中文摘要：

采用改进的基于扩散函数的内集-外集模型，分析1949—2007年辽宁省旱灾受灾指数，评价辽宁省的旱灾风险。结果表明，辽宁省旱灾发生频繁，平均每1．7a一遇且严重干旱多发，这与辽宁省旱灾的实际情况基本吻合。研究表明，内集-外集模型计算结果的意义清楚，对防灾减灾有一定指导作用。

英文摘要：

The present paper would like to introduce an improved drought risk evaluation model based on the interior-outer-set theory by taking Liaoning as a case study. As is known, Liaoning is a province often hit by drought, which takes place unexpectedly at a rather low pace and often lasts for many days or even months. As a province of important commodity grain-producing base of the country, Liaoning suffers serious drought either due to the shortage of rain fall or due to the instable and non-uniform distribution of rain fall in different seasons and different places. In addition, serious drought has also led to the shortage of water for irrigation systems, on which the province heavily relies to sustain its regular agricultural production. It is just from this need that the given paper intends to work out a drought risk evaluation model based on the so-called interior-outer-set theory in hoping to facilitate the drought prevention. In doing our research, we have studied the traditional methods for risk evaluation by using theories of probability and statistics. As a matter of fact, since possibility-probability distribution of the drought risk can make sense in expressing some kind of rough suggestibility or fuzzy probability, it is likely to be one of the effective approaches to evaluating the drought risks. However, the possibility-probability distribution theory can probably be more useful for large-scaled drought risk evaluation than for smaller-scaled evaluation. Therefore, based on the method of information/ data distribution, we have worked out an interior-outer-set model, which can be optimized for smaller samples and represent the imprecision of probability estimation when it is used to calculate a possibility-probability distribution. But the interior-and-exterior set model, which is established based on the method of information distribution, has also had its inadequacy. For example, between 0.5 - 1, there is no value, and the distribution of data/information proves too concentrated. Therefore, we have dev