Potential predictability associated with nonlinear regimes in an atmospheric model

Numerous previous studies addressed mid-latitude atmospheric flow patterns, or regimes, that persist for periods of time exceeding typical lifetimes of weather systems (that is, a few days). This enhanced persistence of the regimes is due to their non-linear dynamics. In this study, output of a realistic atmospheric model is analyzed to examine potential medium-range predictability associated with regime behavior. The regimes here were defined as the regions of the model phase space characterized by excess probability of persistence relative to a benchmark linear statistical model, for two-dimensional, as well as zonally averaged flow patterns. Three regimes were identified in each case, but only four out of six regimes turned out to be statistically distinct. All of the four regimes correspond to certain phases of well-known teleconnection patterns. Probabilities of lagged co-occurrences between the regimes suggest two preferred regime transition paths. While the regimes are defined in terms of deviations from linear-model-based persistence, it turns out that linear model forecast skill is maximum in the regime regions of the phase space. This is the consequence of enhanced regime persistence caused by nonlinear processes, which makes higher signal-to-noise ratio regimes easier to predict upon the regime onset. We developed a more comprehensive measure of a predictive model “nonlinear” skill, which involves transition probabilities between the regimes rather than the quantities related to regime maintenance.