Estimating volume of large slow-moving deep-seated landslides in northern Canada from DInSAR-derived 2D and constrained 3D deformation rates

Large slow-moving deep-seated landslides are observed in two different regions of northern Canada with advanced Differential Synthetic Aperture Radar (DInSAR). Two-dimensional vertical and horizontal east-west deformation rates and time series are computed from ascending and descending Sentinel-1 imagery acquired during 2017–2022. The landslides' east-west deformation rate is significantly larger than the vertical deformation rate, so it is better suited for landslide characterization. The deformation rates remain nearly constant and unaffected by seasonal changes during the entire period, suggesting substantial landslide thickness. Two large landslides in Alberta and the largest landslide in the Northwest Territories are studied in detail to demonstrate various advanced value-added products produced from DInSAR results. From ascending and descending line-of-sight deformation rates, Surface-Parallel Flow (SPF) and Aspect-Parallel Flow (APF)-constrained three-dimensional (3D) deformation rates are computed. Landslides thicknesses are then estimated from the APF-constrained 3D deformation rates, and the limitations of these techniques are discussed. The estimated thickness of the Northwest Territories landslide reaches 100 m, suggesting that the entire permafrost block may be sliding above the non-permafrost ground. The described techniques allow for mapping slow-moving deep-seated landslides in harsh conditions in areas affected by seasonal land cover changes, as in northern Canada. The decomposition of landslide motion into two or three components in certain conditions allows us to derive landslide thickness and volume and improve the estimation of a potential hazard posed by landslides.