Based on the experiment data of water and soil loss in different land farming and different slope in the Jingyan county of Sichuan province, it is used for neural network method, it is build neural network model of sediment yield on slope for different land farming, and obtained the mechanism of water and soil loss of slopes.

Based on measured data, it is analysed that runoff and sediment from hill slope in granite area is influenced by rainfall and maximum ten minutes rainfall density and forecast model of runoff rate and sediment yield from hill slope is established.

Critical topographic condition for slope erosion in Hilly-gully region of Loess Plateau;