基于改进Newmark模型的同震滑坡易发性研究  

The Susceptibility Study of Coseismic Landslides Based on the Improved Newmark Mode

作　　者：席传杰 胡卸文 何坤 罗刚 周瑞宸 胡亚运 XI Chuanjie;HU Xiewen;HE Kun;LUO Gang;ZHOU Ruichen;HU Yayun(Faculty of Geosciences and Environment Engineering,Southwest Jiaotong University,Chengdu 611756,China;Sichuan Province Engineering Technology Research Center of Ecological Mitigation of Geohazards in Tibet Plateau Transportation Corridors,Chengdu 611756,China;Zhejiang Engineering Survey and Design Institute Group Co.,Ltd.,Ningbo 315012,China)

机构地区：[1]西南交通大学地球科学与环境工程学院,成都611756 [2]四川省环青藏高原交通廊道地质灾害生态化防治工程技术研究中心,成都611756 [3]浙江省工程勘察设计院集团有限公司,浙江宁波315012

出　　处：《三峡大学学报（自然科学版）》2024年第6期32-40,共9页Journal of China Three Gorges University：Natural Sciences

基　　金：国家自然科学基金项目(42377170,42277143);四川省自然科学基金面上项目(2024NSFSC0100)。

摘　　要：Newmark模型被广泛应用于同震滑坡快速评估,然而传统模型受限于参数(如黏聚力、内摩擦角)的不确定性.本文从概率建模的角度,基于蒙特卡罗方法提出改进Newmark模型,该模型从特定分布(如高斯分布)中抽取随机样本以解释参数不确定性.以2017年九寨沟地震为例阐述了改进Newmark模型在同震滑坡易发性快速评估中的应用,并与不同采样策略下3种机器学习模型相对比.结果表明:改进Newmark模型极高易发区面积占比为5%,包含滑坡数量占比为59.6%,频率比值为128.4,不同采样策略下模型平均AUC值为0.70,表明改进Newmark模型在同震滑坡易发性评估中取得较好效果.本文所提出的模型由特定参数驱动,与数据驱动方法的区别在于不依赖历史滑坡数据,其成果可应用于震后滑坡快速评估,或在给定地震情境下模拟区域滑坡失稳概率以服务于工程早期选址.The Newmark model was widely adopted in the rapid assessment of coseismic landslide.However,the typical Newmark model is limited by the parameter uncertainty(e.g.cohesion and friction angle).This study proposed and developed an improved Newmark model by using the Monte Carlo simulation.This model account for the parameter uncertainties by random sampling from the specific distribution such as Gaussian distribution.The Jiuzhaigou earthquake occurred in 2017 is selected as a case study to introduce the application of the improved Newmark model in the rapid susceptibility assessment of coseismic landslides,and assess the model performance based on the different sampling strategies.The results show that the average AUC value for different sampling methods is 0.70,and frequency ratio of very high susceptibility is 128.4,which contain 5%area of study area and 59.6%landslides,indicating the model has good performance.The proposed model is driven by specific parameters,thus does not rely on the historical landslides,which differ from data-driven model.The model can be used for rapid landslide assessment after an earthquake or the risk modeling in the given seismic scenario for early site selection of engineering.

关 键 词：地震滑坡 Newmark模型 不确定性模拟 蒙特卡罗 

分 类 号：P642.23[天文地球—工程地质学]