To meet the demand of constructing the rainfall-runoff simulation system coupling the land-surface hydrologic model and the numerical atmospheric model, three types of high-resolution precipitation data based on weather radar and rainfall station observations are acquired, namely radar estimated precipitation, rainfall station interpolated precipitation and radar and rainfall station merging precipitation. The key runoff parameters of the land-surface hydrological model WRF-Hydro are calibrated using both rainfall station interpolated precipitation and radar and rainfall station merging precipitation. The applicability of WRF-Hydro driven by different precipitation for runoff simulation in small and medium scale catchments in the mountainous regions of northern China is explored. The results show that the error indices of merging precipitation are improved by different degrees compared with that of traditional interpolated precipitation, and the spatial and temporal accuracy of merging precipitation is also improved. Both rainfall station interpolated precipitation and merging precipitation can effectively improve the key runoff generation and concentration parameters of WRF-Hydro, and the results of flood simulation with the merging precipitation rate parameter set are better than that of rainfall station interpolated precipitation. Driven by high-resolution precipitation data, more accurate WRF-Hydro key runoff generation and concentration parameters can be acquired, which effectively improves the applicability of the coupled WRF/WRF-Hydro system in the studied basin.