Efficient tropospheric correction in national and continental-scale Sentinel-1 InSAR measurements

The availability of regular SAR data acquired by the Sentinel-1 mission provides enormous opportunities for deformation mapping over extensive areas. However, it brings significant technical challenges that need to be properly addressed to extract high-quality deformation products. One of the major challenges in this regard is the tropospheric contribution in the interferometric phase. In this study, we investigate three different approaches for reducing the tropospheric effect from large stacks of Sentinel-1 interferograms and apply them for large-scale displacement mapping in three study areas in Iran, Germany, and Chile.

In the first approach, we use the tropospheric phase delay estimated from a network of GNSS data to correct the interferograms. This approach is evaluated for Germany, where a dense network of GNSS is available. With the current distribution of GNSS stations, we can reduce the tropospheric phase from the interferograms by approximately 60%.

For the other two study areas in Iran and Chile, such dense GNSS as in Germany is not available. Therefore, we use the ECMWF weather model data provided by Generic Atmospheric Correction Online Service for InSAR (GACOS) to correct the individual interferograms for the tropospheric phase. Although this method appears to be successful in many cases, it has difficulties in correcting interferograms with data acquisition times a few hours separated from the temporal intervals of ECMWF data. Therefore, we develop an adaptive method that uses the interferometric phase itself to filter out the topography-correlated phase and broad-scale phase changes as unwanted phases from the interferograms. Because in large-scale interferograms, the spatial variations of tropospheric delay are significant, we split the interferograms into smaller patches to adaptively estimate and remove the unwanted phases. Then, the remaining tropospheric contribution, mainly from the turbulent atmosphere, are reduced through time series analysis.

In each study area, three years of Sentinel-1 data are analyzed using Small Baseline Subset (SBAS) approach. We compare the effectiveness of the three methods in reducing the tropospheric contribution and discuss the potentials and limitations of them in retrieving anthropogenic and tectonic deformation in different study areas.