Landslide surface monitoring based on UAV-and ground-based images and terrestrial laser scanning: Accuracy analysis and morphological interpretation

摘要

The spatial and temporal monitoring of the kinematics, morphological features (fissures, scarps, lobes, etc.), and soil humidity of active landslides can provide valuable information to understand their origin and mechanisms for hazard assessment and eventually early-warning. For the surficial investigation of landslides, several surveying methodologies can be applied. Point-based measurements like GNSS, inclinometer, extensometer, and tiltme-ter (e.g., Malet et al., 2002) are complemented by area-based surveys such as photogrammetry, laser scanning (LiDAR, e.g., Jaboyedoff et al., 2010), or interferometric synthetic aperture radar (InSAR, e.g., Fruneau et al., 1996). In recent years, the very high-resolution (VHR, mm-few cm) and multi-temporal mapping of Earth's surface using terrestrial laser scanning (TLS), ground-based optical imagery, and low-cost UAV (Unmanned Aerial Vehicle)-based aerial imagery has grown in importance. For the latter, advances in computer vision provide freely available Structure-from-Motion (SfM) and Multi-View Stereo (MVS) algorithms (e.g., Snavely et al., 2006, Furukawa and Ponce, 2010) which automatically process large numbers of varying, overlapping, disordered, oblique imagery from consumer-grade cameras to an accurate, dense and coloured 3D point cloud. To date such algorithms are implemented in web services and open source software packages. Production of dense point clouds by MVS from UAV photography are described by Niethammer et al.,(2010, 2012), Dandois and Ellis (2010), Neitzel and Klonowski (2011), Turner et al., (2012), Harwin and Lucieer (2012), Verhoeven et al., (2012); for ground-based imagery see James and Robson (2012), Westoby et al., (2012).