Long-Range Prediction and the Stratosphere
Numerous studies have demonstrated that an improved representation of the stratosphere in Earth system models and forecast systems improves subseasonal and decadal predictions and multi-decadal projections. The state of the stratospheric polar vortex, including sudden stratospheric warmings, affect surface prediction, especially of the North Atlantic Oscillation (NAO) on subseasonal timescales. The stratospheric quasi-biennial Oscillation (QBO) affects the Madden-Julian Oscillation (MJO) on monthly timescales and the NAO and El Nino Southern Oscillation (ENSO) teleconnections on monthly to annual timescales. Stratospheric volcanic aerosols, solar variability, ozone, and greenhouse gas concentrations become important for longer-term predictions.
The lack of adequate representation of the stratosphere, and under-estimation of stratospheric-tropospheric teleconnections leads to missed windows of opportunity for improved prediction of the key modes of variability and surface variability and change.
In this study, we provided a historical overview and recent advances in quantifying the role of the stratosphere on prediction from monthly to multi-decadal timescales. The vast majority of Earth system models exhibit the ‘signal-to-noise paradox’, which could be a contributor to the underrepresentation of surface impacts of stratospheric-tropospheric teleconnections, such as the QBO-MJO.