摘要
快速准确地获取地震滑坡分布和评估灾害损失,对应急救援和安置规划至关重要。该文提出了一种针对应急响应、中期安置和重建三个阶段的地震触发滑坡应急评估方法,并以2022年M_(S)6.8泸定地震为例验证其适用性。在震时响应阶段,应用全国模型进行快速评估,AUC值为0.80。第二阶段,基于不完整数据构建评估模型,AUC值提升至0.88,但震中西侧滑坡多发区被低估。第三阶段,融合完整数据,模型AUC值超过0.94,高危险区域预测与实际高度结果一致。结果显示,该方法能够迅速地为地震灾区提供滑坡信息,满足应急响应与灾民安置的实际需求。
Rapid and accurate acquisition of earthquake-induced landslide distribution and disaster loss assessment is critical for emergency rescue and resettlement planning.An emergency landslide hazard assessment framework is proposed,tailored to three key post-earthquake phases:immediate emergency response,mid-term resettlement,and long-term reconstruction.The applicability of the proposed method is demonstrated using the 2022 M_(S) 6.8 Luding earthquake as a case study.In the initial response phase,a nationwide model is employed for rapid assessment,achieving an AUC of 0.80.In the second phase,a model based on incomplete inventory data improves the AUC to 0.88,although landslide-prone areas west of the epicenter are underestimated.In the third phase,by incorporating a complete landslide inventory,the model's AUC exceedes 0.94,with predicted high-hazard zones closely matching actual observations.The results indicate that the proposed approach can rapidly deliver landslide information to affected areas,effectively supporting emergency response and resettlement efforts.
作者
邵霄怡
许冲
马思远
谭庆全
薄涛
高会然
SHAO Xiaoyi;XU Chong;MA Siyuan;TAN Qingquan;BO Tao;GAO Huiran(National Institute of Natural Hazards,Ministry of Emergency Management of China,Beijing 100085,China;Key Laboratory of Compound and Chained Natural Hazards Dynamics,Ministry of Emergency Management of China,Beijing 100085,China;Key Laboratory of Active Tectonics and Volcano,Institute of Geology,China Earthquake Administration,Beijing 100029,China;Beijing Earthquake Agency,Beijing 100080,China)
出处
《灾害学》
北大核心
2025年第4期71-76,共6页
Journal of Catastrophology
基金
北京市科技计划项目“基于人工智能的地震地质灾害风险评估与预警模型研究”(Z231100003823035)
国家自然科学基金青年基金项目“考虑发震断层效应的区域地震滑坡规模参数预测研究”(42407275)。
关键词
应急评估方法体系
2022年泸定地震
地震滑坡
机器学习模型
emergency assessment methodology system
2022 Luding earthquake
earthquake-induced landslides
machine learning model
作者简介
第一作者:邵霄怡(1993-),女,汉族,陕西西安人,副研究员,主要从事地震滑坡危险性研究.E-mail:xiaoyishao@ninhm.ac.cn;通信作者:许冲(1982-),男,汉族,河南周口人,研究员,主要从事滑坡地震地质学研究.E-mail:xc11111111@126.com。
