Individuazione di dissesti associati ad evento meteorico intenso attraverso l'analisi di immagini satellitari multi-spettrali.

In this paper we expose the project activities related the analysis of intense rainfall events and identification of triggered landslides through semi-automatic multi-temporal satellite images. It is analyzed the event of 24-25/10/2011 in Lunigiana, through two subsets of RapidEye satellite scenes (multispectral sensor, 5-band high-resolution, pixel size on the ground equal to 5 m). After pre-processing, it was realized a multi-temporal layer stack on which they were carried out change detection techniques using different indices (spectral angle, NDVI differencing, principal components analysis, Independent components analysis). The analysis of these indices has led to the identification of 90 clusters of pixels of the images associated with landslides. The checks carried out through inspections and aerial photographs of aid has led to confirmation of 46 landslides related to the event.