Reconstituting past floods and assessing scenarios by integrating hydraulic modelling and citizen science: a study case in the Upper Paranapanema River Basin

Quantifying flood hazards requires complex models based on historical data, often scarce in developing countries. This study demonstrates an innovative approach that integrates citizen science with hydrodynamic modeling to assess flood risks in data-limited regions. We applied the HEC-RAS model to reconstruct past flood events in a small town in the Upper Paranapanema River Basin, São Paulo State (southeastern Brazil). Validation against resident observations and satellite imagery showed strong agreement, with a Kappa statistic of 0.7 and NSE up to 0.79. Collective memory also supported statistical analysis of peak flows to develop flood scenarios with various return periods. We further assessed the effect of earth dikes, originally built to protect upstream rural areas, and found negligible influence on flooding within the urban area. Our findings highlight the value of combining local knowledge and hydrodynamic modeling, showing that citizen science is a practical alternative when official records are limited.