Evaluation of Global Numerical Weather Prediction Model Extended-Range Forecasts

The accuracy of weather forecasts from global numerical weather prediction (NWP) models has been steadily improving over the last several decades and is currently very high for forecasts up to 7 days, especially with implementations of data assimilation of the Geostationary (GEO) and Low Earth Orbit (LEO) satellite observations.  The NWP model extended-range forecasts are now in high demand by the public and private sectors.  This paper is intended to present the preliminary results on the evaluation of the NPW model extended-range forecasts using the WMO Year of Tropical Convection/Madden Julian Oscillation (YOTC/MJO) 20-day forecasts and the WMO/WWRP/THORPEX-WCRP Subseasonal to Seasonal Project (S2S) datasets. These archived datasets are ideal for our purposes because almost all current available global NPW operational/research models and their coupled ensemble systems are involved. The Arctic Oscillation index (AO) and Antarctic oscillation index (AAO) have been applied to analyze the skill of these individual systems and their ensemble forecasts.  Results indicate that overall the global NPW models demonstrate extended-range forecast skill, while the ensemble systems outperform deterministic system at long forecast ranges.  The multi-model ensemble forecasts present the most promise for skillful extended-range forecasts.