Ensemble Forecast

An ensemble forecast runs a numerical model many times with perturbed initial conditions and model physics to sample forecast uncertainty rather than producing a single deterministic track. The spread among members estimates the probability distribution of future states, so a tight ensemble signals confidence and a wide spread signals a poorly constrained situation. ECMWF runs a 51-member ensemble; the GEFS and others run similar member counts. Probabilistic products such as the chance of exceeding a wave height or wind threshold come directly from member counts. Mariners use ensemble spread to judge how firm a routing forecast is before committing to a departure window.