Remote Sensing Of Phytoplankton Size Glasses And Phytoplankton Population Structures In The East China Sea

Marine phytoplankton is the basis of food web and the most important primary producer in aquatic ecosystems,and the changes in the biomass and compostion of phytoplankton species can affect the function and structure of marine ecosystem.Therefore,knowledge on the distribution and variation of phytoplankton size classes(PSCs)and phytoplankton population structures(PPSs)could provide valuable information when studying marine biogeochemical processes and ecosystems.Compared with traditional methods using field observations,satellite remote sensing data can provide synoptic observations,which are ideal for investigating PSCs and PPSs at large spatial and temporal scales.However,the estimation of PSCs and PPSs in the East China Sea(ECS)using satellite data is a challenge.Consequently,the dynamics of phytoplankton size classes and population structure in the ECS at different spatial and temporal scales and their mechanisms are still poorly understood.Thus,it is of great scientific significance to investigate the PSCs and PPSs in the ECS by remote sensing.Based on multi-year MODIS satellite data and navigation observations,our study proposed a model for the PSCs estimation in the ECS,and investigated previously unkown spatiotemporal dynamics of the PSCs and discussed their mechanisms in the ECS using long-term time series of MODIS satellite data.Meanwhile,according to wavelength characteristic of MODIS sensor,a model for estimating PPSs using phytoplankton absorption(a_ph)was designed in this study.Additionally,we assessed the performance of the model and discussed satellite application of the PPS model.The main conclusions of this study were as follows:(1)The quantitative relationship between measured a_ph shape and PSCs was found by analyzing the local measured dataset from the ECS.Base on this relathinship,we proposed the PSC model with good applicability to estimate PSCs from a_ph data.The retrieval accuracy of the PSC model was validated using an independent dataset.The estimated PSCs values showed reasonably good agreement with the measured data,and most of the samples were within the error range of±20%.Meanwhile,based on the measured dataset,the PSC model was applied to remote sensing reflectance(R_rs)by coupling QAA algorithm,to examined the feasibility of the PSC model for satellite observations.Quasi-analytical algorithm(QAA)was used to retrieve a_ph from the measured R_rs.The obtained results clearly demonstrated that QAA can produce accurate estimates of a_ph at 412,443,469,488,531,and 547 nm.Then,the PSCs values were inferred from the retrieved a_ph using the PSC model,and agreed with the measured results.These results indicate that the proposed PSC model has great potential for satellite application.(2)On the basis of improving the observation accuracy of MODIS R_rs data,we optimized the PSC model for estimating the PSCs in the ECS using satellite data.The accuracy of MODIS R_rs data was assessed using the synchronous field measurements from satellite matchup dataset.The low accuracy was observed at 412 and 443 nm,which may introduce uncertainty in the retrieved a_ph data,and further increase the uncertainty of satellite-derived PSCs.Thus,the accuracy of the satellite R_rs data at 412 and 443 nm were improved through modeling them from R_rs between 469 and 555 nm using multiple regression analysis.Then,the QAA algorithm and PSC model were applied to the reconstructed satellite R_rs data to obtain the satellite-deirved a_phand PSCs results.The use of reconstructed satellite R_rs data could improve the accuracy and percentage of valid points for the satellite-derived a_ph,and thereyby significantly improve the performance of the PSC model on satellite observations and yielded reasonable satellite-derived PSCs data,which were better than those derived from original satellite observations.These results show that the proposed PSC model in this study could produce the reasonable satellite-derived PSCs results in the ECS using MODIS data,which can be used to analyze the spatiotemporal variability of PSCs in the ECS.(3)The proposed PSC model was applied to 15 years(2003-2017)of MODIS monthly R_rsdata to obtain satellite products of PSCs,to analyze the variability of the PSCs in the ECS.Generally,the PSCs distribution in the ECS was heterogeneous in both temporal and spatial scales.Micro-phytoplankton were dominant in coastal waters throughout the year,especially in the Changjiang estuary.For the middle and outside shelf region,the dominance of nano-phytoplankton were observed.Pico-phytoplankton were the dominant size class in the areas deeper than 200 m.Additionally,the different subareas with different geographical locations and hydrological conditions showed the distinct PSCs dynamics.The study showed the PSCs variations in the ECS were probably affected by a combination of the water column stability,upwelling,sea surface temperature,wind and Kuroshio current.Additionally,human activity and riverine discharge might also influence the PSCs distribution in the ECS,especially in the coastal region.(4)Different phytoplankton populations showed distinct a_ph spectra by analyzing the local measured dataset.Bosed on this characteristic,this study designed a model for PPSs estimation with a_phaccording to MODIS band setting,and it had good performance of the model in Case I waters,i.e.,the Arctic ocean.The result of CHEMTAX analysis based on pigment concentration data indicated that the Arctic Ocean were dominated by six phytoplankton groups,namely,diatom,dinoflagellate,c3-flagellate,hapto-7,prasino-2,and prasino-3.For the dominant groups in the Arctic Ocean,the PPS model was used to obtain the specific absorption spectra of the phytoplankton populations with high reliability and the biomass concentration of the phytoplankton populations with high accuracy,and MAPE values for all groups were both less than 35%,except for diatoms with MAPE value of about 45%.(5)The proposed PPS model also had good applicability in the East China Sea as Class II waters,and has great feasibility and potential for satellite application.The result of CHEMTAX analysis using pigment concentration data showed that the ECS were dominated by six phytoplankton groups,namely,diatom,dinoflagellate,prymnesiophytes,cryptophytes,chlorophytes and cyanobacteria.The proposed PPS model in our study was applied in the ECS,and the performance of the model was assessed.Our study found that for the ECS with complex optical properties,the PPS model can also generate the reasonable specific absorption spectra and biomass concentration results of these dominant phytoplankton populations,suggesting that our PPS model had good universality and applicability.In addition,the use of a_phdata at the first six MODIS wavebands can yield reasonable the PPSs results with an acceptable level of accuracy using our PPS model,wich indicated that our PPS model has great feasibility and potential for satellite application.