Modelling of Significant Wave Height Using Wavelet and Genetic Programming

Sajjad Shahabi, Mohammad-Javad Khanjani

Monday 29 june 2015

17:45 - 17:48h at Amazon (level 1)

Themes: (T) Water engineering, (ST) River and coastal engineering, Poster pitches

Parallel session: Poster pitch: 3C. Coastal Engineering

Accurate modelling of significant wave height (SWH) is an essential issue for most of the ocean and coastal operations. Traditional models are basically linear and assume that the data are stationary. Some models based on artificial neural network (ANN) has been developed and employed for SWH modelling (Tsai et al., 2002; Markarynsky et al., 2005; Londhe and Panchang, 2006). For some limitation of ANN, some other approaches such as genetic programming (GP) which is evolutionary computing method has been presented and developed (Gaur and Deo, 2008). In spite of advantages of the ANN and GP methods for SWH modelling, there is a limitation when signal are highly nonstationary. To cope with this problem, some researchers hybridized wavelet technique with other methods (Nourani et al., 2014; Deka and Prahlada, 2012). The hybrid wavelet-GP (namely WGP) model was employed to rainfall-runoff modelling and drought forecasting (Nourani et al., 2012; Danandeh Mehr et al., 2014). In this paper, a WGP model was employed to forecasting of SWH in the west of North Atlantic Ocean.