Calculation of soil moisture affects the energy and water balances of landing surface models and is therefore key for core calculation of such models. The Noah land surface model (NOAH-LSM) is used by the Global-Regional Assimilation and Prediction System (GRAPES) to simulate soil water content. However, because this model cannot effectively express changes in runoff yield area, it does not provide a complete description of the hydrological cycle. This study offers improvements for the hydrological profile of GRAPES that include adding repletion of storage, which considers the change in runoff yield within the grid area for more accurate simulation of runoff in each grid, and combining flow concentration values for considering the redistribution of soil moisture in a two-dimensional level to improve the simulation accuracy of the surface runoff. Simulations were conducted to study feedback of the land surface water cycle in the atmosphere from August to September 2009. Results indicate that the meteorological elements simulated by the improved model in the surface layer influenced by land surface water cycle processes such as soil moisture and temperature are closer to the observed values. Therefore, these improvements have a definite impact on the regional precipitation.