Seasonal predictability of the southern annular mode due to its association with ENSO.

摘要

Predictability of the southern annular mode (SAM) for lead times beyond 1-2 weeks has traditionally been considered to be low because the SAM is regarded as an internal mode of variability with a typical decorrelation time of about 10 days. However, the association of the SAM with El Nino-Southern Oscillation (ENSO) suggests the potential for making seasonal predictions of the SAM. In this study the authors explore seasonal predictability and the predictive skill of SAM using observations and retrospective forecasts (hindcasts) from the Australian Bureau of Meteorology dynamical seasonal forecast system [the Predictive Ocean and Atmosphere Model for Australia, version 2 (POAMA2)]. Based on the observed seasonal relationships of the SAM with tropical sea surface temperatures, two distinctive periods of high seasonal predictability are suggested: austral late autumn to winter and late spring to early summer. Predictability of the SAM in the austral cold seasons stems from the association of the SAM with warm-pool (or Modoki/central Pacific) ENSO, whereas predictability in the austral warm seasons stems from the association of the SAM with cold-tongue (or eastern Pacific) ENSO. Using seasonal hindcasts for 1980-2010 from POAMA2, it is shown that the observed relationship between SAM and ENSO is faithfully depicted and SST variations associated with ENSO are skillfully predicted. Consequently, POAMA2 can skillfully predict the phase and amplitude of seasonal anomalies of the SAM in early summer and early winter for at least one season in advance. Zero-lead monthly forecasts of the SAM are furthermore shown to be highly skillful in almost all months, which is ascribed to predictability stemming from observed atmospheric initial conditions.