Because El Nino/Southern Oscillation (ENSO) produces relatively small shifts in the mean of atmospheric variables, previous studies on the predictability of seasonal-to-interannual variations have focused on seasonal mean anomalies. We show that ENSO effects on daily variance are not small, have implications for the daily risk of extreme events, and may be predictable.

The present study estimates the predictability of synoptic (2-8 day) variance anomalies associated with anomalous storm tracks and compares this to the predictability of seasonal mean precipitation and 500 mb height. We make quantitative estimates of the predictability based on a previously-derived analytical expression that is valid for any distribution. NCEP/NCAR Reanalysed data (1948-2000) and general circulation model (GCM) simulations of northern hemisphere winter (January through March) are used. The NCEP atmospheric GCM is integrated with prescribed seasonally-evolving sea surface temperatures for warm, cold, and neutral ENSO conditions. A large number of seasonal integrations, differing only in initial condition, are made for observed climatological mean SSTs, SSTs for an observed warm event (1987), and SSTs for an observed cold event (1989). A total ensemble of 540 members is generated, 180 members each for warm, cold, and neutral conditions. With this large ensemble, we show the potential predictability of seasonal means and storm track anomalies even in regions not usually associated with an ENSO effect.