Abstract: Drought is a natural phenomenon whose occurrence is linked to a sustained lack of precipitation. When a lack of precipitation causes a substantial hydrological imbalance, there is a drought event. Recently, the consideration of global warming has been paid more attention, the northwest China is very sensitive to the climate change. Therefore, more and more meteorologists are attracted to study droughts. By using linear regression, K-means, Mann-Kendall nonparametric statistical test and Morlet-Wave analysis, the characteristics of precipitation and drought in the target area were analyzed based on long-term(1961-2015)daily precipitation data from 51 meteorological stations. The results show an uptrend on precipitation and rainy days over the past 55 years, and the correlation analysis indicates that the increase rainy day mainly caused the increase in precipitation. Regional precipitation characteristic was obtained by K-means cluster, the precipitation in mountainous area (III cluster)was much higher than that in north Xinjiang(II cluster), and south Xinjiang(I cluster). Also, the relationships between elevation and precipitation in both north and south Xinjiang were analyzed respectively. It revealed only significant correlation in north Xinjiang. Furthermore, correlation analysis shows that SPI and Pa were significantly correlated and there was a downward trend in drought frequency, scope and severity. The spatio-temporal variation of drought relief showed an increasing trend from south to north throughout the target area, but there was some heterogeneity at local scales, drought became more severe at some individual sites in north Xinjiang. Three periodic oscillation of droughts were found, including 4-year period, 8-year period and 16-year period; 8-years period was dominant.

Key words: standardized precipitation index(SPI), precipitation anomaly(Pa), K-means cluster, Xinjiang