Spatio-temporal variations of chlorophyll from satellitederived data and CMIP5 models along Indian coastalregions

DHANYA JOSEPH, G ROJITH, P U ZACHARIA*, V H SAJNA, S AKASH andGRINSON GEORGE

ICAR-Central Marine Fisheries Research Institute, Kochi, Kerala, India.*Corresponding author. e-mail: zachariapu@gmail.com

MS received 2 October 2020; revised 21 April 2021; accepted 22 April 2021

Comparison of chlorophyll data of three sets of CMIP5 models for RCP 4.5 (MPI-ESM-MR, HadGEM2-ES and GFDL-ESM2M) and RCP 6.0 (IPSL-CM5A-LR, HadGEM2-ES and GFDL-ESM2M) were donewith satellite derived data (OC-CCI) for the period of 1998–2017 along four Indian coastal regions. Themonthly, yearly and zone-wise seasonal comparison between model and satellite data were carried out.Analysis of monthly variations of chlorophyll during 1998–2017 reveals that the satellite data showmaximum value of 0.53 mg/m3 in September, whereas all other models show maximum in August. Yearlyanalysis indicates maximum satellite data in the year 2004, while minimum was observed in 2015.HadGEM2-ES exhibited maximum model value and the lowest was found for IPSL-CM5A-LR. It wasobserved that the maximum chlorophyll value of 2.56 mg/m3 for satellite data was in the monsoon seasonand the lowest value of 0.14 mg/m3 was in the pre-monsoon. Seasonal analysis reveals no clear matchamong model and satellite values in any of the coastal regions. In northwest and northeast regions, thesatellite values were found higher than the model values in most of the years, whereas in other regions, themodel values were found Cuctuating with the satellite values. Owing to the mismatch of the model and thesatellite values, the work cautions to apply biases or corrections on usage of RCP model data for regionalmarine climate change research.

Keywords. Chlorophyll; CMIP5; Indian Ocean; climate change; Indian coastal regions; RCP scenarios.

1. Introduction

Climate change has significant impacts on marineecosystems (IPCC 2001; Hoegh-Guldberg andBruno 2010; Brennan et al. 2016; Hoegh-Galbraithet al. 2017; BindoA et al. 2019), as well as induceschanges in the abundance and distribution ofcommercial Bsh species (Doney et al. 2011). Oceanclimate plays key role in the sustenance of marineproductivity, nutrient availability, survival ofmarine organisms (Herr and Galland 2009) and the

pattern of marine species richness (Macpherson2002; Vivekanandan 2011). The mounting eAect oncoastal Bsheries poses serious concern (Zachariaet al. 2016; Barange et al. 2018; Palomares andPauly 2019), which warrants scientiBc interven-tions to monitor and project the oceanographicvariable changes. A decline of around 8.6% of oceanprimary productivity due to climate change by theend of this century was forecasted (Bopp et al.2013). Changes in the chlorophyll (Chl) pattern ofglobal ocean was reported by Gregg et al. (2017),

J. Earth Syst. Sci. (2021) 130:153 � Indian Academy of Scienceshttps://doi.org/10.1007/s12040-021-01663-6 (0123456789().,-volV)(0123456789().,-volV)

while several regional studies report the impacts ofchlorophyll variations on marine Bsh resources(Bhaskar et al. 2016; Sajna et al. 2019; Bharti et al.2020). ScientiBc community widely uses the Cou-pled Model Intercomparison Project 5 (CMIP5)(Taylor et al. 2012) for global and regional climaterelated studies. The climate change projections ofIndian Ocean based on CMIP5 model under Rep-resentative Concentration Pathways (RCPs) sce-narios were reported for sea surface temperature,sea surface salinity, sea level rise, precipitation andpH (Akhiljith et al. 2019). Regional climatic pro-jection data are used to unravel the eAect ofchlorophyll on commercial Bsh species (Sajna et al.2019).Representative concentration pathways (RCPs)

represent classes of mitigation scenarios that pro-duce emission pathways following various assumedpolicy decisions which would inCuence the timeevolution of the future emissions of GHGs, aero-sols, ozone, and land use/land cover. The four RCPscenarios (RCP 2.6, RCP 4.5, RCP 6.0 and RCP8.5) are based on the change in radiative forcing atthe tropopause by 2100 relative to pre-industriallevels. RCP 4.5 and RCP 6.0 are considered asstabilization scenario, while RCP 2.6 is consideredas ideal and RCP 8.5 as uncontrolled changes(Taylor et al. 2012). CMIP5 includes different cli-mate models, and it follows identical forcingpathways for historical and projections. Differencebetween models may arise due to variations inmodel structure, complexity, spatial resolution andinitial conditions. The resolution of models devel-oped by each institutes may also vary.India is the second largest country in global Bsh

production (Hand book on Bsheries statistics 2018)and a total Bsh production of 12.59 metric tonneswas recorded during the period of 2017–2018. Theimpacts of climate change on Indian marine Bsh-eries sector create more ripples in the nationaleconomy owing to the livelihood dependency onthis sector. It is imperative to safeguard the Indianmarine Bsheries sector from the wrath of oceano-graphic variations.ScientiBc reports are available on the paucities,

biases and the shortcomings of the climate modelsimulations and projections upon which IPCCdepends. The General Circulation Models (GCM)inherit uncertainty (Olesen et al. 2007) and modelthe earth process in a coarse scale, which areinsufBcient for precise agricultural studies. Theperformance variations of several GCMs of CMIP5have also been reported (Gu et al. 2015; Khan et al.

2018), which imply the necessity to validate themodel outputs with real or observed values of theregion so as to improve the suitability and accuracyof projections. Liu et al. (2013) compared BveCMIP5 models and satellite observations forchlorophyll data of global and Indian Ocean, andfurther identiBed that the performance of theselected CMIP5 models shows maximum Chl con-centration in western part of Arabian Sea andwestern side of Bay of Bengal. Significant ampli-tude diversity among models during the repro-duction of chlorophyll concentrations over theregions was further reported. Many of the CMIP5models fail to represent spatial variability of IOmainly in the western region which is a robust(coastal and open ocean) upwelling zone (Roxyet al. 2016). The downscaling (Segu�ı et al. 2010)and bias correction (Hawkins et al. 2013) methods(Jalota et al. 2018) are commonly used to reducethe model variations in regional scales and toderive more localized realistic model outputs.Misrepresentation of local physical process is oneamong the reasons for model biases (Wang et al.2014), which is of concern in coast-wise marineresearch. Cyclones, upwelling, ENSO are amongthe localized ocean related phenomena that maycause bias to the models. Bias corrections focus tomake climate model outputs more realistic(Navarro-Racines et al. 2020) by the application ofany of the methods such as statistical techniques(Ehret et al. 2012; Hawkins et al. 2013), incorpo-ration of change factor derived from GCM ontothe historical observations (Hawkins et al. 2013;Gebrechorkos et al. 2019), representation ofnudging factor to the climate model output (Segu�ıet al. 2010; Jakob Themeßl et al. 2011), or byquantile mapping of climate model outputs ontoobservations (Segu�ı et al. 2010; Jakob Themeßlet al. 2011).Investigations on the spatial and temporal pat-

tern changes of the oceanographic variables help toframe the region-speciBc habitat and resourcemanagement plans. As successful simulation ofparticular region does not essentially represent agood model for another region (Schneider et al.2008; Anav et al. 2013), validation of the modeldataset with observed or real-time values is nec-essary to elucidate the variations and biases inregional downscaling. Indian coastal zone basedtime series analysis of model and satellite valueswere done for sea surface temperature (Dhanyaet al. 2019) and the present paper compares thesatellite data of chlorophyll variations in the

153 Page 2 of 12 J. Earth Syst. Sci. (2021) 130:153

northern Indian Ocean with CMIP5 global climatemodel projections.

2. Data and methodology

The seasonal and region-wise satellite derived data(SDD) of chlorophyll taken from European SpaceAgency’s Ocean Colour-Climate Change Initiative(OC-CCI) (Sathyendranath et al. 2019) werecompared with six CMIP5 model derived data(MDD) pertaining to 1998–2017. The OC-CCIsatellite data is of 4 km resolution. Based on theregion-wise characteristic variations in IndianOcean, the region bounded by 50�–100�E longitudeand 5�S–30�N latitude were selected as study area.The region was further subdivided into two regionsin eastern part of India as southeast (SE 1�S–15�N;75�–87�E) and northeast (NE 14�–25�N; 80�–90�E)and other two regions from western side of thecountry as southwest (SW 1�S–15�N; 65�–76�E)and northwest (NW 14�–25�N; 63�–73�E). Thespatial and temporal analysis of chlorophyll vari-ations in the four coastal regions in three differentseasons of pre-monsoon (February–May), monsoon(June–September) and post-monsoon (October–January) were done. The chosen time period for allanalysis were categorized as historical RCP (dur-ing 1998–2005) and projected RCP (during2006–2017). Multiple climatic projection modelsare available for different RCPs (Moss et al. 2010;Van Vuuren et al. 2011) under the CMIP5 (VanVuuren et al. 2011; Taylor et al. 2012) experiment(http://cmip-pcmdi.llnl.gov/cmip5data˙portal.html). Two intermediate emission scenarios ofRCP 4.5 and RCP 6.0 were used for comparison ofhistorical RCP and projected RCP of selectedmodels. The selected six models used in this studyfor the two RCP scenarios are as mentioned intable 1 and analysis were also done with theensemble mean. Two ensemble means, EM1 andEM2 were created as the average value of threemodels from both RCP scenarios. In CMIP5project, ensemble members are named in the rip-nomenclature, where r represents realization,i represents initialization and p signiBes physicsfollowed by integers (Taylor et al. 2009) and allthe selected models in this study have ensemblecharacteristics of r1i1p1. For uniformity in com-parison, each of the selected models was interpo-lated to one degree resolution and further timeseries analysis of satellite data with each modelswere done.

3. Results and discussions

The comparative study of 20 years of chlorophylldata of six different models unravelled the varia-tions of CMIP5 model values with satellite value.The inter-annual, seasonal and regional analyses inspatial and temporal scale inferences are presentedin the following sections.

3.1 Spatial variations of Chl during 1998–2017

As shown in Bgure 1, EM1 and EM2 were selectedinstead of comparing each model with SDD. As perSDD, Chl distribution is more in coastal regions,whereas for both model ensemble means, thecoastal regions do not exhibit much data as that ofSDD values with exception in Oman coast, whichshows data points for the ensemble mean of his-torical and projected models. Chlorophyll distri-bution is the variation of chlorophyll concentrationwithin a selected geographical location. It impliesspatial variation within the selected region(50�–100�E and 5�S–30�N) as represented inBgures 1 and 2. In the spatial plot of satellite, thechlorophyll concentrations are clearly observed incoastal regions, whereas the ensemble mean of twoset of models EM1 and EM2 for historical andprojected does not clearly exhibit chlorophyllpresence. In other words, the values of ensemblemean were found lower than that of satellite datafor the region. The coastal region of Oman showshigher values of chlorophyll during 1998–2005 than2006–2017. The study of satellite and model pre-dictions by Roxy et al. (2016) indicates thereduction of marine productivity in western Ara-bian Sea due to the upper ocean warming. How-ever, it is worthy to note the contrary assumptionreported by Praveen et al. (2016) as per which thenet upwelling increases in the next century, due tothe eAect of northward shift in the monsoon low-level jet under RCP 8.5, which may cause thecoastal wind to angle against the Oman coast.Therefore, the possibility of increased marine pro-ductivity in the next century could not be ruledout. Chlorophyll variations of Arabian Sea mainlydepend on the southwest and northeast monsoonseasons (Banse and English 2000; Marra andBarber 2005; Sarma et al. 2012). Liu et al. (2013)report that the maximum chlorophyll regions andthe seasonal variations are simulated by CMIP5,but the simulated chlorophyll concentration fromeach individual model are quite different. The

J. Earth Syst. Sci. (2021) 130:153 Page 3 of 12 153

present work also reports that though the maxi-mum chlorophyll regions could be identiBed inCMIP5 models, differences in chlorophyll concen-tration values were observed among individualmodels. The spatial pattern of all six MDD used inthis comparison study are represented in Bgure 2,in which (a, b and c) represent the variations inthree models of historical RCP 4.5 and (d, e and f)represent the variations in three models for his-torical RCP 6.0.Historical RCP 4.5 of GFDL-ESM2M model

shows chlorophyll variations (0.27–0.29 mg/m3) inthe study region of Indian Ocean, whereas othertwo models exhibit variations in the range of0.63–0.69 mg/m3 for HadGEM2-ES and the rangefor MPI-ESM-MR was 0.42–0.48 mg/m3. It couldbe further inferred that among the three models,MPI-ESM-MR is relatively better as it has good

simulated values and data availability, followed byHadGEM2-ES and GFDL-ESM2M. In case of his-torical RCP 6.0, the GFDL-ESM2M model data isnot available in the eastern side, and the modelexhibits no significant chlorophyll variations in thewestern coast of India, whereas the IPSLCM5A-LRmodel has low data values (0.04–0.05 mg/m3). TheHadGEM2-ES exhibits data variations on westernand eastern regions of India. In case of projecteddata under RCP 4.5 scenario, the GFDL-ESM2Mmodel does not have data for eastern side, and themodel exhibits variations ranging from 0.26 to0.29 mg/m3 for western side, whereas MPI-ESM-MR could be seen as better (0.42–0.48 mg/m3)among the three models followed by HadGEM2-ESwith values ranging from 0.59 to 0.66 mg/m3. Dueto unavailability of MPI-ESM-MR model in RCP6.0, the IPSL-CM5A-LR model was chosen, which

Table 1. Details of CMIP5 climate model datasets from RCP 4.5 and RCP 6.0 used in the present study (https://portal.enes.org/data/enes-model-data/cmip5/resolution).

Sl. no. Model id Expansion Selected scenario Spatial resolution Ensemble mean

1 MPI-ESM-MR Max Planck Institute for Meteorology

(MPI-ESM-MR-M)

RCP 4.5 1.875 9 1.85 EM1

2 HadGEM2-ES Met ODce Hadley Centre RCP 4.5 1.875 9 1.25

3 GFDL-ESM2M Geophysical Fluid Dynamics Laboratory RCP 4.5 2.5 9 2.0

4 GFDL-ESM2M Geophysical Fluid Dynamics Laboratory RCP 6.0 2.5 9 2.0 EM2

5 HadGEM2-ES Met ODce Hadley Centre RCP 6.0 1.875 9 1.25

6 IPSL-CM5A-LR Institute Pierre-Simon Laplace RCP 6.0 3.75 9 1.875

Figure 1. The annual average of Chl for satellite and model ensembles of two RCP scenarios. (a) Annual average of satellite dataduring 1998–2005. (b) Annual average of historical EM1. (c) Annual average of historical EM2. (d) Annual average of satelliteduring 2006–2017. (e) Annual average of projected EM1 and (f) Annual average of projected EM2.

153 Page 4 of 12 J. Earth Syst. Sci. (2021) 130:153

shows very low values (0.04–0.05 mg/m3) as com-pared to other MDD values with no significantvariations in the study regions. The HadGEM2-ESmodel is found to be comparatively better(0.59–0.66 mg/m3) than other two models for RCP6.0. The projected data are as shown in Bgure 2, inwhich (g, h and i) represent RCP 4.5, while (j, kand l) for RCP 6.0. The mismatch between themodels and observed data arise due to severalreasons such as variations in model structure,complexity, spatial resolution and initial condi-tions. Possibility exist that CMIP5 models canunderestimate multi-decadal variations in the trueresponse of the climate system to external forcingor misrepresent the forcing (Booth et al. 2012;Murphy et al. 2017).

3.2 Temporal variations of Chlduring 1998–2017

Figure 3 represents the inter-annual variations ofChl distributions and is elusive that the SDD shows

increase since 2001 with maximum value (0.47 mg/m3) in 2004, followed by relative decrease till 2012.The EM1 is higher than SDD of OC-CCI, whereasEM2 is less than SDD during 2001–2012 and higherin other years. Estimation of each MDD shows nosignificant yearly variations (0.04–0.05 mg/m3) forIPSL-CM5A-LR model, whereas the two models ofHadGEM2-ES from each scenario shows similarvalues (0.59–0.69 mg/m3) higher than the SDD.The two models of GFDL-ESM2M from each sce-nario show lesser values (0.24–0.29 mg/m3 forRCP 4.5 and 0.26–0.29 mg/m3 for RCP 6.0) thansatellite data.Monthly variation of historical and projected

mean chlorophyll values during 1998–2017 is asshown in Bgure 4. SDD and the MDD exhibitedsimilar trend of variations during the historicalperiod of 1998–2005 as shown in Bgure 4a. Themaximum value of SDD (0.56 mg/m3) wasobserved in September, whereas all MDD exhibitedmaximum in August. Among the six models, IPSL-CM5A-LR exhibited very low chlorophyll values,

Figure 2. Spatial pattern of historical and projected RCPs of selected models of CMIP5 of Indian Ocean (50�–100�E and5�S–30�N). Left panel (a, d, g, j) represents the GFDL-ESM2M. Middle panel (b, e, h, k) shows HadGEM2-ES. Right panelshows (c, i) MPI-ESM-MR of RCP 4.5 and (f, l) IPSL-CM5A-LR of RCP 6.0.

J. Earth Syst. Sci. (2021) 130:153 Page 5 of 12 153

while the two models of HadGEM2-ES under RCP4.5 and RCP 6.0 exhibited identical values. GFDL-ESM2M models from two RCPs show almost sim-ilar variations with identical values for Febru-ary–July as well as for September, but smalldeviations in values were observed for othermonths.Historical value of MPI-ESM-MR of RCP 4.5

exactly matches with EM1 simulations for July andSeptember, while other months show deviations.Similarly, MPI-ESM-MR shows uniform variationswith SDD for February–May as well as for Octoberand November, while the MDD was higher thanSDD for other months. Figure 4b representsthe monthly variations of chlorophyll during

2006–2017 and the IPSL-CM5A-LR model showsvery low values compared to other models andSDD. The maximum value of SDD (0.49 mg/m3)was observed in September, whereas all MDDexhibited maximum in August. The satellite aswell as model data exhibited similar trend. As theprojected datasets are from different RCP scenar-ios, minor deviations were observed for HadGEM2-ES and GFDL-ESM2M models. In comparisonwith the SDD values, the two models of Had-GEM2-ES were found to be higher, whereas thetwo GFDL-ESM2M model data were observed aslower. The MPI-ESM-MR data match with SDDduring February–April and exhibit higher valuethan the SDD, and reaches the maximum

Figure 3. Inter-annual variations of chlorophyll during 1998–2017.

Figure 4. Monthly variations of Chl during (a) 1998–2005 and (b) 2006–2017.

153 Page 6 of 12 J. Earth Syst. Sci. (2021) 130:153

(0.83 mg/m3) in August; and further decrease intrend was observed. The EM1 data show highervalues (0.29–0.74 mg/m3) than SDD for allmonths, whereas the EM2 correctly matches withSDD for June to August as well as for December,while for other months the EM2 values were foundto be lower than satellite values. Models are sce-nario dependent with possibilities of over-estima-tion and under-estimation. The plot shows thatensemble means are good choice than the individ-ual models. The IPSL-CM5A-LR exhibited verylow values for each year, while MPI-ESM-MR ofRCP 4.5 model is relatively better. It has beenreported that the IPSL-CM5A-LR simulationsmodel exhibits many biases considered as longstanding systematic biases of many coupledocean–atmosphere models (Dufresne et al. 2013),while MPI-ESM-MR has been reported to simulatethe mean climatology of chlorophyll concentrationswith a relatively lower bias and realistic spatialdistribution in the Indian Ocean (Roxy et al. 2016).

3.3 Seasonal variations of chlorophyll in fourcoastal zones

SDD shows higher variations and values in north-ern regions during 1998–2017 for pre-monsoonseason as shown in Bgure 5. In NW region, SDDwas observed to be higher than MDD with mini-mum value recorded as 0.71 mg/m3 in 2016 andmaximum value recorded as 1.15 mg/m3 in 2008.The chlorophyll variations of SDD for NE regionwere within the range of 0.28–0.74 mg/m3, whereasthe MDD exhibited the maximum at 0.50 mg/m3

for HadGEM2-ES. Lower values for SDD wereobserved in 2013 (0.29 mg/m3) and 2016 (0.28 mg/m3). The SDD of SW region was observed withinthe range of 0.20–0.37 mg/m3 with maximumvalue recorded in 2003, and further decrease intrend was observed. Very low SDD were found forSE region with exception to IPSL-CM5A-LRmodel. For all regions, the IPSL-CM5A-LR modelvalues were found to be very much lower than SDDas well as other MDDs.For monsoon season, as shown in Bgure 6, the

SDD shows higher values than MDD in northernregions and southwest region, whereas for south-east region, the MDD of two HadGEM2-ES mod-els, MPI-ESM-MR and EM1 were found to behigher than the SDD. During 1998–2017, thenorthwest region exhibited the maximum value of2.56 mg/m3 for SDD in the year 2008. Significantinter-annual variations (0.47–2.56 mg/m3) were

found for SDD of NW region, while the MDD andensemble means exhibit lower variations.In NE region the SDD, MPI-ESM-MR model

and EM1 shows significant variations annually,with highest annual values recorded as 0.96 mg/m3

(2006), 1.03 mg/m3 (2008) and 0.73 mg/m3 (2008),respectively, above an arbitrary value of 0.4 mg/m3. In SW region, the MPI-ESM-MR and EM1show significant annual variations in line with theSDD, whereas other MDD exhibit relatively lesservariations. Maximum SDD was recorded as2.09 mg/m3 in 2013, whereas for MDD, the highestwas recorded as 1.13 mg/m3 in 2011 for MPI-ESM-MR. In SE region, the satellite chlorophyll valuesare relatively lesser than other regions and themaximum of 0.55 mg/m3 was recorded in 2016. InSW region of India, which is considered as a pro-ductive region globally (Madhupratap et al. 1996;Sarangi et al. 2008), a noteworthy change(0.26–0.43 mg/m3) in the value of the IPSL-CM5A-LR model is observed during the southwestmonsoon season. However, IPSL-CM5A-LR modelshows no significant variations in the pre-monsoonand the post-monsoon seasons.The NW region, as shown in Bgure 7, record

higher chlorophyll values for SDD during post-monsoon season with maximum as 1.21 mg/m3 in2005, and further decrease in trend was observed.Annual variations of the MDD were relativelymuch lower than the SDD values of the region. Innortheast region, the SDD mostly shows highervalues than MDD with highest value of 0.72 mg/m3 recorded in 2007 and further shows decreasingtrend. The HadGEM2-ES model data of the regionrecords highest of 0.63 mg/m3 in 2007 for bothRCP scenarios. IPSL-CM5A-LR model also exhi-bits the noticeable data variations (0.07–0.14 mg/m3) in the NE region. Satellite-derived data showssignificant variations in the SW region and showsmostly higher values than MDD with maximumrecorded value as 0.82 mg/m3 in 2016. In the SEregion, SDD shows very low values (0.21–0.34 mg/m3) than MDD except for IPSL-CM5A-LR. TheGFDL-ESM2M models are relatively closer to theSDD, while HadGEM2-ES models exhibit highervalues for the region. Figures 6 and 7 representchlorophyll distribution in monsoon and post-monsoon seasons and hence the change. In both theseasons, the chlorophyll distribution are more inthe selected region, especially in western and SEregion of India. The IPSL-CM5A-LR model wasfound to respond and simulate only in case ofhighest chlorophyll concentration.

J. Earth Syst. Sci. (2021) 130:153 Page 7 of 12 153

Higher chlorophyll concentrations were observedin the study area during monsoon (JJAS) and post-monsoon season (ONDJ) with maximum concen-tration for MDD and SDD in monsoon season. Thehigher chlorophyll values and variations wereobserved in the NW region consisting of Kutch,Narmada and Mahi delta and Gulf of Cambay. TheSW region exhibited variations in the range of0.81–2.09 mg/m3, while the NE region of India

shows Chl variations in the range of 0.43–0.96 mg/m3. The lowest variations were observed as0.26–0.28 mg/m3 in the SE region. Seasonal andregional chlorophyll variations were also reportedin several studies (Sarangi et al. 2008; Bhushanet al. 2018; L�evy et al. 2007). The regional andseasonal variations of chlorophyll concentrationscould be attributed to the variations in theoceanographic parameters (Nieto and M�elin 2017;

Figure 5. Annual mean of chlorophyll distribution of six models, EM1, EM2, and satellite observations are shown in four regions(SW, NW, SE and NE) during 1998–2017 for pre-monsoon.

Figure 6. Annual mean of chlorophyll distribution of six models, EM1, EM2, and satellite observations are shown in four regions(SW, NW, SE and NE) during 1998–2017 for monsoon.

153 Page 8 of 12 J. Earth Syst. Sci. (2021) 130:153

Dunstan et al. 2018) as one among the contributingfactors. Comparison of satellite data with EM1,EM2 and six individual model values reveals theSDD as higher than MDD in all seasons fornorthern regions and the MDD shows only minordifference among them for pre-monsoon season. Insouthern region, the SDD and MDD shows similartrend, though not a close match.Several studies have been reported on the devia-

tions of models from observations (Qian et al. 2016;Dhanya et al. 2019) and climate model projectionsare commonly associated with uncertainties andbiases (Taylor et al. 2012; Lupo et al. 2013). Kravt-sov (2017) suggested the possibility of differences inthe climate models due to the misrepresentation ofCMIP5 models, attributed to the variations in theinternal climate system dynamics. Internal vari-ability of ocean is also an important concept to havesubstantial inCuence on the particular region (Che-ung et al. 2017). Hogan and Sriver (2019) reportedthat the difference between the two models showsuncertainty due to the internal variability of theocean. Spatio-temporal analysis of SDD and CMIP5model data of chlorophyll elucidates the inconsis-tency among them, specifically for regional scale,which could be attributed to the multiple factorssuch as data resolution, unavailability of data,regional and seasonal climatological forcing fre-quencies, inCuences of regional oceanographic pro-cesses and biases or errors in the preliminary

development stage of global models (L�evy et al.2007; Nieto and M�elin 2017; Dunstan 2018). Themodel simulations have limitations to reCect theregional climatic variabilities, and may be a con-tributing factor to the deviations in global to regio-nal downscaling process. This paper could establishthe chlorophyll deviations of satellite and CMIP5models, for Indian coastal zones, which could befurther quantiBed by marine researchers as per thescope of concerned studies. Region-speciBc modelswith validations are needed to accurately depict theregional level chlorophyll variations. Climatechange projections related research essentiallyshould incorporate model corrections, to attainaccurate zone-wise chlorophyll model valuepredictions.

4. Conclusions

The spatio-temporal analysis of chlorophyll varia-tions specifically for the coastal regions of Indiareveals the regional and seasonal distributionalvariations of chlorophyll during 1998–2017. Thesignificant chlorophyll variations were mainly dis-tinguishable in coastal regions than open Ocean,with satellite-derived data indicating significantvariations and values in northern coasts thansouthern coasts. The inter-annual variation depictsthe maximum chlorophyll values, obtained for

Figure 7. Annual mean of chlorophyll distribution of six models, EM1, EM2, and satellite observations are shown in four regions(SW, NW, SE and NE) during 1998–2017 for post-monsoon.

J. Earth Syst. Sci. (2021) 130:153 Page 9 of 12 153

satellite and model data in different years. Theanalysis unravels the chlorophyll maximum valuesfor satellite data in September, whereas the maxi-mum values for models were found in August.Comparison of the 20 years of satellite data withsix CMIP5 data reveals that the maximumchlorophyll values and variations are in thenorthwest region of India. The discrepanciesbetween satellite and CMIP5 models of RCP 4.5and RCP 6.0 were established, which cautions toapply model corrections for regional forecast studies.

Acknowledgements

This research was funded by the ICAR sponsoredNational Innovations in Climate Resilient Agri-culture Project (NICRA) carried out at ICAR-Central Marine Fisheries Research Institute Kochi.The authors are grateful to the Director ICAR-CMFRI, Kochi, for providing facilities. CMIP5data employed in this study are available at theProgram Climate Model Diagnosis and Intercom-parison (PCMDI) and satellite chlorophyll datafrom European Space Agency’s Ocean Color-Climate Change Initiative (OC-CCI) and theauthors are grateful to the data providers. Theauthors are also thankful to the research scholars inthe NICRA project for the support extended.

Author statement

Dr Dhanya designed the research, performed thedata analysis, result inferences and prepared theinitial manuscript. Dr Rojith contributed to theresearch framework reBnement, result inferencesand rewrote the manuscript. Dr Zacharia super-vised the research, reviewed and approved themanuscript. Sajna and Akash contributed to thedata analysis. Dr Grinson provided insights toRCP scenarios and performed the Bnal review ofthe manuscript.

References

Akhiljith P J, Liya V B, Rojith G, Zacharia P U, Grinson G,Ajith S, Lakshmi P M, Sajna V H and Sathianandan T V2019 Climatic projections of Indian Ocean during 2030,2050, 2080 with implications on Bsheries sector; J. Coast.Res. 86(SI) 198–208.

Anav A, Friedlingstein P, Kidston M, Bopp L, Ciais P, Cox P,Jones C, Jung M, Myneni R and Zhu Z 2013 Evaluating the

land and ocean components of the global carbon cycle in theCMIP5 earth system models; J. Climate 26(18) 6801–6843.

Banse K and English D C 2000 Geographical differences inseasonality of CZCS-derived phytoplankton pigment in theArabian Sea for 1978–1986 Deep Sea Research Part II;Topical Studies in Oceanography 47(7–8) 1623–1677.

Barange M, Bahri T, Beveridge M C, Cochrane K L, Funge-Smith S and Poulain F 2018 Impacts of climate change onBsheries and aquaculture: Synthesis of current knowledge,adaptation and mitigation options; United Nations’ Foodand Agriculture Organization, Rome.

Bharti V, Jayasankar J, Shukla S P, George G, Ambrose T V,Augustine S K, Sathianandan T V and Shafeeque M 2020Study on sea surface temperature and chlorophyll-a concen-tration along the south-west coast of India; Indian J. Geo-Mar. Sci. 49 51–56.

Bhaskar T U, Jayaram C and Rao K H 2016 Spatio-temporalevolution of chlorophyll-a in the Bay of Bengal: A remotesensing and bio-argo perspective; In: Remote sensing ofthe oceans and inland waters: Techniques, applications,and challenges; Int. Soc. Optics Photonics 987898780Z1–98780Z6.

Bhushan R, Bikkina S, Chatterjee J, Singh S P, Goswami V,Thomas L C and Sudheer A K 2018 Evidence for enhancedchlorophyll-a levels in the Bay of Bengal during early north-east monsoon; J. Oceanogr. Mar. Sci. 9 15–23.

BindoA N L, Cheung W W L, Kairo J G, Ar�ıstegui J, GuinderV A, Hallberg R, Hilmi N, Jiao N, Karim M S, Levin L,O’Donoghue S, Purca Cuicapusa S R, Rinkevich B, Suga T,Tagliabue A and Williamson P 2019 Changing ocean,marine ecosystems, and dependent communities; In: IPCCSpecial Report on the Ocean and Cryosphere in a ChangingClimate (eds) P€ortner H O, Roberts D C, Masson DelmotteV, Zhai P, Tignor M, Poloczanska E, Mintenbeck K,Alegr�ıa A, Nicolai M, Okem A, Petzold J, Rama B, WeyerN M, Intergovernmental Panel on Climate Change,Switzerland, pp. 477–587.

Booth B B B, Dunstone N J, Halloran P R, Andrews T andBellouin N 2012 Aerosols implicated as a prime driver oftwentieth century North Atlantic climate variability; Na-ture 484 228–232.

BoppL,Resplandy L,Orr JC,Doney SC,Dunne J P,GehlenM,HalloranP,HeinzeC, IlyinaT, SeferianR andTjiputra J 2013Multiple stressors of ocean ecosystems in the 21st century:Projections with CMIP5 models; Biogeosci. 10 6225–6245.

Brennan C E, Blanchard H and Fennel K 2016 Puttingtemperature and oxygen thresholds of marine animals incontext of environmental change: A regional perspective forthe Scotian Shelf and Gulf of St. Lawrence; PloS One11(12) e0167411.

Cheung A H, Mann M E, Steinman B A, Frankcombe L M,England M H and Miller S K 2017 Comparison of low-frequency internal climate variability in CMIP5 models andobservations; J. Climate 30(12) 4763–4776.

Dhanya J, Liya V B, Rojith G, Zacharia P U, Sajna V H andGrinson G 2019 Time series analysis of CMIP5 Model andobserved sea surface temperature anomaly along IndianCoastal Zones; J. Coast. Res. 86(SI) 239–247.

Doney S C, Ruckelshaus M, DuAy J E, Barry J P, Chan F,English C A, Galindo H M, Grebmeier J M, Hollowed A B,Knowlton N and Polovina J 2011 Climate change impactson marine ecosystems; Ann. Rev. Mar. Sci. 4 11–37.

153 Page 10 of 12 J. Earth Syst. Sci. (2021) 130:153

Dufresne J L, Foujols M A, Denvil S, Caubel A, Marti O,Aumont O, Balkanski Y, Bekki S, Bellenger H, Benshila Rand Bony S 2013 Climate change projections using theIPSL-CM5 Earth System Model: From CMIP3 to CMIP5;Clim. Dyn. 40(9) 2123–2165.

Dunstan P K, Foster S D, King E, Risbey J, O’Kane T J,Monselesan D, Hobday A J, Hartog J R and Thompson P A2018 Global patterns of change and variation in sea surfacetemperature and chlorophyll a; ScientiBc Reports 8(1) 1–9.

Ehret U, Zehe E, Wulfmeyer V, Warrach-Sagi K and Liebert J2012 HESS opinions ‘‘Should we apply bias correction toglobal and regional climate model data?’’; Hydrol. EarthSyst. Sci. 16(9) 3391–3404.

Galbraith E D, Carozza D A and Bianchi D 2017 A coupledhuman-Earth model perspective on long-term trends in theglobal marine Bshery; Nature Commun. 8(1) 1–7.

Gebrechorkos S H, H€ulsmann S and Bernhofer C 2019Statistically downscaled climate dataset for East Africa;ScientiBc Data 6(1) 1–8.

Gregg W W, Rousseaux C S and Franz B A 2017 Globaltrends in ocean phytoplankton: A new assessment usingrevised ocean colour data; Remote Sens. Lett. 8(12)1102–1111.

Gu H, Yu Z, Wang J, Wang G, Yang T, Ju Q, Yang C, Xu Fand Fan C 2015 Assessing CMIP5 general circulation modelsimulations of precipitation and temperature over China;Int. J. Climatol. 35(9) 2431–2440.

Hand book on Bsheries statistics 2018 Fishery Survey of Indiaon behalf of Department of Fisheries Onlooker Press,Mumbai.

Hawkins E, Osborne T M, Ho C K and Challinor A J 2013Calibration and bias correction of climate projections forcrop modelling: An idealised case study over Europe; Agr.Forest Meteorol. 170 19–31.

Herr D and Galland G R 2009 The ocean and climate change:Tools and guidelines for action; IUCN, Gland, Switzerland,72p.

Hoegh-Guldberg O and Bruno J F 2010 The impact of climatechange on the world’s marine ecosystems; Science328(5985) 1523–1528.

Hogan E and Sriver R L 2019 The eAect of internal variabilityon ocean temperature adjustment in a low-resolutionCESM initial condition ensemble; J. Geophys. Res.: Oceans124(2) 1063–1073.

IPCC 2001 Climate Change 2001, Synthesis Report; AContribution of Working Groups I, II, and III to the ThirdAssessment Report of the Intergovernmental Panel onClimate Change; Cambridge University Press, Cambridge,UK.

Jakob Themeßl M, Gobiet A and Leuprecht A 2011 Empirical-statistical downscaling and error correction of daily precip-itation from regional climate models; Int. J. Climatol.31(10) 1530–1544.

Jalota S K, Vashisht B B, Sharma S and Kaur S 2018Understanding climate change impacts on crop productivityand water balance; Academic Press, pp. 55–86, ISBN9780128095201, https://doi.org/10.1016/B978-0-12-809520-1.00002-1.

Khan N, Shahid S, Ahmed K, Ismail T, Nawaz N and Son M2018 Performance assessment of general circulation modelin simulating daily precipitation and temperature usingmultiple gridded datasets; Water 10 1793–1810.

Kravtsov S 2017 Pronounced differences between observedand CMIP5-simulated multidecadal climate variability inthe twentieth century; Geophys. Res. Lett. 44(11)5749–5757.

L�evy M, Shankar D, Andr�e J M, Shenoi S S C, Durand F andde Boyer Mont�egut C 2007 Basin-wide seasonal evolution ofthe Indian Ocean’s phytoplankton blooms; J. Geophys.Res.: Oceans 112(C12).

Liu L, Feng L, Yu W, Wang H, Liu Y and Sun S 2013 Thedistribution and variability of simulated chlorophyll con-centration over the tropical Indian Ocean from Bve CMIP5models; J. Ocean Univ. China 12(2) 253–259.

Lupo A, Kininmonth W, Armstrong J S and Green K 2013Global climate models and their limitations; In: Climatechange reconsidered II; Phys. Sci. 9 7–148.

Macpherson E 2002 Large-scale species-richness gradients inthe Atlantic Ocean; Proc. Roy. Soc. London Ser. B: Biol.Sci. 269(1501) 1715–1720.

Madhupratap M, Kumar S P, Bhattathiri P M A, Kumar M D,Raghukumar S, Nair K K C and Ramaiah N 1996Mechanism of the biological response to winter cooling inthe northeastern Arabian Sea; Nature 384(6609) 549–552.

Marra J and Barber R T 2005 Primary productivity in theArabian Sea: A synthesis of JGOFS data; Prog. Oceanogr.65(2–4) 159–175.

Moss R H, Edmonds J A, Hibbard K A, Manning M R, Rose SK, Van Vuuren D P, Carter T R, Emori S, Kainuma M,Kram T and Meehl G A 2010 The next generation ofscenarios for climate change research and assessment;Nature 463(7282) 747–756.

Murphy L N, Bellomo K, Cane M and Clement A 2017 Therole of historical forcings in simulating the observedAtlantic multidecadal oscillation; Geophys. Res. Lett.44(5) 2472–2480.

Navarro-Racines C, Tarapues J, Thornton P, Jarvis A andRamirez-Villegas J 2020 High-resolution and bias-correctedCMIP5 projections for climate change impact assessments;ScientiBc Data 7(1) 1–4.

Nieto K and M�elin F 2017 Variability of chlorophyll-a concen-tration in the Gulf of Guinea and its relation to physicaloceanographic variables; Prog. Oceanogr. 151 97–115.

Olesen J E, Carter T R, Diaz-Ambrona C H, Fronzek S,Heidmann T, Hickler T, Holt T, Minguez M I, Morales P,Palutikof J P and Quemada M 2007 Uncertainties inprojected impacts of climate change on European agricul-ture and terrestrial ecosystems based on scenarios fromregional climate models; Climatic Change 81(1) 123–143.

Palomares M L D and Pauly D 2019 Coastal Bsheries: Thepast, present, and possible futures; In: Coasts and Estuar-ies; Elsevier, pp. 569–576.

Praveen V, Ajayamohan R S, Valsala V and Sandeep S 2016IntensiBcation of upwelling along Oman coast in a warmingscenario; Geophys. Res. Lett. 43(14) 7581–7589.

Qian Y, Jackson C, Giorgi F, Booth B, Duan Q, Forest C,Higdon D, Hou Z J and Huerta G 2016 UncertaintyquantiBcation in climate modeling and projection; Bull.Am. Meteorol. Soc. 97(5) 821–824.

Roxy M K, Modi A, Murtugudde R, Valsala V, Panickal S,Prasanna Kumar S, Ravichandran M, Vichi M and L�evy M2016 A reduction in marine primary productivity driven byrapid warming over the tropical Indian Ocean; Geophys.Res. Lett. 43(2) 826–833.

J. Earth Syst. Sci. (2021) 130:153 Page 11 of 12 153

Sajna V H, Zacharia P U, Liya V B, Rojith G, Somy K, JosephD and Grinson G 2019 EAect of climatic variability on theBshery of Indian oil sardine along Kerala coast; J. Coast.Res. 86(SI) 184–192.

Sarangi R K, Nayak S and Panigrahy R C 2008 Monthlyvariability of chlorophyll and associated physical parame-ters in the southwest Bay of Bengal water using remotesensing data; Indian J. Mar. Sci. 37(3) 256–266.

Sarma Y V B, Al Azri A and Smith S L 2012 Inter-annualvariability of chlorophyll-a in the Arabian Sea and its Gulfs;Int. J. Mar. Sci. 2 1–11.

Sathyendranath S, BrewinR J, BrockmannC, BrotasV, CaltonB, Chuprin A, Cipollini P, Couto A B, Dingle J, Doerffer Rand Donlon C 2019 An ocean-colour time series for use inclimate studies: The experience of the ocean-colour climatechange initiative (OC-CCI); Sensors 19(19) 4285.

Schneider B, Bopp L, Gehlen M, Segschneider J, Fr€olicher TL, Cadule P, Friedlingstein P, Doney S C, Behrenfeld M Jand Joos F 2008 Climate-induced interannual variability ofmarine primary and export production in three globalcoupled climate carbon cycle models; Biogeosci. 5 597–614.

Segu�ı P Q, Ribes A, Martin E, Habets F and Bo�e J 2010Comparison of three downscaling methods in simulating the

impact of climate change on the hydrology of Mediter-ranean basins; J. Hydrol. 383 111–124.

Taylor K E, Stouffer R J and Meehl G A 2009 A summary ofthe CMIP5 experiment design; PCDMI Rep, 33p.

Taylor K E, Stouffer R J and Meehl G A 2012 An overview ofCMIP5 and the experiment design; Bull. Am. Meteorol.Soc. 93(4) 485–498.

Van Vuuren D P, Edmonds J, Kainuma M, Riahi K, ThomsonA, Hibbard K, Hurtt G C, Kram T, Krey V, Lamarque J Fand Masui T 2011 The representative concentration path-ways: An overview; Clim. Change 109 5–31.

Vivekanandan E 2011 Marine Bsheries policy brief-3; climatechange and Indian marine Bsheries; CMFRI Spec. Publ. 1051–97.

Wang C, Zhang L, Lee S K, Wu L and Mechoso C R 2014 Aglobal perspective on CMIP5 climate model biases; NatureClimate Change 4(3) 201–205.

Zacharia P U, Gopalakrishnan A, George G, Muralidhar Mand Vijayan K K 2016 Climate change impact on coastalBsheries and aquaculture in the SAARC region: Countrypaper – India; In: Climate Change Impact on CoastalFisheries and Aquaculture in South Asia, SAARC Agricul-ture Centre (SAC), Dhaka, pp. 1–25.

Corresponding editor: MARIPI DILEEP

153 Page 12 of 12 J. Earth Syst. Sci. (2021) 130:153