Prediction skill of the NAO and PNA from daily to seasonal time scales

The skill of state-of-the-art operational dynamical models in predicting the two most important modes of variability in the Northern Hemisphere extratropical atmosphere, the North Atlantic Oscillation (NAO) and Pacific-North American (PNA) teleconnection patterns, is investigated at time scales ranging from daily to seasonal. Two uncoupled atmospheric models used for deterministic forecasting in the short to medium range as well as eight fully coupled atmosphere-land-ocean forecast models used for monthly and seasonal forecasting are examined and compared. For the short to medium range, the level of forecast skill for the two indices is higher than that for the entire Northern Hemisphere extratropical flow. The forecast skill of the PNA is higher than that of the NAO. The forecast skill increases with the magnitude of the NAO and PNA indices, but the relationship is not pronounced. The probability density function (PDF) of the NAO and PNA indices is negatively skewed, in agreement with the distribution of skewness of the geopotential field. The models maintain approximately the observed PDF, including the negative skewness, for the first week. Extreme negative NAO/PNA events have larger absolute values than positive extremes in agreement with the negative skewness of the two indices. Recent large extreme events are generally well forecasted by the models. On the intraseasonal time scale it is found that both NAO and PNA have lingering forecast skill, in contrast to the Northern Hemisphere extratropical flow as a whole. This fact offers some hope for extended range forecasting, even though the skill is quite low. No conclusive positive benefit is seen from using higher horizontal resolution or coupling to the oceans. On the monthly and seasonal time scales, the level of forecast skill for the two indices is generally quite low, with the exception of winter predictions at short lead times.