Articles | Volume 22, issue 4
Research article
26 Apr 2022
Research article |  | 26 Apr 2022

Machine-learning blends of geomorphic descriptors: value and limitations for flood hazard assessment across large floodplains

Andrea Magnini, Michele Lombardi, Simone Persiano, Antonio Tirri, Francesco Lo Conti, and Attilio Castellarin

Data sets

MERIT DEM: Multi-Error-Removed Improved-Terrain DEM Yamazaki Lab

River Flood Hazard Maps at European and Global Scale European Commission

Short summary
We retrieve descriptors of the terrain morphology from a digital elevation model of a 105 km2 study area and blend them through decision tree models to map flood susceptibility and expected water depth. We investigate this approach with particular attention to (a) the comparison with a selected single-descriptor approach, (b) the goodness of decision trees, and (c) the performance of these models when applied to data-scarce regions. We find promising pathways for future research.
Final-revised paper