Extraction of critical hydrologic features from high-resolution topographic data is challenging using existing grid coarsening approaches, such as surface flow path, river network, and slope, which limits the application of hydrological models. In this research, the influence of various grid coarsening techniques on the prediction outcomes was measured by a numerical experiment based on the integrated hydrological model ParFlow-Common Land Model (ParFlow.CLM). Three grid coarsening methods (Nearest Neighbor Coarsening, Majority Coarsening, Hydrography-Driven Coarsening) were applied to simulate evapotranspiration(E), soil temperature(ST), streamflow, soil moisture(SM) and latent(LE) heat fluxes in central China’s Sixi Valley, a classic example of a karstic basin. As a result, the three grid coarsening methods perform uniform in simulating latent heat fluxes and soil temperature. However, their ability to predict soil moisture surface flow and evapotranspiration are more diverging. The hydrography-driven coarsening extracts significantly more accurate valleys, rivers network, and slopes closer to the actual terrain than existing coarsening strategies. Slopes derived from hydrography-driven coarsening methods can be used to predict more accurately the top soil moisture, evapotranspiration, and streamflow dynamics processes. This study stresses that a hydrography-driven coarsening strategy is advocated for all those cases in which topographic slope extracted using a coarse-grid digital elevation model is an important influence on the ParFlow.CLM simulation of essential hydrographic features.