REPRODUCIBLE FORCED MODES IN AGCM ENSEMBLE INTEGRATIONS AND POTENTIALPREDICTABILITY OF ATMOSPHERIC SEASONAL-VARIATIONS IN THE EXTRATROPICS

An approach to assess the potential predictability of the extratropica l atmospheric seasonal variations in an ensemble of atmospheric genera l circulation model (AGCM) integrations has been proposed in this stud y by isolating reproducible forced modes and examining their contribut ions to the local ensemble mean. The analyses are based on the monthly mean output of an eight-member ensemble of 10-yr Atmospheric Model In tercomparison Project integrations with a T42L18 AGCM. An EOF decompos ition applied to the ensemble anomalies shows that there exist some fo rced modes that are less affected by the internal process and thus app ear to be highly reproducible. By reconstructing the ensemble in terms of the more reproducible forced modes and by developing a quantitativ e measure, the potential predictability index (PPI), which combines th e reproducibility with the local variance contribution, the local ense mble mean over some selective geographic areas in the extratropics was shown to result primarily from reproducible forced modes: rather than internal chaotic fluctuations. Over those regions the ensemble mean i s potentially predictable. Extratropical potentially predictable regio ns are found mainly over North America and part of the Asian monsoon r egions. Interestingly. the potential predictability over some preferre d areas such as Indian monsoon areas and central Africa occasionally r esults primarily from non-ENSO-related boundary forcing. although ENSO forcing generally dominates over most of the preferred areas. The qua ntitative analysis of the extratropical potential predictability with PPI has shown chat the preferred geographic areas have obvious seasona lity. For the 850-hPa temperature, for example, potentially predictabl e regions during spring and winter are confined to Alaska, northwest C anada, and the southeast United States, the traditional PNA region, wh ile during summer and fall they are favored over the middle part of No rth America. It has also been shown that the boreal summer season (Jun e-August) possesses the largest potentially predictable area, which se ems to indicate that it is a favored season for the extratropical pote ntial predictability. On the contrary, boreal winter (December-Februar y) appears to have a minimum area of extratropical potential predictab ility. The results have been compared with the more traditional statis tical tests for potential predictability and with observations from th e National Centers for Environmental Prediction reanalysis, which indi cates that the PPI analysis proposed here is successful in revealing e xtratropical potential predictability determined by the external forci ng.