Abstract: The hydrologic regime of the Poyang Lake has changed due to the impact of climate change and human activities. The influence of different driving factors on the hydrologic regime evolution of the Poyang Lake needs to be quantitatively evaluated. With a series of hydrological indicators characterizing the hydrological situation of Poyang Lake, a BP neural network model was constructed to simulate the water level of the Poyang Lake. The rates of contribution of flow rate of mainstream Yangtze River, inflow from sub-basins to the Poyang Lake, and topographic changes to the variation of hydrological indicators are studied through scenario analysis. Results manifest that the flow rate of mainstream Yangtze River is the major driving factor for the decline of the average water level of Poyang Lake from July to October, with a contribution rate of 52%-67%; the flow rate of mainstream Yangtze River cut obviously the annual maximum extreme water level while boosted the annual minimum extreme water level. Topography is the main driving factor for the decrease of average water level of the Poyang Lake from December to next March, with a contribution rate of 92% to 185%. Topography also brought down the minimum extreme water level apparently. In addition, some suggestions are put forward for the management and regulation of water resources in the Poyang Lake. The operation of the Three Gorges Reservoir should be optimized, and the water storage of the Three Gorges Reservoir should be appropriately advanced in time according to actual situation. The construction of water conservancy projects in the Poyang Lake sub-basins should be strengthened, and the discharge from five rivers during the impoundment of the Three Gorges Reservoir can be enhanced. Sand mining must also be regulated.

Key words: hydrologic regime evolution, BP neural network model, hydrologic factor, topographic factor, contribution rate, Poyang Lake