基于确定性系数与证据权模型的滑坡灾害敏感性评估:以金沙江龙开口——期纳段流域为例

发布时间：2018-09-18 18:42

【摘要】：金沙江流域内滑坡灾害频发,对大型工程及交通设施均造成了极大的损害。采用确定性系数(CF)与证据权(WOE)模型,选取坡度、坡向、NDVI、海拔、岩石性质、纵剖面曲率、距道路的距离、距河流的距离、距断层的距离与第四纪沉积物类型10个指标,分不同组合进行区域性滑坡灾害敏感性评估。CF与WOE在多种不同指标组合下经过累积频率曲线下面积(AUC)检验,结果表明10种指标组合下的CF模型表现最佳,相比WOE模型所得的最高准确率(74.52%)与预测率(69.89%),CF的准确率与预测率分别高达83.40%与74.43%。CF计算获得的敏感性指数通过自然间断点法将研究区域分为滑坡极难发生区、滑坡较难发生区、滑坡较易发生区与滑坡极易发生区4类区域。评估结果指示程海断裂带南北延伸带与金沙江中段主要干支流沿线为区域的滑坡灾害易发区,表明断裂带及河流与该区域的滑坡发生具有一定的相关性。
[Abstract]:Landslide disasters occur frequently in Jinsha River basin, causing great damage to large engineering and transportation facilities. Using deterministic coefficient (CF) and evidential weight (WOE) model, 10 indexes of slope, slope direction, elevation, rock property, longitudinal profile curvature, distance from road, distance from river, distance from fault to Quaternary sediment type are selected. The sensitivity assessment of regional landslide hazard. CF and WOE are tested by area (AUC) under cumulative frequency curve under different index combinations. The results show that the CF model with 10 index combinations is the best. Compared with the WOE model, the highest accuracy (74.52%) and the prediction rate (69.89%) of CF are as high as 83.40% and 83.40%, respectively. The sensitivity index calculated by 74.43%.CF can divide the study area into the extremely difficult area of landslide and the area of difficult occurrence of landslide by natural discontinuous point method. There are 4 types of landslide prone areas and landslide prone areas. The evaluation results indicate that the north and south extension zone of Chenghai fault zone and the main trunk and tributaries of Jinsha River are prone to landslide disasters, which indicates that the fault zone and river have certain correlation with the occurrence of landslide in this area.
【作者单位】：中国地质大学(北京)地球科学与资源学院;中国地质科学院地质力学研究所;
【基金】：中国地质调查局项目(12120114001501)
【分类号】：P642.22

【相似文献】