Detection of glaciers displacement time-series using SAR

Glaciers are sensitive indicators of climate change. Particularly, glacier surface velocity constitutes a key parameter for estimating ice volume variations as response to global warming and its incidence in sea level rise. Several methodologies based in remotely sensed data have been employed for estimating ice velocity fields. They are mostly based in cross-correlating pairs of images in order to track features displacement between two dates. High ice flux velocity, which can reach more than 1 km/year, constitute a challenge for the existing methodologies, in practice limiting to a few days the time span between useful data. In this work we present an extension of the known Pixel Offset - Small Baseline Subsets (PO-SBAS) technique, that profit a set of successive Synthetic Aperture Radar (SAR) scenes for computing displacement time series and ice velocity fields. The algorithm is guided by a preliminary ice velocity model estimated from the data itself, which significantly improves the results reliability and reduces the overall computational cost. Furthermore, it implements a processing scheme that considers the displacement estimations (PO) quality in order to decide which pixels are included in the time-series inversion. The proposed technique is applied to 22 COSMO-Skymed SAR images of Viedma Glacier (Southern Patagonian Icefield, Argentina) spanning roughly a year. The results obtained are robust and make profit of the whole available dataset. Resulting mean velocity field and displacement time series show the algorithm suitability for retrieving and characterizing complex ice motion patterns.