Search

The stratosphere can have a signi_cant impact on winter surface weather on subseasonal to seasonal (S2S) timescales. This study evaluates the ability of current operational S2S prediction systems to capture two important links between the stratosphere and tropo sphere: (1) changes in probabilistic prediction skill in the extratropical stratosphere by precursors in the tropics and the extratropical troposphere and (2) changes in surface predictability in the extratropics after stratospheric weak and strong vortex events. Prob abilistic skill exists for stratospheric events when including extratropical tropospheric precursors over the North Paci_c and Eurasia, though only a limited set of models captures the Eurasian precursors. Tropical teleconnections such as the Madden‐Julian Oscillation, the Quasi‐Biennial Oscillation, and El Nin~o Southern Oscillation increase the probabilistic skill of the polar vortex strength, though these are only captured by a limited set of models. At the surface, predictability is increased over the USA, Russia, and the Middle East for weak vortex events, but not for Europe, and the change in predictability is smaller for strong vortex events for all prediction systems. Prediction systems with poorly resolved stratospheric processes represent this skill to a lesser degree. Altogether, the analyses indicate that correctly simulating stratospheric variability and stratosphere‐troposphere dynamical coupling are critical elements for skillful S2S wintertime predictions.

The stratosphere has been identified as an important source of predictability for a range of processes on subseasonal to seasonal (S2S) timescales. Knowledge about S2S predictability within the stratosphere is however still limited. This study evaluates to what extent predictability in the extratropical stratosphere exists in hindcasts of operational prediction systems in the S2S database. The stratosphere is found to exhibit extended predictability as compared to the troposphere. Prediction systems with higher stratospheric skill tend to also exhibit higher skill in the troposphere. The analysis also includes an assessment of the predictability for stratospheric events, including early and mid‐winter sudden stratospheric warming (SSW) events, strong vortex events, and extreme heat flux events for the Northern Hemisphere, and final warming events for both hemispheres. Strong vortex events and final warming events exhibit higher levels of predictability as compared to SSW events. In general, skill is limited to the deterministic range of one to two weeks. High‐top prediction systems overall exhibit higher stratospheric prediction skill as compared to their low‐top counterparts, pointing to the important role of stratospheric representation in S2S prediction models.