基于鲁棒M估计协方差矩阵相位优化的DSInSAR矿区形变监测研究  

Study on deformation monitoring in mine by DSInSAR based on the phase optimization method under robust M-estimator covariance matrix

作　　者：曾祥凯 孙凤娜 陈东兴 段文再 滕文龙 ZENG Xiangkai;SUN Fengna;CHEN Dongxing;DUAN Wenzai;TENG Wenlong(The Second Exploration Team of Shandong Coalfield Geological Bureau,Jining 272000,China;Yangquan High-tech Industrial Development Zone Management Committee,Yangquan 045000,China;School of Environment and Spatial Informatics,China University of Mining and Technology,Xuzhou 221116,China)

机构地区：[1]山东省煤田地质局第二勘探队,山东济宁272000 [2]阳泉高新技术产业开发区管理委员会,山西阳泉045000 [3]中国矿业大学环境与测绘学院,江苏徐州221116

出　　处：《中国矿业》2024年第3期142-151,共10页China Mining Magazine

基　　金：中国博士后科学基金项目资助(编号:2022M723376)。

摘　　要：针对时序干涉合成孔径雷达(Interferometric Synthetic Aperture Radar,InSAR)技术在矿区所处区域内耕地、裸地等自然地表地物上无法识别有效监测点信息,致使矿区地表形变信息不足及形变解译困难等问题,提出一种基于鲁棒M估计协方差矩阵特征值分解相位优化的DSInSAR技术,并基于34景Sentinel-1A影像获取了霄云煤矿2022年的地表时序形变。推导基于鲁棒M估计协方差矩阵估计公式,提高对样本异常值及异质像素的鲁棒性,开展基于鲁棒M估计协方差矩阵特征值分解的相位优化处理,分析了上述优化模型与通用相位优化模型的相似性,最终通过对优化估计相位开展相位信息解译处理来获取最终的形变信息。实验结果表明:基于鲁棒M估计协方差矩阵特征值分解相位优化的DSInSAR技术较常规SBAS技术及基于最大似然估计协方差矩阵的DSInSAR技术在监测点密度上分别提高了约11.4倍、0.2倍,且与水准数据对比具有最小的均方根误差,约22 mm;此外,霄云煤矿共包含三个主要形变场,其地表沉降的时序变化呈现出较显著的非线性趋势,煤矿内地表的视线向最大沉降量约418 mm。研究成果为矿区地表形变规律反演及矿区灾害防控提供重要数据支撑。The time series Interferometric Synthetic Aperture Radar(InSAR)cannot identify effective monitoring point information on natural surface features such as cultivated land and bare land in the mining area,resulting in insufficient surface deformation information and difficulty in deformation interpretation.To address these problems,a distributed scatterers InSAR(DSInSAR)method based on robust M-estimator covariance matrix eigenvalue decomposition is proposed,and the surface time series deformation of Xiaoyun Coal Mine in 2022 is obtained based on 34 Sentinel-1A images.Firstly,a sample covariance matrix based on the M-estimator is derived to improve robustness to sample outliers and heterogeneous pixels.Then,phase optimization processing based on eigenvalue decomposition for this robust M-estimator covariance matrix is performed,and the similarity between the above optimization model and the general phase optimization model is also analyzed.The deformation information is finally obtained by interpreting the phase information based on the optimized estimated phase.The experimental results show that the DSInSAR based on robust M-estimator covariance matrix eigenvalue decomposition phase optimization increased by about 11.4 times and 0.2 times in monitoring point density compared to conventional SBAS and DSInSAR based on maximum likelihood estimator covariance matrix,respectively,and has the smallest root mean square error,about 22 mm,compared to level data.In addition,the Xiaoyun Coal Mine consists of three main deformation fields,and the time series variation of surface subsidence shows a significant nonlinear trend.And the maximum subsidence in line of sight of the surface in the coal mine is about 418 mm.The research results provide important data support for the inversion of surface deformation rules and disaster prevention and control in mining areas.

关 键 词：DSInSAR技术 鲁棒协方差矩阵 特征值分解 相位优化 矿区形变监测 

分 类 号：TD325.4[矿业工程—矿井建设]