Water Level Observations and Short-Term Predictions Including Meteorological Events for Entrance of Galveston Bay, Texas

Abstract

This paper shows that conventional harmonic analysis alone does not adequately predict the coastal water level variation at the entrance to Galveston Bay when strong meteorological forcing is present. The water level anomalies (the difference between the observed water level and that predicted by harmonic analysis) are shown to be as large as the tidal range itself. The water level anomaly at the entrance to Galveston Bay is primarily due to the east-west directed wind speed, and a simple linear model is shown to predict the anomaly based on the locally measured wind and a nine hour lag between the wind forcing and water level response. The model is further refined using a neural network approach with east-west and north-south winds, barometric pressure, and the previously observed water level anomaly and without assuming previous knowledge of the phase lag between wind and water level. Both linear and neural network models are shown to improve significantly short-term