Intensification and deepening of the Arabian Sea Oxygen Minimum Zone in response to increase ...

34 Pages·2017·5.67 MB·English

BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
Intensification and deepening of the Arabian Sea Oxygen Minimum
Zone in response to increase in Indian monsoon wind intensity
ZouhairLachkar1,MarinaLevy2,andShaferSmith1,3
1TheCenterforPrototypeClimateModeling,NewYorkUniversityinAbuDhabi,AbuDhabi,UAE.
2SorbonneUniversité(UPMC,Paris6/CNRS/IRD/MNHN),LOCEAN-IPSL,Paris,France.
3CourantInstituteofMathematicalSciences,NewYorkUniversity,NewYork,USA.
Correspondenceto:Z.Lachkar([email protected])
Abstract.Thedeclineinoxygensupplytotheoceanassociatedwithglobalwarmingisexpectedtoexpandoxygenminimum
zones(OMZs).Thisglobaltrendcanbeattenuatedoramplifiedbyregionalprocesses.IntheArabianSea,theWorld’sthickest
OMZishighlyvulnerabletochangesintheIndianmonsoonwind.Evidencefrompaleorecordsandfutureclimateprojections
indicate strong variations of the Indian monsoon wind intensity over climatic timescales. Yet, the response of the OMZ to
5 thesewindchangesremainspoorlyunderstoodanditsamplitudeandtimescaleunexplored.Here,weinvestigatetheimpacts
ofperturbationsinIndianmonsoonwindintensity(from-50%to+50%)onthesizeandintensityoftheArabianSeaOMZ,and
examinethebiogeochemicalandecologicalimplicationsofthesechanges.Tothisend,weconductedaseriesofeddy-resolving
simulationsoftheArabianSeausingtheRegionalOceanicModelingSystem(ROMS)coupledtoanitrogenbasedNutrient-
Phytoplankton-Zooplankton-Detritus(NPZD)ecosystemmodelthatincludesarepresentationoftheO cycle.Weshowthat
2
10 theArabianSeaproductivityincreasesanditsOMZexpandsanddeepensinresponsetomonsoonwindintensification.These
responsesaredominatedbytheperturbationofthesummermonsoonwind,whereasthechangesinthewintermonsoonwind
play a secondary role. While the productivity responds quickly and nearly linearly to wind increase (i.e., on a timescale of
years),theOMZresponseismuchslower(i.e.,atimescaleofdecades).OuranalysisrevealsthattheOMZexpansionatdepth
isdrivenbyincreasedoxygenbiologicalconsumption,whereasitssurfaceweakeningisinducedbyincreasedventilation.The
15 enhanced ventilation favors episodic intrusions of oxic waters in the lower epipelagic zone (100-200m) of the western and
centralArabianSea,leadingtointermittentexpansionsofmarinehabitatsandamorefrequentalternationofhypoxicandoxic
conditionsthere.TheincreasedproductivityanddeepeningoftheOMZalsoleadtoastrongintensificationofdenitrificationat
depth,resultinginasubstantialamplificationoffixednitrogendepletionintheArabianSea.Weconcludethatchangesinthe
Indianmonsooncanaffect,onlongertimescales,thelarge-scalebiogeochemicalcyclesofnitrogenandcarbon,withapositive
20 feedbackonclimatechangeinthecaseofstrongerwinds.
1 Introduction
ThecombinationofstrongorganicmatterdecompositionandpoorventilationexplainsthepresenceoflargeOxygenMinimum
Zones(OMZs)intheintermediateoceanoftheeasterntropicalPacificandAtlanticOceansaswellasinthenorthernIndian
Ocean.Atlowoxygenconcentrations,hypoxia-sensitivemarinespeciesaresubjecttovaryingenvironmentalstressesthatcan
1
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
affecttheirgrowthandreproductivesuccessandultimatelycausetheirdeath(Vaquer-SunyerandDuarte,2008).Nearcomplete
oxygendepletion,suboxicconditionsfavoranaerobicremineralizationoforganicmatterviadenitrification:aprocessthrough
which nitrate is used as an alternate oxidant. This not only depletes the oceanic inventory of bioavailable nitrogen that is
essentialforphytoplanktongrowth,butalsoreleasesN O,amajorgreenhousegasthatalsocontributestostratosphericozone
2
5 depletion (Codispoti et al., 2001). Thus, the OMZs not only shape marine ecosystem habitats, but also impact and regulate
climate.
Dissolved O is predicted to decline in the future in response to upper ocean warming and increased stratification, which
2
mayresult intheexpansion ofOMZs(Bopp etal.,2002; Keelingetal., 2010).Thesechanges willhavedeleterious impacts
onmarinehabitatsandmayleadtodisruptionofkeybiogeochemicalcycles(Keelingetal.,2010;Gruber,2011;Doneyetal.,
10 2012).Observationalevidencesuggeststhattheoceanicoxygeninventoryhasalreadybeendecliningovertherecentdecades
(Helmetal.,2011;Schmidtkoetal.,2017)andthatOMZshaveexpandedinseverallocations(Strammaetal.,2008,2012).
Globalmodelprojectionsshowconsistentlydecliningoxygeninventoriesthatarecommensuratewiththewarminganomaly,
butdisagreeonthefutureevolutionofOMZs(e.g.Coccoetal.,2013).Thismayresultinpartfromthemisrepresentationof
thedynamicsofOMZsinglobalmodels(Boppetal.,2013;Coccoetal.,2013;Cabréetal.,2015;Longetal.,2016)andthe
15 sensitivityofOMZstoregionalclimateperturbationsthatarepoorlyrepresentedinglobalsimulations.
An example of such perturbations is regional wind changes (e.g. Deutsch et al., 2014). Located nearby major coastal up-
wellingsystems,theOMZsareindeedhighlyvulnerabletochangesinalongshorewinds.InmostEasternBoundaryUpwelling
SystemsaswellasinthewesternArabianSea,upwelling-favorablewindintensificationhasbeenshowntooccurunderwarm-
ingscenarios(Wangetal.,2015;deCastroetal.,2016).Bakun(1990)haslinkedthistoincreasedland-seathermalgradient
20 underwarmerclimates,strengtheningtheland-seapressuregradient,andhencealongshorewinds.Yet,previousobservational
andmodel-basedstudiesshowthattheamplitudeofsuchupwellingintensificationstronglyvariesfromonesystemtoanother
(McGregor et al., 2007; Narayan et al., 2010; García-Reyes and Largier, 2010; Gutiérrez et al., 2011; Barton et al., 2013;
Sydemanetal.,2014;Varelaetal.,2015).
IntheArabianSea,thesummermonsoonsouthwesterlywindsdrivestrongupwellingoffthecoastsofOmanandSomalia,
25 giving rise to one of most productive coastal upwelling ecosystems in the world (Ryther and Menzel, 1965). This high pro-
ductivity in conjunction with a sluggish circulation sustains the World’s thickest OMZ, responsible for up to 40% of global
pelagicdenitrificationdespiteoccupyinglessthan2%oftheWorldOceanarea(Bangeetal.,2005).Futureclimateprojections
suggestthatIndiansummermonsoonmayintensifyinthefutureunderawarmerclimate(e.g.Wangetal.,2013;Sandeepand
Ajayamohan, 2015; Praveen et al., 2016; deCastro et al., 2016). Praveen et al. (2016) found that the future upwelling inten-
30 sificationwillaffectessentiallythecoastofOman.Usinganensembleofglobalandregionalmodelsimulationsforthe21st
century,deCastroetal.(2016)showastrengtheningoftheSomalicoastalupwellingthatincreaseswithlatitude.Furthermore,
theseauthorsshowthattheintensificationofupwellinginthewesternArabianSeaisevenhigherthanwhatispredictedforthe
EasternBoundaryUpwellingSystems.Thisstudyalsofindsanintensificationofland-seathermalandpressuregradientsthat
is consistent with the Bakun (1990) hypothesis. Other studies reveal that summer monsoon wind intensification has already
35 beenobservedrecently(Goesetal.,2005;Wangetal.,2013).Goesetal.(2005)reportedanincreaseinupwellingfavorable
2
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
windsoffSomaliaovertheperiodbetween1997and2005andhaveshownanassociatedincreaseinproductivityoverthesame
period. Wang et al. (2013) found a global intensification of the Northern Hemisphere summer monsoon since the late 1970s
andattributedthesetrendstotheinteractionofmega-ElNiño/SouthernOscillationandtheAtlanticMultidecadalOscillation,
inadditiontohemisphericalasymmetricglobalwarming.
5 WhilethereisstillnoconsensusonthemagnitudeandthedriversoftherecentandfutureIndianmonsoonwindchanges,
evidencefrompaleoclimaterecordsoverwhelminglysuggestsastronglinkbetweennorthernhemispheretemperaturesandthe
Indianmonsoonwindintensityontimescalesrangingfromdecadestothousandsofyears(e.g.Altabetetal.,2002;Schulzetal.,
1998;Guptaetal.,2003;Ivanochkoetal.,2005).Generally,enhancedIndiansummermonsoonintensitywasrecordedduring
northernhighlatitudeswarmperiods(e.g.,Dansgaard-Oeschgerevents)whilereducedsummermonsoonintensitywasfound
10 tocorrespondtocoldperiods(e.g.,Heinrichevents)(Schulzetal.,1998).Ivanochkoetal.(2005)hasestablishedarelationship
between millennial-scale oscillations in summer monsoon intensity and the position of the Intertropical Convergence Zone
(ITCZ), where the ITCZ moves northward during warm periods (interstadials) and southward during cold periods (stadials).
Thisrelationshipbetweenthenorthernhigh-latitudeclimateandtheIndianmonsoonseenfromamultitudeofproxiesduring
the last glacial period was also found valid during the Holocene (Gupta et al., 2003). This includes the most recent climate
15 changes from the Medieval Warm Period and the Little Ice Age. The analysis of paleo proxy records also suggests a strong
couplingbetweennorthernclimateexcursionsandchangesinproductivityanddenitrificationintheArabianSeaatdifferent
timescales(e.g.Altabetetal.,1995,1999,2002;Guptaetal.,2003;Singhetal.,2011).Forinstance,Altabetetal.(1999)found
thatdenitrificationwasgreatestduringinterglacialperiodsandwasprobablynotactiveduringmostglacialphases.Altabetetal.
(2002)foundastrongcorrespondencebetweenchangesintheproductivityanddenitrificationoftheArabianSeaandcentury-
20 scaleDansgaard-Oeschgereventsduringthelastglacialperiod.Thesestudiesshowthatdenitrificationincreasesduringwarm
periodsconcurrentwiththesummermonsoonandproductivityintensificationanddecreasesduringcoldphases.Otherstudies
havehighlightedthatthefluctuationsindenitrificationandtheintensityoftheOMZcanalsobedrivenbychangesinwinter
monsoonwindintensity(Reichartetal.,2004;KlöckerandHenrich,2006).
Despitethese previousstudies, theamplitude ofthe OMZsensitivity topotential windchanges remainslargely uncertain.
25 Indeed,whetherinthecontextofpastclimatefluctuationsorunderfutureclimatechange,theresponseofOMZtoupwelling-
favorablewindintensificationisdifficulttopredictassuchaperturbationmayincreasebothoxygensupplythroughenhanced
ventilation and oxygen demand via increased biological productivity, thus leading to an uncertain net effect. In the Arabian
Sea,thepictureismadeevenmorecomplicatedbytheseasonalreversalofwindsandthepotentialimportanceofchangesin
wintermonsoonmixing.Additionally,thelargedenitrificationfluxesintheArabianSeacombinedwiththepotentialfeedback
30 of denitrification on biological productivity, and hence on oxygen consumption, further add to the intricacy of the problem.
Finally, the question of the OMZ response timescales is essential but remains unanswered. Here we address these questions
andexplorethemechanismsbywhichtheArabianSeaecosystemrespondstomonsoonwindchangesusingaregionaleddy-
resolving model. We examine how idealized changes in summer and winter monsoon wind intensity affect the productivity
andthevolumesofhypoxicandsuboxicwaterintheArabianSeaandexplorethebiogeochemicalandecologicalimplications
35 of these changes. We show that the productivity increases on a timescale of years while the OMZ expands and deepens on
3
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
a timescale of decades. This response is essentially driven by summer monsoon wind intensification and results from an
enhanced biological consumption of oxygen opposed by increased ventilation near the surface. The enhanced upper ocean
ventilationleadstointermittentexpansionsofhabitatsintheepipelagiczonewiletheOMZintensificationatdepthincreases
denitrification, thus amplifying the depletion of bioavailable nitrogen in the Arabian Sea. We conclude that changes in the
5 Indianmonsooncanaffectthelarge-scalenitrogenmarinebudgetondecadaltocentennialtimescales,withapositivefeedback
onwarminginthecaseofstrongerwinds.
2 Methods
2.1 Models
We use the Regional Oceanic Modeling System (ROMS)-Agrif (documented at http://www.romsagrif.org/) configured for
10 theArabianSearegion.ROMSsolvestheprimitiveequationsandhasfreesurfaceandgeneralizedterrain-followingvertical
coordinates(ShchepetkinandMcWilliams,2005).Advectionisrepresentedusingarotated-splitthird-orderupstreambiased
operatordesignedtolimitdispersiveerrorsandpreservelowdiffusion(Marchesielloetal.,2009).Thesubgridverticalmixingis
representedusingthenonlocalK-ProfileParameterization(KPP)scheme(Largeetal.,1994).Numericaldiffusivityallowsthe
dissipationofsmall-scalenoiseasnoadditionalbackgroundlateraldiffusivityisused.Theecological-biogeochemicalmodel
15 isanitrogen-basednutrient-phytoplankton-zooplankton-detritus(NPZD)model(Gruberetal.,2006;Lachkaretal.,2016).It
isbasedonasystemofordinarydifferentialequationsdescribingthetimeevolutionofthefollowingtracers:nitrate(NO−3),
ammonium(NH+),oneclassofphytoplankton,oneclassofzooplankton,twoclassesofdetritusandadynamicchlorophyll-
4
to-carbon ratio. The two pools of detritus represent respectively large organic matter particles that sink fast (10md 1) and
−
smallparticlesthatsinkatslowerspeed(1md 1).Thesmallparticlescancoagulatewithphytoplanktontoformfastsinking
−
20 large detritus. The sinking of the particles is represented explicitly in the model, thus allowing all tracers to be advected
laterallyincludingintheeuphoticzone.Whensinkingparticlesreachtheseafloor,theyareremineralizedbackintoammonium
at a much slower rate (0.003 d 1) than in the water column (0.03 d 1 for small detritus and 0.01 d 1 for large detritus).
− − −
Finally,themodeliscoupledtoamodulethatdescribesthebiologicalsourcesandsinksofoxygenfollowingLachkaretal.
(2016).Furthermore,atverylowoxygenconcentrations(O <4mmol/m3),nitrificationandaerobicremineralizationareshut
2
25 downanddenitrificationissettoconsumenitrateinsteadofoxygen.Finally,benthicdenitrificationisparameterizedfollowing
Middelburgetal.(1996)(seefurtherdetailsofthemodelinLachkaretal.(2016)).
2.2 Experimentaldesign
Themodeldomainextendsinlatitudefrom5 Sto30 Nandinlongitudefrom34 Eto78 E,thusencompassingthewhole
◦ ◦ ◦ ◦
ArabianSeaOMZaswellasthekeydynamicalfeaturesofthecirculationoftheregionsuchastheSomaliandOmaniupwelling
30 systemsandtheGreatWhirl.Themodelhorizontalgridhasaresolutionof1/12◦ withanaveragegridspacingthatisabout
5to20timessmallerthanthelocalfirstbaroclinicRossbydeformationradius(Cheltonetal.,1998).Thispermitsanexplicit
4
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
representationofalargefractionoftheregioneddies.Theverticalgridconsistsof32sigmalayerswithanimprovedresolution
near the surface. The seafloor topography is derived from the ETOPO2 dataset supplied by the National Geophysical Data
Center(SmithandSandwell,1997).
Themodelisforcedwithamonthlyclimatology.Theabsenceofinterannualvariabilityintheatmosphericforcingenables
5 us to quantify the role of internal variability associated with mesoscale eddies. The temperature, salinity and currents are
initializedandlaterallyforcedusingtheSimpleOceanDataAssimilation(SODA)oceanreanalysis.Oxygenandnitrateinitial
and boundary conditions are derived from the World Ocean Atlas (2009) dataset. The atmospheric boundary conditions for
heatandfreshwaterfluxesarebasedontheComprehensiveOcean-AtmosphereDataSet(COADS)(daSilvaetal.,1994).We
appliedasurfacerestoringtoCOADSobservedsurfacetemperaturesandsalinitiesusingkinematicfluxcorrectionsdescribed
10 inBarnieretal.(1995).ThemodelisforcedwithwindstressdatafromtheQuikSCAT-derivedScatterometerClimatologyof
OceanWinds(SCOW)(RisienandChelton,2008).
The model is first spun up for 12 years and then is run for an additional 50 years using 9 different wind stress scenarios.
Inthecontrolrunthewindstressisleftunperturbed.Inafirstsetoffourperturbedmonsoonsimulationsthewindstresswas
uniformlyincreasedordecreasedby20%and50%,respectively.Thisamountstowindspeedperturbationsofaround10and
15 20%,respectively.Twoadditionalrunswereconductedwheretheperturbationsofthesummerandwintermonsoonwindsare
settobeantagonistic,i.e.,a50%increase(resp.decrease)ofthesummermonsoonwindstressthatisconcomitantwitha50%
decrease(resp.increase)inthewintermonsoonwindstress.Finally,twosimulationswithgradualincrease(resp.decrease)of
windstressatarate+1%peryearareperformed.
Althoughtheserunsexploredifferentwindperturbationscenarios,theyarehighlyidealizedbynatureandarenotintended
20 tomimicfutureprojectionsorrealisticfuturetrajectories,butratheraimatexploringthesensitivityoftheArabianSeaOMZ
tomonsoonwindintensitychangesandimprovingourunderstandingofthekeymechanismsthatcontroltheOMZresponse
anditstimescales.Formodelevaluationweusethelast10yearsofthecontrolrun.
2.3 Modelevaluation
Weuseavailablesatelliteandin-situobservationstoassessthemodelabilitytoreproduceobservedphysicalandbiogeochem-
25 ical properties in the Arabian Sea domain. To this end, we evaluate the model in terms of surface currents and eddy kinetic
energy (EKE), sea surface height (SSH) anomalies, sea surface temperature (SST) and surface chlorophyll-a concentrations.
Furthermore,weexaminethedistributionsoftemperature,salinity,oxygenandnitrateintheupperoceaninthemodelandin
theWorldOceanAtlas(2013)dataset.
The model simulates successfully the surface eddy kinetic energy (EKE) as it reproduces quite accurately the spatial dis-
30 tributionoftheobservedsurfaceeddyfield(Fig.1).Inparticular,theobservedeast-westgradientinEKEiscapturedbythe
model.However,themodeltendstoslightlyunderestimatethemagnitudeoftheeddyactivityintheeasternandnortheastern
Arabian Sea. This might be due to the resolution of the model that does not permit yet the representation of the full eddy
spectruminthisregion(Cheltonetal.,1998;Lachkaretal.,2016).Themodelalsocapturesthemainpatternsoftheobserved
surfacecirculationbothinsummerandwinterseasons(AppendixA:Supplementaryfigures,Fig.A1).Inparticular,themodel
5
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
Model
Data
(Aviso)
Data
(drifter)
Figure1.Surfaceeddykineticenergy.Surfaceeddykineticenergy(incm2s−2)assimulatedinthemodel(top)andfromdatabasedon
Avisosatellitealtimeter(middle)andsurfacedrifterclimatologyofLumpkinandJohnson(2013)(bottom).
6
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
Winter Model Winter Data
(JFM) (JFM)
Summer Model Summer Data
(JFM) (JFM)
Figure2.Surfacedistributionofchlorophyll-a.Surfacechlorophyll-aconcentrations(inmgm−3)assimulatedinROMS(left)andfrom
SeaWiFS(right)duringwinter(top)andsummer(bottom)months.TheSeaWiFSclimatologyiscomputedovertheperiodfrom1997to
2009.
reproducestheNortheastMonsoonCurrentandtheSouthEquatorialCountercurrentinwinterandtheenergeticGreatWhirl
andSouthernGyreinsummer(SchottandMcCreary,2001).Furthermore,thestrengthandtheseasonalreversaloftheSomali
Current are also correctly reproduced in the model. Finally, the model reproduces the observed seasonal sea surface height
(SSH) anomalies in both winter and summer seasons (Appendix A: Supplementary figures, Fig. A2). In particular, both the
5 spatialdistributionandthemagnitudeofcoastalSSHgradientsareaccuratelycaptured.
Themodelcapturesthemainpatternsoftheobservedseasurfacetemperature(SST)fromAVHRRsatellitedatainwinterand
summerseasons(AppendixA:Supplementaryfigures,Fig.A3).Inparticular,themodelreproducestheeast-westandnorth-
southtemperaturegradientsacrossthemodeldomain,aswellasthestrongsurfacegradientsbetweentheArabianSeaandits
marginal seas (i.e., the Gulf and the Red Sea). Furthermore, the model also captures the cold SST tongue that characterizes
10 summerupwellingoffthecoastsofOmanandSomalia.
However,anexaminationofthenorth-southverticaldistributionoftemperaturerevealsthatthemodel:(i)tendstounderes-
timatethesubsurfacewatertemperatureinthenorthernArabianSeaparticularlyinwinterand(ii)slightlyoverestimatesthe
mixedlayerdepthinbothseasons(AppendixA:Supplementaryfigures,Fig.A4).Similarmismatchwithobservationschar-
7
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
Model Data
z=200m z=200m
Model Data
m)
h (m) pth (
ept De
D
o o
longitude=62 E longitude=62 E
Figure3.Horizontalandverticaldistributionsofoxygen.Distributionofannualmeanoxygen(inmmolm−3)(top)at200mand(bottom)
intheupper1000malong62◦EassimulatedinROMS(left)andfromWorldOceanAtlas(2009)dataset.
acterizes the simulated salinity fields (Fig. A3 and Fig. A4). Although the model generally reproduces the observed surface
distributionofsalinity,itunderestimatesthedeeppenetrationofsaltysurfacewatersinthenorthernregionoftheArabianSea.
Thishigh-salinitywaterischaracteristicoftheGulfwateranditsunderestimationinourmodelindicatesthatthemodelmay
underestimatethesubductionoftheGulfwaterintotheArabianSea.Thismightbeduetomodellimitationssuchasthelack
5 ofadequateverticalandhorizontalresolutionsaswellasduetopotentialinconsistenciesinatmosphericforcingdataoverthe
Gulfregion.
ThemodelsuccessfullysimulatesthehighchlorophyllconcentrationsinthenorthernandwesternArabianassociatedwith
thewinterandsummerblooms,respectively(Fig.2).Themodelalsoreproducesthehighchlorophyllconcentrationsthatare
observedofftheIndianwestcoastinbothseasons.However,themodelsubstantiallyoverestimatestheobservedchlorophyll
10 concentrations off the Somali coast south of 4◦N in both seasons. This is likely due to our nitrogen based NPZD model
8
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
that does not account for potential limitation of productivity by silicic acid and iron in this region as documented by Koné
etal.(2009).Besidesthismodeldeficiency,thelarge-scaledistributionofchlorophyllisfairlywellrepresentedinmostofthe
ArabianSeadespiteatendencytounderestimatechlorophyllintheopenocean.Thismighthavetodowiththemodel’slackof
smallphytoplanktonfunctionalgroupsthataremoreadaptedtooligotrophicconditions.Thesimulatedsurfacedistributionof
5 nitrateisconsistentwithobservationsfromtheWorldOceanAtlasdatasetinbothseasons(AppendixA:Supplementaryfigures,
Fig.A5).Inparticular,thehighsurfaceconcentrationsassociatedwiththesummerupwellinginthewesternArabianSeaandthe
winterconvectioninthenorthernArabianSeaarecapturedbythemodel.Finally,thesimulatedoxygendistributionsatdepth
comparegenerallywellwithobservations(Fig.3).Inparticular,thelocation,thesizeandthedepthoftheArabianSeaoxygen
minimum zone (OMZ) (defined here by O < 60 mmolm 3) are correctly reproduced. However, the model overestimates
2 −
10 the intensity of the OMZ in the northern Arabian Sea and off the Indian west coast. This might be driven by the model
underestimatededdyactivitythatmayleadtounderestimatedOMZventilationasshownbyLachkaretal.(2016).Thismight
also partly result from the misrepresentation and likely underestimation of the injection of the Gulf waters into the northern
ArabianSeaatintermediatedepths.
A more quantitative evaluation of model skill is performed using in-situ observations from the World Ocean Database
15 (2013)thatwebinnedintoa0.5◦ monthlyclimatology.Nospatialinterpolationismadeandthemodelandobservationsare
comparedattheobservationpoints.TheresultsofthisevaluationaresummarizedgraphicallyusingTaylordiagrams(Taylor,
2001) quantifying the similarity between the observations and model in terms of their correlation, the amplitude of their
standard deviations and their centered root-mean-square (RMS) difference (Fig. 4). Our statistical analysis shows that the
simulated and observed SSH anomalies have similar standard deviations and correlate strongly with correlation coefficients
20 of 0.77 and 0.83 in winter and summer, respectively (Fig. 4a). Similarly, the quantitative comparison of the simulated and
observedSSTclimatologiesshowssimilarstandarddeviationsandsmallRMSerrorsaswellasveryhighcorrelations(r>0.92)
in both seasons, confirming the visual comparison conclusions (Fig. 4a). Likewise, the simulated and observed chlorophyll
distributionsshowrelativelylowRMSerrorsinbothseasonswithdecentcorrelationsrangingfrom0.69insummerto0.74in
winter(Fig.4a).DespitetheunderestimationofthepenetrationofthewarmandsaltysurfacewatersinthenorthernArabian
25 Seaatdepth,thelargescaledistributionsoftemperatureandsalinityintheupperoceanarecapturedbythemodelasindicated
bylowRMSerrorsandveryhighcorrelationswithin-situobservations(r>0.91)forbothtemperatureandsalinity(Fig.4b).
Additionally, the model shows an overall high skill in reproducing the observed large-scale distributions of both nitrate and
oxygen,withcorrelationswiththeWorldOceanDatabaserangingintheupper200mfrom0.88fornitrateinsummermonths
toaround0.93foroxygeninbothseasons(Fig.4b).
30 Insummary,despitesomelocalbiases,themodelgenerallyshowsreasonableskillinreproducingthelarge-scalefeaturesof
thecirculationintheArabianSeaaswellastheseasonaldynamicsofphytoplanktonbloomsintheregion.Moreimportantly,
itreproducesfairlywellthelocationandstructureoftheArabianSeaoxygenminimumzone.Thediscussionofthepotential
impactofthemodellimitationsonourresultswillbeaddressedindetailinthediscussionsection.
9
BiogeosciencesDiscuss.,doi:10.5194/bg-2017-146,2017
ManuscriptunderreviewforjournalBiogeosciences
Discussionstarted:2May2017
(cid:13)c Author(s)2017.CC-BY3.0License.
Taylor Diagrams
(a) sea surface (b) upper ocean
Winter (JFM)
Summer (JJAS)
Figure4.Taylordiagramdisplayingstatisticalcomparisonofmodeledandobservedfields.Taylor(2001)diagramofsimulated(a)sea
surfacetemperature(datapointslabeled“1”),seasurfaceheightanomalies(datapointslabeled“2”)andsurfacechlorophyll-a(datapoints
labeled “3”) in winter (blue) and summer (red) months and (b) temperature (data points labeled “1”), salinity (data points labeled “2”),
oxygen(datapointslabeled“3”)andnitrate(datapointslabeled“4”)intheupper1000mandinthetop200minwinter(blue)andsummer
(red)months.ThereferencepointoftheTaylordiagramcorrespondsto(a)SeaWiFSobservationsforchlorophyll,AVHRRdataforsurface
temperatureandAVISOclimatologyforseasurfaceheightanomaliesand(b)WorldOceanDatabase2013forallfourvariables.Theradius
(distance to the origin point) represents the modeled standard deviation relative to the standard deviation of the observations. The angle
betweenthemodelpointandtheX-axisindicatesthecorrelationcoefficientbetweenthemodelandtheobservations.Finally,thedistance
fromthereferencepointtoagivenmodeledfieldrepresentsthatfield’scenteredrootmeansquare(RMS)withrespecttoobservations.
3 Results
To explore the sensitivity of the Arabian Sea ecosystem to changes in the intensity of monsoon winds, we consider various
scenarios of idealized wind perturbations. A first set of scenarios consists in increasing (resp. decreasing) the wind stress
over the whole domain and throughout the year by 20% and 50%, respectively. In a second set of experiments, we increase
5 the wind stress by 50% in summer (resp. winter) and decrease it by 50% in winter (resp. summer). This is to account for
perturbationscenarioswheresummerandwintermonsoonwindsevolveinoppositedirectionsassuggestedbymultiplepaleo
records(e.g.KlöckerandHenrich,2006)showingintensificationofsummermonsoontobeconcomitantwiththeweakeningof
wintermonsoonandviceversa.Thiswillalsoallowustodisentangletheimpactsofthechangesineachmonsoonseasonand
determinetheirrespectivecontributionstotheoverallresponse.Thewindperturbationsareappliedinstantlyandmaintained
10 over 50 years for each scenario (abrupt perturbation). The impact of perturbations entailing gradual changes in the wind
intensitywillbeaddressedlaterinsection4.
10