On the exploitation of ETAD data to filter out atmospheric artifacts from Sentinel-1 medium-resolution DInSAR products: a performance evaluation analysis

Differential Synthetic Aperture Radar Interferometry (DInSAR) products are often contaminated by atmospheric contributions, frequently referred to as atmospheric artifacts or atmospheric phase screen (APS) signals [1]. Specifically, the velocity and, consequently, the path length of the radar signal propagating through the inhomogeneous atmosphere layers can vary among repeat-pass SAR acquisitions. Therefore, these variations can be erroneously interpreted as due to deformation signals. Accordingly, filtering out the APS component from DInSAR products, for correctly retrieving deformation measurements, represents a challenging issue.

In this work we present a statistical analysis to investigate the performance of the Sentinel-1 (S1) Extended Timing Annotation Dataset (ETAD) data for mitigating the APS component in both DInSAR interferograms and deformation time-series. The ETAD product consists of different correction layers that provide the azimuth and range timing shifts for each S1 burst to achieve a precise geolocation with centimetre accuracy [2]. It is worth noting that, although the ETAD dataset is not primarily designed for the interferometric phase filtering, some of its correction layers, specifically, the tropospheric, the ionospheric, and the solid Earth tidal displacement ones, may be effectively used to filter the APS signal component affecting the DInSAR measurements [2].

For the presented experimental analisys we used 104 S1 images acquired along ascending orbits over Central/Southern Italy, during the 2018-2020 time-span. The exploited DInSAR products have been generated at medium spatial resolution (about 40 m) and are represented by 278 interferograms and by the corresponding displacement time series retrieved through the P-SBAS processing chain [3].

To quantitively evaluate the ETAD APS correction performance, we analyzed first their impact on DInSAR interferograms. In particular, we considered the following statistical metrics for the entire dataset of unwrapped interferograms, before and after the application of the ETAD APS correction: i) standard deviation of the interferograms at different spatial scales and ii) empirical variograms of the unwrapped phase fitted with a parametric exponential model. Subsequently, to investigate the impact of the ETAD APS correction on the deformation time-series, we analyzed the behavior of their variance before and after the application of the ETAD correction.

The measured metrics results, although still preliminary, demonstrate the effectiveness of the ETAD APS corrections for both the DInSAR interferograms and the deformation time-series in more than 90% of the cases.

[1]  I. Zinno et al., "On The Exploitation of ETAD Data for the Atmospheric Phase Screen Filtering of Medium/High Resolution DInSAR Products," IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Pasadena, CA, USA, 2023, pp. 7882-7885.

[2]  C. Gisinger et al., "The Extended Timing Annotation Dataset for Sentinel-1Product Description and First Evaluation Results,"  IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1-22, 2022, Art no. 5232622

[3]  M. Manunta et al., "The parallel SBAS approach for Sentinel-1 interferometric wide swath deformation time-series generation: Algorithm description and products quality assessment," IEEE Transactions  on Geoscience and Remote Sensing, vol. 57, no. 9, pp. 6259-6281, 2019