Storm surge forecast through a combination of dynamic and neural network models

摘要

An operational surge forecast system based on a combination of a hydrodynamic model and an artificial neural network is described. The system runs at the Centre for sea level forecasting and flood warnings of the Venice Municipality (ICPSM, Istituzione Centro Previsioni e Segnalazioni Maree) and is focused on the prediction of the surge near Venice. The hydrodynamic model provides a five-day forecast for the Mediterranean Sea. Then, results near Venice are extracted and improved using a neural network. The method exposed in this paper reduced by half the average error of the hydrodynamic model for the first day forecast and maintains good performances also for longer forecasts. Moreover it is able to reduce the bias error of the model.