Does low-frequency atmospheric variability exhibit regime-like behavior?

If the general circulation exhibits nonlinear, regime-like behavior, it should be reflected in the frequency distributions of the state of the atmospheric in phase space either in the form of bi- (or multi) modality or skewness. In order to determine whether the Northern Hemisphere winter circulation exhibits these characteristics, we examine frequency distributions of (1) the inner product or spatial covariance (2) the (spatial) “anomaly correlation” and (3) the root mean squared difference or “distance” between each map and each other map in phase space. Our distributions are based on a sample of 702 10-day lowpass filtered, hemispheric 500 mb height anomaly maps (18 maps at 5-day intervals for the equivalent of 39 winter seasons). Excluding pairs of maps from the same winter season, the sample size for the distributions is 240,156. All the frequency distributions of (1) and (2) for the hemispheric maps exhibit a close fit to the corresponding randomly generated distributions based on 20 spatial degrees of freedom. There is no evidence of bimodality. However, the numbers of large positive spatial covariances and anomaly correlations greatly exceed the numbers of their counterparts in the negative tails of the frequency distributions (i.e. “analogues” outnumber “antilogues” of comparable quality). Using a simple form of cluster analysis we show that three rather distinctive blocking patterns account for a disproportionate share of the largest positive values in the frequency distributions. The asymmetries in our frequency distributions may be a reflection of regime-like behavior, but it is also conceivable that they could be the signature of more subtle forms of nonlinearity in the low frequency dynamics.