The MIRAGE project will provide methods and tools for the semi-automated detection, classification and kinematic characterisation of alpine mass movements through the combination of geomorphological data, spaceborne InSAR and deep learning techniques. 

The construction of libraries of expert-interpreted InSAR phase signal features, corresponding to different mass movements, will allow training and validating a Convolutional Neural Network (CNN). This will be tested with routinely generated SAR interferograms, to produce automated multi-temporal maps able to detect and classify mass movements, evaluate their displacement rates and temporal evolution. 

The main considered gravitational processes will be slow deep-seated slope deformations, rockslides, earthflows and rockglaciers, and the study areas will be the Italian Alps and the Apeninnes. 

These results will provide policy making and engineering communities with new tools to quantify hazards related to specific processes, anticipate their critical evolution and support the prioritisation, planning and design of risk mitigation measures.