Abstract
This study investigates winter forecasts of the northern stratosphere and troposphere using seasonal hindcast experiments of the Japan Meteorological Agency (JMA). A main focus is placed on the seasonal forecasts of the December–February (DJF)-mean northern annular mode (NAM) when the forecasts are initialized in late fall. Results demonstrate that the hindcast data have significant skill for both ensemble-mean and category (probability) forecasts of the NAM but only in the stratosphere. Probability forecasts for DJF major stratospheric sudden warmings (MSSWs) are also suggested to be significant (higher probabilities for actual MSSW years) near the 90% confidence level. The forecast skill of the stratospheric NAM changes with the observed phase of the quasi-biennial oscillation (QBO), although the QBO is not simulated but is only included in initial conditions. The skill is higher for the easterly phase, characterized by hits of negative NAM states (weaker-than-normal polar vortex), whereas it is lower for the westerly phase, reflecting misses of positive NAM states. It is finally shown that a verification score for category forecasts of the stratospheric and tropospheric NAM tends to covary as a whole. The tropospheric forecast skill is significant when the stratosphere has large NAM anomalies in the real world and they are well forecasted. In contrast, the tropospheric forecasts are sometimes poor when the stratospheric forecasts fail to capture observed NAM conditions. It is speculated that stratospheric and tropospheric forecasts could be improved together through the stratosphere–troposphere coupling for such cases, that is, by successfully forecasting anomalous vortex states in the stratosphere.
© 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).
