Persistent Anomalies of the Extratropical Northern Hemisphere Wintertime Circulation: Geographical Distribution and Regional Persistence Characteristics

Abstract

We have studied the geographical and regional persistence characteristics of wintertime Northern Hemisphere 500 mb height anomalies, focusing particular attention on the behavior of strong anomalies that persist beyond the durations associated with synoptic-scale variability (?persistent anomalies?). We have also examined the persistence characteristics of certain dominant regional patterns of low-frequency variability. There are three major regions for the occurrence of persistent anomalies: the North Pacific to the south of the Aleutians, the North Atlantic to the southeast of Greenland, and from the northern Soviet Union northeastward to over the Arctic Ocean. These regions have relatively high numbers of both persistent positive anomaly and persistent negative anomaly cases. For moderate magnitudes and durations, the numbers of positive and negative cases in each region are about the same; however, for larger magnitudes and longer durations, the number of positive cases exceeds the corresponding number of negative cases. Analyses with data that have been low-pass filtered (removing periods of less than 6 days) reveal that part (but not all) of the discrepancy between positive and negative cases results from the relatively greater likelihood that negative anomalies will experience brief interruptions by transient disturbances. For durations beyond about 5 days, the probability that an anomaly which has lasted n days will last at least one more day is nearly constant. This nearly constant probability of continuation resembles the behavior obtained for a linear first-order autoregressive process (red noise). Nevertheless, there are significant differences in persistence between the positive and negative anomalies and red noise, particularly at large magnitudes, with the positive anomalies typically more persistent than either the negative anomalies or red noise. A simple nonlinear autoregressive model is described that simulates many of the observed deviations from red noise, and possible physical sources for the nonlinearities are discussed. Relationships between the initial anomaly value and its subsequent 12 h change are then studied. The height changes are decomposed into two parts: a mean change and a deviation from the mean change. Mean change variations are examined for evidence of multiple ?quasi-equilibria? (multiple anomaly values having mean changes of zero). Mean change variations are also determined for the temporal coefficients of certain dominant regional patterns of low-frequency variability. Although the temporal fluctuations of the patterns exhibit considerably more persistence than found for the corresponding local height anomalies, neither the patterns nor the local anomalies display convincing evidence of multiple quasi-equilibria.