Use of identification techniques to develop a water table prediction model

A water table prediction model has been developed which uses identification techniques for predicting water table elevations. Data recorded near Aurora in the North Carolina coastal plains over a two-year period were used to develop and test a model. Rainfall and water table elevations were recorded continuously at this site and the observed water table elevations were compared to predicted day end values. The identification of both linear and non-linear difference equation models is described to represent the relationship between the three inputs (rainfall, potential evaporation, and ditch water elevation) and the output (water table elevation). The model was developed using measured results from one year (1975) and tested against observed data for a second year (1976). Agreement between predicted and observed results was comparable to that obtained with the physically based DRAINMOD model