基于固定秩克里金的SBAS-InSAR大气快速改正方法

A Fast Atmospheric Correction Method of SBAS-InSAR based on Fixed Rank Kriging

  • 摘要: 在合成孔径雷达干涉测量( Interferometric Synthetic Aperture Radar, InSAR)技术监测地表形变过程中,大气延迟误差制约着形变监测结果的精度。传统小基线集 InSAR( Small Baseline Subsets,SBAS-InSAR)技术通常利用干涉相位与高程关系的统计分析和时空滤波等策略抑制大气效应,但前者仅能去除垂直分层大气,忽略了湍流大气的影响,后者对滤波窗口的选取则具有较强的主观性。另一方面, SAR影像中百万乃至千万数量级的数据点为计算过程带来了不小的负担。为了提高时序 InSAR技术的形变监测精度并保证高效的计算效率,提出了基于固定秩克里金( Fixed Rank Kriging, FRK)和干涉图自身的 SBAS-InSAR大气快速改正方法,通过构建多尺度空间模型估计干涉图大气随机效应,并利用空间降维大幅提高了大气改正的效率。选用覆盖山西省大同市的 25景 Sentinel-1 SAR影像验证了该方法的有效性:结果表明: 1)区域内真实 GNSS数据验证结果表明,相较传统 SBAS-InSAR方法,新方法的 RMSE从 0.83cm降低至 0.43cm; 2)在相同数据量下,基于固定秩克里金的大气改正效率是普通克里金的 15倍。

     

    Abstract: Objectives: In the process of monitoring surface deformation using Interferometric Synthetic Aperture Radar (InSAR) technology, the atmospheric delay error constrains the accuracy of the deformation monitoring results. The traditional small baseline subsets InSAR (SBAS-InSAR) technology usually utilizes the statistical analysis of the interferometric phase-height relationship and spatio-temporal filtering to suppress the atmospheric effect, but the former only removes the vertically stratified atmosphere and ignores the effect of turbulent atmosphere, while the latter is highly subjective in the selection of the filtering window. On the other hand, the millions or even tens of millions of data points in SAR images bring a considerable burden to the calculation process. Methods: In order to improve the deformation monitoring accuracy of time-series InSAR technology and ensure efficient calculation efficiency, a SBAS-InSAR atmospheric rapid correction method based on Fixed Rank Kriging (FRK) and the interferogram itself is proposed. Starting from spatial correlation, a multi-scale spatial model is constructed to estimate the atmospheric random effects of the interferogram, and spatial dimensionality reduction is used to greatly improve the efficiency of atmospheric correction. Finally, a real experiment was conducted using 25 scenes of Sentinel-1 SAR images covering Datong City, Shanxi Province, to verify the effectiveness and performance of this method. Results: The results show that: (1) Compared with the deformation results obtained by traditional SBAS-InSAR, the root mean square error (RMSE) of this method is reduced by 48% on average; (2) The atmospheric estimation efficiency of the fixed-rank kriging method is 15 times higher than that of the ordinary kriging method with the same amount of data. Conclusions: The proposed method can significantly improve the deformation monitoring accuracy of the SBAS-InSAR technique without relying on external data, and greatly improve the efficiency of atmospheric correction.

     

/

返回文章
返回