耦合时序InSAR与XGBoost模型的渝东南滑坡识别及易发区分布研究

Research of integrated temporal InSAR and XGBoost model for landslide identification and distribution of susceptible areas in southeastern Chongqing

摘要

为实现渝东南高山峡谷滑坡灾害早期防治，本文提出融合合成孔径雷达干涉(interferometric synthetic aperture radar，InSAR)技术与极致梯度提升(eXtreme gradient boosting，XGBoost)模型开展研究区滑坡隐患广域识别与易发性评价研究。首先，采取时序小基线集(small baseline subset，SBAS)-InSAR技术获取地表形变速率，继而依据卫星飞行参数计算地表可视性与形变监测敏感性，联合形变速率阈值与局部地形坡度构建约束条件初步筛选异常形变区域，并借助光学影像精准识别滑坡隐患。同时引入InSAR形变结果作为动态评价因子与其他静态评价因子共同运用XGBoost模型开展滑坡灾害易发性评价，与滑坡隐患早期识别结果互为补充，综合划定地表不稳定区。研究结果表明，基于SBAS-InSAR技术获取研究区监测时段内地表形变速率为-57~5.36 mm/a，结合上述约束条件初步探查出360处显著异常形变区域，经验证后确认其中186处为潜在滑坡隐患。运用XGBoost易发性评价模型显示研究区5类滑坡易发性占比分别为极低(20.5%)、低(34.7%)、中(26.7%)、高(10.4%)、极高(7.7%)，受试者工作特性(receiver operating characteristic，ROC)曲线的曲线下面积(area under curve，AUC)值为0.886，表明模型具有良好的预测性能，综合滑坡隐患识别结果共划定79处地表不稳定区，可为当地滑坡灾害防治提供科学参考。

Abstract

To achieve early prevention and control of landslide disasters in alpine canyon regions of southeastern Chongqing, this paper proposed an integrated approach combining interferometric synthetic aperture radar (InSAR) with eXtreme gradient boosting (XGBoost) models for identifying wide-area potential landslide hazards and evaluating their susceptibility in the study area. First, time-series small baseline subset InSAR (SBAS-InSAR) analysis was employed to acquire surface deformation rates. Then, satellite orbital parameters were used to calculate surface visibility and deformation monitoring sensitivity, and constraint criteria combining deformation rate thresholds and local slope gradients were established to preliminarily screen anomalous deformation zones. Finally, optical images were used to accurately identify potential landslide hazards. Furthermore, InSAR-derived deformation results were introduced as a dynamic evaluation factor, working together with other conventional static evaluation factors to conduct landslide susceptibility assessment using the XGBoost model, which complements the early identification of landslide hazards, and the combined results are used to delineate surface instability zones. The results show that the SBAS-InSAR technique revealed surface deformation rates ranging from -57 to 5.36 mm/a during the monitoring period. Based on the constraint criteria, 360 areas of significant abnormal deformation were initially identified, and 186 of these were confirmed as potential landslide hazards. The XGBoost susceptibility assessment model showed that the susceptibility rates for the five landslide categories in the study area were extremely low (20.5%), low (34.7%), medium (26.7%), high (10.4%), and extremely high (7.7%), respectively. The area under curve (AUC) of the receiver operating characteristic (ROC) curve was 0.886, indicating good predictive performance. Based on the comprehensive landslide hazard identification results, 79 unstable surface zones were delineated, providing a scientific reference for local landslide disaster prevention and control.

关键词

渝东南 / 滑坡隐患 / SBAS-InSAR / XGBoost模型 / 易发性评价

Key words

southeastern Chongqing / landslide disasters / SBAS-InSAR / XGBoost model / susceptibility assessment

参考文献

[1] Dai Keren, Li Zhenhong, Tomás R, et al. Monitoring activity at the Daguangbao mega-landslide (China) using Sentinel-1 TOPS time series interferometry[J]. Remote Sensing of Environment, 2016, 186: 501-513.

[2] 国家统计局. 2024年中国统计年鉴[M]. 北京: 中国统计出版社, 2024.

[3] National Bureau of Statistics. China statistical yearbook, 2024[M]. Beijing: China Statistics Press, 2024. (in Chinese)

[4] 韩子夜, 薛星桥. 地质灾害监测技术现状与发展趋势[J]. 中国地质灾害与防治学报, 2005, 16(3): 138-141.

[5] Han Ziye, Xue Xingqiao. Status and development trend of monitoring technology for geological hazards[J]. The Chinese Journal of Geological Hazard and Control, 2005, 16(3): 138-141. (in Chinese)

[6] Fruneau B, Achache J, Delacourt C. Observation and modelling of the Saint-Étienne-de-Tinée landslide using SAR interferometry[J]. Tectonophysics, 1996, 265(3/4): 181-190.

[7] 许强, 董秀军, 李为乐. 基于天-空-地一体化的重大地质灾害隐患早期识别与监测预警[J]. 武汉大学学报(信息科学版), 2019, 44(7): 957-966.

[8] Xu Qiang, Dong Xiujun, Li Weile. Integrated space-air-ground early detection, monitoring and warning system for potential catastrophic geohazards[J]. Geomatics and Information Science of Wuhan University, 2019, 44(7): 957-966. (in Chinese)

[9] 王洪明, 师芸, 平继伟, 等 . 蓄泄水期水电站流域滑坡灾害识别与易发性评价[J]. 自然灾害学报, 2025, 34(1): 85-94.

[10] Wang Hongming, Shi Yun, Ping Jiwei, et al. Identification and susceptibility evaluation of landslide hazard in drainage period of hydropower station basin[J]. Journal of Natural Disasters, 2025, 34(1): 85-94. (in Chinese)

[11] 廖军, 吴彩燕, 王立娟, 等 . 基于时序InSAR的滑坡早期识别: 以贵州省水城县为例[J]. 自然灾害学报, 2022, 31(3): 251-260.

[12] Liao Jun, Wu Caiyan, Wang Lijuan, et al. Early detection of landslides based on time series InSAR technologies: a case study of Shuicheng County of Guizhou Province[J]. Journal of Natural Disasters, 2022, 31(3): 251-260. (in Chinese)

[13] 张路, 廖明生, 董杰, 等 . 基于时间序列InSAR分析的西部山区滑坡灾害隐患早期识别: 以四川丹巴为例[J]. 武汉大学学报(信息科学版), 2018, 43(12): 2039-2049.

[14] Zhang Lu, Liao Mingsheng, Dong Jie, et al. Early detection of landslide hazards in mountainous areas of West China using time series SAR interferometry: a case study of Danba, Sichuan[J]. Geomatics and Information Science of Wuhan University, 2018, 43(12): 2039-2049. (in Chinese)

[15] Fialko Y, Simons M, Agnew D. The complete (3-D) surface displacement field in the epicentral area of the 1999 M_W7.1 Hector Mine Earthquake, California, from space geodetic observations [J]. Geophysical Research Letters, 2001, 28(16): 3063-3066.

[16] Notti D, Herrera G, Bianchini S, et al. A methodology for improving landslide PSI data analysis[J]. International Journal of Remote Sensing, 2014, 35(6): 2186-2214.

[17] 古腾飞, 段平, 王伟鹏, 等 . 滑坡易发性评价方法综述[J]. 自然灾害学报, 2024, 33(6): 1-16.

[18] Gu Tengfei, Duan Ping, Wang Weipeng, et al. Review of methods for landslide susceptibility assessment[J]. Journal of Natural Disasters, 2024, 33(6): 1-16. (in Chinese)

[19] 李成林, 刘严松, 赖思翰, 等 . 基于逻辑回归和支持向量机耦合模型的滑坡易发性分析[J]. 自然灾害学报, 2024, 33(2): 75-86.

[20] Li Chenglin, Liu Yansong, Lai Sihan, et al. Landslide susceptibility analysis based on the coupling model of logistic regression and support vector machine[J]. Journal of Natural Disasters, 2024, 33(2): 75-86. (in Chinese)

[21] 黄智杰, 沈佳, 简文彬, 等 . 基于XGBoost模型的降雨诱发阶跃型滑坡位移预测[J]. 自然灾害学报, 2023, 32(2): 217-226.

[22] Huang Zhijie, Shen Jia, Jian Wenbin, et al. Displacement prediction of rainfall-induced step-like landslide based on XGBoost model[J]. Journal of Natural Disasters, 2023, 32(2): 217-226. (in Chinese)

[23] 胡璐宇. 渝东南地区构造特征及其对页岩气富集的影响[D]. 徐州: 中国矿业大学, 2017.

[24] Hu Luyu. Tectonic characteristics and its effect on shale gas enrichment in the regional of southeastern Chongqing[D]. Xuzhou: China University of Mining and Technology, 2017. (in Chinese)

[25] 戴可人, 铁永波, 许强, 等 . 高山峡谷区滑坡灾害隐患InSAR早期识别: 以雅砻江中段为例[J]. 雷达学报, 2020, 9(3): 554-568.

[26] Dai Keren, Tie Yongbo, Xu Qiang, et al. Early identification of potential landslide geohazards in Alpine-canyon terrain based on SAR interferometry: a case study of the middle section of Yalong River[J]. Journal of Radars, 2020, 9(3): 554-568. (in Chinese)

[27] Notti D, Davalillo J C, Herrera G, et al. Assessment of the performance of X-band satellite radar data for landslide mapping and monitoring: upper Tena Valley case study[J]. Natural Hazards and Earth System Sciences, 2010, 10(9): 1865-1875.

[28] Colesanti C, Wasowski J. Investigating landslides with space-borne Synthetic Aperture Radar (SAR) interferometry[J]. Engineering Geology, 2006, 88(3/4): 173-199.

[29] Plank S, Singer J, Minet C, et al. Pre-survey suitability evaluation of the differential synthetic aperture radar interferometry method for landslide monitoring[J]. International Journal of Remote Sensing, 2012, 33(20): 6623-6637.

[30] Wang Yian, Liu Donglei, Dong Jie, et al. On the applicability of satellite SAR interferometry to landslide hazards detection in hilly areas: a case study of Shuicheng, Guizhou in Southwest China[J]. Landslides, 2021, 18(7): 2609-2619.

[31] Berardino P, Fornaro G, Lanari R, et al. A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms[J]. IEEE Transactions on Geoscience and Remote Sensing, 2002, 40(11): 2375-2383.

[32] Dai Keren, Deng Jin, Xu Qiang, et al. Interpretation and sensitivity analysis of the InSAR line of sight displacements in landslide measurements[J]. GI Science & Remote Sensing, 2022, 59(1): 1226-1242.

[33] Chen Tianqi, Guestrin C. XGBoost: A scalable tree boosting system[C] // Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2016: 785-794.

[34] Lin Kaixiong, Jiapaer G, Yu Tao, et al. Identification of potential landslides in the Gaizi Valley section of the Karakorum Highway coupled with TS-InSAR and landslide susceptibility analysis[J]. Remote Sensing, 2024, 16(19): 3653.

基金资助

国家自然科学基金项目(42174045)