Remote Sensing Of Suspended Particulate Matter In Typical Estuarine And Inland Waters Based On Optical Classification

Suspended particulate matter(SPM)is the main carrier for carbon,nutrients,and pollutants in water and thus is crucial in the biogeochemical cycle.Among various algorithms for retrieval of SPM concentration,satellite remote sensing techniques have been widely used to monitor SPM concentrations in vary kinds of water because of the advantages of large spatial coverage,high spatial resolution,high revisit frequency,and low cost compared to the traditional field sampling methods.In this study,field sampling and indoor experiments were carried out with the Daihai Lake and the Yellow River Estuary as the study area.Based on the obtained measured spectral data,the optical classification index was used to classify the waters in the Daihai Lake and the Yellow River Estuary,the spectral reflectance of each optical water type was analyzed,and the empirical algorithm and random forest algorithm for remote sensing retrieval of SPM concentration were established based on the classification results,and the advantages and disadvantages of the two models were discussed.Sentinel-2 MSI data were used to obtain the SPM concentration spatiotemporal distribution maps of Daihai and Yellow River Estuary.The main conclusions of the study are as follows:(1)Based on the field measurement data,the Forel-Ule(FUI)water types in the study area are concentrated in FUI 7 and FUI 8.Among them,the Yellow River Estuary is distributed with eleven optical water types from FUI 7 to FUI 17,with FUI 7 being the most and FUI 12 being the least;the Daihai Lake is distributed with six optical water types from FUI6 to FUI 11,with FUI 7 being the most and FUI 11 being the least.Maximum Chlorophyll Index(MCI)water types are concentrated in MCI1,among which nearly 80% of the waters in the Daihai Lake are MCI 1,and more than 60% of the Yellow River Estuary are MCI 1.In addition,according to the FUI water classification method,the waters in the study area are basically mesotrophic,among which,most of the waters in the Daihai Lake are mesotrophic,and only a small portion of the waters are oligotrophic and eutrophic;60% of the water bodies in the Yellow River estuary are mesotrophic,the rest are eutrophic,and there is no oligotrophic water.(2)Four retrieval algorithms for SPM concentration were established,namely: empirical algorithm for SPM concentration without optical classification(r = 0.82,RMSE = 147.57 mg/L,MAPE = 65.81%),empirical algorithm for SPM concentration based on FUI index(r = 0.97,RMSE = 50.82 mg/L,MAPE = 34.27%),random forest algorithm for SPM concentration without optical classification(r = 0.96,RMSE =70.50 mg/L,MAPE = 32.52%),and MCI index-based random forest algorithm for SPM concentration(r = 0.96,RMSE = 70.13 mg/L,MAPE= 28.19%).The MCI index-based random forest algorithm for SPM concentration performed best,and then used this algorithm for further remote sensing retrieval.(3)Based on the remote sensing reflectance of the Sentinel-2 MSI Level-2A surface reflectance products,a random forest algorithm based on MCI index for SPM concentration(r = 0.58,RMSE = 9.95 mg/L,MAPE = 49.91%)was established to obtain the spatial and temporal distribution of SPM concentration in the Yellow River estuary and Daihai.High concentrations of SPM in the Yellow River Estuary were mainly distributed near the estuary and coastal areas,and low concentrations were in offshore waters.The trend of high concentration of SPM was shown at the edge of Daihai Lake,and the concentration of SPM was generally lower in the center of the lake.