Author(s): Goncalo Jesus; Rodrigo Garcia; Antonio L. Antunes; Joao Fernandes; Francisco Da Cruz
Linked Author(s):
Keywords: No Keywords
Abstract: Flash floods in urban basins are characterized by short response times, strong rainfall intensities, and high uncertainty propagation in hydrological–hydraulic models. This paper presents the results of the INFLOOD project, showcasing the AI-based early warning system developed for the Ribeira das Vinhas basin (Cascais, Portugal). The system integrates real-time multi-sensor monitoring (including precipitation, water levels, wind and temperature), soft-coupled HEC-HMS and HEC-RAS modelling [1], and machine learning-based forecasting models. By combining dependable hybrid physics-based and AI data modelling, the flood prediction framework enables scenario-based analysis for flash-flood-prone urban basins.
Year: 2026