A continuous rainstorm event in the middle of Shanxi Province during 1-3 August 2015 was analyzed by using meteorological observations and NCEP/NCAR reanalysis data. Based on the test of ECMWF ensemble forecast products, the joint probability method was used to improve the ensemble forecast of rainstorm. The results are as follows: (1) The precipitation was characterized obviously by convection in this event, which could be divided into two stages. The first stage was the sudden rainstorm in the front of the cold front. In the second stage, the heavy rainfall was caused by the regional strong convective weather, which was affected by the high trough, the low shear line, and the surface cold front. (2) The strong upward motion caused by the convergence of the lower level made many mesoscale convective cloud clusters merge and develop continually, which resulted in rainstorm directly. In the first stage, the convergence layer at the middle and low level was deeper, the suction effect at high level was stronger, and the precipitation rate was greater. In the second stage, the dynamic and energy disturbances at the lower layer were more obvious, and the precipitation area was more concentrated. (3) Mesoscale baroclinic diagnosis showed that, in the first stage, the convergence and frontogenesis of the wet Q-vector at the lower layer increased many times and caused the convective motion stronger and the range larger, and formed many rain peaks of the precipitation. While in the second stage, the baroclinic development of mesoscale system was stronger, but the thickness of instability decreased, therefore the precipitation intensity decreased. (4) The improved ensemble forecast product of the joint probability showed that the range of high probability coincided with the center of rainstorm.

The fifth generation PSU /NCAR’s non - hydrostatic mesoscalemodel ( MM5 ) is app lied to simulatemeteorological fields of the dustweather affecting Beijing in North China on 21March 2001. The results show the model is capable for simulating and p redicting spatial and temporal variations of dustweather in Northern China. Low - p ressure system and its vertical structure of high - level divergence and low - level convergence resulted in strong updraft causing dust emission, and strong wind was dynamic condition for transmission of sand dust. The trajectorymodel is app lied to simulate the air particle trajectory, which commendably reflects source area and transmission trajectory of sand dust. The p redicted area of dust storms is in good agreementwith those ofmeteorological observation and satellite images. The software is p rogrammed to demonstrate dynamic p rocess of the meteorological fields.

Based on the fifth generation PSU /NCAR’s non - hydrostatic mesoscale model (MM5) and comp rehensive air qualitymodel with extensions (CAMx, a 3 - D Eulerian regional air qualitymodel) a air quality simulation system has been comp leted, which is app lied to simulate sandstorm transmission affecting Beijing in North China in sp ring 2001. The results show the system is capable for simulating and p redicting sandstorm transmission and concentration distribution in North China. Significant correlation was found between simulation value of TSP and L idar observation, which confidence levelwas above 99%. The contribution rate ofMongolia, Inner Mongolia and Hebei to dust concentration in Beijing are 30% - 50% , 30% - 60% and under 29% , respectively.