西吉县地处宁夏回族自治区南部山区，地质环境条件复杂，滑坡数量多、规模大，阻碍当地的经济发展，因此对西吉县开展滑坡易发性研究具有重要的实际意义。以西吉县滑坡为研究对象，在分析研究区地质环境条件、滑坡发育特征及分布规律的基础上，应用ArcGIS、WEKA和SPSS等软件，开展了不同滑坡表征方法下（滑坡点、缓冲圆和滑坡周界）基于不同机器学习模型（LR、NB和RBF Network模型）的滑坡易发性评价研究，并进行了模型适用性及评价精度分析，取得主要成果如下：

（1）基于研究区地质环境条件，分析了西吉县滑坡发育类型、发育特征、分布规律及影响因素。结果表明：境内滑坡以大中型为主，受降雨、地震及人类工程活动作用明显；主要分布于地质环境条件复杂的黄土丘陵区，在黄土覆盖较厚的区域呈滑坡群（链）分布，阴坡滑坡发育数量多于阳坡，具有明显的时空分布规律；

（2）遵循系统性、独立性、可操作性及继承性原则，结合指标相关性和多重共线性分析，对初步选定的评价指标进行筛选和分析，最终选定包括高程、坡度、坡向、平面曲率、剖面曲率、距道路距离、距河流距离、距断层距离、降雨量、土地利用类型、岩土体类型和地震动峰值加速度等12个对滑坡影响程度显著的指标，建立了研究区的滑坡易发性评价指标体系，应用频率比法分析了滑坡与各指标之间的关系；

（3）分别建立了基于上述三种表征方法的LR、NB和RBF Network评价模型，得到了相应的滑坡易发性评价结果图。经统计分析不同易发等级区间内的频率比值，表明：三种表征方法在上述三种模型下均得到了较好的分区结果，符合合理性检验标准；

（4）分别采用ROC曲线、频率比精度及Kappa系数对不同表征方法下三种模型评价结果的预测精度和一致性检验进行了对比分析，得到：研究区最优表征方法与评价模型的组合为基于滑坡周界的NB模型。该组合成功率曲线AUC=0.965，预测率曲线AUC=0.886，FRA=0.873，k=0.710，预测精度与一致性检验均达到了较高水平，评价结果与实地调查情况吻合度高。

Xiji County is located in the mountainous region in the south of Ningxia Hui Autonomous Region, with complex geological and environmental conditions and a large number and scale of landslides, which hinder the local economic development, so it is of great practical significance to carry out a study on landslide susceptibility in Xiji County. Based on the analysis of the geological and environmental conditions, landslide development characteristics and distribution patterns in the study area, a study on landslide susceptibility evaluation based on different machine learning models (LR, NB and RBF Network models) under different landslide characterization methods (Point, Circle and Polygon) was carried out using ArcGIS, WEKA and SPSS software, taking the landslides in Xiji County as the research object. The analysis of the applicability and accuracy of the models was carried out and the main results are as follows:

(1) Based on the geological and environmental conditions of the study area, the types of landslide development, development characteristics, distribution patterns and influencing factors in Xiji County were analyzed. The results show that the landslides in the territory are mainly large and medium-sized, and are significantly affected by rainfall, earthquakes and human engineering activities. They are mainly distributed in the loess hilly areas with complex geological and environmental conditions, and are distributed in groups (chains) of landslides in areas with thick loess coverage, with more landslides developing on shady slopes than on sunny slopes, and have obvious spatial and temporal distribution patterns.

(2) Following the principles of systematicity, independence, operability and inheritance, combined with the correlation of indicators and multiple covariance analysis, the initially selected evaluation indicators were screened and analyzed, and 12 indicators were finally selected, including elevation, slope, slope aspect, plane curvature, profile curvature, distance to roads, distance to rivers, distance to faults, rainfall, land use, lithology and peak acceleration of ground motion, which have significant influence on landslides, and established the evaluation indicator system of landslide susceptibility in the study area, and applied the frequency ratio method to analyze the relationship between landslides and each indicator.

(3) The LR, NB and RBF Network evaluation models based on the above three characterization methods were established respectively, and the corresponding landslide susceptibility evaluation result maps were obtained. The statistical analysis of the frequency ratios within different susceptibility intervals showed that: the three characterization methods obtained good zoning results under the above three models and met the reasonableness test.

(4) The ROC curve, Frequency Ratio Accuracy and Kappa coefficient were used to compare and analyse the prediction accuracy and consistency test of the evaluation results of the three models under different characterization methods, and it was obtained that: the combination of the optimal characterization method and evaluation model in the study area is the NB model based on landslide perimeter. The success rate curve AUC=0.965, the prediction rate curve AUC=0.886, FRA=0.873, k=0.710. The prediction accuracy and consistency test of this combination have reached a high level, and the evaluation results are in good agreement with the field survey situation.

[1] 金帅. 基于3S技术及多模型集成的滑坡易发性评价 [D]. 廊坊: 防灾科技学院, 2021.

[2] 赵国君, 赵祺彬, 申文金. 废弃矿山地质环境治理模式探讨 [J]. 资源与产业, 2018, 20(5): 77-82.

[3] 施紫越, 朱海燕, 王晶菁, 等. 耦合模型视角下的湘西州土质滑坡易发性探讨 [J]. 水土保持研究, 2021, 28: 377-383.

[4] Roccati A, Paliaga G, Luino F, et al. GIS-based landslide susceptibility mapping for land use planning and risk assessment [J]. Land, 2021, 10(2): 162-189.

[5] Huang F M, Yan J, Fan X M, et al. Uncertainty pattern in landslide susceptibility prediction modelling: Effects of different landslide boundaries and spatial shape expressions [J]. Geoscience Frontiers, 2022, 13(02): 101317.

[6] 李巍林. 基于AHP-信息量法的柳林县地质灾害风险评价及风险管控研究 [D]. 石家庄: 河北地质大学, 2022.

[7] 尚慧, 倪万魁, 刘海松. 宁夏回族自治区西吉县地震诱发型黄土滑坡发育特征与分布规律 [J]. 水土保持通报, 2012, 32(5): 28-31.

[8] 林荣福. 基于优化支持向量机模型的滑坡易发性评价 [D]. 阜新: 辽宁工程技术大学, 2021.

[9] Wang X J, Huang F M, Fan X M, et al. Landslide susceptibility modeling based on remote sensing data and data mining techniques [J]. Environmental Earth Sciences, 2022, 81(2): 1-19.

[10] Sun D L, Xu J H, Wen H J, et al. Assessment of landslide susceptibility mapping based on Bayesian hyperparameter optimization: A comparison between logistic regression and random forest [J]. Engineering Geology, 2021, 281: 105972-105983.

[11] 辛存林, 施紫越, 任文秀, 等. 甘肃天水市北山地质灾害危险度区划 [J]. 兰州大学学报(自然科学版), 2020, 56: 16-24.

[12] Tang Y M, Feng F, Guo Z Z, et al. Integrating principal component analysis with statistically-based models for analysis of causal factors and landslide susceptibility mapping: A comparative study from the loess plateau area in Shanxi (China) [J]. Journal of Cleaner Production, 2020, 277: 124159.

[13] Chae B G, Park H J, Catani F, et al. Landslide prediction, monitoring and early warning: a concise review of state-of-the-art [J]. Geosciences Journal, 2017, 21(06): 1033-1070.

[14] Chen W, Pourghasemi H R, Panahi M, et al. Spatial prediction of landslide susceptibility using an adaptive neuro-fuzzy inference system combined with frequency ratio, generalized additive model, and support vector machine techniques [J]. Geomorphology, 2017, 297: 69-85.

[15] Martha T R, Roy P, Jain N, et al. Geospatial landslide inventory of India—an insight into occurrence and exposure on a national scale [J]. Landslides, 2021, 18(6): 2125-2141.

[16] Zhu L, Huang L, Fan L, et al. Landslide susceptibility prediction modeling based on remote sensing and a novel deep learning algorithm of a cascade-parallel recurrent neural network [J]. Sensors, 2020, 20(6): 1576.

[17] Fang G H, Yuan T, Zhang Y, et al. Integrated study on soil erosion using RUSLE and GIS in Yangtze River Basin of Jiangsu Province (China) [J]. Arabian Journal of Geosciences, 2019, 12(5): 173.

[18] Süzen M L, Doyuran V. Data driven bivariate landslide susceptibility assessment using geographical information systems: a method and application to Asarsuyu catchment, Turkey [J]. Engineering Geology, 2004, 71(3-4): 303-321.

[19] Nefeslioglu H A, Gokceoglu C, Sonmez H, et al. Medium-scale hazard mapping for shallow landslide initiation: the Buyukkoy catchment area (Cayeli, Rize, Turkey) [J]. Landslides, 2011, 8(4): 459-483.

[20] 黄发明, 胡松雁, 闫学涯. 基于机器学习的滑坡易发性预测建模及其主控因子识别 [J]. 地质科技通报, 2022, 41(2): 79-90.

[21] Dagdelenler G, Nefeslioglu H A, Gokceoglu C. Modification of seed cell sampling strategy for landslide susceptibility mapping: an application from the Eastern part of the Gallipoli Peninsula (Canakkale, Turkey) [J]. Bulletin of Engineering Geology and the Environment, 2015, 75(2): 575-590.

[22] 黄发明, 曹昱, 范宣梅. 不同滑坡边界及其空间形状对滑坡易发性预测不确定性的影响规律 [J]. 岩石力学与工程学报, 2021, 40(S2): 3227-3240.

[23] Beaty C B. Landslides and slope exposure [J]. The Journal of Geology, 1956, 64(1): 70-74.

[24] Wang K, Zhang S J, Delgadotéllez R, et al. A new slope unit extraction method for regional landslide analysis based on morphological image analysis [J]. Bulletin of Engineering Geology and the Environment, 2018, 78(6): 4139-4151.

[25] 许国庆, 周宇. 基于地貌单元的小区地质灾害易发性分区方法研究 [J]. 世界有色金属, 2017, (11): 137-138.

[26] Ba Q Q, Chen Y M, Deng S S, et al. A comparison of slope units and grid cells as mapping units for landslide susceptibility assessment [J]. Earth Science Informatics, 2018, 11(3): 373-388.

[27] 张曦, 陈丽霞, 徐勇. 两种斜坡单元划分方法对滑坡灾害易发性评价的对比研究 [J]. 安全与环境工程, 2018, 25(1): 12-17+50.

[28] Deng N D, Li Y X, Ma J Q, et al. A comparative study for landslide susceptibility assessment using machine learning algorithms based on grid unit and slope unit [J]. 2022, 10: 2188.

[29] 曹文霞. 基于组合赋权法的交口县地质灾害易发性评价研究 [D]. 太原: 太原理工大学, 2021.

[30] Tien Bui D, Tuan T A, Klempe H, et al. Spatial prediction models for shallow landslide hazards: a comparative assessment of the efficacy of support vector machines, artificial neural networks, kernel logistic regression, and logistic model tree [J]. Landslides, 2015, 13(2): 361-378.

[31] Hong H Y, Pradhan B, Xu C, et al. Spatial prediction of landslide hazard at the Yihuang area (China) using two-class kernel logistic regression, alternating decision tree and support vector machines [J]. Catena, 2015, 133: 266-281.

[32] Yoo Y, Baek T, Kim J, et al. A Comparative Study of the Frequency Ratio and Evidential Belief Function Models for Landslide Susceptibility Mapping [J]. Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, 2016, 34: 597-607.

[33] 沈玲玲, 刘连友, 许冲. 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, 2016, 24(1): 19-28.

[34] Tien Bui D, Shahabi H, Shirzadi A, et al. Landslide detection and susceptibility mapping by AIRSAR data using support vector machine and index of entropy models in Cameron Highlands, Malaysia [J]. Remote Sensing, 2018, 10(10): 1527.

[35] 张向营, 张春山, 孟华君. 基于Random Forest和AHP的贵德县北部山区滑坡危险性评价 [J]. 水文地质工程地质, 2018, 45(4): 142-149.

[36] Nguyen T, Liu C C. A new approach using AHP to generate landslide susceptibility maps in the Chen-Yu-Lan Watershed, Taiwan [J]. Sensors, 2019, 19(3): 505.

[37] Chen W, Chen Y Z, Tsangaratos P, et al. Combining evolutionary algorithms and machine learning models in landslide susceptibility assessments [J]. Remote Sensing, 2020, 12(23): 3854.

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

[39] 刘增源. 安康市汉滨区滑坡敏感性评价研究 [D]. 西安: 西安科技大学, 2021.

[40] 刘阳, 尚慧, 占惠珠, 等. 评价单元对地质灾害易发性评价的影响 [J]. 科学技术与工程, 2022, 22: 15536-15545.

[41] Chiessi V, Toti S, Vitale V. Landslide susceptibility assessment using conditional analysis and rare events logistics regression: A case-study in the Antrodoco area (Rieti, Italy) [J]. Geoscience and Environment Protection, 2016, 4(12): 1.

[42] 田述军. 基于不同评价单元的滑坡易发性评价对比研究 [J]. 自然灾害学报, 2019, 28(6): 137-145.

[43] 刘任鸿, 李明辉, 邓英尔. 基于GIS的华蓥市地质灾害易发性评价 [J]. 沉积与特提斯地质, 2021, 41(1): 129-136.

[44] Dou J, Yunus A P, Bui D T, et al. Assessment of advanced random forest and decision tree algorithms for modeling rainfall-induced landslide susceptibility in the Izu-Oshima Volcanic Island, Japan [J]. Science of the Total Environment, 2019, 662: 332-346.

[45] Chen W, Chen X, Peng J B, et al. Landslide susceptibility modeling based on ANFIS with teaching-learning-based optimization and Satin bowerbird optimizer [J]. Geoscience Frontiers, 2021, 12(1): 93-107.

[46] Xiao L M, Zhang Y H, Peng G Z. Landslide susceptibility assessment using integrated deep learning algorithm along the China-Nepal highway [J]. Sensors, 2018, 18(12): 4436.

[47] Chen C Y, Chang J M. Landslide dam formation susceptibility analysis based on geomorphic features [J]. Landslides, 2016, 13(5): 1019-1033.

[48] Chen W, Chai H C, Zhao Z, et al. Landslide susceptibility mapping based on GIS and support vector machine models for the Qianyang County, China [J]. Environmental Earth Sciences, 2016, 75(6): 472-486.

[49] 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: 西安科技大学, 2021.

[50] 王秀英, 聂高众, 王登伟. 汶川地震诱发滑坡与地震动峰值加速度对应关系研究 [J]. 岩石力学与工程学报, 2010, 29(1): 82-89.

[51] Booth G D, Niccolucci M J, Schuster E G, et al. Identifying proxy sets in multiple linear-regression - an aid to better coefficient interpretation [J]. Journal of Applied Physiology, 1994: 470-476.

[52] Hair J F, Black B, Babin B J, et al. Multivariate data analysis [M]. Multivariate Data Analysis, 2011.

[53] Liao D, Valliant R. Variance inflation factors in the analysis of complex survey data [J]. Survey methodology, 2012, 38(1): 53-62.

[54] Berhane G, Kebede M, Alfarah N, et al. Landslide susceptibility zonation mapping using GIS-based frequency ratio model with multi-class spatial data-sets in the Adwa-Adigrat mountain chains, northern Ethiopia [J]. Journal of African Earth Sciences, 2020, 164: 103795.

[55] Van Westen C J, Seijmonsbergen A C, Mantovani F. Comparing Landslide Hazard Maps [J]. Natural Hazards, 1999, 20(2): 137-158.

[56] Moosavi V, Talebi A, Shirmohammadi B. Producing a landslide inventory map using pixel-based and object-oriented approaches optimized by Taguchi method [J]. Geomorphology, 2014, 204: 646-656.

[57] Kalantar B, Pradhan B, Naghibi S A, et al. Assessment of the effects of training data selection on the landslide susceptibility mapping: a comparison between support vector machine (SVM), logistic regression (LR) and artificial neural networks (ANN) [J]. Geomatics, Natural Hazards and Risk, 2017, 9(1): 49-69.

[58] Oh H J, Kadavi P R, Lee C W, et al. Evaluation of landslide susceptibility mapping by evidential belief function, logistic regression and support vector machine models [J]. Geomatics, Natural Hazards and Risk, 2018, 9(1): 1053-1070.

[59] Pham B T, Tien Bui D, Pourghasemi H R, et al. Landslide susceptibility assesssment in the Uttarakhand area (India) using GIS: a comparison study of prediction capability of naïve bayes, multilayer perceptron neural networks, and functional trees methods [J]. Theoretical and Applied Climatology, 2015, 128(1-2): 255-273.

[60] Chen W, Yan X S, Zhao Z, et al. Spatial prediction of landslide susceptibility using data mining-based kernel logistic regression, naive bayes and RBFNetwork models for the Long County area (China) [J]. Bulletin of Engineering Geology and the Environment, 2018, 78(1): 247-266.

[61] Lei X X, Chen W, Pham B T. Performance evaluation of GIS-based artificial intelligence approaches for landslide susceptibility modeling and spatial patterns analysis [J]. 2020, 9(7): 443.

[62] 韩明. 概率论与数理统计教程(统计类) [M]. 上海: 同济大学出版社, 2014.

[63] Chen W, Xie X S, Wang J L. A comparative study of logistic model tree, random forest, and classification and regression tree models for spatial prediction of landslide susceptibility [J]. Catena, 2016, 151: 147-160.

[64] 汪云云, 陈松灿. 基于AUC的分类器评价和设计综述 [J]. 模式识别与人工智能, 2011, 24: 64-71.

[65] 尚慧. 宁南山区地质灾害形成机理研究 [D]. 西安: 长安大学, 2010.