中文摘要：

干旱和洪水灾害是新疆主要的自然灾害，其发生频率高，分布广且危害严重。洪旱灾害空间聚集研究对防灾减灾、救灾物资科学配置的意义重大。以近30年新疆各县市洪旱灾害次数为基础，运用全局Moran’sⅠ系数、Moran散点图和Anselin局部Moran’sⅠ系数分析了新疆县域尺度洪旱灾害的空间自相关性，揭示了新疆洪旱灾害的局部聚集和局部异常特征。全局空间自相关结果表明，县域尺度干旱、暴雨与升温混合型洪水、融雪型洪水和山洪Moran’sⅠ z值大于0．01，置信水平为2．54，表明洪旱灾害在行政单位间存在很强的空间自相关性，空间分布呈聚集状态。局部空间自相关结果表明，干旱聚集区比较分散，主要分布在东疆哈密地区和北疆伊犁地区、昌吉回族自治州、阿勒泰地区和南疆和田地区；1984—1989年干旱聚集程度最高，研究时段内聚集整体呈下降趋势，各时段均有少数几个县市与邻近县市局部空间自相关显著，局部区域呈集聚或离散的空间格局。各类型洪水“高一高”聚集空间差异显著，融雪型洪水聚集在塔城地区和克州，山洪聚集在哈密地区哈密市、伊吾和巴里坤县与周边木垒县；暴雨与升温混合型洪水聚集在博州的温泉、博乐和精河县以及伊犁地区的霍城县。

英文摘要：

Drought and flood are the main natural disasters in Xinjiang. They are widely distributed, happen fre- quently, and cause serious damage Study on spatial clustering of flood and drought is of great significance for dis- aster prevention and mitigation, and the scientific allocation of disaster relief supplies. Based on the disaster fre- quency in cities and counties of Xinjiang in past 30 years, this paper analyzes the spatial autocorrelation of flood and drought on a county scale in Xinjiang and revealed the local clustering and the local anomaly characteristics by using Moran＇s I coefficient, Moran scatter-plot and Anselin local Moran＇s I coefficient, reveals local cluster and lo- cal abnormal characteristics of drough and flood in Xinjiang. The result of global spatial autocorrelation shows ： the Moran＇s I Z values of drought, floods caused by rainstorm and heating-up, snow-melt floods and mountain torrentson a county scale are greater than 0.01 and the confidence level is 2.54. It shows that there is a strong spatial au- tocorrelation of flood and drought among administrative units, and the spatial distribution is in a state of clustering. The results of local spatial autocorrelation shows： drought clustering areas are disperse and are mainly distributed in Hami Prefecture in eastern Xinjiang, Ill Prefecture in northern Xinjiang, Hui Autonomous Prefecture of Changji, Ahay Prefecture and Hotan Prefecture in southern Xinjiang. During 1984 to 1989, the drought aggregation degree is the highest, the aggregate entirety is on the decline, the local spatial autocorrelation between few cities and counties and neighboring counties and cities is remarkable in different time period. And parts of area show clustering or dis- persing spatial patterns. The ＂High-High＇clustering spatial difference of different types of floods is significantly dif- ferent. Snow-melt floods are tain torrents are aggregated Autonomous County nearby city, Jinghe County in Bole aggregated in Tacheng Prefecture and Kergez Autonomous