Abstract

Accurate numerical simulations of floods provide precise hazard estimations and can improve risk management strategies. Unfortunately, flood events of extraordinary magnitude -needed for risk assessment- are rare and difficult to measure. Hence, numerical models are typically calibrated and validated using a few measured scenarios of moderate magnitudes. Flood social cartography, a participatory mapping method, identifies historically flood-prone areas based on the residents' experience and knowledge. This research proposes using social cartographies to calibrate and validate hydraulic flood models. A numerical flood simulation model of the lower reach of the unregulated Carampangue River, Chile, was calibrated and validated using available information from gauges, satellite images, and social cartographies for the towns of Arauco and Ramadillas. The discharge magnitude of floods recorded in the social cartographies was determined. The quality of both models, calibrated with gauge data and with social cartography flood maps, was determined in each case by comparing computations with depth gauge data and satellite images through the root mean square error and the critical success index. Results show that social cartography participants recognized flooded areas corresponding to a discharge with a return period of 35 years. The calibration with social cartographies delivered a set of Manning's roughness coefficients similar to those obtained using gauge data. The validation cases show that water depth and the extension of floods computed with the model calibrated using social cartographies is comparable to that computed using gauge data in the calibration process. RMSE from both calibrated models is of order 0.1 m. Consequently, the social cartography of floods constitutes a useful data source of extreme flood scenarios for calibrating and validating numerical flood models.