中文摘要：

小范围大比例尺地质灾害易发性评价面临时空数据精度或质量不满足理论要求的现状。如何利用有限的数据实现相对可靠的评价结果,是论文研究的主要目的。论文选取湖北省五峰县渔洋关镇为研究区域,考虑水文、地形、岩性分布、历史灾害规模等因素讨论了斜坡单元的划分方法;将证据权模型分析所得滑坡易发性值转换至各斜坡单元,绘制基于斜坡单元的研究区易发性评价分级图;分别采用ROC曲线和野外核查手段,评价了易发性结论的成功率和预测率。研究表明,斜坡单元的划分一方面要符合实际地形,另一方面更应体现单元的地质差异性;栅格计算结论转化为斜坡单元的表达成果、精度分析的注重和野外核查工作的开展是保证分析结果可靠性的有效途径;湖北省五峰县渔洋关镇易发性评价成功率和预测率分别为84%和74%,野外验核与理论分析结果的一致性较高。

英文摘要：

Data accuracy and quality does not satisfy small area-large scale landslide susceptibility mapping in mountainous area. In this paper,it aims to propose a methodology of susceptibility mapping using limited data for landslides in a town named Yuyangguan,located at Wufeng county,west Hubei province,China. Firstly,slope unit map was generated by a method considering hydrology, topography, lithology and historical landslide magnitude. Weights-of-evidence model was applied for grid-based landslide susceptibility mapping,which was later converted to slope unit results using average susceptibility value. Then the success rate and prediction rate was evaluated by Receiver Operating Characteristic（ ROC） curve and filed check. The results showed that slope unit division should consider not only terrain and hydrology,but also geological differences,such as lithology. Gridbased susceptibility map can be converted to slope unit map for the convenience of risk management practice. The reliability can be measured by not only ROC curves but also field investigation. For Wufeng county,the success rate and prediction rate of landslide susceptibility were 84% and 74%,approving a high agreement with field work.