Based on the hourly precipitation observation data of 105 national stations and 1240 regional stations and 8 Doppler weather radars data in Xinjiang from 2010 to 2018, from the perspective of operational forecasting, the definition of shortterm heavy precipitation process in Xinjiang was proposed and 468 shortterm heavy precipitation processes were selected, and circulation configuration of influence system and radar observation characteristics were analyzed. The results show that there were four kinds of influence systems including the central Asia trough (vortex), the west Siberian trough (vortex) and low level northwest jet stream. The convective storms resulting in shortterm heavy rainfall were combined enhancement type, train effect type and isolated convective cell type, and the combined enhancement type was 45.1%, isolated convective cell type was 34.8%, and the train effect type was 20.2%, respectively. The parameter thresholds of the maximum reflectivity factor intensity (Zmax), the maximum height of the strong echo (Dmax), the maximum height of the echo top (ET) and the maximum vertical accumulated liquid water content (VIL) observed by Doppler radar during shortterm heavy precipitation processes in southern Xinjiang were less than those in northern Xinjiang, and in Yili prefecture they were largest and in Aksu they were smallest. The shortterm heavy precipitation in Yili prefecture was dominated by low centroid echo, while in other regions it was dominated by low centroid and high centroid echo.

A heavy rain process occurred in Yining of Yili valley on 30 September 2019. Based on the observation data of micro rain radar (MRR), terrestrial OTT-PARSIVEL raindrop spectrometer and rain gauge at Yining station, the reliability of MRR data was compared and tested. And on this basis, the vertical distribution characteristics of microphysical quantities retrieved by MRR were investigated at different stages of the heavy rain process. The results show that the precipitations from three kinds of instruments were close, and their change trends were consistent. The rainfall intensity of MRR near ground level (35, 70 and 105 m) had a good correlation with the observation of OTT-PARSIVEL raindrop spectrometer, the coefficients were 0.9233, 0.9289 and 0.9186, respectively, and the convergence degree of rainfall intensity between two kinds of instruments was higher when the rainfall intensity was less than 3 mm·h^-1. The vertical distributions of microphysical quantities from MRR were different at different stages of rainfall intensity. At the low rainfall intensity stage, the environmental humidity was lower and evaporation was stronger during the initial period of rainfall, the radar reflectivity factor, liquid water content and rain intensity from MMR decreased with decrease of height, while the environmental humidity was higher and evaporation was weaker during the middle period of rainfall, and the vertical changes of radar reflectivity factor, liquid water content and rain intensity weren’t obvious. During the late period of rainfall, the rain intensity near ground decreased significantly under the low rainfall intensity condition because the supply of water vapor and power was lack. At the moderate and high rainfall intensity stages, the vertical distribution of particles falling velocity was stable, while the radar reflectivity factor, liquid water content and rain intensity increased with decrease of height due to the collision and merger interaction between raindrops. The small raindrops dominated during the heavy rain process in Yining, the percentage of average number concentration of small raindrops to total number concentration exceeded 90%, and it decreased with decrease of height. The contribution of medium raindrops to total rain intensity was the greatest, the contribution rates at different stages of rainfall intensity were 60% and above, and it increased with decrease of height. The number concentration of large raindrops was the smallest in proportion of total number concentration, and its contribution to total rain intensity was the smallest.

Based on conventional ground observation data, TBB data of FY-2G satellite and NCEP/NCAR reanalysis data, the synoptic circulation pattern and mesoscale systems of the heavy rainstorm in Yili area of Xinjiang from 16 to 17 June 2016 were analyzed. And on this basis, the formation of the heavy rainstorm was analyzed in detail by using the output data of WRF model with high resolution. The results show that the main weather systems of the heavy rainstorm process were low trough in Central Asia, westerly jet in upper level, shear lines and convergence lines in lower layer. The multiple mesoscale cloud clusters sustained in Tianshan mountains area of northern Yili for a long time under the effect of terrain lifting, which caused strong rainfall continuously. The simulation result showed that the mesoscale convective cells moving continuously to the Tianshan mountains in northern Yili directly caused the heavy rainfall in Yili region, and the development of convective cells was closely related with low-level jets, low-level wind convergence and terrain. The dynamic convergence enhanced with increase of low-level jets, which triggered the release of CAPE (convective available potential energy), further led to the maintenance and rapid development of vertical motion together with terrain uplift, and caused the enhancement of convective system near convergence line in lower level, which were beneficial to the occurrence of heavy rainfall along Tianshan mountain of northern Yili.

Based on the NCEP/NCAR 0.25 ° × 0.25 ° reanalysis data, ERA-Interim 0.5°×0.5°reanalysis data, hourly and daily precipitation data, ground-based GPS precipitable water vapor data (PWV) from March 2016 to February 2017  at three stations in the Yili River Valley, temporal variation characteristics of PWV and its relationship with precipitation at three stations were analyzed. The evolution characteristics of PWV at different stations under different precipitation conditions and in different seasons were clarified. The results were as follows: (1) PWV had an obvious monthly variation, and it presented single-peak distribution at each station with the lowest value in January and the peak in July. The variation of PWV at Yining station was in consistent with the precipitation change from February to September (except July) but it was opposite from October to December. (2) The diurnal variation of PWV at each station presented double-peak distribution in spring and summer, the maximum  PWV appeared at 17:00 BST and 00:00 BST in spring, and it was 2-4 h later than that in summer. In autumn, the diurnal variation of PWV presented single peak distribution at Xinyuan station but it presented double peak distribution at  other two stations，while diurnal variation of PWV at three stations presented single-peak distribution in winter. As the altitude of the station increased, the variation of daily PWV increased gradually. (3) The maximum PWV was 0-3 h, 5 h, and 7-9 h ahead of precipitation, which was the highest  frequency. The occurrence time of precipitation in four seasons mainly was 0-3 h, 0-2 h and 5-7 h, 0-3 h and 7-9 h, and 0-1 h later than that of the maximum PWV, respectively. There were significant differences for PWV values in three stations under different precipitation conditions, and the higher the altitude, the less significant the difference was. (4) Humidification of PWV was related to the transport and inflow of water vapor in the lower troposphere before rainfall occurrence. During the precipitation period, the peak value of PWV and humidification time were different under different influence system and different water vapor transportation. The start time of precipitation had a good correspondence with peak  value of PWV. During the precipitation period, there was  a significant vertical transport of water vapor over the rainstorm area, which resulted in a gathering area of ice water formed in clouds in the middle and upper troposphere, PWV transition and short-term heavy rainfall at the station.