滑坡灾害的发生对人类生活造成极大影响，对区域进行合理、有效的滑坡灾害易发性评价能够为防灾减灾提供参考。山阳县位于陕西南部山区，地形复杂、地质条件脆弱、降雨量充沛、人类工程活动强烈，导致滑坡灾害频发。基于此，本文对山阳县滑坡灾害易发性评价展开研究，通过收集资料、分析研究区背景、进行实验建模，利用ArcGIS技术、R语言、Matlab语言以及SPSS软件进行山阳县滑坡灾害易发性评价，主要研究内容和成果如下：

（1）分析研究区背景资料和地质图件，选取高程、坡度、坡向、曲率、降水、植被指数、水系、道路、断层、地层岩性、地形地貌和土地利用类型12类评价因子，并利用ArcGIS软件制作评价因子图层。通过统计各评价因子不同状态分级下的分级面积、分级比、灾害点数量、灾害比以及灾害点密度，对选取的指标因子进行相关性分析与多重共线检核，剔除曲率因子与土地利用类型因子，对剩余10类评价因子建立易发性评价指标体系。

（2）分别选取频率比模型(Frequency Ratio Model，FRM)、逻辑回归模型(Logistic Regression Model，LRM)、BP神经网络模型(Back PropagationNeural Network Model，BPNNM)以及随机森林模型(Random Forest Model，RFM）开展了山阳县滑坡易发性评价研究。将建模得到的各模型易发性指数导入ArcGIS中按照自然间断法划分为不易发区、低易发区、中易发区、高易发区和极高易发区五个易发等级，最后得出了山阳县不同模型下的易发性等级分区图。根据各模型分区结果得到：基于频率比、逻辑回归、BP神经网络、随机森林模型预测所得的高—极高易发区面积分别占山阳县总面积的22.61%、30.60%、29.38%、30.96%；滑坡灾害点占总灾害点数的：57.36%、72.36%、74.38%、93.93%，极高易发区的灾害点密度分别为：3.36个/km²、3.71个/km²、3.17个/km²、6.28个/km²。结果表明，分区结果与滑坡灾害点实际位置分布一致，评价结果与实际情况相符。

（3）对四种模型的ROC曲线精度进行验证，结果显示：频率比模型的训练集正确率和验证集预测率分别为78.9%和82.6%、逻辑回归模型的训练集正确率和验证集预测率分别为81.3%和85.3%、BP神经网络模型的训练集正确率和验证集预测率分别为89.4%和90.5%、随机森林模型的训练集正确率和验证集预测率分别为98.2%和97.6%，四种模型均能较好的进行预测，随机森林模型较其余三种模型性能更优，更适用于研究区的滑坡易发性评价。

The occurrence of landslide disasters has a great impact on human life, and a reasonable and effective assessment of the susceptibility assessment of landslide disasters in a region can provide a reference for disaster prevention and mitigation. Shanyang County is located in the mountainous area of southern Shaanxi Province, with complex terrain, fragile geological conditions, abundant rainfall, and intense human engineering activities, resulting in frequent lanslide disasters. Based on this, this paper studies the landslide susceptibility evaluation in Shanyang County, by collecting data, analyzing the background of the research area, and conducting experimental modeling, ArcGIS technology, R language, Matlab language and SPSS software are used to evaluate the susceptibility of Shanyang County. The main research contents and results are as follows:

(1) Analyze the background data and geological maps of the study area, select 12 evaluation factors of elevation, slope, aspect, curvature, precipitation, vegetation index, river, road, fault, stratigraphic lithology, topography and land use type, and use ArcGIS software Make an evaluation factor layer. By calculating the classification area, classification ratio, number of disaster points, disaster ratio and disaster point density under different status classifications of each evaluation factor, correlation analysis and multi-collinearity check are carried out on the selected index factors, and the curvature factor and land use type are eliminated. factor, and establish a susceptibility evaluation index system for the remaining 10 types of evaluation factors.

(2) The Frequency Ratio Model (FRM), logistic Regression Model (LRM), BP Neural Network Model (BPNNM) and Random Forest Model (RFM) were selected to evaluate the susceptibility of landslides in Shanyang County. The susceptibility index of each model obtained by modeling was imported into ArcGIS and divided into five susceptibility levels: non-susceptible area, low-susceptibility area, medium-susceptibility area, high-susceptibility area and extremely high-susceptibility area according to the natural discontinuity method. Zoning map of susceptibility grades under different models in Shanyang County. According to the results of each model partition, the areas of high-high-prone areas predicted by frequency ratio, logistic regression, BP neural network, and random forest model account for 22.61%, 30.60%, 29.38%, and 30.96% of the total area of Shanyang County, respectively; Landslide disaster points account for 57.36%, 72.36%, 74.38%, and 93.93% of the total disaster points. The density of disaster points in the extremely high-prone areas are: 3.36/km², 3.71/km², 3.17/km², 6.28/km² . The results show that the zoning results are consistent with the actual location distribution of landslide disaster points, and the evaluation results are consistent with the actual situation.

 (3) The ROC curve accuracy of four models are verified, and results showed that the the training set accuracy and validation set prediction rate of the frequency ratio model were 78.9% and 82.6%, the training set accuracy rate and the verification set prediction rate of the logistic regression model were 81.3% and 85.3%, the training set accuracy and validation set prediction rate of the BP neural network model were 89.4% and 90.5%, and the training set accuracy rate and the verification set prediction rate of the random forest model were 98.2% and 97.6%. The four models can make better predictions, and the random forest model has better performance than the other three models, which is more suitable for landslide susceptibility evaluation in the study area.

[1] 李媛, 孟晖, 董颖, 胡树娥. 中国地质灾害类型及其特征——基于全国县市地质 灾害调查成果分析[J]. 中国地质灾害与防治学报, 2004(02): 32-37.

[2] 陈晓岚, 吴晓宾. 重庆市綦江县地质灾害风险评价[J]. 山西建筑, 2010, 36(03): 135-137.

[3] 全国地质灾害防治“十三五”规划印发[J]. 地质装备, 2017, 18(02): 3-4.

[4] 中华人民共和国国土资源部. 全国地质灾害通报[R]. 北京: 中国地质环境监测院, 2020.

[5] 郭子正, 殷坤龙, 付圣, 黄发明, 桂蕾, 夏辉. 基于GIS与WOE-BP模型的滑坡易发性评价[J]. 地球科学, 2019, 44(12): 4299-4312.

[6] 郑苗苗. 黄土高原陕甘宁地区地质灾害数据库建设与危险性评价[D]. 长安大学, 2017.

[7] 谭景峰. 关中盆地结构及三维地质建模[D]. 西安石油大学, 2017.

[8] 郭帅. 陕西秦巴山区地质灾害特征分析及敏感性评价[D]. 西北大学, 2020.

[9] 郭有金. 基于集成学习算法的西安市滑坡灾害易发性评价[D]. 西安科技大学, 2020.

[10] 李利峰, 张晓虎, 邓慧琳, 韩六平. 基于SVM-LR融合模型的滑坡灾害易发性评价——以山阳县为例[J]. 科学技术与工程, 2020, 20(26): 10618-10625.

[11] 吴伟. 山阳县降雨诱发堆积层滑坡形成演化机理研究[D]. 长安大学, 2019.

[12] 潘奕婷. 陕西省山阳县“8·12”突发山体滑坡[J]. 中华灾害救援医学, 2015, 3(09): 476.

[13] 倪晓娇, 南颖. 基于GIS的长白山地区地质灾害风险综合评估[J]. 自然灾害学报, 2014, 23(01): 112-120.

[14] 李福建, 马安青, 丁原东, 焦俊超, 刘乐军. 基于RS与GIS技术的地质灾害危险性评价——以青岛市崂山区为例[J]. 中国海洋大学学报(自然科学版), 2010, 40(06): 47-52+152.

[15] 陈玉, 郭华东, 王钦军. 基于RS与GIS的芦山地震地质灾害敏感性评价[J]. 科学通报, 2013, 58(36): 3859-3866.

[16] 霍艾迪, 张骏, 卢玉东, 成玉祥, 姚以亮. 地质灾害易发性评价单元划分方法——以陕西省黄陵县为例[J]. 吉林大学学报(地球科学版), 2011, 41(02): 523-528+535.

[17] Romstad B, Etzelmuller B. Mean-curvature watersheds: A simple method for segmentation of a digital elevation model into terrain units[J]. Geomorphology, 2012, 139: 293-302.

[18] Calvello M, Cascini L, Mastroianni S. Landslide zoning over large areas from a sample inventory by means of scale-dependent terrain units[J]. Geomorphology, 2013, 182(JAN.15): 33-48.

[19] He A, Pan P, Dai L, et al. Application of kernel-based Fisher discriminant analysis to map landslide susceptibility in the Qinggan River delta, Three Gorges, China[J]. Geomorphology, 2012.

[20] Ohlmacher G C. Plan curvature and landslide probability in regions dominated by earth flows and earth slides[J]. Engineering Geology, 2007, 91(2-4): 117-134.

[21] J Minár, Evans I S. Elementary forms for land surface segmentation: The theoretical basis of terrain analysis and geomorphological mapping[J]. Geomorphology, 2008, 95(3-4): 236-259.

[22] Ding H, Jia-Ming N A, Huang X L, et al. Stability analysis unit and spatial distribution pattern of the terrain texture in the northern Shaanxi Loess Plateau[J]. 山地科学学报：英文版, 2018, 15(3): 13.

[23] 郑玲静, 李秀珍, 徐瑞池. 基于斜坡单元的区域滑坡敏感性评价 ——以云南省小江流域为例. 科学技术与工程, 2021, 21(28): 12322-12329.

[24] Ba Q, Chen Y, Deng S, et al. A comparison of slope units and grid cells as mapping units for landslide susceptibility assessment[J]. Earth Science Informatics, 2018.

[25] 唐川, 马国超. 基于地貌单元的小区域地质灾害易发性分区方法研究[J]. 地理科学, 2015(1): 8.

[26] Van Den Eeckhaut, M., Reichenbach, P., Guzzetti, F., Rossi, M., & Poesen, J. (2009). Combined landslide inventory and susceptibility assessment based on different mapping units: an example from the Flemish Ardennes, Belgium. Natural Hazards and Earth System Sciences, 9(2), 507–521.

[27] 王志旺. 基于GIS技术的区域滑坡分形特征分析与危险性评价[D]. 中国地质大学, 2010.

[28] 郑迎凯, 陈建国, 王成彬, 程潭武. 确定性系数与随机森林模型在云南芒市滑坡易发性评价中的应用[J]. 地质科技通报, 2020, 39(6): 14.

[29] 武雪玲, 杨经宇, 牛瑞卿. 一种结合SMOTE和卷积神经网络的滑坡易发性评价方法[J]. 武汉大学学报·信息科学版, 2020, 45(8): 10.

[30] 范强, 巨能攀, 向喜琼, 等. 基于结果验证的信息量法地质灾害易发性评价——以贵州省开阳县为例[J]. 人民长江, 2015, 46(15): 4.

[31] Jebur M N, Pradhan B, Tehrany M S. Optimization of landslide conditioning factors using very high-resolution airborne laser scanning (LiDAR) data at catchment scale[J]. Remote Sensing of Environment, 2014, 152:150-165.

[32] Feizizadeh B, Blaschke T. Landslide Risk Assessment Based on GIS Multi-Criteria Evaluation: A Case Study in Bostan-Abad County, Iran[J]. Journal of Earth ence & Engineering, 2011, 1(1): 67-72.

[33] MARIO MEJíA-NAVARRO, Wohl E E. Geological Hazard and Risk Evaluation Using GIS: Methodology and Model Applied to Medellín, Colombia[J]. Environmental & Engineering Geoscience, 1994.

[34] Beaty C B. Landslides and Slope Exposure[J]. Journal of Geology, 1956, 64(1): 70-74.

[35] 毛华锐, 孙小飞, 周颖智. 基于频率比-投影寻踪模型的渝东北三峡库区滑坡敏感性制图[J]. 科学技术与工程, 2022, 22(05): 1803-1813.

[36] 张俊, 殷坤龙, 王佳佳, 等. 三峡库区万州区滑坡灾害易发性评价研究[J]. 岩石力学与工程学报, 2016, 35(2): 13.

[37] 孙琳, 任娜娜, 李云安, 等. 基于证据权法的公路路基岩溶塌陷危险性评价[J]. 中国地质灾害与防治学报, 2019, 30(3): 7.

[38] 杨光, 徐佩华, 曹琛, 等. 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, 2019, 27(5): 11.

[39] 刘阳. 延长县滑坡地质灾害风险评估和管理研究[D]. 长安大学, 2009.

[40] 曹璞源, 胡胜, 邱海军, 等. 基于模糊层次分析的西安市地质灾害危险性评价[J]. 干旱区资源与环境, 2017, 31(8): 7.

[41] 胡致远, 罗文强, 晏鄂川, 等. 基于改进层次分析法的英山县地质灾害易发性评价[J]. 安全与环境工程, 2018, 25(4): 6.

[42] Pourghasemi H R, Pradhan B, Gokceoglu C. Application of fuzzy logic and analytical hierarchy process (AHP) to landslide susceptibility mapping at Haraz watershed, Iran[J]. Natural Hazards, 2012, 63(2): 965-996.

[43] Zhang T Y, Mao Z A, Wang T. GIS-based evaluation of landslide susceptibility using a novel hybrid computational intelligence model on different mapping units[J]. 山地科学学报·英文版, 2020, 17(12): 13.

[44] Guo-Liang D U, Zhang Y S, Javed I, et al. Landslide susceptibility mapping using an integrated model of information value method and logistic regression in the Bailongjiang watershed, Gansu Province, China[J]. Journal of Mountain Science, 2017(14): 249-268.

[45] Youssef A M, Pourghasemi H R, Pourtaghi Z S, et al. Landslide susceptibility mapping using random forest, boosted regression tree, classification and regression tree, and general linear models and comparison of their performance at Wadi Tayyah Basin, Asir Region, Saudi Arabia[J]. Landslides, 2016, 13(5): 839-856.

[46] Chen, Wei, Xie, et al. A comparative study of logistic model tree, random forest, and classification and regression tree models for spatial prediction of landslide susceptibility[J]. Catena Giessen Then Amsterdam, 2017.

[47] 刘坚, 李树林, 陈涛. 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, 2018, 43(7): 7.

[48] 李远远, 梅红波, 任晓杰, 等. 基于确定性系数和支持向量机的地质灾害易发性评价[J]. 地球信息科学学报, 2018, 20(12): 11.

[49] Kavzoglu T, Sahin E K, Colkesen I. Landslide susceptibility mapping using GIS-based multi-criteria decision analysis, support vector machines, and logistic regression[J]. Landslides, 2014, 11(3): 425-439.

[50] Pradhan B, Youssef A M, Varathrajoo R. Approaches for delineating landslide hazard areas using different training sites in an advanced artificial neural network model[J]. 地球空间信息科学学报（英文）, 2010(2):10.

[51] 吴常润, 角媛梅, 王金亮, 等.基于频率比-逻辑回归耦合模型的双柏县滑坡易发性评价[J]. 自然灾害学报, 2021, 30(4): 12.

[52] 陈飞, 蔡超, 李小双, 等. 基于信息量与神经网络模型的滑坡易发性评价[J]. 岩石力学与工程学报, 2020, 39(S01): 12.

[53] 王珂, 郭长宝, 马施民, 等. 基于证据权模型的川西鲜水河断裂带滑坡易发性评价[J]. 现代地质, 2016, 30(3): 11.

[54] 刘艳芳, 方佳琳, 陈晓慧, 等. 基于确定性系数分析方法的秭归县滑坡易发性评价[J]. 自然灾害学报, 2014(6): 9.

[55] Pham B T, Pradhan B, Bui D T, et al. A comparative study of different machine learning methods for landslide susceptibility assessment: A case study of Uttarakhand area (India)[J]. Environmental Modelling & Software, 2016, 84(OCT.): 240-250.

[56] 姜琪文. 基于ArcGIS的区域滑坡危险性评价[D]. 成都理工大学, 2005.

[57] 谷天峰, 王家鼎, 付新平. 基于斜坡单元的区域斜坡稳定性评价方法[J]. 地理科学, 2013, 33(11).

[58] 马啸. 府谷县崩-滑地质灾害发育特征及其易发性评价[D]. 西安科技大学, 2021.

[59] 崔阳阳. 基于不同评价单元的滑坡易发性评价方法研究[D]. 西安科技大学, 2021.

[60] 唐晓娜. 基于卷积神经网络和综合指数模型的吕梁市滑坡灾害易发性评价[D]. 太原理工大学, 2019.

[61] 刘睿, 施婌娴, 孙德亮, 许嘉慧. 基于GIS与随机森林的巫山县滑坡易发性区划[J]. 重庆师范大学学报(自然科学版), 2020, 37(03): 86-96.

[62] 王念秦, 朱文博, 郭有金. 基于PSO-SVM模型的滑坡易发性评价[J]. 长江科学院院报, 2021, 38(04): 56-62.

[63] Chen X, Chen W. GIS-based landslide susceptibility assessment using optimized hybrid machine learning methods[J]. Catena, 2021, 196: 104833.

[64] 李树兴. 山阳县地质灾害特征及风险评价研究[D]. 长安大学, 2021.

[65] 武斌, 王帅帅. 频率比法在地质灾害危险性评估的应用[J]. 中国科技博览, 2015(44): 1.

[66] Rasyid A R, Bhandary N P, Yatabe R. Performance of frequency ratio and logistic regression model in creating GIS based landslides susceptibility map at Lompobattang Mountain, Indonesia[J]. Geoenvironmental Disasters, 2016, 3(1): 19.

[67] 胡涛, 樊鑫, 王硕, 等. 基于逻辑回归模型和3S技术的思南县滑坡易发性评价[J].地质科技通报,2020,39(2):9.

[68] Liu R, Li L, Pirasteh S, et al. The performance quality of LR, SVM, and RF for earthquake-induced landslides susceptibility mapping incorporating remote sensing imagery[J]. Arabian Journal of Geosciences, 2021, 14(14): 259.

[69] Sun D, Xu J, Wen H, et al. Assessment of landslide susceptibility mapping based on Bayesian hyperparameter optimization: A comparison between logistic regression and random forest[J]. Engineering Geology, 2020, 281: 105972.

[70] TANG R, YAN E, CAI J, et al. Back analysis of initial ground stress based on back-propagating neural network[J]. Electronic Journal of Geotechnical Engineering, 2013, 18: 5839–5856.

[71] 张世睿,李心科, ZHANG, 等. 基于模拟退火的BP网络隐藏层节点估算算法[J]. 合肥工业大学学报: 自然科学版, 2017, 40(11): 4.

[72] Breiman L. Bagging Predictors[J]. Machine Learning, 1996.

[73] Ho T K. The random subspace method for constructing decision forests[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 1998, 20(8): 832-844.

[74] Lu W, Zhou Z, Liu T, et al. Discrete Element Simulation Analysis of Rock Slope Stability Based on UDEC[J]. Advanced Materials Research, 2012, 461: 384-388.

[75] 曹正凤. 随机森林算法优化研究[D]. 首都经济贸易大学.

[76] 商洛市政府. 商洛市人民政府办公室关于印发商洛市地质灾害防治十四五规划的通知[EB/OL]. http://www.shangluo.gov.cn/info/1687/99337.htm, 2021-11-12.