Atmospheric flow regimes are usually defined as large-scale circulation patterns associated with statistical equilibria in phase space, in which the dynamical tendencies of the large-scale flow are balanced by tendencies due to non-linear interactions of high-frequency transients. The existence of states with such properties can be verified in a rigorous way in numerical simulations with simplified numerical models (as in the pioneering study of Reinhold and Pierrehumbert, 1982, or in the experiments by Vautard and Legras, 1988). By contrast, the existence of flow regimes in the real atmosphere has been strongly debated. The detection of regimes in the observational record of the upper-air field is indeed a complex task, which has been approached by a number of research groups with a variety of sophisticated statistical methods (see Section 14.3).
Although the regime classifications provided by the different observational studies were not identical, a ‘core’ number of regimes were consistently detected in most studies devoted to a specific spatial domain. For example, the three northern-hemisphere clusters found by Cheng and Wallace (1993) were also identified by Kimoto and Ghil (1993a), Corti et al. (1999) and Smyth et al. (1999). However, consistency does not necessarily imply statistical significance, and one may question whether the level of confidence attached to these regime classifications is sufficiently high.
The search for regimes in the real atmosphere is also made complex by the fact that, unlike in simple dynamical models, the sources of energy and momentum at the lower boundary display variations on seasonal, interannual and interdecadal timescales.