The Poyang Lake is an important wetland ecosystem in China,which plays an important role in regulating the change of water and sediment.Because of large area and large internal differences of the Poyang Lake, a single suspended sediment inversion model is not enough to inverse the suspended sediment concentration.Therefore, this paper used a inversion model based on the classification with the measured reflectance spectra data and the sediment concentration data. It is said that according to the classification of spectral morphological characteristics of the measured data from 5 typical types of water bodies. On this basis, the classification of the various types of water were established respectively for the inversion of the appropriate inversion model. The results show that the empirical model retrieval based on water classification achieves satisfactory results, the average absolute error is 0.00217g/L, and the average relative error is 3.022%. Therefore, the empirical inversion model based on classification is applied to monitor the distribution of suspended sediment concentration in the Poyang Lake. This study will help to grasp the spatial distribution and change of the Poyang Lake sediment concentration, and provide decision basis for the sustainable development and utilization of the Poyang Lake resources.