Spatial Variation Characteristics and Remote Sensing Retrieval of Total Suspended Matter in Surface Water of the Yangtze River

Abstract

Total suspended matter(TSM) is an important parameter that describes optical characteristics and water pollution level. Remote sensing was used to obtain suspended matter concentrations in the planar water body, which can monitor and analyze the reliability of turbidity and pollution levels. In this study, a semi-empirical model of suspended matter along the Yangtze River mainstream(from Chongqing to Shanghai) was constructed based on the field measured hyperspectral data and suspended matter concentration and its components, with reference to MODIS(1-4 band), Landsat 8(1-5 band), Sentinel 2(2-6 band), HJ-B1(1-4 band), and other commonly used band ranges. The results show that:① Turbidity and TSM of the Yangtze River mainstream increased from Chongqing to Shanghai during the dry season, whereas the proportion of OSM to TSM(OSM/TSM) decreased from Chongqing to Shanghai. ② The spectral properties of chlorophyll-a were not obvious; however, it showed a bimodal spectral characteristic of suspended matter concentration. Due to the low concentration of suspended matter in the water(<114.68 mg·L-1), the first peak value was significantly higher than the second peak. However, peak reflection and position had an obvious "red shift" phenomenon in the direction of the wavelength. ③ The reflectance values of the Yangtze River mainstream were most sensitive to the concentration of suspended matter in the 600-710 nm and 475-550 nm bands. Most of these fall in the green and red band of MODI, Landsat 8, Sentinel 2, and HJ-1B satellites. Therefore, the commonly used satellites have a band setting to detect the inversion of turbidity degree index of the Yangtze River main stream water body. The Sentinel 2 satellite band is most suitable for the inversion of suspended matter concentration in the Yangtze River mainstream. The mean absolute percentage error(MAPE) and root mean square error(RMSE) of the optimal exponential model were 10.23 mg·L-1 and 23%, respectively.