The APS filtering approach of the Parallel Small BAseline Subset DInSAR processing chain: methodology and performance analysis

Atmospheric Phase Screen (APS) is one of the main error sources in the DInSAR measurements retrieval procedures. In this work, we present a performance analysis carried out on the multi-step atmospheric filtering approach integrated into the Parallel Small BAseline Subset (P-SBAS) DInSAR processing chain, which makes use of both ECMWF ERA-5 data and interferometric data-driven techniques, to filter out the APS signal from the retrieved displacement time series. The proposed approach combines diverse filtering techniques to estimate the different APS contributions. In particular, as a first step for the estimation and removal of the topography-related atmospheric phase component, we compare the effectiveness of two alternative solutions. The former uses the quasi-linear phase-elevation relationship to estimate the APS stratified component from the DInSAR data. The latter makes use of the ERA-5 data, which are particularly effective in mitigating the atmospheric contributions correlated with the height. Then, we analyze the impact of an iterative spatial filtering step used to estimate the spatially-correlated atmospheric components at different spatial scales. Finally, we investigate the effectiveness of the last temporal filtering step, allowing us to mitigate the residual high-frequency atmospheric signals.We show the results of the experimental analysis, based on the processing of a large dataset of 403 Sentinel-1 images acquired along ascending orbits during the 2016-2024 time-span over the Central/Southern Italy through the P-SBAS approach, aimed at investigating, on a statistical basis, the performance of each step of the atmospheric filtering procedure and the related achieved accuracy.