New Ensemble Forecasting Model Claims 30% Improvement in Swell Accuracy

Researchers at the Scripps Institution of Oceanography have published results showing a new ensemble forecasting approach that improves open-ocean swell height accuracy by approximately thirty percent compared to current operational models, with particular gains in long-range predictions beyond five days.

The system blends output from four global wave models — NOAA's WaveWatch III, the European Centre's ECMWF wave model, and two regional high-resolution models — using a machine-learning weighting algorithm trained on fifteen years of buoy validation data.

The practical implication for surf forecasting is meaningful. Current operational models can diverge significantly on swell height at the seven-to-ten day horizon, making trip planning uncertain. The new approach reduces that spread while also improving directional accuracy, which determines whether a given swell will have the period and angle to activate specific breaks.

Surfline and Windguru have both expressed interest in incorporating the methodology. A technical paper is under review at the Journal of Geophysical Research.