The impact of high-resolution topography on landslide characterization using DInSAR

Differential interferometric synthetic aperture radar (DInSAR) can measure surface deformation at the centimeter level and, as a result, has been used to investigate a wide variety of natural hazards since the 1990s. In general, short spatial and temporal baselines are selected to reduce decorrelation and the effect of incorrect removal of the topographic component in differential interferograms. The nearly global coverage of the Shuttle Radar Topography Mission (SRTM) digital elevation models (DEMs) significantly simplified and improved the modelling and removal of topography for differential interferometric applications. However, DEMs are produced today at much finer resolutions, although with varying availability and cost. SRTM DEMs are freely available at 30 m resolution world-wide and 10 m resolution in the US. The TanDEM-X mission has produced a worldwide DEM at 12 m, although it is not generally free of cost. Light Detection and Ranging (LiDAR) DEMs can provide better than 1m resolution, but are expensive to produce over limited extents. Finally, DEMs from optical data can be produced from Digital Globe satellite images over larger regions at resolutions of less than 1 m, subject to various restrictions. It can be shown that the coherence quality of a DInSAR image is directly related to the DEM resolution, improving recovery of the differential phase by significantly reducing the geometric decorrelation, and that the number of recovered pixels significantly increases with higher resolutions, particularly in steep topography. In this work we quantify that improvement for varying resolutions, from 1 to 30 m, and slopes and investigate its effect on the characterization of landslides in different regions and with a variety of surface conditions, including Greenland, Alaska, California, and the Canary Islands.