Mapping SAR geometric distortions and their stability along time: a new tool in Google Earth Engine based on Sentinel-1 image time series

Operational services based on SAR data from satellite missions are showing to have the potentialities of becoming a real scenario; nevertheless, the complexity of data pre-processing remains one of the main reasons for its slow uptake by a wider user community. Google Earth Engine (GEE) web-based platform allows an immediate access to SAR imagery (namely, Sentinel-1 - S1) making users able to directly focus on the expected application. SAR side-looking acquisition mode generates many geometric distortions within recorded images, especially in mountain areas, determining a different degree of reliability of deductions. Consequently, a mapping of these areas is desirable for a correct interpretation of derived information. In this work a trigonometry-based method for mapping was implemented in GEE. With reference to a time series made of 60 S1 images covering the whole Piemonte Region (NW Italy) in 2020, some maps of distortions were generated using the 30 m gridded SRTM DTM as topographic surface descriptor. S1 images, belonging to the analyzed time series, were acquired from both ascending and descending orbits. In particular, active/passive shadows, active/passive layover and foreshortening masks were computed and mapped. Distortion maps were finally intersected with land cover classes to test the correspondent degree of analysability by SAR data. The results show that such methodology can be proficiently used to mask unreliable observations, making possible to a priori be informed about the areas of a given territory that can be reasonably and reliably monitored by SAR data.