New insights on the Salmon Falls Creek Canyon landslide complex based on geomorphological analysis and multitemporal satellite InSAR techniques

Abstract

Multitemporal satellite interferometric synthetic aperture radar (InSAR) techniques can characterize line-of-sight displacements of active landslide areas with resolution (mm scale) and accuracy comparable to or higher than differential GPS, sensor network, or photogrammetry techniques. This study improves understanding of the rate of movement and the lateral extent of the active domain of a landslide complex within Salmon Falls Creek Canyon near Twin Falls, Idaho. Specifically, we were able to estimate displacement of yearly motion rates in early and late stages of the event by analyzing a collection of archived radar satellite imagery. Small baseline subset (SBAS) InSAR performed better than persistent scatterer (PS) InSAR for analyzing distributed scatterers because of its ability to capture strongly nonlinear displacement rates. In addition, comparison with GPS field measurements showed agreement with InSAR-derived displacements. Geostatistical analysis was used to describe surface and morphometric characteristics of two separate landslides within Salmon Falls Creek Canyon. Each was divided into representative geomorphologic units, and morphometric analysis focused on two key parameters: topographic texture and mean slope. Scarp units are topographically rough because of their greater relief and steep slopes, while a displaced headwall block has retained a smooth topography. Each landslide upper body has a higher topographic texture than the corresponding body unit. Both landslides display a progressive decrease in mean slope from upper body to toe to body. InSAR SBAS results showed that headwall block and transverse scarp of the landslide complex at Salmon Falls Creek Canyon had the highest mean annual velocity in the radar line-of-sight (LOS) direction. Velocity of movement in each landslide toe and body was less, signifying that LOS movement was more active in the upper reaches of the landslides, although lateral translation may have been greater in the body and toe compared to the headwall region due to the curved shape of the landside detachment surface.