Remote Sensing Quantitative Retrieval Of Phytoplankton Size Classes In Eastern China Seas

Phytoplankton size structure is an important physiological parameter in many marine ecological and biogeochemical processes,and the knowledge of its composition and distribution is greatly helpful in understanding the functional role of phytoplankton in marine ecosystem.The study of phytoplankton size classes has been relied on the traditional in situ measurements from cruise campaigns,which are limited by the given sampling areas and environmental conditions.With the high spatial-temporal resolution and high revisit period,ocean colour remote sensing has increasingly become an important tool in monitoring the size structure,making remote sensing retrieval of phytoplankton size classes become a research hotspot in the marine ecology and earth observation fields.In recent years,related researches are well developed in global oceanic waters.However,limited researches are carried out in coastal and estuarine waters.Under the influence of high turbidity,various compositions of water components,and complex bio-optical properties,the applicability and limitations of the existing remote sensing quantitative inversion models in eastern China seas are required to be investigated,validated,improved,and innovated.Chlorophyll-a concentration and phytoplankton absorption spectra are two common ocean colour parameters that can be used to retrieve phytoplankton size structure indirectly and directly.Chlorophyll-a has been well developed in recent decades,and phytoplankton absorption spectra has shown increasingly potential in retrieving phytoplankton composition in recent years.Therefore,this thesis is focus on developing two types of approaches,one is abundance-based,and the other is absorption-based,for remotely sensed estimating and monitoring phytoplankton size classes in eastern China seas.The main purposes and results are as follows.(1)This study has investigated the spatial distributions and variations of in situ phytoplankton size classes in eastern China seas,which provide a comprehensive data base for the development of the abundance-based approach.A large dataset of phytoplankton pigments,bio-optical properties,and environmental parameters are collected through 7 cruise campaigns from 2015 to 2019 in the eastern China seas.The weights of diagnostic pigments are optimized firstly for improving the accuracy of in situ measurements of the size structure,with the correlation between the weighted estimations and measured chlorophyll-a concentration at 0.93.Differences of spatial distribution of phytoplankton size classes are observed:with the increasing distance from the shore,there are trends of decreasing contributions of larger cell size,and increasing contributions of smaller cell size,where temperature and salinity are both influential factors.(2)This study has explored the inherent optical properties and apparent optical properties of waters dominated with different phytoplankton size classes,which provides a theoretical foundation for the development of absorption-based approach.Due to the size structure,the shape and magnitude of the absorption spectra varies,with higher values in absorption spectra,flatter shapes in normalized absorption spectra,and lower values in chlorophyll-specific absorption spectra,for larger size classes.Among different waters,the contributions of phytoplankton,non-algae particles,and chromophoric dissolved organic matter to the total absorption at 443 nm are 10.34–23.04,8.42–21.27,and 54.39–75.45%,respectively.Phytoplankton and non-algae particles contribute more in waters dominated by larger size classes.Differences in normalized total absorption are significantly higher than those in normalized phytoplankton absorption,with relative percentage differences of 42.03 and 22.35%,and the difference in the normalized remote sensing reflectance is 41.00%.(3)This study has improved and developed a highly accurate abundance-based approach for turbid eastern China seas,and revealed the multi-temporal spatial variations of phytoplankton size classes,as well as their responses to the environmental and climatic changes.Validation with the independent satellite match-ups shows that the accuracy of the improved model is significantly higher,with correlations are 0.742,0.663 and 0.328,and root mean square errors are 0.507,0.436 and 0.162 mg/m~3,for micro-,nano-,and picoplankton,respectively.A significant improvement in model performance is achieved by introducing the dependence of model parameters on sea surface temperature,especially for nano-and picoplankton,with decreasing root mean square error of 0.416 and 0.145 mg/m~3.Remotely sensed diurnal variations of phytoplankton size classes are investigated:in central south Yellow Sea,diurnal variation of size-fractionated percentages are 24.26–25.85,62.34–63.59,and 11.81–12.14%throughout the day,which might due to the diurnal variation of light and temperature;while in the area outside the Changjiang Estuary,the ranges of variation are 63.24–76.22,20.79–31.88,and 2.99–4.88%,respectively,where the turbidity caused by tides might also be the influential factor.Based on the 22-year satellite data,the spatial-temporal distributions of phytoplankton size classes show monthly variations:in central south Yellow Sea,due to the formation of Yellow Sea Cold Water Mass,monthly percentages of microplankton are positively correlated with sea surface temperature,sea level anomaly,and zonal wind speed,while negatively correlated with mixed layer depth,meridional wind speed and total wind speed,and opposite correlations are found between these variables and nano-and picoplankton.The interannual variations of phytoplankton size classes in eastern China seas are affected by El Ni(?)o events,and their responses are different to two types of El Ni(?)o.When Eastern Pacific El Ni(?)o occurs,microplankton percentages tend to show negative anomalies,while nano-and picoplankton have positive anomalies,where correlations between Eastern Pacific El Ni(?)o index and anomalies of size-fractionated percentages are-0.357,0.317,and 0.252,respectively,while the Central Pacific El Ni(?)o events have the opposite effects on the anomalies of three size classes,with correlations of 0.336,-0.394,and-0.089,respectively.(4)This study has developed an absorption-based approach,where two constraints,optimal bands selection and initial inputs of the size composition related to different absorption properties,are introduced in the model firstly to improve the accuracy.Validated with the independent in situ dataset,the absorption-based model is applicable in eastern China seas,however,the accuracy is slightly lower than those retrieved from the abundance-based model,with root mean square errors are 0.908,0.496,and 0.123mg/m~3,and 0.280,0.239,and 0.060 mg/m~3 for micro-,nano-,and picoplankton,respectively.Remotely sensed estimations of phytoplankton size classes show that at the 36°N section in south Yellow Sea,the ranges of differences of chlorophyll-a concentration retrievals between the absorption-based and abundance-based models for three size classes are 0.116–2.645,0.161–3.576,and 0.011–0.330 mg/m~3,respectively,and differences are obviously higher in the nearshore areas than those in offshore areas.Overall,the achievements in this study contribute to the development of remote sensing monitoring of the functional diversity of phytoplankton effectively,and are of great scientific importance in improving the understanding of phytoplankton size classes in marine ecological and biogeochemical processes.