The Development of a Weather-Type Statistical Downscaling Model for Wave Climate Based on Wave Clustering

Reliable long-term wave data are significant for understanding changes and variability of ocean waves, which has important implications for coastal engineering, land erosion, and coastal flooding. This study develops a regression-guided weather-type statistical method for modelling regional or global significant wave height Hs. The model classifies the atmospheric circulation patterns (predictor) through the regression-guided clustering approach, linking the atmospheric circulation with clustered regional Hs (predictand). It is applied in the Chinese marginal seas and also the global ocean. A comprehensive skill assessment shows robust skill and computationally efficiency of the model in capturing climatology and variability of both mean and extreme Hs in the Chinese marginal seas and global oceans. Furthermore, the reconstructed global Hs data show similar seasonal trends as the ERA5 data, with a gradual decrease in Hs observed during boreal summer in the central Pacific and western North Atlantic regions at lower latitudes. This method proves to be robust for both regional and global Hs reconstruction, and also applicable for Hs climate prediction and projections.

Ocean Engineering