Differentiating the Effects of Streamflow and Topographic Changes on the Water Level of Dongting Lake, China, Using the LSTM Network and Scenario Analysis

Abstract

Dongting Lake is the second largest freshwater lake in China and an internationally critical habitat for migratory birds. However, after 2004, the multiyear mean water levels of West Dongting Lake (WDL), South Dongting Lake (SDL), and East Dongting Lake (EDL) in the high-water stage decreased by 1.05 m, 1.15 m, and 1.32 m, respectively. Different areas of Dongting Lake experienced various degrees of shrinkage. It is necessary to study the dominant driving factors and their contributions to the falling water level. In this study, the water level changes in Dongting Lake were analyzed, and a long short-term memory neural network model was constructed to simulate the water level of Dongting Lake. Moreover, the contribution of changes in streamflow and topographic conditions to the water level changes in different areas of Dongting Lake was estimated with scenario analysis. The research results show that the changes in the streamflow were the main driving factors for the water level decline of WDL, SDL, and EDL in the high-water stage, and their contributions were 0.74 m, 0.97 m, and 1.16 m, respectively. The topographic changes had a great falling effect on the water level of Dongting Lake, and the falling effect on the water levels from October to June of the following year was the strongest in EDL (0.81 m), followed by WDL (0.49 m), and the weakest in SDL (0.3 m). These results can provide a scientific reference for the management of the water resources of Dongting Lake.