Cui Zhou / Central South University of Forestry and Technology
Zhiwei Li / Central South University
It is important to update digital elevation model (DEM) products in time for ensuring the performance of application of DEM products. However, it is still challenging to accurately and frequently update large-scale DEMs using the existing satellite photogrammetry and/or interferometric synthetic aperture radar (InSAR) techniques so far. In this study we proposed a new algorithm for frequently updating large-scale DEMs from multi-track repeat-pass interferograms using a robust variance component estimation (RVCE). Firstly, a new quantitative strategy for guiding the scientific selection of available multi-track interferograms based on the inflection curvature of the error propagation function of phase errors into DEM estimation was presented. Then, a functional model involving DEM residuals (i.e., including surface relief changes and original DEM errors) and the unwrapped phases of the selected multi-track repeat-pass interferograms is constructed. Thirdly, a RVCE estimator was then utilized to determine the stochastic model of the heterogeneous multi-track unwrapped phases, with the main aim at reducing the propagation of random errors and outliers in multi-track unwrapped phases into DEM residual estimates. By adding the estimated DEM residuals to the original DEM product, one can obtain an updated DEM. Finally, the proposed algorithm was tested over Hambach open-pit mine of Germany, where surface relief changes quietly due to open-pit mining and dumping activities. In which, Sentinel-1 interferograms from four tracks over the Hambach open-pit mine were used to update the DEM products every three months. The results show that the mean accuracy of the updated DEMs is about 8.7 m, indicating an improvement by 60% than the updated DEM using the existing single-track repeat-pass InSAR techniques. Owing to the booming of repeat-pass InSAR satellites, the proposed algorithm offers a new view and tool for frequently update large-scale (e.g., global) DEMs, especially over those areas with extensive relief changes.