Augmented chaos-multiple linear regression approach for prediction of wave parameters

Prediction of wave parameters is one of the significant component for several coastal applications; for instance, coastal erosion, inshore and offshore structures, wave energy and others. The current research investigates the potential of the Chaos theory integrated with multiple linear regression (Chaos-MLR) in prediction of wave heights and wave periods. The wave data were collected at four moorings in the coastal environment of Tasmania. In the first stage, reconstructing the phase space and determine the input data for Chaos-MLR model, the delay time and embedding dimension are computed using average mutual information and false nearest neighbors’ analyses. The presence of chaotic dynamics in the used data is identified by the correlation dimension methods. In the second stage, the Chaos-MLR and pure MLR models are constructed for prediction model. Absolute error and best-fit-goodness diagnostic indicators are utilized to inspect the proficient of the proposed model in comparison with the pure MLR model. The inter-comparisons demonstrated that the Chaos-MLR and pure MLR models yield almost the same accuracy in predicting the significant wave heights and the zero-up-crossing wave periods. Whereas, the augmented Chaos-MLR model is performed better results in term of the prediction accuracy vis-a-vis the previous prediction applications of the same case study.