Abstract: Orography influences atmospheric circulation on a variety of spatial and temporal scales. The representation of its impact in numerical weather prediction models remains a challenging issue since the orographic spectrum can only be partially resolved in models. As numerical atmospheric models develop towards running in sub-kilometer resolutions, the need for accurate depiction of orography details becomes increasingly important. In this study, a new method to process orography is implemented in the CMA-MESO model by incorporating a new high-resolution orographic database ASTER-1s and an improved orography filter. The new method can remove harmful noises and retain more detailed small-scall orography features in the model, which greatly improves the representation of orographic effects. This new orography processing method is evaluated in the CMA-MESO based on simulations in June and December 2020. Comparison with observations collected at more than 20000 sites indicates that using ASTER-1s data without changing the filter does not significantly improve the prediction of 2 m temperature and 10 m wind speed. Using ASTER-1s data together with a new filter can greatly improve the prediction, resulting in a reduction of mean root mean square errors by 6.4% and 4.9% for the monthly mean 2 m temperature and 10 m wind speed, respectively. The prediction of monthly mean precipitation is also improved but not as significantly as that for the temperature and wind speed. The energy spectrum analysis shows that the new orography processing method does not show unrealistic energy accumulation at high frequencies, indicating the reliability of this method. Results of the study indicate that the new orography processing method can significantly improve the accuracy of near-surface temperature and wind speed forecast and is numerically stable and reliable.