Neuro-fuzzy approach for the spatial estimation of ocean wave characteristics

Abstract

Real time applications of oceanographic data and statistical calculations for climatological studies suffer from missing data. To overcome this problem, spatial correlation of significant wave height from multiple stations is considered. Here significant wave height is taken as regionalized variable to establish a relationship through the use of fuzzy logic. There are other techniques in the literature to explore spatial dependency such as regression, Kriging and artificial neural networks. Spatial variability of wave characteristics is also important for ocean wave energy applications. It is shown that significant wave heights at the location of one buoy can be predicted using the neuro-fuzzy approach and the data from adjacent buoys. The proposed approach is compared with the multiple linear regression model which includes some restrictive assumptions. The implementation of the methodology is performed for the buoys located in the Gulf of Mexico.