Nudging Climate Models to the Real Quasi-Biennial Oscillation Reveals Weak Stratospheric Links

High above the equator, some 16 to 35 kilometres into the atmosphere, the winds of the tropical stratosphere perform one of the most regular dances in the climate system. Roughly every 28 months, easterly winds give way to westerlies and then back again, each new wind regime descending slowly downwards before dissipating near the tropical tropopause. This rhythm is the Quasi-Biennial Oscillation, or QBO, and for more than four decades scientists have suspected that it reaches far beyond the tropics, tugging at the polar vortex, reshaping jets and even nudging winter weather over Europe and North America. A major new multi-model study, published in the journal Weather and Climate Dynamics, has now put that suspicion to one of its most rigorous tests yet, and the results are both encouraging and sobering.

The research, led by Martin Andrews of the Met Office Hadley Centre in the United Kingdom, brought together twelve climate models from eleven modelling centres around the world under the umbrella of the Quasi-Biennial Oscillation initiative, known as QBOi, part of the World Climate Research Programme’s activities on atmospheric processes. The team ran three coordinated experiments covering the period 1979 to 2020. In the first, the models generated their own QBOs freely, as they normally would. In the second, the tropical stratospheric winds were nudged, or gently relaxed, towards the winds observed in the ERA5 reanalysis dataset, effectively forcing every model to carry the real QBO. In the third, the QBO was suppressed altogether by relaxing the winds towards a climatological average, erasing the oscillation entirely. Comparing the three allowed the researchers to isolate, with unusual clarity, what a realistic QBO does and does not deliver in terms of remote atmospheric influence.

The headline target was the Holton-Tan Effect, the classic teleconnection first described in 1980, in which the phase of the QBO modulates the strength of the Northern Hemisphere winter polar vortex, the swirling westerly jet known as the Polar Night Jet that circles the Arctic stratosphere. Observations show that during easterly QBO winters the polar jet weakens, because the zero-wind line separating tropical and extratropical flow shifts poleward, channelling large-scale planetary waves into the vortex where they deposit easterly momentum and slow it down. During westerly QBO winters the waves spread more freely and the vortex strengthens. The observed signal is dramatic: in January the difference between easterly and westerly composites reaches about 20 metres per second in the reanalysis data.

What the models produced was far more modest. In the free-running experiment, the ensemble mean captured a statistically significant but weak Holton-Tan-like response, peaking at only about 2 metres per second, roughly a tenth of the observed magnitude, and it entirely missed the reversal of the signal seen in observations in March. When the models were nudged towards the observed QBO, the response roughly doubled, peaking near 5 metres per second in December and January, and the number of models showing a negative polar jet response during easterly phases rose from five out of eleven to ten out of twelve. Yet even this improved response amounted to only about 20 per cent of the observed signal. Statistical sub-sampling showed that the nudged and free-running distributions of polar jet responses were not significantly different from each other, suggesting that the improvement came mainly from correcting the too-weak amplitude of the simulated QBOs rather than from strengthening the underlying teleconnection mechanism itself.

The study also probed how well the QBO phase influences sudden stratospheric warmings, the dramatic events in which the polar vortex breaks down and stratospheric temperatures can leap by tens of degrees within days. Observations indicate roughly 0.9 warmings per easterly-QBO winter compared with about 0.5 per westerly-QBO winter, and warmings tend to arrive about a month earlier during easterly phases. The models captured only a small and marginally significant increase in warming frequency during easterly phases, and even then only when the analysis was restricted to models with a significant Holton-Tan Effect. More strikingly, neither the free-running nor the nudged experiments reproduced the observed separation in warming timing between the two QBO phases, nor the observed tendency for final warmings of spring to arrive earlier after westerly-QBO winters. The team notes that models with weaker climatological polar vortices produce more warmings overall, a bias that may be muddying the picture.

One genuinely encouraging result concerned predictability. When the nudged ensemble was compared directly with the observed January polar jet over the 42-year period, the correlation was a highly significant 0.61, implying that up to 30 to 40 per cent of the observed year-to-year variability of the January vortex may be attributable to the equatorial stratosphere, including the QBO. In the experiment with the QBO erased, the correlation collapsed to an insignificant 0.06, despite identical sea surface temperatures, sea ice and atmospheric forcings. This is a powerful demonstration that a forecast system carrying the true phase and amplitude of the QBO holds real predictive information about the winter polar vortex, information that vanishes when the oscillation is removed. The authors suggest that the combination of reasonable skill with a too-weak predicted signal may reflect the well-known signal-to-noise paradox that afflicts seasonal-to-decadal forecasting.

The connection to surface weather proved harder to pin down. The North Atlantic Oscillation, the seesaw of pressure between the Azores and Iceland that steers winter storms across Europe, is only weakly correlated with the QBO even in observations, with correlations of about 0.2 to 0.3 in January. In the model ensembles the correlation was smaller still, generally below 0.1 and statistically insignificant, accounting for less than one per cent of the NAO’s variability. Even when the analysis was restricted to the subset of models with a significant Holton-Tan Effect, only a weak positive January correlation emerged, and its significance vanished when an alternative two-station NAO index was used. Because the downward pathway from a strong or weak polar vortex to the tropospheric circulation is itself well captured by the models, the failure appears to lie in the weak QBO-to-vortex link upstream, which starves the surface response of amplitude.

In the North Pacific, the picture was more nuanced. Observations show that the QBO phase shifts the subtropical jet, with easterly-QBO winters producing a weakening of the jet around 30 degrees north and a strengthening near 50 degrees north, a signature of a slight northward displacement. The nudged ensemble captured a significant wind anomaly on the equatorward flank of the jet, something the free-running ensemble failed to produce, and it also reproduced observed patterns of tropical precipitation anomalies over the Pacific that may act as a tropospheric bridge for QBO influences. However, the model response poleward of 35 degrees north actually degraded with nudging, and the entire latitudinal pattern appeared shifted about five degrees too far north, a systematic bias the authors could not fully explain. The confounding influence of the El Nino-Southern Oscillation loomed large here: filtering to only ENSO-neutral winters changed both the sign and the spatial structure of the wind and precipitation signals, underscoring how entangled these tropical drivers are.

The overall message is a careful one. Bias-correcting the QBO in climate models demonstrably helps: it strengthens the polar jet response, improves the representation of tropical precipitation and subtropical wind anomalies, and unlocks genuine predictive skill for the January vortex. But a realistic QBO alone is not sufficient. The models’ internal machinery, the way planetary waves propagate, break and deposit momentum in response to the equatorial winds, remains too feeble, and key observed signatures such as the timing of stratospheric warmings and the March reversal of the Holton-Tan signal remain out of reach. Whether the remaining gaps stem from biases in the extratropical stratosphere or from a fundamentally incomplete representation of the coupling mechanisms is, the authors conclude, a central question for the next generation of experiments. For now, the QBO’s whisper to the mid-latitudes is real, measurable, and only partially heard by the models that try to simulate it.

Subject of Research: The influence of the equatorial Quasi-Biennial Oscillation on extratropical stratospheric and tropospheric circulation in nudged multi-model climate experiments

Article Title: Extratropical teleconnections in a multi-model ensemble nudged towards the observed QBO

Article References: Andrews, M. B., Butchart, N., Anstey, J. A., Bednarz, E., Elsbury, D., García-Franco, J. L., Kumar, V., Palmeiro, F. M., Trencham, N. E., Yoshida, K., Chai, Z., Hong, D.-C., Huang, K., Jaison, A. M., Kawatani, Y., Knight, J. R., Lin, P., Lott, F., Lu, Y., … Xie, J. (2026). Extratropical teleconnections in a multi-model ensemble nudged towards the observed QBO. Weather and Climate Dynamics, 7(3), 1797-1820. https://doi.org/10.5194/wcd-7-1797-2026

Image Credits: AI Generated

DOI: 10.5194/wcd-7-1797-2026

Keywords: Quasi-Biennial Oscillation, stratosphere, polar vortex, Holton-Tan Effect, sudden stratospheric warming, North Atlantic Oscillation, subtropical jet, climate models, teleconnections, QBOi, nudging, ERA5 reanalysis