| A series of experiments with the GFDL R15 spectral model is used to address aspects of atmospheric variability on time scales ranging from weeks to decades. The experiments are performed in a perpetual January mode.; The Northern Hemisphere low-frequency variability in the control run, in which sea surface temperatures are kept fixed, is dominated by a nearly zonally symmetric mode. The variability associated with this mode exhibits red-noise characteristics on time scales up to about one year. The addition of a midlatitude slab mixed layer in a second model experiment results in a slight increase in low-frequency variability, particularly in that associated with the zonally symmetric mode.; In two subsequent experiments, realistic, El Nino-Southern Oscillation (ENSO)-like sea surface temperature variability is specified in the tropical Pacific. It is shown that the atmospheric response pattern associated with the ENSO forcing corresponds well to the observed. The time series associated with the midlatitude response pattern is somewhat redder than the forcing time series, hinting at the possibility that the predominantly interannual ENSO forcing is instrumental in the generation of interdecadal variability in the extratropics. The addition of a slab mixed layer in midlatitudes to the tropical forcing in the fourth experiment does not appear to increase the amplitude of the instantaneous atmospheric response; it does, however, increase the persistence of the response and the total variance associated with the ENSO response pattern.; The 100,200 day long record from the control run is used to investigate the possible existence of bimodal probability density functions in the planetary-scale flow. It is shown that while relatively short segments of such an index, calculated from the control experiment, exhibit hints of bimodality similar to those found in the observations, the index for the entire run does not show any clear signs of multiple modes. This suggests that published observation-based results purporting to have demonstrated the existence of bimodality are not statistically significant. |
