Mapping slow-moving alpine landslides by UAV — Opportunities and limitations

Abstract

Acquisition of unmanned aerial vehicle (UAV)-based imagery with resolution down to a few centimeters is challenging in alpine conditions. In recent years, UAV-based images have been used in an increasing number of case studies to monitor landslides. Processing of multitemporal high-resolution aerial images provides multitemporal 3D point clouds and multitemporal orthomosaics, which provide substantial information about slope dynamics. Surface processes deciphered from 3D point clouds and orthomosaic temporal sequences supplement on-site geophysical measurements of subsurface structures and dynamics by bringing key spatial and temporal constraints for process interpretation. However, accurate spatial mapping and successful analysis of surface dynamics are functions of the original optical image quality, which in turn relies on the flight system and configuration, camera characteristics and settings, as well as weather conditions and object texture. We evaluate the capabilities and limitations of UAV-based landslide dynamic mapping using illustrative examples gathered from a pioneering low-cost UAV-based imagery data set acquired at the slow-moving (mm-cm/d) Super-Sauze landslide (southeastern France) in 2008 and 2010, which included surface deformations (fissure development and kinematics) and surface conditions (soil moisture, erosion, and sedimentary processes) with very different certainty levels. In a few promising instances, processes detected in the 3D point clouds and orthomosaics could be correlated to geophysical data (passive seismic monitoring of landslide endogenous seismicity and electrical resistivity tomography for water content mapping), thereby validating the UAV analytical approach while providing constraints that improve interpretation of landslide dynamics.