空间数据挖掘技术在滑坡敏感性评价中的应用  

Application of Spatial Data Mining Technology in the Landslide Susceptibility Evaluation

作　　者：夏添[1] 唐川[1] 常鸣[1] 

机构地区：[1]成都理工大学环境与土木工程学院,四川成都610059

出　　处：《人民黄河》2012年第9期79-81,132,共4页Yellow River

基　　金："十一五"国家科技支撑计划项目(2011BAK12B01)

摘　　要：基于空间数据挖掘技术对滑坡敏感性评价进行研究,利用遥感技术、地形图矢量化和数字高程模型(DEM)生成等方法获取空间数据,建立研究区的滑坡灾害空间数据库,在全面分析滑坡与高程、坡度、坡向等地貌因子的关系后,采用Auto Logistic回归模型和信息统计量模型对研究区进行震后浅层滑坡的敏感性评价,并对不同的评价方法产生的敏感性评价结果进行了精度分析。结果表明:信息统计量模型的AUC为0.713,Auto Logistic回归模型的AUC为0.836,两个模型可以相互验证,Auto Logistic模型的预测精确度较高。The landslide sensitivity was evaluated based on spatial data mining technology, and spatial database of landslide disaster was established using spatial database obtained via remote sensing technology, topographic map vector quantization and digital elevation model （DEM） generation method. After comprehensive analysis of relationship between the landslide and elevation, gradient, slope direction and the other physiognomy fac- tors ,the landslide sensitivity after earthquake in superficial lamella of research area was evaluated using improved Auto Logistic regression model and information statistic model, and the precision of evaluated results was analyzed also. The results show that the AUC of information statistic model is 0. 713 ; the AUC of Auto Logistic regression model is 0. 836. Two models can be mutual checked and the prediction precision of Auto Logistic model is higher.

关 键 词：滑坡敏感性评价 空间数据挖掘 遥感技术 ROC曲线 

分 类 号：P642.22[天文地球—工程地质学]