Using feed forward neural networks to model the effect of precipitation on the water levels of the Northeast Cape Fear river

The impact of major flooding events in the United States points to a need to discover an effective method of forecasting changes in river flow which could lead to area flooding. Proper modeling of rainfall and runoff is important, but first-principles modeling is difficult and not plastic. Neural networks provide a data-driven modeling tool capable of capturing the relationship between rainfall and river flow. The work reported here indicates that neural networks are capable of making reliable forecasts of river flow.