基于层次分析与受试者工作特征曲线下面积耦合算法的2022年泸定地震滑坡易发性评估

LANDSLIDE SUSCEPTIBILITY ASSESSMENT FOLLOWING THE 2022 LUDING EARTHQUAKE:A COUPLED ANALYTIC HIERARCHY PROCESS AND AREA UNDER THE RECEIVER OPERATING CHARACTERISTIC CURVE ALGORITHM

地震滑坡是由于地震的触发作用而导致接近或恰好处于临界状态的坡体发生提前滑动的地质事件。2022年9月5日,四川省甘孜藏族自治州泸定县发生6.8级地震,诱发了大量滑坡,造成严重的人员伤亡和财产损失。本文采用野外调查和遥感解译相结合的方法对泸定地震极震区内的地震滑坡进行解译,并基于ArcGIS地理信息系统平台分析了高程、坡度、坡向、曲率、地层岩性、与断层距离、地震峰值加速度、与水系距离和与道路距离等9个因子与地震滑坡的空间分布规律。通过引入受试者工作特征曲线(ROC)来改进层次分析法(AHP),克服了AHP方法采用专家打分确定影响因子权重的主观性,并采用改进的AHP方法对泸定地震滑坡易发性进行了评估。结果表明,泸定地震在极震区内共诱发5008处滑坡,总面积达17.28 km^(2)。高程、与水系距离和与道路距离是诱发泸定地震滑坡的主控因素。地震滑坡的极高易发区和高易发区面积约占研究区总面积的26%,主要分布在大渡河河谷两岸。采用ROC曲线下面积(AUC)对评估结果进行检验,AUC值为0.8505,表明评估模型具有较高的准确性。本研究可为地震滑坡易发性评估方法的改进提供参考,并为泸定地震灾后重建规划提供科学支撑。

Coseismic landslides represent geological phenomena where slopes,either nearing critical states or already in precarious conditions,undergo premature sliding triggered by seismic activity.On 5 September 2022,a M_(S)6.8 earthquake struck Luding County,Garze Tibetan Autonomous Prefecture,Sichuan Province,China,triggering numerous landslides and causing severe casualties and property damage.Drawing from field investigations and remote sensing analyses,this paper delves into the interpretation of coseismic landslides within the Luding earthquake s epicentral area.The ArcGIS geographic information system was employed to analyze the spatial distribution pattern of landslides in correlation with diverse factors such as elevation,slope gradient,slope aspect,curvature,lithology,distance to faults,peak ground acceleration,distance to rivers,and distance to roads.To enhance the precision of the Analytic Hierarchy Process(AHP),we introduced Receiver Operating Characteristic(ROC)curves,effectively mitigating the inherent subjectivity in AHP′s determination of controlling factor weights through expert scoring.An improved AHP method was thus employed to assess the susceptibility of coseismic landslides in Luding.The results reveal that the Luding earthquake triggered a total of 5,008 landslides within the extreme seismic zone,covering an area of 17.28km^(2),predominantly situated along the slopes of the Dadu River valley.Notably,elevation,distance to rivers,and distance to roads emerged as the pivotal factors dictating the spatial distribution of coseismic landslides.Approximately 26%of the study area was identified as extremely high or high susceptibility zones for coseismic landslides.The reliability of the proposed model was evaluated by calculating the Area Under the ROC Curve(AUC),yielding a value of 0.8505,indicating the model s robustness.This study contributes valuable insights for refining methods to assess the susceptibility of earthquake-induced landslides and provides theoretical underpinning for post-disaster reconstruction planning in Luding.