Variations in Arctic ozone during winter and spring are influenced by anthropogenic chemical loss and dynamical resupply. Interannual variability is driven by meteorology, with colder, more isolated vortices associated with less ozone resupply and greater chemical loss. Colder vortices result from a weaker Brewer-Dobson Circulation, reduced planetary-wave activity, and lower eddy heat flux in the extratropical lower stratosphere. Ozone loss occurs after chlorine activation on cold sulfate aerosols and supercooled ternary solution droplets (STS), and on nitric acid trihydrate (NAT) particles or water ice in exceptionally cold conditions. The volume of air cold enough for PSC formation (V_PSC) shows a near-linear relationship with column ozone loss. Record V_PSC values have been observed in recent winters. Early GCM simulations suggested that decreased planetary wave activity due to rising GHGs would lead to colder Arctic vortices. More recent simulations indicated future cooling of the Arctic lower stratosphere due to direct radiative cooling and indirect effects from declining Arctic sea ice and rising sea surface temperatures. This study examines trends in PSC formation potential (PFP), using data from four meteorological centers and GCM output from CMIP6 simulations, to assess the future evolution of Arctic ozone loss.

Previous research has established a strong link between Arctic ozone depletion and meteorological conditions, particularly the temperature and extent of polar stratospheric clouds (PSCs). Studies have shown that colder winters, characterized by stronger and more isolated polar vortices, lead to greater ozone loss due to enhanced heterogeneous chemical reactions involving chlorine and bromine. The interannual variability in ozone depletion has been linked to the strength of the Brewer-Dobson circulation, planetary wave activity, and the resulting transport of ozone from lower latitudes. Earlier modeling studies, using general circulation models (GCMs) with coupled active chemistry (CCMs), suggested that increasing greenhouse gas (GHG) concentrations would lead to changes in stratospheric dynamics, potentially impacting Arctic ozone. These studies hinted at a potential future increase in cold conditions favorable for ozone loss. However, the complexity of the interactions between dynamics, chemistry, and radiative forcing necessitated a more comprehensive investigation to predict the future of Arctic ozone.

This study analyzes trends in PSC formation potential (PFP), defined as the number of days a volume of air equal to the polar vortex is exposed to PSC conditions (temperatures below the threshold for NAT formation). Data from four meteorological centers (ERA5/ERA5.1/ERA5 BE, CFSR/CFSv2, MERRA-2, JRA-55) were used to calculate PFP. The Iterative Selection Approach (ISA) was developed as a more robust method to identify statistically significant trends in the local maxima (LM) of PFP (PFP_LM) time series, compared to previous methods like the Maximum in Interval Method (MIM) and the Value Above Sigma (VAS). The analysis also incorporated output from 26 GCMs in CMIP6, focusing on models that submitted results for the SSP5-8.5, SSP3-7.0, SSP2-4.5, and SSP1-2.6 runs. To account for temperature biases in the GCMs, a temperature offset was applied to each model's PSC threshold to match the observed magnitude of PFPLM during the modern satellite era. The chemical loss of ozone was modeled using the ATLAS Chemistry and Transport Model. Projections of stratospheric halogen loading and humidity were combined with GCM-based temperature forecasts to evaluate the evolution of Arctic ozone loss as a function of future GHG levels and stratospheric H₂O. The impact of future increases in stratospheric H₂O, driven by increasing tropospheric CH₄ and warming of the tropical tropopause, was considered. Monte Carlo simulations were used to assess the statistical significance of trends in PFP_LM. The study also assessed GCM performance by examining their representation of the quasi-biennial oscillation (QBO).

The study found positive, statistically significant trends in PFP_LM over the past four decades using data from four meteorological centers. The slope of the LM of PFP (SPFP-LM) ranged from 3.85 ± 0.40 d decade⁻¹ (MERRA-2) to 4.77 ± 0.48 d decade⁻¹ (CFSR), with a mean of 4.26 ± 0.45 d decade⁻¹. Analysis of CMIP6 GCM output showed positive trends in PFP_LM for all 26 SSP5-8.5 simulations, ranging from 0.62 ± 0.09 d decade⁻¹ to 3.66 ± 0.16 d decade⁻¹. The majority of slopes lay between 1.0 and 2.5 d decade⁻¹, with statistical significance at better than the 2σ level for 16 of the 26 runs. The relationship between column ozone loss (ΔO₃) and ozone loss potential (OLP), defined as OLP(yr) = [EESC(yr)^1.2/EESC_MAX] × PFP(yr), where EESC_MAX is the maximum yearly value of equivalent effective stratospheric chlorine (EESC), showed high correlation (r² = 0.89 for measured ΔO₃ and 0.96 for modeled ΔO₃). Projections of OLP from GCMs, assuming constant stratospheric H₂O, showed that for SSP5-8.5 and SSP3-7.0, OLP remained well above the 1980 level until the end of the century. When the impact of rising H₂O due to both CH₄ oxidation and tropical tropopause warming was considered, OLP and ΔO_3REG were even higher than contemporary values for the SSP5-8.5 and SSP3-7.0 simulations. Conversely, for SSP2-4.5 and SSP1-2.6, OLP and ΔO_3REG declined toward 1980 levels or below. Analysis of CMIP5 GCM output showed similar trends. The observationally based trend in SPFP-LM over 1980-2020 lay near the upper 1σ of GCM values, possibly due to the influence of internal variability and tropospheric climate shifts. The study also examined the models' representation of the QBO, and while CMIP6 models generally showed improved representation, deficiencies remained, particularly below 20 hPa.

The findings strongly suggest that rising GHG levels are driving increased PFP_LM, leading to conditions favorable for enhanced ozone loss. The positive trends in PFP_LM observed in both reanalysis data and GCM projections highlight the influence of climate change on stratospheric temperatures. The strong dependence of OLP on radiative forcing indicates the potential for significant, prolonged chemical ozone loss despite the expected decline in stratospheric halogens due to the Montreal Protocol. The inclusion of projected increases in stratospheric H₂O further exacerbates this potential. The study acknowledges limitations in GCM representation of certain stratospheric processes, particularly the QBO, and the use of prescribed ozone fields in some models. Despite these limitations, the consistent trends across multiple models and datasets strengthen the conclusions. While the study focuses on chemical loss, it acknowledges that future total column ozone will be influenced by a complex interplay of chemical and dynamical factors.

The study concludes that climate change, driven by rising GHGs, could lead to a prolonged period of significant chemical ozone loss in the Arctic stratosphere. Even with the decline in stratospheric halogens due to the Montreal Protocol, the potential for substantial ozone loss could persist and potentially exceed contemporary levels by the end of the century, particularly under high GHG emission scenarios (SSP5-8.5 and SSP3-7.0). Future research should focus on improving the representation of stratospheric dynamics, including the QBO, and interactive chemistry within GCMs to refine projections of Arctic ozone.

The study acknowledges limitations stemming from GCM biases in representing stratospheric temperatures and dynamics, particularly the QBO. The use of prescribed ozone fields in some GCMs might also influence the accuracy of the results. The simplified approach to modeling stratospheric H₂O increase due to CH₄ oxidation and tropopause warming could also affect the precision of future projections. The analysis relies heavily on statistical methods for trend detection, which might not capture the full complexity of the year-to-year variability in Arctic stratospheric conditions. Finally, the focus is primarily on chemical loss, while future ozone levels will be influenced by a complex interplay of chemical and dynamical processes.