Sentinel-1 InSAR measurements of deformation over discontinuous permafrost terrain, Northern Quebec, Canada

Abstract

Differential synthetic aperture radar interferometry (D-InSAR) has been applied in permafrost environments to detect surface deformation caused by freeze-thaw processes in the active layer and underlying permafrost. The effectiveness of Sentinel-1 InSAR in monitoring ground surface deformation over continuous permafrost terrain above the treeline has been proven. The heterogeneous landscape and developed vegetation cover increase the difficulty of applying D-InSAR in sub-Arctic discontinuous permafrost terrain. The potential of Sentinel-1 InSAR in such an environment has not been fully explored. In this study, we explore the capabilities and limitations of applying Sentinel-1 time series data for monitoring surface deformation over discontinuous permafrost terrain. The interferometric coherence time series from September 2016 to April 2018 were analyzed over typical landscapes in discontinuous permafrost environments and their thaw subsidence curves are revealed by the small-baseline subset (SBAS) InSAR technique. The seasonal thaw subsidence in the summer of 2017 was in the range of 15–80 mm in the study area. The land cover types with thaw subsidence magnitudes from low to high are exposed land, peatland, lichen–low shrub, lichen-dominated and wetland low vegetation. The difference in displacement pattern between lichen-dominated and wetland-low vegetation-dominated permafrost terrains is especially clear at the end of the thawing stage, in September and October. The differences in thaw subsidence magnitude and pattern reveal the influence of the soil water content in the active layer and permafrost properties on the thaw subsidence patterns. We also compared the Sentinel-1 retrieved cumulative displacement with the X-band TerraSAR-X and L-band ALOS PALSAR results. The difference of retrieved deformation magnitude using the three sensors is amplified when shrubs are more developed. The findings indicate that Sentinel-1 time series with a 6-day or 12-day span work well over discontinuous permafrost terrain above the tree line (i.e., tundra, tundra wetlands and less developed shrub-tundra environments) during the thawed season, but the results and accuracy are not promising over developed shrub-tundra and, especially forest-tundra environments.