Abstract: Water level monitoring is one of the key tasks in flood control and drought resistance. Global Navigation Satellite System-Reflectometry (GNSS-R ) remote sensing technology has been proven to be capable of high-precision, all-weather, and all-day monitoring of water level changes. China's BeiDou Navigation Satellite System (BDS) has rich frequency band data and has broad application prospects in the field of GNSS-R remote sensing. Based on the GNSS-R experimental station built in the Liuyang River Basin, this paper further studies the performance of BDS in water level inversion. The spectrum analysis and nonlinear fitting methods are used to measure the water level changes of the station, and the experimental results are compared and analyzed with the measured data of the nearby hydrological station. The results show that the nonlinear fitting method has a significantly improved accuracy compared with the spectrum analysis method, with an average RMSE reduction of 66%. After the fusion of multi-frequency GNSS data, the RMSE reaches 1.8 cm. Overall, the water level inversion performance of BDS is comparable to that of GPS.