Interannual Variability in the Stratosphere–Troposphere Exchange of Air Mass and Ozone in Chemistry Climate Models

Stratosphere–troposphere exchange (STE) governs the transport of air mass and chemical constituents, including ozone, across the tropopause and therefore plays an important role in coupling the stratosphere and troposphere. While the annual-mean STE has been well characterized, its interannual variability and the processes that control it remain uncertain. Recent studies based on ERA5 and MERRA2 reanalysis datasets show that large-scale climate modes such as the El Niño–Southern Oscillation (ENSO), the Quasi-Biennial Oscillation (QBO), and variability in the Brewer–Dobson circulation (BDC) explain only a small fraction of STE variability. Moreover, these two reanalyses can explain only about 33% of each other’s variance in global ozone STE monthly anomalies. This raises the question of whether the inferred drivers of STE variability are robust or instead reflect limitations in the underlying reanalysis datasets. Herein we investigate the climatology and interannual variability of the STE of air mass and ozone using six different models from Phase 2 of the Chemistry–Climate Model Initiative (CCMI). Models provide dynamically consistent representations of stratospheric transport and chemistry, offering an independent framework for evaluating the dynamical controls on STE variability. We apply a lagged multiple linear regression framework to monthly anomalies to quantify the contributions of ENSO, the QBO, and BDC variability. To isolate BDC variability independent of ENSO and QBO, we regress out ENSO and QBO signals from the BDC index. The relative importance of these drivers varies by region and between air mass and ozone exchange. Across the models on a global basis, ENSO explains the largest fraction of variance in ozone STE, with a mean contribution of 40% (range: 24–65%). The BDC contributes a smaller but non-negligible share, with a mean of 13% (range: 7–21%), while the QBO accounts for a mean contribution of 8% (range: ~0–29%). Nevertheless, a substantial fraction of variability in global ozone STE remains unexplained, ranging from 19 to 62%, with a mean of 39%. Applying the same methodology to the ERA5 and MERRA-2 reanalyses yields larger unexplained variance of 71% and 81%, respectively, in global ozone STE. Despite the large unexplained residuals in the reanalyses and the wide range across CCMI models, the relative response of ozone STEs exhibits consistent signs: positive for the BDC, negative for the QBO, and negative for ENSO in the tropics but positive over the extratropics, based on both reanalyses and most CCMI models.