Suspended sediments, one of the most important factors affecting the water environments of inland lakes, are closely related to the migration and interaction of various pollutants. Existing studies show that the suspended sediment concentration can be accurately predicted based on assimilation methods coupled with hydrodynamic models. However, in the hydrological assimilation simulation process, the existing perturbation generation methods consider the perturbation error to follow a random Gaussian distribution, which does not consider the spatial variation characteristics of errors. Thus, in this paper, we proposed a new method that generates a hybrid perturbation field for the assimilation simulation instead of using random error. The proposed approach was validated through assimilation simulations of the suspended sediment concentration of Taihu Lake, China, and five assimilation experiments were conducted. The proposed method was compared with the existing methods for generating the perturbation field. After three days and 72 time steps of assimilation simulation based on the hybrid perturbation field, the proposed assimilation method provided results that were more consistent with the buoy-measured data. These findings demonstrate that the proposed method for generating a hybrid perturbation field has a higher simulation accuracy than other methods and is therefore effective and provides a new idea for the assimilation simulation of suspended sediment concentrations in inland lakes.