Analisi della qualità delle carte della suscettività da frana a grande scala topografica implementando algoritmi ad apprendimento automatico e l'analisi spaziale

The Autorità  di Bacino is the Italian agency in charge of landslides prevention. It is responsible for redacting the P.S.A.I. (Piani Stralcio per L'Assetto Idrogeologico) and to guarantee its updates through landslide reshaping survey. This is undertaken through reporting submitted from the local authorities within the relevant drainage basin. The objective of this research is to optimize the landslide susceptibility assessment in the P.S.A.I. redacting and their updates. The Autorità  di Bacino implement the susceptibility assessment of landslide events through GIS technologies, employing spatial analysis techniques based on Boolean algebra (overlay). This study proposes a new methodology to improve the resolution (quality) of the landslide susceptibility, based upon the Machine Learning Approach. This new methodology has been labelled Geo-S.Co.Ma.L. and was realised using the SVM algorithm, on which geological constraints) were applied by means of geoprocessing techniques. Through the comparisons of different releases of the ex AbiSele P.S.A.I., the Geo-S.Co.Ma.L. methodology appears to be capable to foresee a higher number of landslide events than the Boolean algebra, and thus to calculate the landslide susceptibility with greater detail.