LANDSLIDE RISK ASSESSMENT OF GONGLIU COUNTY IN XINJIANG BASED ON MULTIPLE COMBINATION MODELS
-
摘要: 新疆巩留县广泛发育冻融降雨型滑坡地质灾害,对其现有的研究多考虑降水,而缺乏温度影响的研究,为此,本文特增加了温度因子来进行巩留县滑坡灾害危险性评价。基于巩留县已发生的682个滑坡灾害点,选取坡度、起伏度、坡向、曲率、温度、距断层距离、距河流距离、距道路距离、工程地质岩组等9个评价因子。采用信息量模型(I)、确定性系数模型(CF)、信息量模型+逻辑回归模型(I+LR)以及确定性系数模型+逻辑回归模型(CF+LR)等4种模型对巩留县滑坡危险性进行了评价,划分为极高、高、中和低4个危险等级分区并进行了精度检验与现场实际验证。结果表明:(1)温度对滑坡有较大的触发作用;(2)耦合模型极高、高危险性分区面积明显低于单一模型极高、高危险性分区面积,其中CF+LR模型的极高、高危险性分区面积最小,低危险性分区面积最大;(3)4种模型ROC精度检验AUC值分别为0.889、0.893、0.895和0.900,均能较为客观地评价巩留县滑坡危险性。CF+LR模型精度最高,且经局部地区现场检验,CF+LR模型评价结果与实际情况也最为相符,研究成果对新疆地区巩留县滑坡地质灾害的预防和治理具有一定的借鉴意义。Abstract: Freeze-thaw rainfall landslide geological disasters are widely developed in Gongliu County, Xinjiang. The existing research in Gongliu County mostly considers the influence of precipitation but lacks on the influence of temperature. Therefore, the temperature factor was added to evaluate the risk of landslide disaster in Gongliu County in this study. Based on the 682 landslide disaster points that occurred in Gongliu County, nine evaluation factors are selected. They are slope angle, undulation, slope direction, curvature, temperature, distance from fault, distance from river, distance from road and engineering geological rock group. The information model(I), the deterministic coefficient model(CF), the information model+logistic regression model(I+LR) and the deterministic coefficient model+logistic regression model(CF+LR)are used to evaluate the landslide risk in Gongliu County. The risk is divided into four risk levels: extremely high, high, medium and low. The accuracy test and field verification are carried out. The results show that: (1)Temperature has a great triggering effect on landslide. (2)The extremely high and high risk zoning area of the coupled model is significantly lower than that of the single model. The extremely high and high risk zoning area of the CF+LR model is the smallest, and the low risk zoning area is the largest. (3)The AUC values of ROC accuracy test of the four models are 0.889, 0.893, 0.895 and 0.9, respectively, which can objectively evaluate the landslide risk of Gongliu County. The results also show that the CF+LR model has the highest accuracy, and the evaluation results of the CF+LR model are most consistent with the actual situation after field inspection in some areas. The research results have certain reference significance for the prevention and control of landslide geological disasters in Gongliu County, Xinjiang.
-
图 2 滑坡分布与影响因子关系
a. 距断层距离;b. 坡度;c. 坡向;d. 曲率;e. 起伏度;f. 工程地质岩组;g. 温度;h. 距河流距离;i. 距道路距离
Figure 2. Relationship between landslide distribution and influence factors: (a) distance from fault; (b) slope; (c) slope direction; (d) curvature; (e) undulation; (f) engineering geological rock group; (g) temperature; (h) distance from river; (i) distance from road
图 3 滑坡危险性评价因子分级图
a. 距断层距离;b. 坡度;c. 坡向;d. 曲率;e. 起伏度;f. 工程地质岩组;g. 温度;h. 距河流距离;i. 距道路距离
Figure 3. Classification chart of landslide risk evaluation factors: (a) distance from fault; (b) slope; (c) slope direction; (d) curvature; (e) undulation; (f) engineering geological rock group; (g) temperature; h. distance from river; (i) distance from road
表 1 评价因子相关性矩阵
Table 1. Correlation matrix of evaluation factors
相关性 温度 距道路距离 距河流距离 起伏度 坡向 坡度 曲率 距断层距离 工程地质岩组 温度 1 0.038 0.144** -0.210** 0.000 -0.077* -0.057 -0.129** 0.018 距道路距离 1 0.245** -0.136** -0.100** 0.069 -0.072 0.023 0.111** 距河流距离 1 -0.099** -0.027 -0.010 -0.014 0.136** 0.249** 起伏度 1 0.062 0.384** 0.055 -0.093* -0.125** 坡向 1 0.009 -0.008 -0.030 0.014 坡度 1 0.017 -0.105** -0.024 曲率 1 -0.055 0.046 距断层距离 1 -0.017 工程地质岩组 1 ** 在0.01级别(双尾),相关性显著;* 在0.05级别(双尾),相关性显著 表 2 评价因子I值和CF值计算结果
Table 2. Calculation results of evaluation factor I value and CF value
评价因子分级 灾害点/个 类别面积/km2 I值 CF值 坡度/(°) 0~8 52 1322.90 -1.438 45 -0.793 91 8~15 86 750.79 -0.368 90 -0.348 41 15~25 207 879.60 0.351 13 0.354 90 25~35 224 681.34 0.685 46 0.594 64 35~45 92 365.63 0.418 04 0.409 50 >45 21 116.86 0.081 40 0.093 70 坡向 北 155 1056.37 -0.121 28 -0.133 86 东北 133 611.86 0.271 72 0.285 17 东 111 491.67 0.309 60 0.319 12 东南 25 298.04 -0.680 46 -0.538 81 南 16 264.24 -1.006 37 -0.675 35 西南 49 334.16 -0.121 92 -0.134 50 西 110 516.33 0.251 62 0.266 63 西北 83 544.47 -0.083 09 -0.094 08 曲率/(°) < -1 101 610.69 -0.001 58 -0.001 90 ~1~1 427 2884.96 -0.112 60 -0.124 99 >1 154 621.49 0.402 72 0.397 31 起伏度/m 0~35 14 1301.35 -2.734 22 -0.945 22 35~70 102 810.47 -0.274 76 -0.274 84 70~105 238 755.47 0.642 82 0.568 33 105~140 228 591.05 0.845 33 0.683 87 140~175 90 408.07 0.286 26 0.298 35 >175 10 250.73 -1.423 91 -0.790 77 距断层距离/m >2000 621 2878.35 -0.164 24 -0.278 32 0~500 23 377.20 0.499 39 0.572 93 500~1000 15 339.94 0.188 47 0.256 51 1000~1500 8 288.82 -0.144 42 -0.353 17 1500~2000 15 232.84 0.322 83 0.367 48 工程地质岩组 块状坚硬花岗岩岩组 49 851.67 -1.057 49 -0.692 52 块状、层状坚硬-较坚硬以砂岩为主的碎屑岩岩组 155 1100.81 -0.162 49 -0.174 55 黄土类土、冰水砾石双层土体 40 172.34 0.337 27 0.343 12 单层结构砾质土体 393 788.64 1.101 38 0.800 13 互层状较坚硬-软弱以砂岩、砾岩、泥岩为主的碎屑岩岩组 45 205.00 0.281 52 0.294 08 黏性土与积砂土、砾石双层土体 0 998.67 0 -1.000 00 温度/(°) ~5.4~2.7 0 481.42 0 -1.000 00 2.7~7.7 3 540.92 -3.396 77 -0.971 91 7.7~12.7 32 523.87 -0.997 62 -0.672 31 12.7~18.3 510 822.57 1.319 86 0.878 32 18.3~22.8 137 1748.36 -0.748 57 -0.571 76 表 3 基于I与CF的巩留县滑坡危险性分区统计表
Table 3. Statistics of landslide risk zoning in Gongliu County based on I and CF
危险性等级 面积/km2 占巩留县面积比例/% 灾害点/处 占灾害总数比例/% 面密度/处·km-2 信息量模型(I) 极高 256.73 6.24 299 43.84 1.1646 高 428.46 10.41 223 32.70 0.5205 中 1004.70 24.40 142 20.82 0.1413 低 2427.25 58.95 18 2.64 0.0074 确定性系数模型(CF) 极高 259.13 6.29 297 43.55 1.1461 高 448.30 10.89 251 36.80 0.5599 中 1033.69 25.11 121 17.74 0.1171 低 2376.02 57.71 13 1.91 0.0055 表 4 评价因子的方差膨胀因子(VIF)计算结果
Table 4. Calculation results of variance inflation factor(VIF) of evaluation factors
评价因子 VIF 评价因子 VIF 评价因子 VIF I模型 CF模型 I模型 CF模型 I模型 CF模型 温度 3.359 1.793 起伏度 2.408 2.015 曲率 1.009 1.058 坡度 1.778 1.723 坡向 1.016 1.054 距断层距离 1.114 1.160 工程地质岩组 2.048 1.682 表 5 评价因子回归系数
Table 5. Regression coefficients of evaluation factors
模型 因子 B S.E. Wals df Sig Exp(B) I+LR 温度 0.384 0.074 26.878 1 0 1.469 起伏度 0.540 0.098 30.331 1 0 1.717 坡度 0.277 0.131 4.496 1 0.034 1.319 距断层距离 0.935 0.138 45.955 1 0 2.548 工程地质岩组 0.621 0.109 32.660 1 0 1.861 常量 -0.368 0.108 11.735 1 0.001 0.692 CF+LR 温度 0.599 0.111 29.240 1 0 1.821 起伏度 0.791 0.171 21.302 1 0 2.206 坡度 0.546 0.181 9.147 1 0.002 1.727 距断层距离 1.258 0.211 35.695 1 0 3.518 工程地质岩组 0.794 0.149 28.363 1 0 2.213 常量 -1.348 0.182 54.663 1 0 0.260 B代表模型中各个因子的回归系数;S.E.为标准误差;Wals为Wald检验统计量;df为自由度;Sig表示显著性 表 6 基于I+LR与CF+LR的巩留县滑坡危险性分区统计表
Table 6. Statistical table of landslide risk zoning in Gongliu County based on I+LR and CF+LR
模型 危险性等级 面积/km2 占巩留县面积比例/% 灾害点/处 占灾害总数比例/% 面密度/处·km-2 I+LR 极高 237.17 5.76 279 40.91 1.1764 高 369.91 8.98 209 30.65 0.5650 中 693.93 16.85 156 22.87 0.2248 低 2816.13 68.40 38 5.57 0.0135 CF+LR 极高 199.06 4.83 250 36.66 1.2559 高 297.21 7.22 215 31.52 0.7234 中 839.99 20.40 200 29.33 0.2381 低 2780.88 67.54 17 2.49 0.0061 表 7 巩留县滑坡危险性分区合理性检验表
Table 7. Gongliu County landslide risk zoning rationality test table
模型 危险区级别 Sai/% Gei/% Rei=Gei/Sai I 低危险区(Ⅰ) 58.95 2.64 0.04 中危险区(Ⅱ) 24.40 20.82 0.85 高危险区(Ⅲ) 10.41 32.70 3.14 极高危险区(Ⅳ) 6.24 43.84 7.03 CF 低危险区(Ⅰ) 57.71 1.91 0.03 中危险区(Ⅱ) 25.11 17.74 0.71 高危险区(Ⅲ) 10.89 36.80 3.38 极高危险区(Ⅳ) 6.29 43.55 6.92 I+LR 低危险区(Ⅰ) 68.40 5.57 0.08 中危险区(Ⅱ) 16.85 22.87 1.36 高危险区(Ⅲ) 8.98 30.65 3.41 极高危险区(Ⅳ) 5.76 40.91 7.10 CF+LR 低危险区(Ⅰ) 67.54 2.49 0.04 中危险区(Ⅱ) 20.40 29.33 1.44 高危险区(Ⅲ) 7.22 31.52 4.37 极高危险区(Ⅳ) 4.83 36.66 7.59 Sai为i等级危险区的面积占整个研究区面积百分比;Gei为落在等级i中的检验点占整个检验样本数量的百分比(i=Ⅰ,Ⅱ,Ⅲ,Ⅳ) -
Cevik E, Topal T. 2003. GIS-based landslide susceptibility mapping for a problematic segment of the natural gas pipeline, Hendek(Turkey)[J]. Environmental Geology, 44 (8): 949-962. doi: 10.1007/s00254-003-0838-6 Du X C. 2020. Hazard assessment of landslide in Dechang based on ArcGIS and SPSS[D]. Mianyang: Southwest University of Science and Technology. Feng H J, Zhou A G, Yu J J, et al. 2016. A comparative study on plum-rain-triggered landslide susceptibility assessment models in West Zhejiang Province[J]. Earth Science, 41 (3): 403-415. Fischer L, Huggel C, Kääb A, et al. 2013. Slope failures and erosion rates on a glaciered high-mountain face under climatic changes[J]. Earth Surface Processes and Landforms, 38 (8): 836-846. doi: 10.1002/esp.3355 Guo G, Chen Y, Li M H, et al. 2013. Statistic relationship between slope gradient and landslide probability in soil slopes around reservoir[J]. Journal of Engineering Geology, 21 (4): 607-612. Huggel C, Salzmann N, Allen S, et al. 2010. Recent and future warm extreme events and high-mountain slope stability[J]. Philosophical Transactions of the the Royal Society, 368(1919): 2345-2459. Klimeš J, Müllerová H, Woitsch J, et al. 2020. Century-long history of rural community landslide risk reduction[J]. International Journal of Disaster Risk Reduction, 51: 101756. doi: 10.1016/j.ijdrr.2020.101756 Li W L, Gong X P, Yu S Y, et al. 2019. Study on the formation mechanism and stability evaluation of Kalatumusuke landslide in Jiergelang Township, Gongliu County, Xinjiang[J]. West-China Exploration Engineering, 1 : 11-12. Liu J, Li S L, Chen T. 2018. Landslide susceptibility assessment based on optimized random forest model[J]. Geomatics and Information Science of Wuhan University, 43 (7): 1085-1091. Liu L Y, Gao H Y. 2023. Landslide susceptibility assessment based on coupling of WOE mode and Logistic regression model[J]. Journal of Engineering Geology, 31 (1): 165-175. Liu X L, Miao C, Tian C S. 2017. Comparative analysis of two methods for assessing hazard of landslide and debris-flow on a regional scale[J]. Journal of Disaster Prevention and Mitigation Engineering, 37 (1): 71-78. Luo D H. 2010. Experiment for loess landside of freezing-thawing[D]. Xi'an: Xi'an University of Science and Technology. Luo L G, Pei X J, Huang R Q, et al. 2021. Landslide susceptibility assessment in Jiuzhaigou scenic area with GIS based on certainty factor and Logistic regression modle[J]. Journal of Engineering Geology, 29 (2): 526-535. Nadim F, Kjekstad O, Peduzzi P, et al. 2006. Global landslide and avalanche hotspots[J]. Landslides 3 (2): 159-174. doi: 10.1007/s10346-006-0036-1 Oztekin B, Topal T. 2005. GIS-based detachment susceptibility analyses of a cut slope in limestone, Rnkara-turkey[J]. Environmental Geology, 49 (1): 124-132. doi: 10.1007/s00254-005-0071-6 Qin X M. 2007. Based on GIS landslide geological disaster hazard evaluation research-taking Gongliu County as example[D]. Urumqi: Xinjiang University. Schicker R, Moon V. 2021. Comparison of bivariate and multivariate statistical approaches in landslide susceptibility mapping at a regional scale[J]. Geomorpology, 161 : 40-57. am. Shen L L, Liu L Y, Xu C, et al. 2016. Multi-model based landslide susceptibility evaluation-illustrated with landslides triggered by Minxian earthquake[J]. Journal of Engineering Geology, 24 (1): 19-28. Shortliffe E H, Buchanan B G. 1975. A model of inexact reasoning in medicine[J]. Mathematical Biosciences, 23 (3): 351-379. Stevenson P C. 1977. An empirical method for the evaluation of relative landslip risk[J]. Bulletin of the International Association of Engineering Geology-Bulletin de l' Association International edeGéologie de l′Ingénieur., 16 (1): 69-72. doi: 10.1007/BF02591451 Tian C S, Liu X L, Wang J. 2016. Geohazard susceptibility assessment based on CF model and Logistic Regression models in Guangdong[J]. Hydrogeology & Engineering Geology, 43 (6): 154-161, 170. van Westen C J. 1997. Statistical landslide hazard analysis[J]. ILWIS, 2 : 73-84. van Westen C J, Rengers N, Terlien M T J. 1997. Prediction of the occurrence of slope instability phenomena through GIS-based hazard zonation[J]. Geol Rundsch, 86 : 404-414. doi: 10.1007/s005310050149 Wang M, Qiao J P. 2010. Application of contribution weight model in regional landslide hazard evaluation[J]. The Chinese Journal of Geological Hazards and Control, 21 (1): 1-6. Yang G, Xu P H, Cao C, et al. 2019. Assessment of reglonal landslide susceptibility based on combined model of certalnty factor method[J]. Journal of Engineering Geology, 27 (5): 1153-1163. Yang Q, Wang G F, Ding W C, et al. 2021. Susceptibility and accuracy evaluation of regional landslide based on multiple hybrid models[J]. Journal of Natural Disasters, 30 (2): 36-51. Yao L. 2020. Risk assessment of geological hazards in the Dadu River Basin based on GIS[D]. Chendu: Chengdu University of Technology. Yesilnacar E, Topal T. 2005. Landslide susceptibility mapping: a comparison of logistic regression and neural networks methods in a medium scale study, Hendek region(Turkey)[J]. Engineering Geology, 79 (3): 251-266. Yin K L, Yan T Z. 1996. Landslide prediction and relevant models[J]. Chinese Journal of Rock Mechanics and Engineering, 15 (1): 1-8. Yu X Y. 2016. Study on the landslide susceptibility evaluation method based on multi-source data and multi-scale analysis[D]. Wuhan: China University of Geosciences. Yue Y. 2008. Rockfall hazard assessment using a GIS-based Method[D]. Shanghai: Tongji University. Zhan H Q. 2016. GIS-based geological hazard susceptibility assessment along Guizhou section of Sichuan-Guizhou Railway[D]. Beijing: China University of Geosciences. Zhang M D. 2021. Fuzzy logic approach for regional landslide susceptibility analysis constrained by spatial characteristics of environmental factors[D]. Chengdu: Southwest Jiaotong University. Zhang X D, Liu X N, Zhao Z P, et al. 2018. Comparative study of geological hazards susceptibility assessment, constraints from the information value Logistic Regression model and the CF+Logistic Regression model[J]. Geoscience, 32 (3): 602-610. 杜晓晨. 2020. 基于ArcGIS和SPSS的德昌县滑坡危险性评价研究[D]. 绵阳: 西南科技大学. 冯杭建, 周爱国, 俞剑君, 等. 2016. 浙西梅雨滑坡易发性评价模型对比[J]. 地球科学, 41 (3): 403-415. https://www.cnki.com.cn/Article/CJFDTOTAL-DQKX201603006.htm 郭果, 陈筠, 李明惠, 等. 2013. 土质滑坡发育概率与坡度间关系研究[J]. 工程地质学报, 21 (4): 607-612. http://www.gcdz.org/article/id/11324 李文亮, 弓小平, 俞帅一, 等. 2019. 新疆巩留县吉尔格郎乡喀拉吐木苏克滑坡形成机制及稳定性评价研究[J]. 西部探矿工程, 1 : 11-12. https://www.cnki.com.cn/Article/CJFDTOTAL-XBTK201901006.htm 刘坚, 李树林, 陈涛. 2018. 基于优化随机森林模型的滑坡易发性评价[J]. 武汉大学学报·信息科学版, 43 (7): 1085-1091. https://www.cnki.com.cn/Article/CJFDTOTAL-WHCH201807017.htm 刘璐瑶, 高惠瑛. 2023. 基于证据权与Logistic回归模型耦合的滑坡易发性评价[J]. 工程地质学报, 31 (1): 165-175. doi: 10.13544/j.cnki.jeg.2020-482 刘希林, 庙成, 田春山. 2017. 区域滑坡和泥石流灾害两种危险性评价方法的比较分析[J]. 防灾减灾工程学报, 37 (1): 71-78. https://www.cnki.com.cn/Article/CJFDTOTAL-DZXK201701012.htm 罗东海. 2010. 冻融期黄土滑坡实验研究[D]. 西安: 西安科技大学. 罗路广, 裴向军, 黄润秋, 等. 2021. GIS支持下CF与Logistic回归模型耦合的九寨沟景区滑坡易发性评价[J]. 工程地质学报, 29 (2): 526-535. doi: 10.13544/j.cnki.jeg.2019-202 秦晓敏. 2007. 基于GIS的滑坡地质灾害危险性评价研究——以巩留县为例[D]. 乌鲁木齐: 新疆大学. 沈玲玲, 刘连友, 许冲, 等. 2016. 基于多模型的滑坡易发性评价——以甘肃岷县地震滑坡为例[J]. 工程地质学报, 24 (1): 19-28. doi: 10.13544/j.cnki.jeg.2016.01.003 田春山, 刘希林, 汪佳. 2016. 基于CF和Logistic回归模型的广东省地质灾害易发性评价[J]. 水文地质工程地质, 43 (6): 154-161, 170. https://www.cnki.com.cn/Article/CJFDTOTAL-SWDG201606025.htm 王萌, 乔建平. 2010. 贡献权重模型在区域滑坡危险性评价中的应用[J]. 中国地质灾害与防治学报, 21 (1): 1-6. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGDH201001002.htm 杨光, 徐佩华, 曹琛, 等. 2019. 基于确定性系数组合模型的区域滑坡敏感性评价[J]. 工程地质学报, 27 (5): 1153-1163. doi: 10.13544/j.cnki.jeg.2019018 杨强, 王高峰, 丁伟翠等, 2021. 多种组合模型的区域滑坡易发性及精度评价[J]. 自然灾害学报, 30 (2): 36-51. https://www.cnki.com.cn/Article/CJFDTOTAL-ZRZH202102004.htm 姚李. 2020. 基于GIS的大渡河流域地质灾害危险性评价[D]. 成都: 成都理工大学. 殷坤龙, 晏同珍. 1996. 滑坡预测及相关模型[J]. 岩石力学与工程学报, 15 (1): 1-8. https://www.cnki.com.cn/Article/CJFDTOTAL-YSLX601.000.htm 于宪煜. 2016. 基于多源数据和多尺度分析的滑坡易发性评价方法研究[D]. 武汉: 中国地质大学. 岳跃. 2008. 基于GIS的崩塌地质灾害危险性评价研究[D]. 上海: 同济大学. 詹华勤. 2016. 基于GIS川黔铁路贵州段地质灾害易发性评价研究[D]. 北京: 中国地质大学. 张曼迪. 2021. 环境因子空间特征约束的区域滑坡敏感性模糊逻辑分析方法[D]. 成都: 西南交通大学. 张晓东, 刘湘南, 赵志鹏, 等. 2018. 信息量模型, 确定性系数模型与逻辑回归模型组合评价地质灾害敏感性的对比研究[J]. 现代地质, 32 (3): 602-610. https://www.cnki.com.cn/Article/CJFDTOTAL-XDDZ201803018.htm -