The EUROS Workpackage Extended weather forecasts in a nutshell
Global weather forecasts are produced on different temporal scales. Global weather forecasts from a few hours up to a few days ahead are called short-range to medium-range forecasts. Extended (long-range) forecast are typically made for 1-45 days ahead and can be used to forecast week-to-week changes in weather. Seasonal forecast produce forecast of weather regimes up to a few months ahead and typically have a spatial resolution of a few tens of kilometers.
Forecasts on the medium-range are available are produced twice daily and include a deterministic forecast and a probabilistic forecast. A deterministic forecasts is obtained by launching a single runs of a weather forecasting model that is run on a high spatial resolution, Probabilistic forecasts are usually done using an Ensemble Prediction System (EPS). In an EPS a number of separate weather forecasts is produced by launching an ensemble of weather forecasting models with a relative low resolution, which differ from each other with respect to the applied initial conditions and the applied model physics. As all weather forecasts within an EPS are all plausible weather forecasts, the forecasts by the ensemble members can be viewed to describe for a certain time and location, the probability density function (pdf) of a weather variable. Both extended and seasonal forecasts are usually produced in the form of an EPS.
To assess the forecastability of relevant wind and wave characteristics on the southern North Sea, we have validated medium-ranged forecasts (up to 10 days ahead) by the European Centre of Medium-Ranged Weather Forecasts against observations taken at the IJmuiden observational tower and observations of wind and waves taken at a number of oil drilling platform in the southern North Sea. Validation is done for both the deterministic high-resolution forecasts and the EPS forecast. It appears that the up to few days ahead, the deterministic forecast outperforms the EPS forecasts, while for longer time horizons the EPS forecasts perform better.