Spatio-temporal linking of multiple SAR satellite data from medium and high resolution Radarsat-2 images

A recent development in Interferometric Synthetic Aperture Radar (InSAR) technology is integrating multiple SAR satellite data to dynamically extract ground features. This paper addresses two relevant challenges: identification of common ground targets from different SAR datasets in space, and concatenation of time series when dealing with temporal dynamics. To address the first challenge, we describe the geolocation uncertainty of InSAR measurements as a three-dimensional error ellipsoid. The points, among InSAR measurements, which have error ellipsoids with a positive cross volume are identified as tie-point pairs representing common ground objects from multiple SAR datasets. The cross volumes are calculated using Monte Carlo methods and serve as weights to achieve the equivalent deformation time series. To address the second challenge, the deformation time series model for each tie-point pair is estimated using probabilistic methods, where potential deformation models are efficiently tested and evaluated. As an application, we integrated two Radarsat-2 datasets in Standard and Extra-Fine modes to map the subsidence of the west of the Netherlands between 2010 and 2017. We identified 18128 tie-point pairs, 5 intersection types of error ellipsoids, 5 deformation models, and constructed their long-term deformation time series. The detected maximum mean subsidence velocity in Line-Of-Sight direction is up to 15 mm yr(-1). We conclude that our method removes limitations that exist in single-viewing-geometry SAR when integrating multiple SAR data. In particular, the proposed time-series modeling method is useful to achieve a long-term deformation time series of multiple datasets.

Automated summary

This research focuses on addressing two challenges in InSAR technology: identifying common ground targets from different SAR datasets in space and concatenating time series when dealing with temporal dynamics. The study achieved dynamic extraction of ground features from multiple SAR datasets by describing geolocation uncertainty and calculating cross volumes using Monte Carlo methods.