Estimation of significant wave height in shallow lakes using the expert system techniques

Abstract

Highlights? GKF approach has the advantages of Genetic and Kalman Filtering methods. ? GKF predicts better than ANN. ? This is the first time performances of GKF and ANN are examined for shallow water. ? The independent data set shows the importance of different seasonal conditions. ? Previous significant wave height and current wind speed affects current significant wave height. Significant wave height is an important hydrodynamic variable for the design application and environmental evaluation in coastal and lake environments. Accurate prediction of significant wave height can assist the planning and analysis of lake and coastal projects. In this study, the Genetic Algorithm (GA) is used as the optimization technique to better predict model parameters. Also, Kalman Filtering (KF) is used for prediction of significant wave height from wind speed. KF technique makes predictions based on stochastic and dynamic structures. The integrated Geno Kalman Filtering (GKF) technique is applied to develop predictive models for estimation of significant wave height at stations LZ40, L006, L005 and L001 in Lake Okeechobee, Florida. The results show that the GKF methodology can perform very well in predicting the significant wave height and produce lower mean relative error and mean-square error than those from Artificial Neural Network (ANN) model. The superiority of GKF method over ANN is presented with comparisons of predicted and observed significant wave heights.