In this work, a data-driven surrogate model for the rapid prediction of flood maps is presented. Tens of future water depth maps can be forecasted thanks to an autoregressive procedure. The “FloodSformer” model is tested to forecast the spatiotemporal evolution of flooding in a physical model of the Toce River valley. The application of the surrogate model for real-time forecasting and early warning is promoted by its negligible computational time and the good accuracy.
A DATA-DRIVEN MODEL FOR REAL-TIME RIVER FLOOD FORECASTING / Pianforini, Matteo; Dazzi, Susanna; Pilzer, Andrea; Vacondio, Renato. - (2024). (Intervento presentato al convegno XXXIX Convegno Nazionale di Idraulica e Costruzioni Idrauliche (IDRA2024) tenutosi a Parma nel 15-18 settembre 2024) [10.5281/zenodo.13584918].
A DATA-DRIVEN MODEL FOR REAL-TIME RIVER FLOOD FORECASTING
Matteo Pianforini
;Susanna Dazzi;Renato Vacondio
2024-01-01
Abstract
In this work, a data-driven surrogate model for the rapid prediction of flood maps is presented. Tens of future water depth maps can be forecasted thanks to an autoregressive procedure. The “FloodSformer” model is tested to forecast the spatiotemporal evolution of flooding in a physical model of the Toce River valley. The application of the surrogate model for real-time forecasting and early warning is promoted by its negligible computational time and the good accuracy.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
