Machine learning approach to predict significant wave height

Abstract

To estimate significant wave height of ocean wave, a machine learning framework is developed. Significant wave height and period can be used by supervised training of machine learning to predict ocean conditions. In this paper we proposed a method to predict significant wave height using Support vector regression (SVR). Buoy dataset taken from the Queensland government open data portal the input from which were aggregated into supervised learning test and training data sets, which were supplied to machine learning models. The SVR model replicated significant wave height with a root-mean-squared-error of 0.044 and performed on the test data with 95% accuracy. Comparing to forecasting with the physics-based model the Machine learning SVR model requires only a fraction (< 1=1200th) of the computation time, to predict Significant wave height.

Keywords

• machine learning
• significant wave height
• support vector regression