Exploiting freely available imagery to improve land cover characterization and shallow landslide detection

Natural hazards and land management issues can benefit nowadays from the increasing availability of free, high-resolution satellite imagery that opens the way to fine scale detailed investigations. In high elevation catchments the analysis of vegetation dynamics deserves particular attention since little climatic modifications can be amplified in such fragile systems. For the same reasons, the dynamic of instability phenomena as response to an input hydrological forcing, requires a meticulous spatial representation in order to better represent the active processes at catchment scale. The present work focuses on the analysis of high-resolution freely available imagery (Microsoft® Bing MapsTM Platform) that enables the characterization of vegetation cover and the automatic mapping of shallow landslides in an alpine catchment. Semi-automatic detection of vegetation is carried out at the fine scale using both orthophotos and freely available satellite imagery. The analysis based on the satellite imagery showed a better accuracy in respect to the one based on the orthophotos. In particular, satellite imagery analysis showed high sensitivity and high specificity even in low illumination conditions, while, for the same circumstances, orthophotos-based analysis shows a significant wrong detection rate. In the framework of a long term, multi-temporal and high-resolution characterization of vegetation cover and for a rapid mapping of shallow instability phenomena, the effectiveness of the proposed approach can speed up the representation of the local conditions towards an improvement of land management strategies and hazard and risk assessment.