Based on high resolution terrain data, terrain experiments using different terrain disposal methods, including WRF synthetic terrain disposal method(TESTC), Chebyshev polynomial filtering method(a representative experiment TEST5) and five points weighed average method(TEST7) have been done for an heavy rainstorm occurred in Sichuan during 20-22 July 2008. The results show as follows: (1) Using terrain data at different resolutions and adopting different terrain filtering methods have different effects on the prediction results of numeric weather prediction model. Comparatively, based on high resolution terrain data, Chebyshev polynomial filtering method of using 90 truncation rank shows a slight predominance over other methods. (2) Three kinds of terrain disposal methods can predict main precipitation area and intensity. Although precipitation intensity simulated by TEST5 is stronger than the real, simulated precipitation evolution and precipitation area are closer to the real than that of TESTC and TEST7. As to overall simulation effect, TEST7 is between TESTC and TEST5. (3) Different terrain deposal schemes bring about changes to the model terrain altitude and slope of basin and surrounding area, and such changes impact on precipitation areas, intensities and evolutions by working on physical elements going with meso-scale systems activities. (4) Due to differences of terrain disposal, in the east basin where terrain gradient is bigger, topographic perturbation boosts up wind field fluctuation and TEST5 brings about bigger east-west wind-shear. The wind convergence strengthens low vortex and ascend motion, inducing more low-layer vapor being transported to upper air. The intensity of low vortex simulated by TEST5 is stronger and the low vortex position is more eastward, leading to stronger precipitation intensity and more eastward precipitation position than TESTC and TEST7.