sharada et al 2008

8/9/2019 Sharada et al 2008

1/19

Role of biology in the airsea carbon fluxin the Bay of Bengal and Arabian Sea

M K Sharada, P S Swathi, K S Yajnik and C Kalyani Devasena

CSIR Centre for Mathematical Modelling and Computer Simulation (C-MMACS),Wind Tunnel Road, Bangalore 560 037, India.

e-mail: [email protected]

A physical-biological-chemical model (PBCM) is used for investigating the seasonal cycle ofairsea carbon flux and for assessing the effect of the biological processes on seasonal time scalein the Arabian Sea (AS) and Bay of Bengal (BoB), where the surface waters are subjected tocontrasting physical conditions. The formulation of PBCM is given in Swathi et al (2000), andevaluation of several ammonium-inhibited nitrate uptake models is given in Sharada et al (2005).The PBCM is here first evaluated against JGOFS data on surface pCO2 in AS, Bay of BengalProcess Studies (BoBPS) data on column integrated primary productivity in BoB, and WOCE I1data on dissolved inorganic carbon (DIC) and alkalinity (ALK) in the upper 500 meters at 9Nin AS and at 10N in BoB in SeptemberOctober. There is good qualitative agreement with localquantitative discrepancies.

The net effect of biological processes on airsea carbon flux on seasonal time scale is determinedwith an auxiliary computational experiment, called the abiotic run, in which the biological processesare turned off. The difference between the biotic run and abiotic run is interpreted as the net effectof biological processes on the seasonal variability of chemical variables. The net biological effecton airsea carbon flux is found to be highest in southwest monsoon season in the northwest AS,where strong upwelling drives intense new production. The biological effect is larger in AS than inBoB, as seasonal upwelling and mixing are strong in AS, especially in the northeast, while coastalupwelling and mixing are weak in BoB.

1. Introduction

Intensification of scientific interest in the roleof oceans in climate change issues has led toJGOFS, WOCE and other observational pro-grammes, including Bay of Bengal Process Studies(BoBPS; Prasanna Kumar et al 2002, 2004;Madhupratap et al 2003), in the north IndianOcean since the mid-nineties. Subsequently, a vari-ety of biogeochemical issues have been investigatedwith dynamical models to understand the underly-ing processes using the available data (see Wiggertet al 2005 for a lucid review). Statistical models(e.g., Sabine et al 2000; Sarma 2003; Sarma et al2003; Bates et al2006a, 2006b) have also been used

to extrapolate the cruise data to infer the seasonalvariability of airsea carbon flux in the tropical andsubtropical Indian Ocean.

The Indian peninsula and the island of SriLanka, in effect, partition the north Indian Oceaninto eastern and western sub-basins which canexchange mass, momentum, energy and tracersonly through 0N 6N south of Sri Lanka andto a smaller extent through a narrow shallowchannel north of Sri Lanka. The eastern and west-ern sub-basins are thus weakly coupled. A sea-sonal positive-feedback control system maintainsa higher SST and supports larger deep convec-tion in the atmosphere over the BoB with precipi-tation exceeding evaporation, and strengthening

Keywords. Carbon flux; Bay of Bengal; Arabian Sea; biotic; abiotic; coupled model.

J. Earth Syst. Sci. 117, No. 4, August 2008, pp. 429447 Printed in India. 429


2/19


3/19

Role of biology in carbon flux 431

Table 1. Salient features of the physical model.

Physical model MOM 2.2

Computational domain 15S27N; 37.6E100.4E

Resolution Latitude Varying from 1 at 15S to 0.4 above 0NLongitude 0.4

Vertical 35 levels, 12 in upper 120 m eters

Bottom topography Scripps 1

Boundary conditions East Closed; no Indonesian through flow. Effects of river and groundwater discharge are modelledindirectly by restoring

South Closed; temperature and salinity restored in15S5S to interpolated Levitus values (timescale: 30 days)

Surface forcing Wind stress COADS monthlyFluxes Heat and salt fluxes from a 50-year climatological

run. Short wave fluxes from Oberhuber (1988).Precipitation is modelled indirectly

Restoring time 30 days for temperature and salinity

Parameterization Vertical mixing Pacanowski and Philander (1981); unstableregions: instantaneous homogenization; mar-ginally stable regions: mixing coefficient increasedto 50cm2 s1

Horizontal mixing Laplacian with eddy diffusivity of 107 cm2 s1 formomentum and heat

Solar penetration Double exponential (Paulson and Simpson 1977)

Advective transport Heat and salt Quicker algorithm of Leonard (1979)

Initial conditions At rest, with temperature and salinity profilesequal to climatological annual average of Levituset al (1994)

Time step 2700 s

Spin-up time 50 years

data create difficulties in prescribing surface heatflux, precipitation and river discharge with suffi-cient accuracy, we have adopted the indirect courseof applying heat and salt flux at the surface gener-ated from 50-year climatological runs and relaxingmodel surface temperature and salinity to monthlyclimatology (Levitus et al 1994) as explained in

Swathi et al (2000). The effectiveness of the indi-rect modelling in capturing the salinity distributionespecially in the BoB is discussed in section 4. Themodel is spun for 50 years from rest with initialtemperature and salinity profiles based on annualclimatological values.

2.2 Biological model

The present biological model is a 7-componentnitrogen-based model incorporating process

models of nonlinear growth, preferential grazing,nutrient recycling by bacteria, predation by highertrophic states, linear mortality and detritus sink-ing. It generally follows the scheme of Swathiet al (2000) and Sarmiento et al (1993) using the

now-classical FDM model (Fasham et al 1990)within the euphotic layer and an empirical modelof breakdown and regeneration below the euphoticlayer (Sarmiento et al1993). We have given a sum-mary of the processes in figure 1 and the equationsin Appendix I for ease of reference.

The FDM model uses a parameterization scheme

for ammonium inhibition in which nitrate uptakeby phytoplankton diminishes exponentially to zerowith increasing ammonium at constant nitrate con-centration (Wroblewski 1977). Numerous labora-tory and field observations surveyed by Dortch(1990) do not support such a strong inhibition.The results of shipboard experiments on nitrogendynamics by McCarthy et al (1999) led us toinvestigate in some depth the effect of ammo-nium inhibition in a mixed layer ecosystem usingseveral parameterization schemes, termed as nitro-

gen kinetic relations in our earlier work (Sharadaet al 2005). The present model uses the YS para-meterization (Yajnik and Sharada 2003; equa-tions (A12) and (A13) in Appendix I) of nitrateand ammonium uptake that has two important


4/19

432 M K Sharada et al

Figure 1. A schematic of the seven-component model of marine ecosystem (adapted from Fasham et al 1990).

properties. With increasing ammonium concentra-tion, the nitrate uptake decreases hyperbolicallyto an asymptotic limit, which need not be zero.Secondly, the results of ammonium enrichmentexperiments at constant nitrate concentration canbe made to collapse on a single curve by scal-ing nitrate uptake with its maximum value, whichis consistent with the experiments of McCarthyet al (1999). These properties, termed as hyper-

bolicity and similarity, found in several disciplineslike enzyme kinetics and boundary layer theory, arebelieved to be of fundamental significance (Yajnikand Sharada 2003). Also, the YS parameteriza-tion gives more realistic annual depth integratedprimary productivity, primary productivity pro-files at JGOFS stations, f-ratio and particulateexport ratio than other parameterization schemes(Sharada et al 2005).

The biological model coupled with physicalmodel (PBM) is started from the state the physi-

cal model (PM) has attained after 50 years. Theinitial condition of the biological variables is takenas in Swathi et al (2000). The biological model isfound to approach an approximately periodic statein about four years.

2.3 Chemical model

Carbon chemistry is modelled broadly followingDrange (1994, 1996) and Peng et al (1987). Twochemical prognostic variables (DIC and ALK) are

introduced to represent the effect of the physicaland biological processes on the carbon chemistryof sea water. The transport terms of their con-servation equations (A21) and (A22) representinghorizontal and vertical advective fluxes, diffusion,

convective turbulent mixing have the same form asthose for heat and salinity.

Figure 2(a) shows biological processes contribut-ing to the source and sink terms in the conservationequation (A21) of DIC. Carbon fixation in new andregenerated production is modelled by the first twosinks. Calcification accompanying detrital sinkingand implied by predation by higher trophic statesremoves CaCO3 and it is modelled by the third DIC

sink. Release of DIC due to regeneration by bac-teria and zooplankton is modelled by two sourceterms.

Consumption of NO3 in new production andrelease of NH+4 in regeneration by zooplanktonand bacteria increase ALK (figure 2b) and theyare modelled by source terms in (A22). Consump-tion of NH+4 in regenerated production and uptakeof NH+4 by bacteria decrease ALK and they aremodelled by sinks of ALK. Finally, calcificationaccompanying sinking of detritus and predation

implies removal of Ca2+

and is modelled by a sinkin (A22). It is noteworthy that new productionreduces DIC and increases ALK, while the regene-rated production decreases both DIC and ALK.

Determination of airsea carbon flux needs com-putation of pCO2 in the surface water. Dissolvedinorganic carbon is partitioned into dissolved CO2,carbonic acid HCO3 and CO

2

3 . Carbon chem-istry of sea water is modelled following Peng et al(1987) with minor modifications (see Swathi et al2000). Boric acid is taken to be proportional to

salinity. Phosphoric and silicic acids are not con-sidered. Four equilibrium equations govern twodissociations of CO23 and one each for dissoci-ation of water and carbonic acid and there arethree mass balances of DIC, ALK and Boron.


5/19


Figure 2. Schematic of effects of biological processes on conservation of(a) DIC and (b) ALK.

The prognostic equations for DIC and ALK (A21and A22) close the otherwise undetermined systemof seven nonlinear equations. As the equilibriumconstants depend nonlinearly on temperature andsalinity, sea surface pCO2 depends nonlinearly ontwo chemical and two physical variables.

The airsea flux of CO2 (FC) is determined bythe equation

FC = kls(pCOair

2 pCOsea

2 ),

where kl is the piston or transfer velocity and sis the solubility of CO2 in sea water. pCO2 forair has been adjusted for 100% relative humidityand piston velocity (regulated by the turbulence atthe airsea interface and chemical reactions inthe liquid phase) is calculated using Wanninkhof(1992) formulation. In contrast to other biologi-cal and chemical species which have zero airsea

fluxes, the flux FC is added to the top grid boxwhile solving the DIC equation.

Calcification is modelled as follows. The netdownward POC (assuming a Redfield ratio of 5.0)flux at each euphotic level is the sum of sinkingdetritus and instantaneously exported fraction ofzooplankton mortality to higher predators. Thisnet POC flux is multiplied by a rain ratio of 0.07and the volumetric (CaCO3) production rate iscomputed as a vertical divergence of this flux in theeuphotic zone. Below the euphotic zone, CaCO3 is

dissolved with a dissolution scale depth of 4000 m.The initial condition of the physical and the bio-logical variables of PBCM is taken to be the state ofthe PBM at the end of four-year spin-up. The DICand ALK fields are initialized by the objectively

interpolated fields from 2 2 atlas of Myrmehland Drange (1998). The atmospheric pCO2 is heldfixed at 360atm. We do not employ virtual fluxesfor biological and chemical components. We reporthere the results for the fourth year. Issues related toadequacy of the spin-up are discussed in section 4.

3. Simulation experiments

Our investigation begins with a PBCM simulationwith the biological parameters given in Sharadaet al (2005). The resulting surface pCO2 values on65E (6466E) transect between 10N and 21Nare shown as Experiment A (whose parameters aregiven in Appendix II) in table 2, which also showsJGOFS cruise data during five months spanningall the seasons. The model values are derived frommonthly snapshot results on the day closest to thecruise period. The noticeable discrepancy betweenthem, it is hypothesized, is primarily due to exces-

sive regeneration of ammonium by zooplankton,which is much larger than regeneration by bacte-ria. A minimalistic parametric study is designed totest the hypothesis and, if true, to reduce the dis-crepancy. Four key parameters are varied to reduceregeneration of ammonium by zooplankton in threedistinct ways:

(1) by reducing zooplankton growth when itsconcentration is high,

(2) by increasing the share of predation inzooplankton mortality, and

(3) by additionally increasing zooplanktonmortality.

In the first alternative (Experiment B), asymp-totic grazing rate g and half saturation constant


6/19


Table 2. Observed and model surface pCO2

(atm) nominally along 65E transect.

Season1 Duration2 US JGOFS3 Exp A4 Exp B4 Exp C4 Exp D4

Late NEM J: 11, 1723 January 359404 484562 455525 422468 402423M: 1 February

SIM J: 16, 2229 March 346365 482545 452477 423457 408426M: 3 April

SWM J: 22 July, 359392 474540 464537 420457 39240928 July3 AugustM: 3 August

FIM J: 4, 5 November, 358384 491558 473535 436461 408418812 NovemberM: 2 November

Early NEM J: 3 December, 358382 497577 474552 432471 404426915 DecemberM: 2 December

1 See the definitions of seasons given in section 4.2 J and M refer to JGOFS and model.3 pCO2 in water measurements along ship track in JGOFS cruise ttn-049 at 8 stations between 10

N20N.4 Model pCO2 values at 25 grid points 0.4

apart in 10N20N.

of grazing k3 are both reduced from 1 to 0.6,thus keeping their ratio constant so as to leavegrazing rate unaltered when zooplankton con-centration is very low, and leaving all otherparameters unchanged. In the second alternative(Experiment C), detrital fraction 4 of zooplank-ton mortality due to predation is increased from

0.33 to 0.5 keeping all other parameters constant.In the third alternative (Experiment D), in addi-tion to increasing 4, zooplankton mortality 5is also increased from 0.05d1 to 0.1d1 withoutchanging any of the remaining parameters. The rel-evant parameter values of the four experiments areshown in Appendix II.

Table 2 gives the surface pCO2 values of allthe four experiments and the JGOFS data. Theyclearly show that Experiment D is the most effec-tive in reducing the discrepancy. We therefore callit the standard run and present its results in the

subsequent sections.

4. Results and discussion

We begin by examining certain aspects of the mod-elling methodology before examining the results ofthe standard run and the abiotic run. We first beginby summarising the PM performance in capturingtemperature and salinity in the upper ocean, whichhas been discussed at length in Sharada et al(2005)

(section 2.3). The restoring heat flux at the surfaceis small (20 meters in July) is probably due to the con-tribution of high frequency monsoon winds beingexcluded in monthly forcing leading to the under-estimation of wind stress curl.

Model surface salinity compares qualitatively

well in January and July (Sharada et al 2005,figure 7) in the AS with Levitus et al (1994).Both models as well as Levitus have the highestsalinity in January in northwest AS. There arehowever local discrepancies. For example, modelsalinity is lower in the northeast BoB by 1 and0.5 psu in January and July. Recent observationsin the BoB have shown transient stratificationeffects in surface waters like fresh water plume(Gopalkrishna et al 2002) and the formation ofbarrier layer (Vinaychandran et al 2002) in northBoB during JulyAugust as major rivers swollen

by heavy rainfall in their catchment areas dischargelarge amounts of fresh water, which then spreadsacross the bay. For example, the latter observa-tions at the station TS2 (89E, 1730N) show asharp drop in salinity (4 psu) followed by slowrise (1 psu) during 27 July 1999 to 6 August1999 in upper 15 meters. Our model shows a fall inmonthly salinity (2.5 psu) in top 20 meters fromJuly to August at 8830E, 1730N. There is also alocal discrepancy by about 1 psu near the southerntip of the west coast of India in January probably

due to the model seasonal transport of low salin-ity water from BoB (Shenoi et al 2005) and itsaccumulation near Lakshadweep islands, althoughLevitus et al (1994) do not reveal this local fea-ture possibly because of interpolation errors with


7/19


limited data, unlike more recent climatologicaldata (World Ocean Atlas 2005).

An ecosystem model using nitrogen as a singlecurrency cannot do justice to modelling of dis-solved organic matter (Anderson and Williams1998; Anderson and Pondaven 2003), which needscarbon as an additional currency; nor can it deal

with iron limitation that needs bioavailable iron asanother currency.The issue of modelling of iron limitation in

the north Indian Ocean needs further discussion.Interest in the iron limitation on phytoplanktongrowth has increased considerably after the ironhypothesis of Martin (1990). Jickells et al (2005)have estimated on the basis of three extensively-validated models that Indian Ocean receives 25%of global dust deposition of 450 Tg year1. Theirresults also show that the average dust depositionrate drops by an order of magnitude (from 2to 0.2gm m2year1) i n 5N25S. Thus mostof the dust deposition occurs in the north IndianOcean.

Moore et al (2002b) have compared observa-tions of dissolved iron in major ocean basins.They find that concentration of dissolved ironin AS (1023N; 5769E) is larger at 4302930 pM than almost all the sites in north Pacific(956N; 122158W), equatorial Pacific (6S9N;105140W), north Atlantic (3659N; 2072W),and all the sites in southern Ocean, notably

excluding two areas (6946

S,

6

W; 7774

S,182168W). Recent surveys of dissolved iron con-centration in the World Ocean also show clearlythat dissolved iron concentrations in the northIndian Ocean are amongst the highest in the world(Moore and Broucher 2007; figure 2).

Recently, Wiggert et al (2006) have inves-tigated iron limitation in the Tropical IndianOcean (25S25N; 40E125E) using an ecosys-tem model coupled to a reduced gravity model with19-layers below the mixed layer. They suggestthat north Indian Ocean has a tendency towards

seasonal iron limitation in west AS (


8/19


as a consequence of the climatological initial con-ditions. Short-duration PBCM runs thus do notpreclude the long-term effects of deep ocean cir-culation. We refrain from spinning the model forlonger periods as the crude modelling of regene-ration below the euphotic layer results in smallbut cumulative errors in nitrate and ammonium,

which slowly spread to the euphotic layer as thesimulation progresses. Also, propagation of errorsof sponge boundary conditions into the interiorgenerally leads to restriction of the duration ofruns of regional models. This restriction is impor-tant in the present work as only temperature andsalinity are relaxed to climatological conditionsin 515S.

We now turn to evaluating the results of thestandard run by comparing them with other obser-vations. Since PBCM based primary productivityhas already been evaluated in the AS against

JGOFS data in Sharada et al (2005), we restricthere to the comparison of the primary producti-vity in the BoB and the observations of BoBPScruises. Figure 4 shows monthly model primaryproductivity from the standard run (Exp D) inte-grated over the euphotic zone (120 m) along the88E transect for April, July and September alongwith BoBPS cruise data. The standard run resultscompare moderately well with the BoBPS databetween 6N and 16N. The discrepancy near 20Nis most probably due to stratification effects result-

ing from large river discharge and precipitationwhich is a well known difficulty that almost allmodels face in the north BoB.

Turning our attention to chemical variables, wenext compare the results of the standard run forthe zonal transect along 9N (8.59.5N) in the ASand 10N (9.510.5N) in the BoB with observa-tions of WOCE I1 cruise along nominally 9N in ASand nominally 10N in BoB, which were analyzedby Goyet et al (1999). We have chosen Septem-ber average as the WOCE I1 cruise took placeduring August 29 to October 16, 1995. Figure 5(a)

gives contours of September average of DIC andALK in the 9N vertical section in AS up to 500 mdepth. They compare quite well, especially in theupper 200 meters, with WOCE I1 data given infigure 3(a) and 3(b) of Goyet et al(1999). An inter-esting feature of model DIC in figure 5(a) is a50-meter deep subsurface layer shoaling eastwardacross which DIC changes by 150 m mol C m3.Figure 3 of Goyet et al (1999) also shows a simi-lar layer at almost the same location and withabout the same gradient. The September average of

meridional velocity at this section, not given here,shows at the same location a shear layer in meri-dional velocity separating upper faster southwardflow from lower slower southward flow, suggestingthat the eastward monsoonal flow east of 60E in

Figure 4. Depth integrated primary productivity(mgC/m2/day) values from model simulations (Exp D)compared with the data of Bay of Bengal Process Studiesduring April (SK191 10 April to 10 May 2003), July(SK166 6 July to 2 August 2001) and September (SK182 14 September to 12 October 2002).

late SWM season brings with it higher DIC andALK. Both observed and model surface ALK showa high around 66/67E. Goyet et al (1999) havenoted linear correlation of alkalinity with salinity,and their figure 6(b) indeed shows salinity higharound 67E.

Similarly, figure 5(b) gives model DIC andALK contours for September along 10N tran-sect in the BoB. It also shows a westward shoal-ing 50 m deep layer across which DIC changesby 200 m mol C m3. The layer across which

ALK rises by

50m eq m

3

becomes thinner as itshoals westward. Comparable layers are also seenin figure 4(a) and (b) of Goyet et al (1999).

As Goyet et al (1999) have pointed out, thezonal contrast of 170/150 m mol C m3, which


9/19


Figure 5. DIC (mmolC/m3) and ALK (meq/m3) along the 9N transect in the Arabian Sea and along the 10N transectin the Bay of Bengal in the upper 500 m.

Figure 6. Seasonally averaged model surface DIC (mmolC/m3) in the north Indian Ocean in the standard run.

both the WOCE I1 observations and modelDIC show, is larger than typical zonal contrast(100 m mol C m3) over comparable distances inother oceans.

Notwithstanding excellent qualitative agreementbetween the model results and observations, wenow examine critically the discrepancies in the Sep-tember values of model and observed DIC, ALK

and salinity with a view to determine where themodel could be improved. While the zonal contrastin model surface DIC is slightly smaller than in theWOCE observations, the contrast in model surface

ALK (

100meqm3

) is significantly lower thanin the WOCE observations (250meqm3). Zonalcontrast in salinity in the standard run (3.7 psu) isalso smaller than in WOCE I1 observations (4 psu).


10/19


11/19


12/19


Figure 9. Spatial and temporal variation of surface pCO2

(atm) obtained from model simulation.

Figure 10. Spatial and temporal variation of surface carbon flux (gC/m2/year) from atmosphere to ocean obtained frommodel simulation (Note that negative values indicate outgassing from the ocean.).

Seasonally averaged carbon fluxes in the northIndian Ocean derived from Takahashi et al (1999)show:

most intense outgassing of carbon off the Omancoast in SWM season,

less intense outgassing in FIM season, with peakoff the Oman coast,

considerably lower outgassing in the other twoseasons with peak near the Somali coast.

The zonal contrast between AS and BoB inTakahashi et al(1999) is roughly similar to PBCMstandard run results. However, maximum out-gassing in the standard run in SWM season islarger by about 100150%. Our results are based on

Wanninkhof (1992) relation for kl and Hellermanand Rosenstein (1983) winds, while Takahashidata are based on a different kl formulation andEsbensen and Kushnir (1981) winds. Given thequadratic dependence of kl on wind speeds, thecoarser spatial resolution (4 5) in Takahashidata could also be a significant factor, as the windaveraged over a larger area would be weaker lead-ing to significant reduction of the peak carbon fluxin Takahashi data, especially in SWM season.

Monthly distribution of airsea carbon flux givenin Bates et al (2006a) shows that whole of north-ern Indian Ocean is outgassing except eastern BoBduring NEM months (i.e., DecemberFebruary).Higher sea to air fluxes are seen in the western


13/19


Figure 11. Spatial and temporal variation of difference in pCO2

(atm) between biotic and abiotic simulations.

AS and southeast coast of India during DecemberFebruary and JuneAugust where upwelling ishigh. Sea-to-air carbon flux is lowest during SIM inAS. All these features are captured by the modelsimulations. Higher sea to air fluxes are obtainedfrom the standard run during SIM and FIM inBoB, but Bates et al (2006a) obtain highest sea toair carbon flux during SWM and FIM in BoB.

5. Role of biology in the airseacarbon transfer

We have seen in section 2.3 how the biologicalprocesses and physical and chemical processes areintertwined in the transport equations of DIC andALK. To isolate the effect of biology, we use themetaphor of solubility and biological pumps andpose the question: What are the relative effects ofthe biological and solubility pumps on the seasonal

carbon cycle in the north Indian Ocean?We devise a modelling device to address the

above question. It consists of a hypothetical PBCMsimulation in which the biological pump is turnedoff from the beginning of the simulation, keepingthe initial conditions on physical and chemical vari-ables the same as in the standard run. We callthis simulation the abiotic run to indicate that thebiological processes do not contribute to seasonalvariability. The seasonal variability in the abioticrun thus depends only on the solubility pump on

a seasonal time scale. Note that since the initialconditions are taken from climatological atlases,the biological effects over all longer time scales areimplicitly built into all the runs of PBCM, includ-ing the abiotic run. Thus the difference between

the standard run and the abiotic run may be consi-dered as a measure of the effect of biological pumpon the seasonal scale. We investigate this measurefor important chemical variables to obtain insightin the role of biological processes in the carboncycle in the north Indian Ocean.

We first note that the standard run and theabiotic run have the same temperature and salinityfields. So the difference in surface pCO2 is entirely

due to the difference in surface DIC and ALK.Figure 11 shows the spatial variability of the dif-ference in surface pCO2 in the four seasons. Theseasonally averaged surface pCO2 in the standardrun is lower than the abiotic run by 100 ppm ormore in high productivity regions, like the coastalregions of Arabia and, to a smaller extent, in thewest and the east coast of India, in all the fourseasons. This expected trend is clearly the conse-quence of vigorous carbon fixation in the high pro-ductivity regions, where upwelling feeds nutrientsto the surface waters.

The effect of biological pump on the seasonalairsea carbon flux varies spatially and season-ally as shown in figure 12. In SWM season, thereis spectacular decrease in the outgassing of car-bon roughly along the axis of Findlater Jet (Smithet al 1998) in the northwestern AS. Here strongupwelling brings up not only nutrients but alsohigher DIC from deeper waters and leads to a verylarge outgassing in the abiotic run. But the bio-logical uptake in the standard run considerablyoffsets the DIC increase in the upwelled waters.

Vigorous carbon fixation increases new productionsubstantially, thereby decreasing DIC and chang-ing ALK marginally. Resulting reduction in pCO2tends to lower the outgassing. But the effect ofupwelling is so strong that even after the reduction


14/19


Figure 12. Spatial and temporal variation of difference in carbon flux (gC/m2/year) between biotic and abiotic simulations.(A positive value of difference indicates decrease in outgassing.).

Table 3. Carbon flux from ocean to atmosphere over a year (PgC/Year).

% DifferenceRegion Abiotic (ab) Biotic (bi) 100(bi-ab)/ab

Annual Model basin 0.26 0.226 13AS 0.207 0.166 19.8BoB 0.053 0.06 13.2

NEM Model basin 1.53 1.3 15AS 1.29 1.01 21.7BoB 0.23 0.29 20.8

SIM Model basin 0.128 0.124 3.1AS 0.094 0.083 11.7BoB 0.034 0.041 20.6

SWM Model basin 0.602 0.522 13.3AS 0.489 0.396 19BoB 0.113 0.127 12.4

FIM Model basin 0.141 0.111 21.3AS 0.105 0.073 30.5BoB 0.036 0.038 5.6

by biological processes; the remaining DIC is largeenough to cause outgassing. The large values ofoutgassing in this region in both the cases is inno small measure due to significant increase inthe mass transfer coefficient at the airsea inter-face, which depends approximately quadraticallyon wind speed, by strong winds in the SWMSeason. Moderate upwelling off the African coast(south of8N), the west Indian coast and the eastIndian coast has a similar but smaller effect. Asthe stratification in most of the BoB restricts the

scope for vertical transport processes, the decreasein DIC and ALK due to biological processes issmall and together they marginally increase surfacepCO2 and outgassing. In other seasons, both the

biological effects, both positive and negative, aremuch weaker in these regions.

Table 3 gives the annual and seasonal sea-to-air carbon fluxes in the AS (78E, 027N) in the standard and theabiotic runs. The biological processes decrease out-gassing in the north Indian Ocean annually by 13%.They also decrease outgassing in all the seasons,the largest decrease being in the FIM season (21and 15%). The biological processes decrease out-gassing in the AS annually by 20%, and increase in

the BoB by 13%, but in absolute terms the lattereffect is much smaller. The decrease in outgassingin the AS occurs in all seasons, the largest (30%)being in the FIM season. The biological processes


15/19


increase outgassing in BoB in all seasons, thelargest increase (21%) being in the NEM season.This rather unexpected effect is due to the domi-nance of regeneration over carbon fixation in largeparts of the BoB due to stratification effects.

6. Concluding remarks

We have presented results of the standard run ofa model synthesized from physical oceanographicmodel (MOM2.2), marine ecosystem model ofFasham et al (1990) and carbon chemical model ofPeng et al(1987) and Drange (1994). It is shown tocapture qualitatively the trends of JGOFS, BoBPSand WOCE I1 observations in the north IndianOcean. Since dust depositional models show sharpmeridional southward drop south of the equator indust deposition rates (Jickells et al 2005) and fieldobservations of dissolved iron in the World Oceanshow that dissolved iron concentrations in thenorth Indian Ocean amongst the highest (Mooreand Braucher 2007; figure 2), it is reasonable toassume that iron limitation is not a serious issueabove the equator. The quantitative discrepanciesprima facie appear to be largely due to forcingerrors in the physical model. The fact that incor-rect physics is a major hurdle for accurate simula-tion of ecological and biogeochemical processes inthe ocean is nicely demonstrated by Friedrichs et al(2006).

If we quantify the zonal contrast between theBoB (>78E, >0N) and the AS (0N)as the difference between zonal averages of anoceanic variable at a given latitude, our modelshows that the surface waters of the AS are, on anannual average, cooler, more saline, more alkalineand they have more dissolved inorganic carbonthan the BoB (figure 8). The zonal contrast ismuted in the equatorial region where the Wyrtki

jet and the equatorial undercurrent provide mech-anisms for reducing the zonal contrast. However,it increases with latitude north of5N. Also, the

zonal contrast persists in all seasons. Using a dif-ferent measure of zonal contrast, Goyet et al(1999)observed that the zonal contrast between the ASand BoB in WOCE I1 (29 August15 October1995) data on 9/10N transect is larger than whatis observed over comparable distances in otheroceans. The zonal contrast in these physical andchemical oceanic variables translates into the zonalcontrast in pCO2 and the sea-to-air carbon flux.

We have conducted a gedanken computationalexperiment, called the abiotic run, in which the

biological processes have been turned off in thenorth Indian Ocean during the simulation withoutaltering the surface forcing or the initial conditionsof physical and chemical variables. The differ-ence between the results of the standard run and

the abiotic run is interpreted as the effect of thebiological processes on the seasonal variability ofsea-to-air flux. This modelling exercise is intendedto give insight into the role of biological processeson the seasonal time scale. The biological processestend to decrease surface pCO2 in the northwest ASin all seasons (figure 11), and on a smaller scale

near the west Indian coast, especially in FIM andNEM Seasons. High winds in the Findlater Jetregion amplify the biological effect on the sea-to-air carbon flux in a spectacular fashion resultingin a marked decrease in the outgassing (sea-to-aircarbon flux) in the northwest AS in the SWMseason (figure 12).

Acknowledgements

This work was carried out as a part of ocean modelling activities at C-MMACS spon-sored by the Department of Ocean Development(DOD) under several projects. The authors thankDOD for their financial support. The authorsthank Dr. Madhuprathap, Dr. Prasannakumar,Dr. Ramaiah and Dr. Sugandha, Scientists ofNational Institute of Oceanography for useful dis-cussions and for providing the cruise data col-lected in Bay of Bengal under Bay of BengalProcess Studies programme. The authors thankDr. Richard Barber, Dr. John Marra and Dr. Lou

Codispoti for providing the access to U.S.JGOFScruise data on U.S.JGOFS home page. The authorsacknowledge the use of SeaWiFS data available onSeaWiFS home page in accord with the ResearchUsers Terms and Conditions Agreement. Theauthors also thank Mr. R P Thangavelu, Scien-tist, C-MMACS for computer support. The authorswish to thank the referees for their suggestions.

Appendix I

Equations of the biological andchemical models

The biological model has seven ecosystem conser-vation equations of Phytoplankton P, ZooplanktonZ, Bacteria B, Nitrate Nn, Ammonium Nr, Dis-solved Organic Nitrogen Nd and detritus Np. Theseconservation equations have advective and diffusiveterms of the same form as in the temperature andsalinity. Their source-minus-sink terms (SMS) aregiven below.

SMS(P) = (11)J(z, t)[Q1(Nn, Nr)+Q2(Nr)]P

1PG1, (A1)


16/19


SMS(Z) = 2(G1 + G2 + G3) (2 + 5)Z,(A2)

SMS(B) = U1 + U2 G2 3B, (A3)

SMS(Nn) = J(z, t)Q1(Nn,Nr)P, (A4)

SMS(Nr) = [32 + (1 4)5]Z+ 3B

J(z, t)Q2(Nr)P U2, (A5)

SMS(Nd) = 1J(z, t)(Q1(Nn,Nr) + Q2(Nr))P

+ (1 3)2Z+ 4Np U1, (A6)

SMS(Np) = 1P+ (1

2)(G1 + G2 + G3)G3 4Np sNp/z. (A7)

Note that predation 45Z is instantly exported tothe bottom of the euphotic layer.

Light limitation model of phytoplanktongrowth (Fasham et al 1990)

I(z, t) = I(0, t)exp(kwz 0z

kcPdz)PAR, (A8)

Vp = abcT, (A9)

J(z, t) =VpI(z, t)

[V2p + (I(z, t))2]0.5

, (A10)

J(z, t) =

1

zk zk+1

zk

zk+1 J(z, t)dz. (A11)

Nitrogen uptake kinetics

Nitrogen uptake kinetics is modelled by YS para-meterization scheme, which is explained in sec-tion 2.2 (Yajnik and Sharada 2003; Sharada et al2005).

Q1(Nn,Nr) =Nn

k1 + Nn

1 + a1Nr

1 + b1Nr, (A12)

Q2(Nr) =Nr

k2 + Nr. (A13)

Grazing functions (Fasham et al 1990)

F = p1P+p2B +p3D, (A14)

G1 =gp1P

2Z

k3F+p1P2 +p2B2 +p3N2p, (A15)

G2 =gp2B

2Z

k3F+p1P2 +p2B2 +p3N2p, (A16)

G3 =gp3N

2pZ

k3F+p1P2 +p2B2 +p3N2p. (A17)

Bacterial growth (Fasham et al 1998)

S= min(Nr, Nd), (A18)

U1 =VbNdB

k4 + S+ Nd, (A19)

U2 =VbSB

k4 + S+ Nd. (A20)

Chemical model (Drange 1994)

The chemical model uses the two conservationequations for Dissolved Inorganic Carbon (DIC)and Alkalinity (ALK). They are essentially thesame as Drange, except for modifications men-tioned in section 2.3.

SMS(DIC) =J(z, t)[Q1(Nn,Nr) + Q2(Nr)]PRp

t(CaCO3) + 3BRb

+ [32 + (1 4)5]ZRp, (A21)

SMS(ALK) = J(z, t)[Q1(Nn,Nr)Q2(Nr)]P

2

t (CaCO3) + 3B

+ [32 + (1 4)5]Z U2. (A22)


17/19


Appendix II

Biological model parameters in Experiments A, B, C and D

Phytoplankton parameters

Initial slope of P-I Curve 0.025 d

1

/(Wm

2

)PAR Photosynthetically active radiation 0.4

kc Light attenuation by P 0.03 m1(mmolN/m

3)1

kw Light attenuation due to water 0.04 m1

a 0.6 d1

b Maximum growth rate parameters 1.066c 1.0 C1

1 Exudation fraction 0.05

1 Specific mortality rate 0.04 d1

k1 Half-saturation for nitrate uptake 1.47 mmolN m3

k2 Half-saturation for Ammonium uptake 0.47 mmolN m3

a1 Ammonium inhibition parameters 1.0 (mmolN)1

b1 3.0

Rp Redfield ratio for P and Z 7.0

Zooplankton parameters

g Maximum growth rate 1.0 d1 Exp A, C, D

0.6 d1 Exp B

2 Assimilation efficiency 0.75

p1 Relative preference of Z for P 0.4

p2 Relative preference of Z for B 0.3

p3

Relative preference of Z for Np

0.3

k3 Half-saturation rate for ingestion 1.0 mmolN m3 Exp A, C, D

0.6 mmolN m3 Exp B

2 Specific excretion rate 0.1 d1

5 Specific mortality rate 0.05 d1 Exp A, B, C

0.1 d1 Exp D

3 Ammonium fraction of Z excretion 0.75

4 Detrital fraction of Z mortality 0.33 Exp A, B0.5 Exp C, D


18/19


Bacterial parameters

Vb Maximum growth rate 2.0 d1

3 Specific excretion rate 0.05 d1

k4 Half-saturation rate for uptake 0.5 mmolN m3

Ammonium/DON uptake ratio 0.6

Detrital parameters

4 Breakdown rate 0.05 d1

s Sinking velocity 4.0 m d1

RB Redfield ratio for B 5.5

Note: The values are common to all experiments, unless otherwise specified.

References

Anderson T R and Pondaven P 2003 Non-redfield carbonand nitrogen cycling in the Sargasso Sea: pelgic imbal-ances and export flux; Deep-Sea Res. I50 573591.

Anderson T R and Williams P J le B 1998 Modelling theSeasonal Cycle of Dissolved Organic Carbon at StationE1 in the English Channel; Estuarine, Coast. Shelf. Sci.46 93109.

Bates N R, Pequignet A C and Sabine C L 2006a Ocean

carbon cycling in the Indian Ocean: 1. Spatiotempo-ral variability of inorganic carbon and airsea CO2gas exchange; Global Biogeochemical Cycles 20 GB3020,doi:10.1029/2005GB002491.

Bates N R, Pequignet A C and Sabine C L 2006b Oceancarbon cycling in the Indian Ocean: 2. Estimates of netcommunity production; Global Biogeochemical Cycles 20GB3021, doi:10.1029/2005GB002492.

Blackford J C and Burkill P H 2002 Planktonic communitystructure and carbon cycling in the Arabian Sea as aresult of monsoonal forcing: the application of a genericmodel; J. Marine Sys. 36 239267.

Dortch Q 1990 The Interaction between Ammonium andNitrate Uptake in Phytoplankton; Mar. Ecol. Prog. Ser.

61 183201.Drange H 1994 An isopycnic coordinate carbon cycle model

for the North Atlantic; and the possibility of disposing of fossil fuel CO2 in the ocean, PhD Thesis, The NansenEnvironmental and Remote Sensing Center and Dept. ofMathematics, University of Bergen, Bergen, Norway.

Drange H 1996 A 3-dimensional isopycnic coordinate modelof the seasonal cycling of carbon and nitrogen in theAtlantic Ocean; Phys. Chem. Earth21(56) 503509.

Esbensen S K and Kushnir Y 1981 The heat budget ofthe global ocean: An atlas based on estimates from thesurface marine observations; Climatic Research InstituteReport #29, Oregon State University, Corvalis, OR,USA.

Fasham M J R, Ducklow H W and McKelvie S M 1990A Nitrogen-based Model of Plankton Dynamics in theOceanic Mixed Layer; J. Mar. Res. 48 591639.

Friedrichs M A M, Hood R R and Wiggert J D 2006 Ecosys-tem model complexity versus physical forcing: Quantifi-

cation of their relative impact with assimilated ArabianSea data; Deep-Sea Res. II53 576600.

Gopalkrishna V V, Murthy V S N, Sengupta D, Shenoy Sand Araligidad N 2002 Upper ocean stratification and cir-culation in the northern Bay of Bengal during southwestmonsoon of 1991; Cont. Shelf Res. 22 791802.

Goyet C, Coatanoan C, Eischeid G, Amaoka T, Okuda K,Healy R and Tsunogai S 1999 Spatial variation of totalCO2 and total alkalinity in the northern Indian Ocean:A novel approach for the quantification of anthropogenic

CO2

in seawater; J. Mar. Res. 57 135163.Hastenrath S, Lamb P and Greischar L 1989 Climatic Atlasof the Indian Ocean (Madison: University of WisconsinPress).

Hellerman S and Rosenstein M 1983 Normal monthly windstress over the world ocean with error estimates; J. Phys.Oceanogr. 13 10931104.

Jickells T D, An Z S, Andersen K K, Baker A R,Bergametti G, Brooks N, Cao J J, Boyd P W, Duce R A,Hunter K A, Kawahata H, Kubilay N, laRoche J, Liss P S,Mahowald N, Prospero J M, Ridgwell A J, Tegen I andTorres R 2005 Global iron connections between desertdust, ocean biogeochemistry and climate; Science 3086771.

Levitus S, Boyer T P and Antonov J 1994 World OceanAtlas; Vol. 4, NESDIS, Washington.

Leonard B P 1979 A stable and accurate convective mod-elling procedure based on quadratic upstream interpo-lation; Computer Methods in Applied Mechanics andEngineering 19 5998.

Madhupratap M, Gauns M, Ramaiah N, Prasanna KumarS, Muraleedharan P M, de Souza S N, Saradesai S andMuraleedharan U 2003 Biogeochemistry of the Bay ofBengal: Physical, Chemical and Primary Productivitycharacteristics of the central and western Bay of Bengalduring Summer Monsoon 2001; Deep-Sea Res. II 50881896.

Martin J H 1990 Glacial-interglacial CO2 change: The ironhypothesis; Paleoceanography5 113.

McCarthy J J, Garside C and Nevins J L 1999 Nitrogendynamics during the Arabian Sea northeast monsoon;Deep-Sea Res. Part II46(89) 16231664.

Moore J K and Braucher O 2007 Observations of dissolvediron concentrations in the World Ocean: implication and


19/19


constraints for ocean biogeochemical models; Biogeo-sciences Discussions 4 12411277.

Moore J K, Doney S C, Kleypas J C, Glover D M andFung I Y 2002a An intermediate complexity marineecosystem model for the global domain; Deep-Sea Res. II49 403462.

Moore J K, Doney S C, Glover D M and Fung I Y 2002bIron cycling and nutrient limitation patterns in sur-face waters of the world ocean; Deep-Sea Res. II 49463508.

Myrmehl C and Drange H 1998 Distribution of total alka-linity, total dissolved inorganic carbon and derived para-meters in the world oceans; Unpublished Manuscript.

Oberhuber J M 1988 An atlas based on the COADS dataset: The budgets of heat, buoyancy and turbulent kineticenergy at the surface of the ocean; Tech. Rep. 15, Max-Planck Institute for Meteorologie, Hamburg. Digital dataavailable at http://ingrid.ldgo.columbia.edu.

Pacanowski R C and Philander G 1981 Parametrization ofvertical mixing in numerical models of the tropical ocean;J. Phys. Oceanogr. 11 14421451.

Pacanowski R C 1995 MOM2 Documentation, Users Guide

and Reference Manual. Ver 1.0, GFDL Ocean Tech.Report, No. 3.Paulson C A and Simpson J J 1977 Irradiance measurements

in the upper ocean; J. Phys. Oceanogr. 7 952956.Peng T H, Takahashi T, Broecker W S and Olafsson J 1987

Seasonal variability of carbon dioxide, nutrients and oxy-gen in the Northern Atlantic surface water: observationsand a model; Tellus 39B 439457.

Prasanna Kumar S, Muraleedharan P M, Prasad T G,Gauns M, Ramaiah N, de Souza S N, Sardesai S andMadhupratap M 2002 Why is the Bay of Bengal lessproductive during summer monsoon compared to theArabian Sea?; Geophys. Res. Lett. 29(24) 2235.

Prasanna Kumar S, Nuncio M, Narvekar J, Kumar A,

Sardesai S, de Souza S N, Gauns M, Ramaiah N andMadhuprathap M 2004 Are eddies natures trigger toenhance biological productivity in the Bay of Bengal?;Geophys. Res. Lett. 31 L07309, doi:10.1029/2003GL019274.

Sabine C L, Wanninkhof R, Key R M, Goyet C andMillero F J 2000 Seasonal CO2 fluxes in the tropical andsubtropical Indian Ocean; Marine Chemistry72 3353.

Sarma V V S S 2003 Monthly variability in surface pCO2and net airsea CO2 flux in the Arabian Sea; J. Geophys.Res. 108(C8) 3255, doi:10.1029/2001JC001062.

Sarma V V S S, Swathi P S, Dileep Kumar M,Prasannakumar S, Bhattathiri P M A, Madhupratap M,Ramaswamy V, Sarin M M, Gauns M, Ramaiah N,

Sardessai S and de Souza S N 2003 Carbon budgetin the eastern and central Arabian Sea: An IndianJGOFS synthesis; Global Biogeochemical Cycles 17(4)1102, doi:10.1029/2002GB001978.

Sarmiento J L, Slater R D, Fasham M J R, Ducklow HW, Togglweiler J R and Evans G T 1993 A seasonalthree-dimensional ecosystem model of nitrogen cycling inthe north Atlantic euphotic zone; Global BiogeochemicalCycles 7(2) 417450.

Sarmiento J L, Monfray P, Maier-Reimer E, Aumont O,Murnane R J and Orr J C 2000 Sea-air CO2 fluxes and

carbon transport: A comparison of three ocean generalcirculation models; Global Biogeochemical Cycles 14(4)12671281.

Schott F A and McCreary J P 2001 The monsoon circulationof the Indian Ocean; Progr. Oceanogr. 51 1123.

Shankar D, Vinayachandran P N and Unnikrishnan A 2002The monsoon currents in the north Indian Ocean; Progr.Oceanogr. 52 63120.

Sharada M K and Yajnik K S 1997 Seasonal variationof chlorophyll and primary productivity in the CentralArabian Sea: A macrocalibrated upper ocean ecosystemmodel; Proc. Indian Acad. Sci. (Earth Planet. Sci.) 1063342.

Sharada M K, Yajnik K S and Swathi P S 2005 Evaluation ofsix relations of the kinetics of uptake by phytoplankton inmulti-nutrient environment using JGOFS experimentalresults; Deep-Sea Res. II1415 18452078.

Shenoi S C, Shankar D, Gopalkrishna V V and Durand F2005 Role of ocean in the genesis and annihilation of thecore of the warm pool in the southeastern Arabian Sea;Mausam 56(1) 147160.

Smith S, Roman M, Prusova I, Wishner K, Gowing M,

Codispoti L A, Barber R, Marra J and Flagg C 1998Seasonal response of zooplankton to monsoonal reversalsin the Arabian Sea; Deep-Sea Res. II45 23692403.

Swathi P S, Sharada M K and Yajnik K S 2000 A coupledphysical-biological-chemical model for the Indian Ocean;Proc. Indian Acad. Sci. (Earth Planet. Sci.) 109(4)503537.

Takahashi T, Wanninkhoff R, Feely R A, Weiss R F,Chipman D W, Bates N, Olafsson J, Sabine C andSutherland S C 1999 Net airsea CO2 flux over globaloceans: an improved estimate based on seaair pCO

2dif-

ference; In: Proceedings of 2nd International Symposiumon CO2 in the oceans (ed.) Nojiri Y; Centre for Globalenvironmental Research, National Institute for Environ-

mental Studies, Tsukuba, Japan, pp. 914.Vinaychandran P N, Murthy V S N and Babu V R 2002Observations of barrier layer formation in the Bay ofBengal during summer monsoon; J. Geophys. Res. 10719 119.

Vinaychandran P N, Chauhan P, Mohan M and Nayak S2004 Biological response of the sea around Sri Lanka tosummer monsoon; Geophys. Res. Lett. 30 L01302.

Wanninkhof R 1992 Relationship between wind speed andgas exchange over the ocean; J. Geophys. Res. 97(C5)73737381.

Wiggert J D, Hood R R, Banse K and Kindle J C 2005Monsoon-driven biogeochemical processes in the ArabianSea; Prog. Oceanogr. 65 176213.

Wiggert J D, Murtugudde R G and Christian J R 2006Annual ecosystem variability in the tropical IndianOcean: Results of a coupled bio-physical ocean generalcirculation model; Deep-Sea Res. II53 644676.

Wroblewski J S 1977 A model of phytoplankton plume for-mation during variable Oregon upwelling; J. Mar. Res.35 357394.

Yajnik K S and Sharada M K 2003 Ammonium Inhibitionof Nitrate Uptake by Phytoplankton: A New RelationBased on Similarity and Hyperbolicity; Curr. Sci. 85(8)11801189.

MS received 19 February 2007; revised 16 April 2008; accepted 16 April 2008

sharada et al 2008

Documents