Prediction of significant wave height; comparison between nested grid numerical model, and machine learning models of artificial neural networks, extreme learning and support vector machines

• Published in: Engineering applications of computational fluid mechanics Vol. 14; no. 1; pp. 805 - 817
• Main Authors: Shamshirband, Shahaboddin, Mosavi, Amir, Rabczuk, Timon, Nabipour, Narjes, Chau, Kwok-wing
• Format: Journal Article
• Language: English
• Published: Hong Kong Taylor & Francis 01.01.2020; Taylor & Francis Ltd; Taylor & Francis Group
• Subjects: Artificial neural networks; Bathymeters; Coastal engineering; deep learning; extreme learning machines; Learning theory; Machine learning; Mathematical models; nested grid; Neural networks; Numerical modeling; Numerical models; Support vector machines; Surface wind; Wave height; wind waves
• ISSN: 1994-2060; 1997-003X
• DOI: 10.1080/19942060.2020.1773932

Estimation of wave height is essential for several coastal engineering applications. This study advances a nested grid numerical model and compare its efficiency with three machine learning (ML) methods of artificial neural networks (ANN), extreme learning machines (ELM) and support vector regression (SVR) for wave height modeling. The models are trained by surface wind data. The results demonstrate that all the models generally provide sound predictions. Due to the high level of variability in the bathymetry of the study area, implementation of the nested grid with different Whitecapping coefficient is a suitable approach to improve the efficiency of the numerical models. Performance on the ML models do not differ remarkably even though the ELM model slightly outperforms the other models.