Human activities and discharged river pollution affect the marine ecological environment. Therefore, it is very important to study how human activities affect the offshore ecological environment by monitoring the fluxes of terrestrial materials from rivers over long-times sequence. Landsat satellite data, which began in the 1970s, are of great value for monitoring changes in water quality in many small and medium-size rivers in China, which lacked field observation data in the early years. Because river water is usually turbid and the composition of water is complex, there is no uniform precise atmospheric correction. In addition, components of inland riverine water are different, so we need to build regional high-precision total suspended matter (TSM) inversion model. Based on Landsat-8/OLI (Operational Land Imager) data, we evaluate the applicability of different atmospheric correction methods (Fast Line-of-sight Atmospheric Analysis of Hypercubes-FLAASH, Dark Subtraction-DS and Quick Atmospheric Correction-QUAC) to Oujiang River. The results show that FLAASH gives good atmospheric correction, with the mean relative error of each band being around 15%. We collected three occurrences of measured data synchronized with the Landsat-8 satellite. Based on the measured TSM and remote sensing reflectance spectrum data, we analyze the correlation between different band combinations and TSM. The results show that the equivalent remote sensing reflectance ratio of OLI (482nm)/OLI (865nm) is highly correlated. Finally, we apply the established regional TSM inversion model to the OLI data in Oujiang River basin and comprehensively analyze the long-term changes of TSM concentration in the Oujiang River.