Integrated displacement and activity analysis at the Valoria landslide (Italian Apennines) through automated topographic monitoring, image correlation velocimetry and surface roughness computation

ISBN 2-95183317-1-5At the Valoria landslide located in the Italian Apennines, surface monitoring data acquired by an automated total station were analyzed. The system provided short-term information on the behavior of the material during crisis and the duration of reactivations. Long-term displacements and morphometric properties were studied using a geographic information system (GIS). Results from monitoring show a strong correlation between rainfall related triggering of mass wasting in the crown zone and subsequent downslope reactiva-tions. We found that crown-to-toe zone reactivations generally occur within only 6 weeks. Further findings indicate that earth slide materials in the head zone fail almost instantaneous conversely to earth flow materials in the track channel. The findings are complemented by a digital image correlation analysis of multi temporal, high-resolution digital elevation models (DEM) acquired in 2006, 2007 and 2009. We utilized this technique to compute velocities and displacements of pixels in the DEMs between 2006 and 2009. It was found that both azimuth and magnitude of displacements could be reconstructed in for the crown zone, the main track channel and the landslide toe. The results from the image correlation compare well to displacements obtained from independent monitoring methods (GPS and interpreting of shaded reliefs maps). Based on the assump-tion that increased materials movements result in higher terrain roughness, we used LiDAR-derived to per-form a raster based roughness analysis. Different roughness calculation methods were applied to 0.5 m eleva-tion grids using different Kernel sizes. The performance of a supervised and an unsupervised approach was evaluated. Findings from the supervised approach showed that the difference between active and dormant landslides is evident in some cases but also that earth flows and dormant landslides tend to have a similar roughness patterns as stable areas. Results from the unsupervised approach demonstrated that landslide roughness is heterogeneous and that non-landslide areas may have a similar morphometric signature. In this study it was demonstrated that near- and remote sensing techniques are crucial, as well as complimentary in the analysis of landslide hazard. Applying both approaches can yield a more complete picture of the defor-mation history of landslides