Data–driven classification of landslide types at a national scale by using Artificial Neural Networks

• Artificial intelligence methods can classify landslide type for national scale inventories. The classification relies on Digital Elevation Models and shape related parameters. spatial distribution of the landslides is considered as an important feature. data-driven does not require any rule setting by user. Classification essential step in risk management, although often missing large Here we propose a novel method that uses easily accessible morphometric geospatial input parameters to at Italy means shallow Neural Network. We achieved overall True Positive Rate 0.76 five-class over 275,000 (1) rockfall/toppling, (2) translational/rotational slide, (3) earth flow, (4) debris (5) complex landslide. In general, model performance very good entire territory, with areas reaching F-score higher than 0.9. be applied polygonal inventory, those produced automatic mapping procedures from Earth Observation imagery, order automatically identify types landslides.