Dynamic adaptive wavelet based fuzzy framework for extended significant wave height forecasting

Significant Wave Height (SWH) represents a dynamic phenomenon with vital implications for effective coastal engineering design, planning, management, and sustainable development. Numerous studies in the literature indicate that prevailing models for predicting hourly SWH tend to exhibit limited accurate prediction lead times. Recently, the integration of the Discrete Wavelet Transform (DWT) as a data pre-processing technique with Artificial Intelligence (AI) based forecasting models has yielded significant enhancements in predicting ocean wave behavior. Further, the employment of the Maximum Overlap Discrete Wavelet Transform (MODWT) offers distinct advantages over the traditional DWT. This study capitalizes on these advantages to enhance predictive outcomes. The SWH dataset undergoes both DWT and the newly introduced MODWT processes to dissect stochastic and deterministic components, facilitating subsequent modeling. The resultant decomposed spectral bands are then incorporated into a Fuzzy framework to forecast SWH values. Visual examination of the predicted SWH data affirms the superior performance of these hybrid models in capturing intricate details. The results demonstrate that the hybrid models outperform the Fuzzy model in terms of predictive accuracy and precision. Notably, the MODWT-Fuzzy model surpasses the DWT-Fuzzy model for all time spans and all monitoring stations. In conclusion, the MODWT-Fuzzy hybrid model excels in predictive accuracy and presents a promising avenue for SWH prediction. The implications of this study's findings extend to diverse domains within ocean and coastal engineering.