Geomorphic flood index 2.0: enhanced tools for delineating flood-prone areas in data-scarce regions

Abstract

Accurately delineating flood-prone areas over large regions remains challenging in areas with limited data, where detailed hydraulic modelling is often impractical. The Geomorphic Flood Index (GFI) is a parsimonious, DEM-based alternative; however, its original formulation may misrepresent inundation near river confluences, where backwater effects and tributary–main stem interactions alter local water levels. This study presents GFI 2.0, an enhanced GFI framework that introduces an iterative confluence module guided by stream hierarchy. This new module accounts for confluence-controlled backwater effects while preserving the original method's simplicity and low computational cost. GFI 2.0 was evaluated across 12 Italian basins using official ISPRA flood hazard maps and multiple DEM products. Additionally, GFI 2.0 was tested with a detailed hydraulic benchmark in the Bradano River basin based on two-dimensional simulations for 30-, 200-, and 500-year return periods. The results show systematic improvements in flood extent delineation, particularly in low-gradient floodplains, wide valleys, and confluence-dominated sectors. Gains are smaller in strongly confined, mountainous basins. A multi-basin analysis indicates that GFI 2.0's added value is greatest when coarser DEMs are used. In the Bradano basin, the area under the curve (AUC) increased from 0.917 to 0.952, from 0.918 to 0.960, and from 0.914 to 0.959 for the 30-, 200-, and 500-year scenarios, respectively. Flood depth estimates improved as well, particularly for larger return periods. RMSE decreased from 5.25 to 3.24 m for the 200-year scenario and from 5.66 to 3.58 m for the 500-year scenario. Overall, GFI 2.0 expands the use of DEM-based flood mapping by offering a scalable, physically informed, and computationally efficient tool for regional screening and preliminary hazard assessment in areas with limited data.