Because the hydrological response of watersheds is comprehensively affected by physical geographical factors, such as geology, topography, landforms, vagetation, and soil, and underlying surface factors, such as land use, hydrological partition was conducted based on the geographical characteristics of underlying surface of watersheds. The underlying surface of watersheds was classified with remote sensing data and information on land cover and land use was obtained. By combining data of watershed agrotype and the topography index, the soil topography index was calculated. Furthermore, taking advantage of the remote sensing data and soil topography index, hydrological partition was conducted, the parameter C_N of the SCS model was calibrated in every hydrological subregion, and hydrological simulation was conducted. The results showed that the relative error was less with the proposed method than with the traditional method in 21 flood runoff simulations in the Tunxi Watershed and 32 flood runoff simulations in the Dongwan Watershed.