10:45 AM - 12:15 PM
[AAS09-P07] Forecast skill and SN ratio for stratospheric polar vortex variations in Northern winter seasonal hindcasts of the C3S systems
Keywords:stratospheric polar vortex, sudden stratospheric warmings, seasonal forecasts
This study explores forecast skill and signal-to-noise (SN) ratio for stratospheric polar vortex (SPV) variations using Northern winter seasonal hindcast (HC) data of six Copernicus Climate Change Service systems initialized around early November. The SN ratio is evaluated by comparing the degree to which the HC data forecast observations to the degree to which the HC data forecast themselves. The HC data are verified in comparison to the JRA-55 reanalysis data. The target period extends from 1993/1994 to 2016/2017.
Analysis results show high skill for DJF mean El Niño/Southern Oscillation (ENSO) and Quasi-Biennial Oscillation (QBO) variations, although HC equatorial stratospheric wind anomalies tend to decay with time. The HC ENSO and QBO variations are also shown to have too high SN ratio, i.e. to be underdispersive. The forecast skill for DJF mean SPV variations is close to 95 % significance, whereas the skill for occurrence or absence of major sudden stratospheric warmings (MSSWs) during DJF is much lower (insignificant at the 95 % level). The SN ratio is reasonable for both quantities.
A further comparison of the ensemble members for each system, when divided into two groups by the strength of the QBO-SPV teleconnection, suggests an increase in the skill for SPV variations for the group of stronger teleconnection, although the HC teleconnection in most ensemble members is much weaker than observed. Such an increase is not shown for the occurrence/absence of MSSW winters, probably due to very weak or absent teleconnection between the QBO and MSSW winters in the HC data. It is suggested that the skill for SPV variations can be improved as the HC QBO-SPV teleconnection in addition to the QBO itself is better simulated.
Analysis results show high skill for DJF mean El Niño/Southern Oscillation (ENSO) and Quasi-Biennial Oscillation (QBO) variations, although HC equatorial stratospheric wind anomalies tend to decay with time. The HC ENSO and QBO variations are also shown to have too high SN ratio, i.e. to be underdispersive. The forecast skill for DJF mean SPV variations is close to 95 % significance, whereas the skill for occurrence or absence of major sudden stratospheric warmings (MSSWs) during DJF is much lower (insignificant at the 95 % level). The SN ratio is reasonable for both quantities.
A further comparison of the ensemble members for each system, when divided into two groups by the strength of the QBO-SPV teleconnection, suggests an increase in the skill for SPV variations for the group of stronger teleconnection, although the HC teleconnection in most ensemble members is much weaker than observed. Such an increase is not shown for the occurrence/absence of MSSW winters, probably due to very weak or absent teleconnection between the QBO and MSSW winters in the HC data. It is suggested that the skill for SPV variations can be improved as the HC QBO-SPV teleconnection in addition to the QBO itself is better simulated.