Improving dust simulation over the Taklimakan Desert through refinement of wind profile parameters in the Revised-MM5 scheme

The Taklimakan Desert, the world's second-largest shifting-sand desert, is a major dust source for northwestern China and East Asia, and its emissions contribute substantially to the global dust cycle. Therefore, improving dust forecasts over this region is essential. To address limited applicability of the wind-profile functions in the Revised MM5 surface-layer scheme over the Taklimakan Desert, we recalibrated the scheme's stability parameters using turbulence and gradient measurements from the Tazhong station, and evaluated the impacts through sensitivity experiments. The estimated von Kármán constant (k) averages 0.41, close to the canonical value (0.40). The recalibrated stability parameters (γm = 19.0 for unstable conditions; βm = 5.8 for stable conditions) fall within expected theoretical ranges. Using the revised parameters improves the correlation coefficient(R) of 10-m wind speed at 62% of stations, while reducing the mean absolute error (MAE) and root mean square error (RMSE) at 76% and 71% of stations, respectively, with particularly notable improvements during the morning and afternoon transition periods during March–May 2020. For dust concentrations, PM10 MAE and RMSE decrease by 1–6.5% across all stations. For PM2.5, the R increases by 1–4%, and MAE and RMSE decrease by 1–6.8%. Aerosol optical depth (AOD) decreases by 0.1–0.2 over the northwestern and eastern parts of the desert, effectively reducing the model's systematic high bias in dust simulation.