Topography exerts major control on land surface processes. To improve representation of topographic impacts on land surface processes, a new topography‐based subgrid structure has been introduced to the Energy Exascale Earth System Model to represent the subgrid heterogeneity of surface elevation. To take advantage of the new subgrid structure for improving land surface modeling, this study explores four topography‐based methods for downscaling grid precipitation to the subgrids. In the first three methods, the deviation of the subgrid precipitation from the grid mean is equal to the grid precipitation multiplied by the ratio of the elevation difference between the subgrid and grid mean to a specified elevation, which is equal to grid elevation, difference between the maximum and minimum subgrid elevation, and maximum subgrid elevation, respectively. The second method limits the ratio to 0.5 to avoid extreme values in mountains, and the third method utilizes the subgrid hypsometric elevation. The fourth method is similar to the third method except that a height rise determined based on an ambient flow regime with the Froude number equals to 1 is used to limit the numerator. The downscaled precipitation is evaluated using the Precipitation‐elevation Regressions on Independent Slopes Model precipitation data over the United States and statistical metrics. Results show that by utilizing the subgrid hypsometric elevation, the third and fourth methods show clear advantages over the first and second methods. Furthermore, introducing the height rise in the fourth method improves downscaling skill slightly compared to the third method consistently across multiple grid sizes in areas with larger subgrid heterogeneity.