Abstract: In the weather research and forecasting model, different parameterization schemes affect significantly the precipitation prediction. It can improve the prediction accuracy to choose the appropriate parameterization schemes. In this paper, the 3.8 version of WRF (weather research and forecasting model) was used to simulate four precipitation cases in Xinjiang using 3-km and 9-km two-layer nested grids and ECMWF (European Centre for Medium Range Weather Forecasts) reanalysis data as the initial fields and boundary conditions. The applicability of Lin scheme, WSM6 scheme, Thompson scheme and WDM6 scheme to precipitation simulation in Xinjiang is evaluated by TS (threat score) score. The results showed that the Thompson scheme was of an advantage in predicting light rain (0.1-5.0 mm) and moderate rain (5.1-10.0 mm), and there were the missing forecasts on rainfall by other three schemes of cloud microphysical parameterization. According to the correlation coefficient, the simulation accuracy of Thompson scheme was slightly higher than that of other schemes, this scheme was suitable for Xinjiang, but the four kinds of schemes were not very good in forecasting heavy precipitation, so the schemes need to be improved for Xinjiang.

Key words: WRF, threat score, cloud microphysics, parameterization scheme, precipitation, simulation capability; Xinjiang