资源环境科技发展态势分析平台

This paper aims to test the ability of the WRF-Chem model to simulate sand storms in Northwest China and improve the simulation results by introducing remote sensing soil moisture data into WRF-Chem. We conducted sensitivity tests, including parameterization scheme tests and soil moisture tests using WRF-Chem. By comparing various combinations of parameterization schemes, which have considerable impacts on sand emissions, the most suitable parameterization scheme for WRF-Chem to simulate sand storms in Northwest China was selected for establishing models in future research. Based on the optimal combination of parameterization schemes, the sensitivity of sand emissions to soil moisture was tested, indicating that soil moisture has a substantial influence on sand emissions and revealing a nonlinear relationship between sand emissions and soil moisture. Although sand simulations are sensitive to soil moisture, the sand emission volume does not always increase with decreasing soil moisture content. Moreover, the soil moisture content simulated by WRF-Chem is quite different from that observed by satellites. Therefore, we selected ten sand storms that occurred in Northwest China in 2009–2018, including weak, moderate and strong sand storms, to explore the responses of sand storms of different intensities to soil moisture. Some experiments were conducted in two scenarios: scenario A was simulated only by WRF-Chem, and scenario B was designed by replacing the initial soil moisture field in WRF-Chem with soil moisture satellite data from AMSR2 and AMSR-E. Comparing the simulation results with in-situ ground-based and satellite-based measurements indicates that WRF-Chem can capture strong sand storms and some moderate sand storms, but the ability to simulate weak sand storms needs to be greatly improved. After introducing remote sensing soil moisture data into WRF-Chem, the simulation of sand emissions becomes more accurate, substantially improving the simulation results. Among the ten simulated sand storms, the AOD (aerosol optical depth) and PM10 simulation accuracies are improved for five and seven sand storms, respectively. After this improvement, the AOD correlation coefficient can reach 0.63, and the PM10 correlation coefficient can reach 0.60.