Short-Term Prediction using Neural Networks. ?
is collaborating with the Department of Computing and Mathematical Sciences at A&M-CC and researchers at Texas A&M University (College Station) to produce neural-network-based models for short-term prediction of water level and currents along the Texas Gulf coast. The new models will use the real-time and historical observations from TCOON coupled with wind forecasts from the National Weather service to predict water elevations and currents in a 1 to 30-hour time horizon. Short-term predictions such as these are needed for navigation, oil spill response, and marine operations. Preliminary results have indicated that for Texas coastal waters a neural-network-based model can significantly outperform forecasts of water levels based on traditional harmonic analysis methods. ?
is also investigating the use of neural networks to make real-time predictions of storm surge and other impacts associated with tropical storms.