赵金彪 傅贵 崔玉龙

摘 要：【目的】伊犁地区以滑坡为主的地质灾害时有发生，给当地带来了人员伤亡和财产损失，灾害防治形势严峻，需要对当地的滑坡易发性展开研究。【方法】以新疆伊犁州某黄土区为研究区，选取高程、坡度、坡向、坡位、距道路距离、距河流距离、距断层距离、多年平均降雨量、NDVI等9个因子作为滑坡影响因子，建立滑坡易发性评价指标体系，分析因子与滑坡的分布规律。运用逻辑回归模型对研究区滑坡灾害进行易发性评价，并通过ROC曲线对评价结果进行验证。【结果】结果表明：研究区滑坡分布于高程1 400～2 000 m、坡度10°～30°、坡向为东方向、坡位为中坡和上坡、距断层0～1 000 m、NDVI指数为0.7～0.8、降雨量为350～360 mm、距河流距离小于300 m的地区；高易发区和极高易发区主要分布在皮里青河和吉尔格朗河之间的一片低山丘陵区域；逻辑回归模型精度较高，ROC曲线的AUC值均在80%以上。【结论】研究获取的滑坡分布规律和易发性评价结果，可为该区域的滑坡防灾减灾和工程建设规划等提供参考。

关键词：黄土滑坡；易发性评价；逻辑回归；伊犁某区域

中图分类号：P642.22     文献标志码：A     文章编号：1003-5168（2024）08-0109-05

DOI：10.19968/j.cnki.hnkj.1003-5168.2024.08.021

Evaluation of Loess Landslide Susceptibility in a Certain Area of Yili Based on Logistic Regression Model

ZHAO Jinbiao1 FU Gui2 CUI Yulong2

（1.Ruicheng Civil Engineering Design Consulting Co.， Ltd.， Hefei 230088， China; 2. School of Civil Engineering and Architecture， Anhui University of Science and Technology， Huainan 232001， China）

Abstract： [Purposes] The frequent occurrence of landslide-based geological disasters in Yili area has brought serious casualties and property losses to the local area， and the situation of disaster prevention and control is very serious. [Methods] In this paper， a loess area in Yili Prefecture， Xinjiang is taken as the research area. Nine factors， including elevation， slope， slope direction， slope position， distance from road， distance from river， distance from fault， average annual rainfall and NDVI， are selected as landslide impact factors. The evaluation index system of landslide susceptibility is established， and the distribution law of factors and landslides is analyzed. The logistic regression model is used to evaluate the susceptibility of landslide disasters in the study area. Finally， the evaluation results are verified by ROC curve. [Findings] The results show that the landslide in the study area is distributed in the area with an elevation of 1 400～2 000 m， a slope of 10°～30°， an eastward slope， a middle slope and an upper slope， a distance from the fault of 0～1 000 m， a NDVI index of 0.7～0.8， a rainfall of 350～360 mm， and a distance from the river less than 300 m. High-prone areas and extremely high-prone areas are mainly distributed in a low hilly area between the Piliqing River and the Gilgerang River; the accuracy of logistic regression model is high， and the AUC value of ROC curve is more than 80 %. [Conclusions] The landslide distribution law and susceptibility evaluation results obtained in this paper can provide reference for landslide disaster prevention and mitigation and engineering construction planning in this area.

Keywords： loess landslide; susceptibilityevaluation; logistic regression; a certain area of Yili

0 引言

我国西部地区滑坡发生较为频繁［1-2］。滑坡灾害对人民的生命财产安全具有极大的威胁，如2008年5月12日，汶川地震触发滑坡，直接造成约2万人死亡［3］；2003年三峡库区千将坪滑坡造成数千人受灾，直接损失为5 735万元［4］。为减少滑坡灾害带来的威胁，许多学者对滑坡的易发性展开研究。如冯杭建等［5］以东南沿海地区浙江省淳安县为研究区，在野外实地调查的基础上，开展滑坡易发性评价。……