Abstract

This study reconstructs the September 2023 Storm Daniel flood along the lower Pineiós (Pinios) River in Thessaly, Greece, using the Delft3D-FLOW hydrodynamic model. A three-dimensional σ-coordinate grid, incorporating available bathymetric surveys and interpolated channel geometry, was forced with ERA5 atmospheric reanalysis to simulate river–floodplain dynamics under the extreme rainfall of 5-7 September. A Python-based workflow was implemented to process meteorological fields and generate spatially and temporally consistent model inputs. To assess model performance, the inundation extent was compared with Copernicus Emergency Management Service (EMS) Rapid Mapping products. A structured geospatial preprocessing pipeline, combining reprojection, grid-based clipping, channel removal and polygon intersection, ensured spatial alignment between model outputs and EMS observations. Validation metrics demonstrate strong agreement, with Success Rates exceeding 0.94, low Failure Rates, and Centroid Skill Scores close to unity for both upstream and downstream sectors. The spatial patterns of inundation reflect the shift from narrow valley sections to the broader floodplain areas of northeastern Thessaly, matching independently mapped impacts of Storm Daniel. The results highlight the capability of Delft3D-FLOW to simulate extreme flood events in data-limited basins and provide a reproducible framework for integrating atmospheric reanalysis with high-resolution hydrodynamic modelling to support flood hazard assessment and regional resilience planning.

Keywords:Hydrodynamic modelling; Delft 3D-FLOW; Storm daniel; Flood mapping; Copernicus EMS; Thessaly