Abstract:Based on the Richard equation of water movement in unsaturated soil and the characteristics of soil water movement with drip irrigation from single point source, the HYDRUS-2D/3D model was used to numerically simulate the soil wetting pattern under different dripper discharges(1, 2, 3 L/h)in 33 kinds of soil textures(seperately belonging to 11 kinds of soil types in USST). Based on the si-mulated results of the soil wetting volume development under different dripper discharges and soil textures, and using irrigation amount and the ratio of soil saturated water conductivity and dripper discharge as the input variables, the artificial neural network was built for predicting the soil wetting volume development under different soil textures and dripper discharges. This artificial neural network, with less input variables and more maneuverable, simulated the infiltration processes which were consistent with the actual infiltration processes, and the correlation coefficient(R2)is over 0.82. Therefore, the developed model can achieve good effect in predicting the migration of soil water for different kinds of soil textures compared with the experimental data.