2026 News & Events

Reanalysis of stratospheric air samples provides compelling evidence that stratospheric circulation has accelerated in recent decades, helping to resolve a long-standing debate about the strength of the stratospheric circulation.

The stratosphere – a stable layer of the atmosphere above the weather‑filled troposphere, extending from roughly 10 to 50 km altitude – hosts a large-scale overturning circulation known as the Brewer–Dobson circulation. This circulation slowly lifts air out of the tropics and transports it polewards through the stratosphere over the course of years, redistributing ozone, greenhouse gases, and other chemically active compounds across the globe. For decades, chemistry–climate models have consistently predicted that this circulation should intensify in response to rising greenhouse gas concentrations and the decline of ozone-depleting substances. Observations, however, appeared to tell a different story: stratospheric air seemed to be getting older rather than younger, suggesting little change or even a slowdown. Writing in Nature Geoscience, lead author Eric Ray, a CIRES scientist working at NOAA CSL, et al. resolve this long-standing discrepancy by identifying a systematic incompatibility between two widely used observational sampling methods. Once this bias is accounted for, the observations align with model expectations, revealing that the stratospheric circulation has in fact been accelerating.

To understand why this matters, it helps to consider how scientists infer the strength of the Brewer–Dobson circulation. Because this circulation is too slow and spatially diffuse to be tracked directly from wind measurements, researchers rely instead on the concept of mean age of air – the average time an air parcel has spent in the stratosphere since it last crossed the boundary between the stratosphere and the troposphere, known as the tropopause. Younger mean ages indicate faster transport, whereas older mean ages imply slower circulation. Mean age is inferred from trace gases whose concentrations have been rising steadily at the Earth's surface for decades, such as sulfur hexafluoride and carbon dioxide. Air that has entered the stratosphere recently carries higher concentrations than air isolated there for many years. Most previous estimates of mean age trends in the Northern Hemisphere pointed towards increasing age, suggesting a weakening of the Brewer–Dobson circulation4 and deepening the apparent mismatch with model predictions.

Ray et al. exploit a well-established physical constraint to test the consistency of observational estimates of stratospheric mean age of air. In the extratropical stratosphere, the Brewer–Dobson circulation both sets how long air has resided in the stratosphere and controls how much nitrous oxide is destroyed by ultraviolet radiation during its transit. In effect, two independent clocks are timing the same journey. Mean age of air and normalized nitrous oxide – defined as the measured nitrous oxide concentration relative to its current surface value – are expected to exhibit a tight, predictable relationship throughout the extratropical stratosphere5,6. The authors use this link between mean age and nitrous oxide as a consistency test for two commonly used sampling approaches: discrete flask samples collected by balloons and analysed in the laboratory, and real-time in situ measurements from aircraft and a new balloon-borne system7. The in situ data, comprising nearly one million individual measurements, fall onto a coherent, physically consistent curve. By contrast, the flask samples show systematically younger mean ages and much greater scatter at the same normalized nitrous oxide levels, with no dependence on time or latitude – effectively ruling out natural atmospheric variability as the cause. The authors therefore exclude the flask-sampled data from their trend analysis, concluding that the two sampling approaches are fundamentally incompatible.

Once flask measurements are set aside, a strikingly different picture emerges. Using in situ observations from the 1990s to the present— supplemented by independent satellite retrievals – Ray et al. find that extratropical mean ages in the Northern Hemisphere have decreased significantly at all sampled altitudes up to 27 kilometres. This trend is opposite in sign to most prior estimates and aligns with the long-standing model prediction of an accelerating Brewer–Dobson circulation since the 1990s. The acceleration is evident across the stratosphere, from the lower to the upper levels. Chemistry–climate models, however, underestimate its magnitude, particularly in the upper stratosphere. Whether this discrepancy reflects a missing forced response to long-term changes in greenhouse gases and ozone-depleting substances, deficiencies in parameterized gravity-wave forcing, unforced multi-decadal variability, or a combination of these factors, remains an open question.

Ray et al. resolve an observational contradiction that has persisted for more than two decades, even though the physical reason why flask-sampled mean ages deviate from in situ measurements remains unclear. By excluding the incompatible flask data, the authors show that the Brewer–Dobson circulation has been accelerating in the Northern Hemisphere since the 1990s, bringing observations into qualitative agreement with the direction of change long predicted by chemistry–climate models8,9. This convergence matters beyond stratospheric dynamics alone. The Brewer–Dobson circulation controls the polewards transport of ozone, regulates the exchange of trace gases between the stratosphere and troposphere, and sets the residence time of aerosols injected into the stratosphere by volcanic eruptions or proposed solar radiation management – all central to projections of ozone recovery and surface climate. Yet identifying the correct sign of the trend is only a first step. A substantial quantitative gap between observed and simulated changes remains, and current constraints are largely limited to the Northern Hemisphere extratropics, leaving Brewer–Dobson circulation changes in the tropics and Southern Hemisphere poorly understood.

Closing the quantitative gap between observed and modelled circulation trends, and extending observational constraints beyond the Northern Hemisphere, will require progress on multiple fronts, including expanded and better coordinated observations across platforms and hemispheres, improved model representations of the processes that govern stratospheric transport, and targeted model-intercomparison efforts to systematically diagnose where current simulations fall short. The results of Ray et al. also prompt a reassessment of previous model-based attribution studies, particularly the relative roles of greenhouse gases and ozone-depleting substances in driving the Brewer–Dobson circulation acceleration. This effort comes at a moment when the imminent decommissioning of key satellite instruments threatens to usher in a data-sparse era for stratospheric monitoring, precisely when the stakes could not be higher.