A weather-type statistical downscaling framework for oceanwave climate

Abstract

Wave climate characterization at different time scales (long-term historical periods, seasonal prediction, and future projections) is required for a broad number of marine activities. Wave reanalysis databases have become a valuable source of information covering time periods of decades. A weather-type approach is proposed to statistically downscale multivariate wave climate over different time scales from the reanalysis long-term period. The model calibration is performed using historical data of predictor (sea level pressure) and predictand (sea-state parameters) from reanalysis databases. The storm activity responsible for the predominant swell composition of the local wave climate is included in the predictor definition. N-days sea level pressure fields are used as predictor. K-means algorithm with a postorganization in a bidimensional lattice is used to obtain weather patterns. Multivariate hourly sea states are associated with each pattern. The model is applied at two locations on the east coast of the North Atlantic Ocean. The validation proves the model skill to reproduce the seasonal and interannual variability of monthly sea-state parameters. Moreover, the projection of wave climate onto weather types provides a multivariate wave climate characterization with a physically interpretable linkage with atmospheric forcings. The statistical model is applied to reconstruct wave climate in the last twentieth century, to hindcast the last winter, and to project wave climate under climate change scenarios. The statistical approach has been demonstrated to be a useful tool to analyze wave climate at different time scales.