文章应用空间自相关模型和GIS技术,测算并绘制出自然灾害受灾面积省域分布Moran’s I散点图,藉此呈现自然灾害高—高、低—低、高—低、低—高集聚4种空间象限关系,并通过LISA集聚图划分了集聚分布区域,发现空间分布并非随机而是具有规律可循的,存在显著空间依赖性特征,并呈现出省域旱灾、风雹等损失类型的带状、团状、点状或离散等空间分布规律和特征。研究发现,自然灾害频发、生态脆弱区域往往与贫困地区存在空间叠加关系,致贫返贫效应长期恶性循环且难以消除。 Based on the spatial autocorrelation model and the GIS technology, the paper calculates and draws the Moran’s I scatter plot of natural disasters affected area, and presents four kinds of high-low, high-low, high-low and high-concentration natural disasters Spatial quadrant relationship, and through the LISA clustering map to divide the distribution of regional distribution and found that the spatial distribution is not random but has a rule-based, there is a significant spatial dependence of the characteristics of the province and showed the loss of drought, wind hail and other types of ribbon , Clumps, punctate or discrete spatial distribution patterns and characteristics. The study found that frequent natural disasters and ecologically fragile regions tend to have spatial overlapping relationship with poor regions, and the poverty and poverty alleviation effect has a long-term vicious circle and is difficult to eliminate.