ANN Applications to Storm Surge Predictions in Apalachicola, Florida

Strom surge inundation is responsible for extensive damage to coastal areas. The prediction on storm surge hydrograph is very important to the protection of coastal properties and human lives. The three-layer feed-forward backpropagation neural network was applied on predicting storm surge during hurricanes in Apalachicola. By removing tidal signals, storm surge hydrographs induced by forces of winds and atmospheric pressure were obtained from the time series of observed water levels for Hurricane Dennis in 2005 and Hurricane Ivan in 2004. Results from model verifications using the data set of Hurricane Ivan indicate that model predictions of storm surge hydrograph reasonably follow the general trend of the observations. The peak elevations of the storm surge were satisfactorily predicted by the ANN model using the atmosphere pressure and wind speed in the study site.