Predictability of Major Stratospheric Sudden Warmings: Analysis Results from JMA Operational 1-Month Ensemble Predictions from 2001/02 to 2012/13

This study investigates the predictability of major stratospheric sudden warmings (MSSWs) with a chief question: how far in advance can MSSWs be forecasted? An average picture and case-to-case variations of the MSSW predictability are revealed by analyzing operational 1-month ensemble prediction data of the Japan Meteorological Agency from 2001/02 to 2012/13 in comparison with the Japanese 55-year Reanalysis Project (JRA-55) data. The variations are further related to planetary wave forcing (PWF) from the troposphere to the stratosphere. A contingency table analysis for nine MSSWs occurring in the period shows that the average percentage of ensemble members that successfully forecast the MSSWs is about 70%, 30%, and 20% for lead times of 5, 10, and 15 days, respectively, when a 3-day time difference between actual and forecasted vortex collapses, or zonal wind reversals, is allowed. Using other measures such as the root-mean-square error and anomaly correlation of 10-hPa geopotential height, forecasts of lead times less than about two weeks are judged to be useful. Results also show large case-to-case predictability variations for lead times of about 10 days. The variations for most MSSWs are explained by the degree to which the strength of PWF, or wave activity flux in the lower stratosphere, is forecasted in comparison to the JRA-55 data. Forecasted PWF in the lower stratosphere is largely determined by that in the upper troposphere for several MSSWs, whereas it is also affected by the planetary wave propagation between the two regions for a few others.