结合InSAR与ICA-VMD-LSTM的南汀河活动断裂地表形变监测与预测

Surface Deformation Monitoring and Prediction of Active Rupture of Nanting River by Combining InSAR and ICA-VMD-LSTM

ES评分 0

DOI 10.14075/j.jgg.2025.04.114
刊名
Journal of Geodesy and Geodynamics
年,卷(期) 2026, 46(3)
作者
作者单位

1 昆明理工大学国土资源工程学院, 昆明, 650093;
2 中交基础设施养护集团有限公司工程勘察设计分公司, 北京, 100010;
3 昆明理工大学公共安全与应急管理学院, 昆明, 650093

摘要
采用SBAS-InSAR技术处理2019—2023年覆盖南汀河活动断裂带与南云高速公路交错区域的123景升轨Sentinel-1影像,获取交错区域的形变速率图,并结合独立成分分析(ICA)和遥感影像得到地表形变特征。选取断裂带与公路交错位置A1~A3及大跨度桥梁位置B作为典型点位,构建ICA-VMD-LSTM形变预测模型。结果表明,研究区大部分区域形变速率在±5 mm/a之间,主要形变模式为突发性形变、周期性形变与长期线性形变。相较于现有的VMD-LSTM模型,ICA-VMD-LSTM模型具有更全面的特征提取能力以及更强的抗噪能力,可以有效提高预测精度。总体来看,ICA-VMD-LSTM模型在MAE和RMSE方面分别平均降低约33.2%和32.4%,R2平均提高23.3%,表明其在误差控制和趋势拟合方面具有明显优势。
Abstract
This study employs the SBAS-InSAR technique to process 123 ascending Sentinel-1 images covering the intersected area of the active Nantinghe fault zone and the Nanyun expressway from 2019 to 2023, obtaining the deformation rate map of the intersected area. Combined with independent component analysis (ICA) and remote sensing images, the characteristics of surface deformation are derived. Subsequently, typical points including the intersection positions A1 to A3 of the fault zone and the expressway, as well as the large-span bridge position B, are selected to construct the ICA-VMD-LSTM deformation prediction model. The results indicate that the majority of the study area exhibits deformation rates within ±5 mm/a. The primary deformation patterns include sudden events, seasonal periodic deformation, and long-term linear trends. Compared with the conventional VMD-LSTM model, the proposed ICA-VMD-LSTM model demonstrates superior feature extraction capability and enhanced robustness to noise, thereby improving predictive accuracy. On average, the ICA-VMD-LSTM model achieved reductions of approximately 33.2% and 32.4% in MAE and RMSE, respectively, and an increase of 23.3% in R2, indicating significant improvements in error control and trend representation.
关键词
SBAS-InSAR; 独立成分分析(ICA); 变分模态分解(VMD); LSTM; 断裂带; 形变监测与预测
KeyWord
SBAS-InSAR; independent component analysis (ICA); variational mode decomposition (VMD); LSTM; fracture zone; deformation monitoring and prediction
基金项目
页码 349-359
  • 参考文献
  • 相关文献
  • 引用本文

张万福, 李素敏, 孙天文, 吴长全, 陈青, 聂晗. 结合InSAR与ICA-VMD-LSTM的南汀河活动断裂地表形变监测与预测 [J]. 大地测量与地球动力学. 2026; 46; (3). 349 - 359.

  • 文献评论

相关学者

相关机构