Preliminary Research on the BP Networks Foreasting Model of Watershed Runoff and Sediment Yielding

In this paper,considering the radical principle of neural networks and acting rainfall condition as the main affecting factors,a back propagation (BP) networks model of watershed runoff and sediment yielding is discussed.The model has satisfactory learning and generalization performance,and it may be used to value the human-action influence on runoff and sediment yielding in a watershed.The results identified by Xihanshui, Datongjiang and Xiangcihe basins's observed data indicate that the model are satisfactory.