The Euro-Atlantic Circulation Response to the Madden-Julian Oscillation Cycle of Tropical Heating: Coupled GCM Intervention Experiments

Intervention experiments using the Coupled Forecast System model, version 2 (CFSv2), have been performed in which various Madden-Julian Oscillation (MJO) evolutions were added to the model's internally generated heating: Slow Repeated Cycles, Slow Single Cycle, Fast Repeated Cycles, and Fast Single Cycle. In each experiment, one of these specified MJO evolutions of tropical diabatic heating was added in multiple ensemble reforecasts of boreal winter (1 November to 31 March for 31 winters: 1980-2010). Since in each experiment, multiple re-forecasts were made with the identical heating evolution added, predictable component analysis is used to identify modes with the highest signal-to-noise ratio. Traditional MJO-phase analysis of total model heating (dominated by internally generated heating) shows that the MJO-related heating structure compares well with heating estimated from observed fast and slow episodes; however, the model heating is larger by a factor of two. The evolution of Euro-Atlantic circulation regimes indicates a clear response due to the added heating, with a robust increase in the frequency of occurrence of the negative phase of the North Atlantic Oscillation (NAO-) after the heating crosses into the Pacific and a somewhat less robust increase in the positive phase of the NAO (NAO+) following Indian Ocean heating. In the Fast Cycle experiments, the model response is somewhat muted compared with the Slow Cycle experiments. The Scandinavian Blocking regime becomes more frequent prior to the NAO- regime. The two leading modes in the predictable component analysis of 300 hPa height (Z300), synoptic scale feedback (DZ300), and planetary wave diabatic heating in all experiments form an oscillatory pair with high statistical significance. The oscillatory pair represents the cyclic response to the particular MJO signal (Fast or Slow, Single, or Repeated Cycles) in each case. The period is about 64 days for the Slow Cycle and 36 days for the Fast Cycle, consistent with the imposed periods. The time series of one of the leading modes of Z300 is highly anti-correlated with the frequency of occurrence of the NAO- in the Repeated Cycle experiments. A clear cycle involving the Z300 and DZ300 leading modes is identified.