The method, effectiveness and correctness of replacement process on Weather Research and Forecasting Model (WRF) underlying surface data were described in quite detail. The comparison analysis on mean bias error, root mean square error and correlation coefficientof two WRF simulation groups between pre-process and post-process was executed. The following conclusions were summarized: (1) Terrain in Heihe River Basin upstream originated from WRF model is rough, and would cause certain error in WRF simulation; (2) Replacement of terrain with high spatial resolution can improve WRF simulation, especially for complex region; (3) In addition to elevation, 2 m temperature as well as 2 m relative humidity is strongly effected by water body, so it is vital to provide the precise information of water body for regional climate model; (4) There is a positive relationship between rainfall and elevation.