Abstract: In order to improve the assimilation effect of automatic weather station (AWS) observation data in the WRF-DA (Weather Research and Forecasting Model-Data Assimilation) system and improve the prediction of the numerical model, the ground observation data assimilation scheme in the WRF-DA is improved and an updated assimilation scheme is developed. Considering the impact of terrain height difference on the assimilation effect, the correctness of ground elements to the model surface are modified in the assimilation scheme to improve the accuracy of the values of ground temperature and wind speed. Analysis of the results of a real-data case experiment on 5 July 2017 and the results of batch experiments for the entire month of July 2017 indicates that compared to the results of the experiments with assimilation using the original scheme, the forecast errors of the experiments using the updated scheme are significantly reduced in the first 9 h. The advantage of the updated scheme can be maintained for 24 h. It shows that the updated scheme has obvious positive effects on the assimilation of ground temperature and wind speed.