Multibaseline Interferometry Based on Independent Component Analysis and InSAR Combinatorial Modeling for High-Precision DEM Reconstruction

Authors:
Publisher website
Google Scholar
Abstract
Digital elevation models (DEMs) are essential for national economic development, disaster management, and military applications. Multi baseline interferometric synthetic aperture radar (MB-InSAR) technology has proven to be an effective method for DEM reconstruction. However, the presence of atmospheric noise and other residual signals introduces unavoidable errors in the phase observations, and most MB-InSAR DEMs are generated using a single empirical mathematical model that ignores the influence of deformation factors. To compensate for these limitations, we propose spatial independent component analysis (sICA) phase separation and interferometric synthetic aperture radar (InSAR) combinatorial modeling (CM) InSAR CM (ISCM). The sICA was used for phase separation, resulting in clear InSAR signals and reducing atmospheric noise and other residual signal interference; then, the effects of linear deformation, seasonal deformation, and environmental factors were considered in the InSAR modeling. In the experiments, a total of 19 TerraSAR-X images from San Diego, USA (SD), and 18 PAZ images from Yan’an, China (YA), were selected to generate DEMs with resolutions of 3 and 6 m, respectively. The accuracy of the DEM generated by ISCM was evaluated using the photogrammetric DEM, and the root-mean-square errors (RMSEs) of the elevation are 3.20 m for SD and 4.41 m for YA, with an improvement of 30.8%–44.9% and 21.9%–38.4%, respectively, compared to the traditional MB-InSAR method. In addition, ICESat/GLAS data collected in YA were used for further validation with an improvement of 13.7%–29.5%. The DEM generated by ISCM has significant advantages in improving accuracy and preserving terrain features, providing theoretical support for global high-precision DEM mapping.
Funding Information
  • National Natural Science Foundation of China (42471456)
  • National Key Research and Development Program of China (2022YFB3902300)
  • Fundamental Research Funds for the Central Universities, China (2042022dx0001)

This publication has 48 references indexed in Scilit: