Long-range terrestrial laser scanning for geomorphological change detection in alpine terrain - handling uncertainties

Long-range terrestrial laser scanning (TLS) is an emerging method for the monitoring of alpine slopes in the vicinity of infrastructure. Nevertheless, deformation monitoring of alpine natural terrain is difficult and becomes even more challenging with larger scan distances. In this study we present approaches for the handling of spatially variable measurement uncertainties in the context of geomorphological change detection using multi-temporal data sets. A robust distance measurement is developed, which deals with surface roughness and areas of lower point densities. The level of detection (LOD), i.e. the threshold distinguishing between real surface change and data noise, is based on a confidence interval considering the spatial variability of TLS errors caused by large laser footprints, low incidence angles and surface roughness. Spatially variable positional uncertainties are modelled for each point according to its range and the object geometry hit. The local point cloud roughness is estimated in the distance calculation process from the variance of least-squares fitted planes. Distance calculation and LOD assessment are applied in two study areas in the Eastern Alps (Austria) using multi-temporal laser scanning data sets of slopes surrounding reservoir lakes. At Finstertal, two TLS point clouds of high alpine terrain and scanned from ranges between 300 and 1800m are compared. At Gepatsch, the comparison is done between an airborne laser scanning (ALS) and a TLS point cloud of a vegetated mountain slope scanned from ranges between 600 and 3600m. Although these data sets feature different conditions regarding the scan setup and the surface conditions, the presented approach makes it possible to reliably analyse the geomorphological activity. This includes the automatic detection of rock glacier movement, rockfall and debris slides, even in areas where a difference in vegetation cover could be observed. Copyright (c) 2016 John Wiley & Sons, Ltd.