Abstract: The land surface provides underlying lower boundary conditions for atmospheric models. Topography plays a crucial role in the accuracy of the model results. Most of current land surface models use the same atmospheric forcing in subgrid units within the same model grid, and don’t not consider the influence of subgrid topography on atmospheric forcing that would impact the simulation of meteorological elements and land-atmosphere interaction. In this paper, a modified Land Surface Subgrid Topographic Parameterization (LSSTP) is proposed to revise the input atmospheric forcing according to its relationship with the subgrid terrain height in the NOAH land surface model . The LSSTP is then introduced into the WRF (Weather Research and Forecasting) model for numerical experiments. Three groups of numerical experiments have been conducted to investigate the effect of this improvement on the simulation results of the WRF model. It is found that the new LSSTP introduced in this paper has clearly improved the simulation of 2m air temperature in the surface over the Tianshan Mountains, the Kunlun Mountains and the southern Qinghai-Tibet Plateau. However, the new LSSTP shows little improvement on precipitation simulation. In contrast, the experiments with higher resolution can well simulate precipitation, which is attributed to the fact that the experiment with higher resolution has been refined in both the land and the atmosphere, while the new LSSTP only considers subgrid effect in the surface. Temperature simulations using the new LSSTP are improved by revising surface upward long-wave flux and surface sensible heat flux, while the experiment with higher resolution has simultaneously refined the grids of both the atmosphere and land, and the improvement of precipitation and temperature simulation is achieved by comprehensively changing the surface energy balance.