Analysis of the Gulf of Mexico water levels as random signals

This paper deals with estimates for the best water level predictions for the shallow waters of the Gulf of Mexico. The predictions are made by Artificial Neural Networks (ANNs) and Statistical Modeling. In this paper we use elements of the theory of stochastic to evaluate the quality of forecasts, the stochastic properties of the inputs as well as the given goodness criteria of predictions. As a result of such investigation we can outline limitations of different methods used for predictions of water levels in the bays and estuaries of the Texas coast.