Development Page--Not for Official Use

Neural Network Application to Water Level Forecasting

Forecasting water levels in coastal bays and estuaries is challenging because a number of factors such as the shape and size of the body of water, changes in sea surface temperatures and heterogeneity of the land cover will affect the wind patterns, sea breezes, and ultimately the water levels. Neural network modeling differs fundamentally from classical modeling in the sense that an accurate knowledge of the boundary conditions, inputs and outputs, and forcing functions is not required. Neural networks use a time series of prior data to model a system. If the appropriate physical parameters are selected over a set of representative weather patterns, the time histories of these parameters contain all the physics of the system.

At the heart of a neural network is the assignment of judicious weights and biases to the elemental neurons of the network. This learning or training process must be based on a large set of prerecorded observations such as the TCOON database.

The ANN selected to forecast water levels was chosen based on previous studies focused on Galveston Bay and Port Aransas. These studies found that a simple ANN is optimal for forecasting water levels under the conditions encountered along the Texas Gulf coast. Previous wind and water level measurements and wind and tidal forecasts are the most important important input variables as determined through ANN training.

The ANN selected for application in Corpus Christi Bay consists of two layers as seen in the figure below. The first layer contains the input variables and the second layer consisting of two neurons defines the inter-relationships between the input variables.

What Limits Accuracy of ANN Prediction?

The quality of the wind forecasts will likely be the limiting factor for the accuracy of the water level forecasts.

1. refer to water level forecast performance comparisons#

<< LinksToMoreInformationAboutArtificalNeuralNetworks | ForecastsSiteMap | WindMeasurements >>|NeuralCCBay|>>