Evaluating Different Techniques for Generating Probabilistic Wind and Wave Guidance

There is increasing demand for probabilistic guidance to inform maritime weather forecasts and warnings issued by the Ocean Prediction Center (OPC). Probabilistic forecasts of winds exceeding marine warning thresholds (Gale, Storm, Hurricane-Force) can be computed from ensemble systems like the Global Ensemble Forecast System (GEFS). This project explored different neighborhood techniques for generating and interpreting such forecasts using GEFS data.
Three techniques for generating probabilistic guidance for wind and wave hazards were analyzed. The Neighborhood Ensemble Probability (NEP) approach averages the binary probability field over a constant neighborhood before calculating an ensemble mean. The Neighborhood Maximum Ensemble Probability (NMEP) approach searches a constant neighborhood for any instance of threshold exceedance and then calculates an ensemble mean. The EAS approach uses spatial neighborhoods of variable size based on the similarity of the ensemble members’ forecasts. These methods were used to compute Day 2 forecasts and were compared against daily composites from the Global Data Assimilation System (GDAS). The EAS method was demonstrated to have sharper and smaller regions of high probability associated with Greenland barrier jets and tip jets and winds in the Gulf of Tehuantepec off the Mexican coast. The EAS method may better highlight the regional extent of orographically forced wind events.