【摘要】 根据地震诱发的崩塌具有数量大和成区分布的特点,将研究区北川苏宝河流域划分为211个子流域,评价选取子流域的平均高程、最大高差、平均坡度、距离河流距离、沟谷密度和岩石类型等指标,采用Logistic回归的统计学方法进行建模,并且使用ROC曲线和Kappa系数对模型的精度进行评价,其ROC曲线下面积为0.926,Kappa系数为0.690。采用同样的评价方法基于像元对研究区进行建模,模型的ROC曲线下面积为0.814,Kappa系数为0.470,均小于基于子流域的评价模型。使用模型对研究区进行敏感性评价,基于子流域的敏感性评价,有90.04%的崩塌位于高敏感和敏感区域内,而基于像元的敏感性评价,高敏感和敏感区域内只占有68.82%的崩塌。但不同敏感性区域面积比例来看,基于子流域的敏感性评价,有高敏感区域面积过大的缺点。研究进一步提出崩塌响应单元这一概念,即针对崩塌的各影响因子,将研究区划分为比子流域更小的单元,而每个单元内部,对崩塌具有相同的敏感度。更多还原

【Abstract】 Landslide is one of the most common hazards induced by the Wenchuan Earthquake for its large quantity and wide distribution.The sub-basin was introduced into the assessment of landslides induced by earthquake.The Subao River Basin,Beichuan,which was chosen as the study area,was divided into 211 sub-basins,and the aver-age elevation,maximum relative elevation,average slope gradient,distance to the river,gully density and stone type were chosen as the assessing factors.The statistical method Logistic regression was used to build the assessing model.The model was checked by ROC(Receiver Operating Characteristic) curve and Kappa.And the AUC(Aare Under Curve) and Kappa was 0.926 and 0.690,respectively,which was much better than the pixel-oriented sus-ceptibility assessing model build by the same method.The AUC and Kappa was 0.814 and 0.470,respectively.From the susceptibility map,about 90.04 percent of landslides located in the high susceptible area.The deficiency of the sub-basin oriented assessment was that the area of high susceptible area is too large.So at last the concept of LRU(Landslide Response Unit) was introduced.The sub-basin could be divided into smaller units by some factors related to landslide.The assessment based on LRU may have a better assessing result.更多还原