Subseasonal Forecast Skill for Weekly Mean Atmospheric Variability Over the Northern Hemisphere in Winter and Its Relationship to Midlatitude Teleconnections

This study assesses the subseasonal predictability of the weekly mean geopotential height anomaly at 500 hPa and its relationship to teleconnections over the Northern Hemisphere in winter. The skill over the North Pacific, Canada, and Greenland is higher than over other areas for weeks 3 and 4 forecasts. These peaks correspond to the centers of action for the Pacific-North American (PNA) pattern and the North Atlantic Oscillation (NAO). PNA (NAO phase) predictions are better for El Nino years at lead times of 3-4 weeks (2-4 weeks). The effects of La Nina forcing on PNA and NAO forecasts are small compared with the El Nino forcing. Numerical models have a negative PNA bias at these lead times in La Nina years. Results suggest that the improvement in the midlatitude upper-level jet rather than in the response to El Nino-Southern Oscillation forcing in the tropics could lead to better S2S predictions. Plain Language Summary Weather forecasts at medium-range timescales are basically skillful because of the evolution of numerical weather prediction systems. However, numerical models generally struggle to make accurate predictions at subseasonal to seasonal (S2S; 2 week to 2 month range) timescales. Atmospheric low-frequency variability and atmosphere-ocean-land coupled processes are potential sources of improvements to S2S forecast skill. This study investigates the forecast skill of weekly mean atmospheric variability and its relationship to dominant teleconnections over the Northern Hemisphere in winter. The subseasonal forecast skill (3-4 weeks) is higher over the North Pacific, Canada, and Greenland than over other areas. These regions correspond to the centers of action of the Pacific-North American (PNA) pattern and the North Atlantic Oscillation (NAO). Subseasonal forecasts of PNA and NAO are better in El Niflo years than in neutral years. This indicates that ENSO conditions can enhance subseasonal predictability through improvements to PNA and NAO forecasts. In contrast, the numerical model used here tends to have a negative PNA bias in La Nina years. Both higher skill in certain regions and the negative PNA bias depend on the predictions of the upper -level jet in midlatitudes. The upper -level jet predictions in midlatitudes rather than the atmosphere-ocean coupling process predictions in tropics play an important role in S2S timescales. Therefore, reducing the bias in midlatitudes would improve the S2S forecasts.