviations, and acronyms andthe pattern of listing references and recom-mendations for additional reading at theconclusion of each chapter, all of which arevaluable features of the book.

The introductory material provides abrief history of U.S. interests in satelliteocean observations, offers a supporting defi-nition of remote sensing, describes commonsatellite orbits and levels of data processing,and itemizes satellite missions with an oceanobservation role through the year 2007. Thisbackground material is effective at summa-rizing a vast and diverse collection of infor-mation on these satellite platforms and sen-sors. The listing of past, present, and futuresatellite missions with ocean observation

applications in Table 1.1 is a handy and use-ful reference, though Martin includes thepost-2003 information at some risk becauseof the lack of maturity of programs and theirsupporting references at the time of thebook’s publication.

The early chapters introduce physicalocean surface phenomena impacting thereflection and emission of radiation and pro-vide the fundamentals of electromagneticradiation, atmospheric properties and radia-tive transfer, and interactions at the air-oceaninterface. A notable strength is the charac-terization of the ocean surface as dynamicand complex, lending practical support tothe remote sensing discussion that follows.Martin wisely chooses to leave most of therigorous derivations to the reader, assumingsome preexisting mathematical competencyin the student and wanting to keep the levelof his treatment synchronized with the othercontents of the book.

I found the discussion on ocean color tobe among the most enjoyable and engagingin the book. The material is easy to read andunderstand, current, well referenced, andsufficient to provide any student with aworking knowledge of the important appli-cations of these observations. The linkagesbetween these observations and some of thewidely accepted critical requirements for tak-ing them are unambiguous: “As the phy-toplankton die, they sink into the abyss andsequester carbon in the deepocean…Because of fossil fuel consumption,the carbon cycle is out of balance…Giventhe concerns about the imbalance of the car-bon cycle, and additional concerns aboutfeeding the growing human population and

hen I was a member of the fac-ulty at the U.S. Naval Academy back in1990, I searched for an appropriate textbookto accompany a course in Satellite Ocean-ography only to discover that there was, atleast at the time, a dearth of texts that wouldappeal to the cross-section of senior under-graduates taking my course. One group oftexts offered rigorous and exhaustive presen-tations of radiative transfer principles accom-panied by limited examples of real-worldapplications, which had little hope of ap-pealing to my Oceanography majors. Con-versely, there were a few texts that providedrelatively high-level descriptions of platformsand sensors of interest and their resultantdata products—mostly imagery—but hadlittle connection to the supporting scienceand technologies. Were I to embark on asimilar search today, I would be pleased andfortunate to find Seelye Martin’s An Intro-duction to Ocean Remote Sensing, a splendidcandidate for filling the gap in coverage. Itstimulates interest in the underlying theoryof ocean remote sensing by providing ap-propriate linkages between the supportingscience, algorithms, and spacecraft and sen-sor architectures and the actual scientific andoperational applications that are of the great-est interest to students in undergraduateocean science and engineering courses.

Despite the title of the book and thebroad definition of ocean remote sensingprovided in the background chapter, thiswork deals exclusively with space-basedocean remote sensing and, in truth, mighthave been better represented if the title in-cluded this caveat. The ocean observationcontributions of aircraft- and surface-based


determining the carrying capacity of theplanet, there is an immediate need to deter-mine the oceanic global and regional distri-bution of chlorophyll and primary produc-tivity.” I noted an absence of discussionconcerning the contribution of bottom scat-tering to the water-leaving radiance, eventhough Martin recognizes that ocean colorradiances in the blue-green portion of thespectrum can be scattered and upwelled fromocean depths as great as 50 meters. One othertopic that might have added value to thischapter is a discussion of, or at least an in-troduction to, hyperspectral remote sensingand the potential payoff this technology hasto ocean optics.

The discussion of infrared observationsof sea surface temperature (SST) emphasizesretrievals acquired using the AVHRR andMODIS instruments. The sections in thischapter define SST and the factors affectingit, describe the instrument bands useful inmaking SST measurements, discuss the asso-ciated emissive and reflective properties of theocean surface, highlight operational algo-rithms and their applications, and provideimagery examples. The section on the use ofsurface matchup data sets is a short but veryuseful addition to this chapter given that itsapplication is frequently underappreciated.The color plates at the conclusion of this chap-ter create a nice point of demarcation and aneffective transition to the microwave-centricchapters that follow.

The introduction to microwave imagersin Chapter 8 starts a six-chapter presenta-tion of subject matter dealing with this por-tion of the electromagnetic spectrum. Theintroductory information on microwaveimaging, antenna characteristics and scan-ning geometries, and passive microwaveimagers is concise, interesting, and presentedat just the right level for the book. A few ofthe errors from Table 1.1 have been inher-ited, the most egregious being the expecta-tion of a CMIS instrument on the NPOESSPreparatory Project (NPP) satellite eventhough this sensor has never been part ofthe NPP instrument suite.

The radar remote sensing material fo-cuses on two specific radar sensor technolo-gies: scatterometers and imaging radars.

Through its description of three primaryinstruments—NSCAT, AMI, andSeawinds—the book provides a goodintercomparison of fan beam Doppler-binned, range-binned, and circular scanningscatterometer technologies and their relativeadvantages and disadvantages. The ex-amples, although limited, complete the link-age from theory to application and are aidedby the inclusion of color plate illustrations.The information on radar altimeters pro-vides a relatively good understanding of thetraditional, high-orbit, exact repeat groundtrack, dedicated altimeter missions repre-sented by the TOPEX/POSEIDON andJASON-1 altimeters, the various errorsources that contribute to the signal, and asubset of applicable examples. Martin em-phasizes basin and global-scale applicationslike large-scale geostrophic flow, seasonal seasurface height variations, and Rossby wavepropagation. One absence is the lack of ref-erence to the importance of a continuoustime-series of sea surface height observationsfor determining trends in global mean sealevel and its relationship to global climatechange. The discussion also neglects theweaknesses inherent in these orbits whenapplied to other ocean observation needs,such as the mesoscale height errors that re-sult from the large distance between repeatground tracks. The material on imaging ra-dars clearly has its roots in, and highlights,the pioneering ocean applications demon-strated by the Canadian RADARSAT satel-lite mission that launched in 1995 and isstill operating well past its designed lifetime.Martin may be forgiven for this bias in lightof the unprecedented global public accessto SAR data that has been realized throughRADARSAT and the resulting increase inthe number of innovative oceanographic, seaice, and other products that can be extractedfrom these data. The discussion of backscat-ter theory, resolution, polarization, interfer-ometry, noise and error sources, imagemodes, and data storage is good and appro-priate for the level of the book.

The illustrations and graphics through-out the text are appropriate in number andare generally well conceived, clear, and easyto understand. The color plates, centrally

located in the book, are sharp, vivid, andexcellent selections for the image conceptsintended to be conveyed. The text is delight-fully free of typos and the layout is agree-able and presents the subject matter in a waythat is not overbearing. Some of the uni-form resource locator (URL) Web site refer-ences did not work. Although Martin in-cludes a clear URL disclaimer on thecopyright page, it might have been benefi-cial to stay with dependable, high-level URLsand avoid the riskier sites that often live afleeting existence on some principalinvestigator’s Web server.

On the whole, An Introduction to OceanRemote Sensing is an excellent book that helpsfill a gap in most remote sensing textbooksby relating ocean remote sensing science andsatellite technologies to real-world applica-tions. The trivial shortcomings in the bookare dwarfed by the breadth of the presenta-tion and the applicability of the material tothe target audience of senior undergradu-ates. Most importantly, the organized andrational writing style effectively weaves thedescriptive and quantitative information sothat the reader’s interest is maintained andcultivated. Each chapter is comprehensiveenough to nearly stand on its own, openingup the possibility for instructors to pick andchoose material most appropriate for theirown needs. I expect this book will enjoywidespread use within college and univer-sity ocean remote sensing curricula. I wouldhave certainly put it to good use back in myfaculty days and am glad to have it now as aready reference.


Edited by Niels WestPraeger, 2004689 pp., $149.95

Reviewed by Stephanie ShowalterSea Grant Law Center, Universityof Mississippi

B O O K R E V I E W

Marine Affairs Dictionary: Terms, Concepts,Laws, Court Cases, and International Conventionsand Agreements

T formative. For example, entire books havebeen written on salvage law and West man-ages to condense the most important pointsinto twenty-five words: “(1) the process ofsaving an endangered vessel from wreck-age; (2) the process of retrieving a wreckedvessel or cargo; or (3) the compensation dueto the salvor.” While admiralty lawyersmight have some objections to this sim-plistic definition, it adequately explains theconcept to individuals unfamiliar with thelaw. In addition to entries on court cases,international treaties, national laws, andscientific terms, the Marine Affairs Dictio-nary contains information on environmen-tal organizations, international and nationalmarine resource-related commissions andcouncils, oil spills, fishing practices, andmany other topics.

The real benefit of the Marine AffairsDictionary to the field, however, is its useas a tool to jumpstart research projects.Users can navigate through the entries anumber of different ways depending onwhat information is needed. Attorneyscould start with case names and use thecross-references to find the definitions ofscientific terms. For example, a law pro-fessor writing a paper on Solid WasteAgency of Northern Cook County v. U.S.Army Corps of Engineers, a U.S. SupremeCourt case ruling on the extent of theArmy Corps of Engineers’ authority overwetlands under the Clean Water Act,

could use the Dictionary to find the defi-nition of vernal pool. Coastal managers,on the other hand, could start with a sci-entific term to determine which interna-tional treaties, laws, and court cases gov-ern the use of a particular resource. Theentry for wetland would lead to SierraClub v. Marsh, which would lead to Envi-ronmental Impact Statement, and finallyto the National Environmental Policy Act.

While the Marine Affairs Dictionary is avaluable reference tool, there are a few draw-backs. First, court cases are listed by nameand may be difficult to find if the case namesare not widely known. With a little effortand practice, however, users should be ableto find cases through the numerous cross-references contained in most entries. Addi-tionally, scientists may find the definitionstoo simplistic for use in scholarly papers andjournals. Scientists, however, are not the tar-get audience for the Dictionary, which is in-tended more for marine affairs profession-als and students.

Black’s Law Dictionary has been a defini-tive legal resource for attorneys since 1891.The new 8th edition contains over 43,000definitions of legal terms and concepts. Westhas admirably adapted this concept to sup-port the work of marine affairs profession-als. If the longevity of Black’s is any indica-tion of the popularity of and need for suchreference materials, the Marine Affairs Dic-tionary will be around for a long time.

he field of marine affairs encompassesa number of scientific disciplines, academicendeavors, and professions. In its broadestsense, anyone dealing with the use of coastaland ocean spaces works in marine affairs.The number of concepts, scientific terms,treaties, and laws, not to mention the num-ber of acronyms, a marine affairs profes-sional must be familiar with is overwhelm-ing. The learning curve for students andprofessionals new to marine policy andmanagement is steep. The climb to the topof that curve, however, just got a little easierwith the publication of the Marine AffairsDictionary by Niels West. Inspired by astudent’s complaint “that there were toomany terms and concepts within the broadrubric of marine affairs that were difficultto search out and identify,” West spent sevenyears compiling descriptions of key marineaffairs concepts and terms. The resulting689-page reference book, unique in its cov-erage of both legal and scientific terms andconcepts, is valuable to both graduate stu-dents and seasoned professionals.

Individuals can use the Marine AffairsDictionary just like any other dictionary tolook up unfamiliar terms such as sour crude(“oil with sulfur content higher than 2.5 %”)or common law (“law that derives its au-thority from legal principles and rules—precedent, customs, previous court cases,and usage—instead of from statutes”).West’s definitions are clear, direct, and in-

Exhibit Boothsand Sponsorships

Available Now

J O U R N A L SAccustic Tracking of Marine Fishes: Implications for

the Design of Marine Protected Areas .......... $20Final Report from the U.S. Commission on Ocean

Policy: Implications and Opportunities ........ $20Underwater Pollution Threats to Our

Nation’s Marine Resources .......................... $20Innovations in Ocean Research Infrastructure to

Advance High Priority Science .................... $20Human-generated Ocean Sound and

the Effects on Marine Life ........................... $20Ocean Observing Systems ................................ $20Science, Technology and Management in the

National Marine Sanctuary Program ............ $20Ocean Energy—an Overview of the State of

the Art ........................................................ $20Marine Archaeology and Technology—

A New Direction in Deep Sea Exploration ... $20Technology in Marine Biology ......................... $20Ocean Mapping—A Focus of Shallow

Water Environment .................................... $20Oceanographic Research Vessels ....................... $20Technology as a Driving Force in the Changing Roles

of Aquariums in the New Millennium ......... $20

Submarine Telecoms Cable InstallationTechnologies ............................................... $20

Deep Ocean Frontiers ...................................... $20A Formula for Bycatch Reduction .................... $16Marine Science and Technology in the Asia Region,

Part 2 ......................................................... $16Marine Science and Technology in the Asia Region,

Part 1 ......................................................... $16Major U.S Oceanographic Research Programs:

Impacts, Legacies and the Future ................. $16Marine Animal Telemetry Tags: What We Learn and

How We Learn It ........................................ $16Scientific Sampling Systems for Underwater

Vehicles ...................................................... $16U.S. Naval Operational Oceanography ............ $16Innovation and Partnerships for Marine Science and

Technology in the 21st Century .................. $16Marine Science and Technology in Russia ........ $16Public-Private Partnerships For Marine Science &

Technology (1995) ...................................... $16Oceanographic Ships (1994–95) ...................... $16Military Assets for Environmental Research

(1993–94) .................................................. $16Oceanic and Atmospheric Nowcasting and

Forecasting (1992) ...................................... $12Education and Training in Ocean Engineering

(1992) ........................................................ $12Global Change, Part II (1991–92) ................... $10Global Change, Part 1 (1991) .......................... $10

M A R I N E E D U C A T I O NEducation and Training Programs In Oceanography

and Related Fields (1995) ....................... $6Operational Effectiveness of Unmanned Underwater

Systems .................................................. $99State of Technology Report—Ocean and Coastal

Engineering (2001) ................................ $7 dom............................................................. $9For.

State of Technology Report—Marine Policy andEducation (2002) ................................... $7 dom............................................................. $9 For.

P R O C E E D I N G SOceans 2003 MTS/IEEE (CDROM) .............. $50

Oceans 2002 MTS/IEEE (CDROM) .............. $50Oceans 2001 MTS/IEEE (CDROM) .............. $50Artificial Reef Conference ................................ $25Oceans 2000 MTS/IEEE CD-ROM ............... $50Oceans 1999 MTS/IEEE ‘99 Paper Copy ........ $80Oceans 1999 MTS/IEEE CDROM ................. $40Ocean Community Conference ‘98 ............... $100Underwater Intervention 2002 ........................ $50Underwater Intervention 2000 ........................ $50Underwater Intervention ‘99 ........................... $50Underwater Intervention ‘98 ........................... $50500 Years of Ocean Exploration

(Oceans ‘97) ............................................. $130Underwater Intervention ‘97 ........................... $50The Coastal Ocean—Prospects for

The 21st Century (Oceans ‘96) ................. $145Underwater Intervention ‘96 ........................... $50Challenges of Our Changing Global

Environment (Oceans ‘95) ........................ $145Underwater Intervention ‘95 ........................... $75Underwater Intervention ‘94 ........................... $95Underwater Intervention ‘93 ........................... $95MTS ‘92 ....................................................... $140ROV ‘92 ....................................................... $105MTS ‘91 ....................................................... $130ROV ‘91 ......................................................... $951991–1992 Review of Developments in Marine

Living Resources, Engineering and Technology. $15ROV ‘90 ......................................................... $90The Global Ocean (Oceans ‘89) ...................... $75ROV ‘89 ......................................................... $65Partnership of Marine Interest (Oceans ‘88) ..... $65The Oceans–An International Workplace

(Oceans ‘87) ............................................... $65Organotin Symposium (Oceans ‘87, Vol. 4) ..... $10ROV ‘87 ......................................................... $40Technology Update–An International

Perspective (ROV ‘85) ................................. $35Ocean Engineering and the Environment

(Oceans ‘85) ............................................... $45Ocean Data: Sensor-to-User (1985) ................. $33Arctic Engineering for the 21st Century (1984) ..... $20Marine Salvage: Proceedings of the 3rd

International Symposium (1984) ................. $21

Marine Technology Society Publications Listing

Please print

_____________________________________________________________________________________________Mr./Ms./Dr. First name Last name

_____________________________________________________________________________________________Address

_____________________________________________________________________________________________City State Zip Country

_____________________________________________________________________________________________Telephone FAX

_____________________________________________________________________________________________Total E-mail

P A Y M E N T: Make checks payable to Marine Technology Society (U.S. funds only)Credit card: ❏ Amex ❏ Mastercard ❏ Visa ❏ Diners Club

_____________________________________________________________________________________________Card Number Exp. Date Signature

The following Marine Technology Societypublications are available for purchase.Prices for 2005 are listed below. Membersare granted a discount of 10% off the pur-chase order.You can purchase our publications by:1. Calling MTS at 410-884-5330 with

your publication(s) order and creditcard number. Please use referencenumber SU39.2 when placing your order.

2. Log on to our website atwww.mtsociety.org, click on iconpublications, sidebar publications list.

3. Circle the items you would like toorder, fill out the form below andmail to MTS.

M T S P U B L I C A T I O N S O R D E R F O R M — R E F : S U 3 9 . 2S E N D A P P L I C A T I O N T O :

Marine Technology Society5565 Sterrett Place, Suite 108Columbia, Maryland 21044

Please call MTS at 410-884-5330 for aMTS Membership Application or seefollowing pages for an Application.

FAX (410) 884-9060

EMail: [email protected]

MTS Member Special Rate4 Volumes of the Journal: Volume 39, #1-4

MTS JOURNAL SUBSCRIPTION (Paper Version)

Member Rate� Journal Subscription (Domestic) $ 25� Journal Subscription (Foreign) $ 50

Non-Member Rate� Journal Subscription (Domestic) $120� Journal Subscription (Foreign) $135

Journal Payment

$_________ Journal Subscription

Make checks payable to the Marine Technology Society (U.S. Funds only) or you can pay by:

CREDIT CARD INFORMATION

� Mastercard � Visa � Diners Club � AMEX

Card No. ____________________________________________________________ Expiration Date__________________

Printed Name __________________________________________ Signature ______________________________________

Mailing Address ________________________________________________________________________________________

City/State/Zip __________________________________________________________________ Country________________

Thank You for Your Support of MTS

Submit payment to: Marine Technology SocietyPublications5565 Sterrett Place, Suite 108Columbia, Maryland 21044410-884-5330 (Phone)410-884-9060 (Fax)

MTS Journal Subscription Rates

Subscribe Today!

CALL FOR PAPERSThe Marine Technology Society

is accepting papers, technicalnotes, and commentaries for

publication in the MTS Journalfor the following issue:

Summer 2006General Issue

Submissions due byMarch 1, 2006

If you have material thatyou would like considered,

please forward it to:MTS Journal, Managing Editor


Or e-mail:[email protected]

Or visit our homepage atwww.mtsociety.org

UPCOMING MTS JOURNAL ISSUESFall 2005State of the Ocean Report:Marine Technology, Policyand Education in 2005

Justin Manley—Mitretek

ADVERTISING RATES

SPACE INSERTION RATE (US$) SPACE DIMENSIONS

1x 2x 4x Width Depth

Full Page 525 475 435 7" 10"

Half Page

Vertical 350 320 300 3 ½” 10"

Horizontal 350 320 300 7" 4 ½”

One-Quarter 205 185 160 3 ½” 4 ½”

One-Eighth 135 115 105 3 ½” 2"

Check the Society Web site for future Journal topics.www.mtsociety.org

MEMBERSHIP CATEGORIES■ MEMBER: $75 Any person with active professional inter-

est in the marine field or a closely related field. Includesthe online version of the Journal.

■ ASSOCIATE: $75 Any person with an interest in themarine field and supporting MTS objectives. Includes theonline version of the Journal.

■ STUDENT: $25 Must be a full-time undergraduate or grad-uate student. Includes the online version of the Journal.

■ PATRON: Any person supporting MTS objectives by con-tributing $100 or more annually. (Contribution above $75dues is tax deductible.) Includes the online version of theJournal.

■ EMERITUS: $40 Any person who is retired from activeprofessional interest in the marine field. Includes theonline version of the Journal.

■ LIFE: $1000 One-time payment. Includes both the onlineversion of the Journal and the paper copy of the Journal.

■ CORPORATE Fortune 500: $2000 Any corporationendorsing the objectives, policies and activities of MTS.May appoint 16 representatives to the Society, who havethe same rights and privileges as Members. Includes theonline version of the Journal for each representative. Themain contact receives a subscription to the paper copy ofthe Journal.

■ CORPORATE Non-Fortune 500: $1000 Any corporationendorsing the objectives, policies and activities of MTS.May appoint 11 representatives to the Society, whohave the same rights and privileges as Members.Includes the online version of the Journal for each rep-resentative. The main contact receives a subscription tothe paper copy of the Journal.

■ BUSINESS: $550 Any business firm whose gross annualincome is less than $4 million endorsing the objectives,policies and activities of MTS. May appoint six represen-tatives to the Society, who have the same rights andprivileges as Members. Includes the online version ofthe Journal for each representative. The main contactreceives a subscription to the paper copy of theJournal.

■ INSTITUTION: $550 Any library, government unit orother qualified nonprofit organization endorsing theobjectives, policies and activities of MTS. May appointsix representatives to the Society, who have the samerights and privileges as Members. Includes the onlineversion of the Journal for each representative. The maincontact receives a subscription to the paper copy ofthe Journal.

MTS JOURNAL PRINT VERSION SUBSCRIPTIONDomestic Subscription■ Member Rate: Journal Subscription........................$ 25■ Non-Member Rate: Journal Subscription................. $120

Foreign Subscription■ Member Rate: Journal Subscription.......................$ 50■ Non-Member Rate: Journal Subscription.................$135

■ Yes, I want to become a member of the Marine Technology Society.■ I wish to purchase a subscription to the print version of the MTS Journal. (MTS Journal online is a free

member benefit not available to non-members.)

■ I’m not ready to join at the present time, but I would like to receive news and information from MTS.Please keep my name on your mailing list. (Please fill in member information section and return to MTS.)

MEMBER INFORMATION (Please fill in the address where MTS publications and correspondence should be sent.)

Name: _______________________________________________Company (if applicable): _____________________________

Title:________________________________________________Email:____________________________________________

Phone:______________________________________________ FAX:_____________________________________________

Address:________________________________________________________________________________________________

_______________________________________________________________________________________________________

Please complete other side

Marine Technology Society Membership Applicationand MTS Journal Subscription Request

Optional Information:

n■ Male n■ Female What is your age? n■ Under 30 n■ 30-40 n■ 41-50 n■ 51-60 n■ Over 60

Marine Technology Society Membership Application (continued)

EDUCATIONAL INFORMATIONPlease fill out the following information about yourself:

Check your highest level of education:n■ High School Diploma n■ Associate (2 yr.) Degree n■ Four Year Degree n■ Graduate Degree n■ Doctorate

Check all that apply: n■ B.S. n■ B.E. n■ B.A. n■ M.S. n■ M.A. n■ M.E. n■ Ph.D. n■ Sc.D.

Do you have a P.E. license? n■ YES n■ NO

BUSINESS/PROFESSIONAL INFORMATION

Name of current employer: _________________________________________________________________________________

Your employer’s primary line of business at your location: ________________________________________________________

If you don’t work for an employer, please identify your business: ___________________________________________________

If military, rank: _________________________________________________________________________________________

Your principal job function/responsibilities:__Engineering Management__Science Management__Sales__Marketing__Administration__Policy Making, Regulatory__Public Affairs__Engineering Design__Mechanical Engineering__Software Engineering__Education/Teaching__Legal__Consulting__Retired__Other (please specify)_______________________

Your job title:

__President/CEO/COO__Owner/Partner__VP, Senior Manager__Project Manager,

Engineering__Project Manager, Other__Corporate VP, Engineering__Engineering Director__Chief/Senior Engineer__Chief/Senior Scientist__Project Manager__Engineer__Operations VP__Scientist__Other (please specify)_______________________

Check areas of interest:

__Autonomous UnderwaterVehicles

__Dynamic Positioning__Manned Underwater

Vehicles__Ocean Energy__Oceanographic

Instrumentation__Remote Sensing__Remotely Operated

Vehicles__Underwater Imaging__Marine Geodesy__Marine Living Resources__Mineral Resources__Ocean Pollution__Oceanographic Ships

__Physical Oceanography &Meteorology

__Seafloor Engineering__Buoy Technology__Cables & Connectors__Marine Archaeology__Diving__Marine Materials__Moorings__Offshore Structures__Ropes & Tension Members__Coastal Zone Management__Marine Education__Marine Law & Policy__Marine Recreation__Merchant Marine__Marine Security__Ocean Economic Potential__Other (please specify)_______________________

Four easy ways to join!Mail: Send application with check or credit card info to:

Marine Technology Society / 5565 Sterrett Place, Suite 108 / Columbia, MD 21044Fax: Fax application to: 410-884-9060 (credit card payments only)Online: Apply online at www.mtsociety.orgPhone: Contact us at: 410-884-5330

MEMBERSHIP AND JOURNAL PAYMENTPayment Method: n■ Check Enclosed n■ Master Card n■ Visa n■ Diners Club n■ Am ExMake checks payable to the Marine Technology Society (U.S. funds only)

Card #: __________________________________________________________ Expiration Date: ________________________

Signature: ________________________________________________________ Date: _________________________________

TOTAL PAYMENT:

Membership: $_____________

Journal: $_____________

TOTAL: $_____________

C O R P O R A T E M E M B E R SAlstom Power Conversion, Inc.

Houston, TexasC & C Technologies, Inc.

Lafayette, LouisianaC-Mar America, Inc.

Houston, TexasCompass Publications

Arlington, VirginiaCortland Cable Company

Cortland, New YorkDynacon, Inc.

Bryan, TexasFMC SOFEC Floating Inc.

Houston, TexasFugro Chance, Inc.

Lafayette, LouisianaFugro Geoservices, Inc.

Houston, TexasFugro-McClelland Marine Geosciences

Houston, TexasFugro Pelagos, Inc.

San Diego, CaliforniaGeneral Dynamics/AIS

Herndon, VirginiaGeospace Offshore Cables

Houston, TexasInterMoor, Inc.

Houston, TexasInnerspace Corporation

Covina, CaliforniaJ. P. Kenny, Inc.

Houston, TexasJDR Cable Systems, Inc.

Houston, TexasKlein Associations, Inc.

Salem, New HampshireKongsberg Maritime, Inc.

Houston, TexasKvaerner Oilfield Products

Houston, TexasLockheed Martin Orincon Defense

Kailua, HawaiiMaersk Line, Limited

Norfolk, VirginiaMaritime Communication Services

Melbourne, FloridaMitsui Engineering and Shipbuilding Co. Ltd.

Tokyo, JapanMohr Engineering & Testing

Houston, TexasNautronix, Inc.

Houston, TexasNavatek, Ltd.

Honolulu, HawaiiOcean Design, Inc.

Ormond Beach, FloridaOceaneering International, Inc.

Houston, TexasOceaneering Advanced Technologies

Upper Marlboro, MarylandOil States Industries, Inc.

Arlington, TexasPegasus International, Inc.

Houston, TexasPerry Slingsby Systems, Inc.

Jupiter, FloridaPhoenix International, Inc.

Landover, MarylandPlanning Systems, Inc.

Reston, Virginia

RD InstrumentsSan Diego, California

Reson, Inc.Goleta, California

SBM-IMODCO, INC.Houston, Texas

Schilling Robotics, LLCDavis, California

Science Applications International Corp.San Diego, California

SeaCon Brantner and Associates, Inc.El Cajon, California

Sippican, Inc.Marion, Massachusetts

Sonsub, Inc.Houston, Texas

SonTek/YSI, Inc.San Diego, California

South Bay Cable Corp.Idyllwild, California

SubConn, Inc.Burwell, Nebraska

Subsea SevenHouston, Texas

TechnipHouston, Texas

The Tsurumi-Seiki Co., Ltd.Yokohama, Japan

Tyco Telecommunications (US) Inc.Morristown, New Jersey

B U S I N E S S M E M B E R SAanderaa Instruments, Inc.

S. Attleboro, MassachusettsApplied Marine Solutions

Kailua, HawaiiAshtead Technology, Inc.

Houston, TexasBennex Subsea, Houston

Houston, TexasBluewater Offshore Production Systems USA, Inc.

Houston, TexasC.A. Richards and Associates

Houston, TexasCEI Maritime, Inc.

Oakland, CaliforniaDeep Marine Technology, Inc.

Houston, TexasDeepsea Power and Light

San Diego, CaliforniaDTC International, Inc.

Houston, TexasFalmat, Inc.

San Marcos, CaliforniaFugro SeaFloor Surveys, Inc.

Seattle, WashingtonGilman Corporation

Gilman, ConnecticutImpulse Enterprise

San Diego, CaliforniaInterOcean Systems, Inc.

San Diego, CaliforniaMakai Ocean Engineering, Inc.

Kailua, HawaiiMarine Desalination Systems, L.L.C.

Washington, DCMatthews-Daniel Company

Houston, TexasNatural Resources Canada

Dartmouth, Nova Scotia, CanadaOceanic Imaging Consultants, Inc.

Honolulu, Hawaii

OceanWorks InternationalHouston, Texas

Odyssey Marine ExplorationTampa, Florida

Physics Materials and Applied MathematicsResearch (PM & AM)

Kailua Kona, HawaiiPrizm Advanced Communication Electronics, Inc.

Baltimore, MarylandRemote Ocean Systems, Inc.

San Diego, CaliforniaSaipem, Inc.

Houston, TexasSeacon Phoenix, Inc.

Westerly, Rhode IslandSeaeye Marine

Fareham, Hampshire, UKSeaLandAire Technologies, Inc.

Jackson, MississippiSonardyne, Inc.

Houston, TexasSound Ocean Systems, Inc.

Redmond, WashingtonTension Member Technology

Huntington Beach, CaliforniaTSC Holdings Group, Inc.

Palm City, FloridaVideoray, LLC

Exton, PennsylvaniaWestney Consulting Group, Inc.

Houston, Texas

I N S T I T U T I O N A L M E M B E R SBritish Embassy

Washington, DCCEROS

Kailua-Kona, HawaiiConsortium for Oceanographic Research and Education

Washington, DCDepartment of Transportation Library/OST

Washington, DCGulf Coast Research Lab, Gunter Library

Ocean Springs, MississippiHarbor Branch Oceanographic Institution, Inc.

Fort Pierce, FloridaInternational Seakeepers Society

Fort Lauderdale, FloridaMBARI

Moss Landing, CaliforniaMitretek Systems

Falls Church, VirginiaNOAA/PMEL

Seattle, WashingtonNaval Facilities Engineering Command

Washington, DCNaval Facilities Engineering Service Center

Port Hueneme, CaliforniaNaval Meteorology and Oceanography Command

Stennis Space Center, MississippiScripps Institution of Oceanography

La Jolla, CaliforniaService Argos, Inc.

Largo, MarylandSPAWAR - San Diego

San Diego, CaliforniaUniversity of British Columbia Library

BC, CanadaUniversity of California Library

Berkeley, California

Marine Technology Society Member Organizations

The Marine Technology Society gratefully acknowledges the critical support of the Corporate, Business, and Institutional members listed.Member organizations have aided the Society substantially in attaining its objectives since its inception in 1963.

general issue - marine technology society international, interdisciplinary society devoted to ocean...

Documents