Regional characterization of rock glacier activity based on DInSAR phase and permafrost extent

Rock glaciers are bodies of frozen debris and ice that move under the influence of gravity in permafrost areas. They are important climatic proxies and can undergo destabilization related to flow of the frontal sectors over steep topography or acceleration related to permafrost degradation and climate change. As consequence, they evolve with complex mechanisms, mirrored by spatial heterogeneity and extremely variable displacement rates. Although a sound quantification of activity is a key component of the study of rock glaciers, only few of them can be characterized by point-like site investigations and ground-based displacement measurements. Their study is thus widely facilitated by remote sensing applications, which proved to be powerful tools for a spatially distributed and temporally continuous characterization on a regional scale

Here, we developed a novel methodology to exploit the potential of spaceborne DInSAR analyses to characterize the state of activity of 516 rock glaciers mapped by Scotti et al., (2013) over an area of approximately 1000km² in the north-eastern sector of Valtellina (Italian Central Alps) and we exploited Landsat-8 thermal imaging to explore their regional distribution according to the land surface temperature.

The original rock glacier inventory, based on orthophotos and DSM mapping, provides a morphological and a dynamic classification (active/inactive vs. relict) of the mapped landforms according to surface evidence. To integrate this dataset with information on the present-day state of activity, we developed a semi-automatic procedure in ArcGIS and Matlab ^TM combining DInSAR products, morphometric data and available permafrost extent information (APIM). To obtain a spatially distributed characterization of rock glacier activity patterns, we processed Sentinel-1 A/B images (2017-2020) with increasing temporal baselines (Bt from 12 to 120 days) and generated 124 interferograms in ascending and descending geometry to account for all the different topographic orientations. We then implemented an analysis of the interferometric phase to achieve a quantification of each rock glacier activity based on four steps: 1) correcting the phase values inside each rock glacier for the modal phase value inside a surrounding stable area; 2) stacking (median phase values) of all the selected interferograms generated with same temporal baselines; 3) extracting frequency distributions of median phase values inside each rock glacier and stable area; 4) calculating the percentage of phase values inside each rock glacier that falls outside the uncertainty ±σ range of the stable area ones. This percentage provides an “Activity Index” that allows defining four classes of rock glacier activity together with the presence (active, inactive) or absence (active debris, relict) of permafrost. Classification results based on DInSAR data at different temporal baselines allow recognizing styles of activity characterized by different ranges of displacement rates and spatial and temporal heterogeneities, possibly correlated with the underlying deformation mechanisms. The integration with land surface temperature finally provides useful insights on the distribution of rock glacier activity classes in different topographic conditions.

Our methodology can be applied to other alpine areas and datasets for a wide-area evaluation of rock glacier activity for climatic studies and possible geohazard hot-spot identification.