The initial tendency approach is used to diagnose systematic errors in global GRAPES （Global/Regional Assimilation Prediction System）, including overly strong westerlies in the northern midlatitudes, cold/warm bias dipoles in the vicinity of the tropopause, and excessively strong southerlies in downstream regions of the Tibetan Plateau. This approach, involving the use of the assimilation system, focuses on the first few time steps of numerical weather forecasts to identify the deficiencies in diabatic forcing. The results show that there is insufficient diabatic dissipation in the upper troposphere and lower stratosphere of the northern midlatitudes and the lower troposphere of most latitudes, which results from the absence of a parameterization of subgrid orographic drag in global GRAPES. A scheme to parameterize the effects of these drags is therefore tested and the experiments indicate that the newly introduced scheme reduces zonal momentum budget residuals, weakens the northern midlatitude westerlies and southerlies in the downstream regions of the Tibetan Plateau, decreases the cold/warm bias dipoles, and leads to improved objective verification scores.
