| Summary: | This study dealt with finding the sequence of the most influential parameters among the factors that affect the offshore wave height. A dataset comprising of four climatic input parameters: sea surface wind speed (U), wind direction (θ), air temperature (Ta), and sea surface temperature (Tw); as well as one output parameter (significant wave heights, Hs) was generated. The offshore field measurements were derived from three buoy stations, deployed in the western part of the North Atlantic Ocean. The primary goal of this study was to identify the predominant input parameters that influence prediction of Hs at each buoy station. In this view, ANFIS (an enhanced type of Takagi-Sugeno-based fuzzy inference system) was implemented on the dataset for variable selection. This process found a subset of the entire set of the observed parameters, which was suitable for prediction purposes. As a result, the following sequence of parameter have the most to least influence on the predictions of Hs, U, Ta, Tw, and θ. In addition, it was found that combination of three variables, namely U, Ta, and θ, forms the most influential set of input parameters with RMSEs of 0.82, 0.44 and 0.62, respectively for the predicted Hs at three stations. Most of the previous studies only employed U and θ to predict Hs using the soft-computing methods. As the first study of its type, the findings from this study suggest that the accuracy of wave height prediction may improve when Ta and Tw are included as inputs along with U and θ. The present study can serve as a gear towards enhancing the accuracy in prediction of wave height at various offshore locations. Wave height; Air temperature; NDBC; North Atlantic Ocean; ANFIS; Variable selection;
|
|---|