Remote Sensing Estimation Of Particulate Organic Matter (POM) Concentration In Inland Lakes

As an important resource related to human survival and social development,the water quality and biogeochemical cycle of inland lakes have been concerned for many years.Particulate organic matter（POM）is an important component of organic matter in inland lakes,and the temporal and spatial variations of its composition,source and concentration can reflect the nutritional status of the lake to a certain extent.The study on the temporal and spatial distribution of POM components and their concentrations in inland lakes is helpful to understand the material cycling process in the whole lake ecosystem and the protection of water environment effectively.Based on the field measured data of Taihu Lake,Chaohu Lake and Hongze Lake,the components and sources of POM in lakes were analyzed by using three-dimensional fluorescence technology combined with parallel factor analysis method.According to the results of analysis,the optical properties of two categories of water bodies with different POM sources（endogenous and exogenous）and the differences of components in water bodies were further explored.A remote sensing inversion algorithm for POM concentration of inland lakes based on the classification of water bodies dominated by different POM sources,and the algorithm is suitable for Sentinel-3 OLCI images.Finally,the algorithm was applied to OLCI images of Taihu Lake,Chaohu Lake and Hongze Lake from 2016 to 2020 to explore the spatial-temporal distribution and variation of POM in the three lakes.The research contents and main conclusions of this paper mainly include the following aspects:（1）Based on the three-dimensional fluorescence data of POM from Taihu Lake,Chaohu Lake and Hongze Lake,it was decomposed into four effective fluorescence components by parallel factor analysis,including one protein component C1 and three humus components C2～C4.The spatial distribution of relative content of each component and its correlation with water quality parameters were analyzed.The results showed that endogenous input was the main factor affecting the fluorescence intensity of POM in lakes.However,in some water areas which were significantly affected by human activities,the main source of POM was exogenous input.（2）According to the value of fluorescence index HIX,the water bodies of inland lakes can be divided into two categories,which indicate that the sources of POM are mainly exogenous and endogenous respectively.On the basis of this classification,a variety of corresponding POM concentration inversion models were constructed for the two types of water.Among the endogenous dominant water types,the highest accuracy POM inversion model showed a MAPE value of 24.49%and a RMSE value of 0.8693mg·L^-1;among the exogenous dominant water types,the MAPE and RMSE of POM inversion model with the highest accuracy are 19.85%and 1.9355 mg·L^-1,respectively.The accuracy verification results of independent validation data set showed that the POM inversion algorithm based on water classification showed a higher inversion accuracy than other existing algorithms.（3）The two-step inversion algorithm proposed in this paper is applied to OLCI images of Taihu Lake,Chaohu Lake and Hongze Lake from 2016 to 2020,and the spatiotemporal variation results of monthly and annual mean POM concentrations in three lakes are obtained.Specifically,the concentration of POM is relatively high in the coastal waters in the northwest of Taihu Lake,the western waters of Chaohu Lake and the northwest of Hongze Lake.In terms of time,the monthly average concentration of POM in the three lakes increased first and then decreased in the whole year.The annual average concentration of POM in Taihu Lake decreased gradually in five years,while the annual average concentration of POM in Chaohu Lake and Hongze Lake increased.