Deriving Global OH Abundance and Atmospheric Lifetimes for Long-Lived Gases: A Search for CH3CCl3 Alternatives

Abstract

An accurate estimate of global hydroxyl radical (OH) abundance is important for projections of air quality, climate, and stratospheric ozone recovery. As the atmospheric mixing ratios of methyl chloroform (CH₃CCl₃) (MCF), the commonly used OH reference gas, approaches zero, it is important to find alternative approaches to infer atmospheric OH abundance and variability. The lack of global bottom-up emission inventories is the primary obstacle in choosing a MCF alternative. We illustrate that global emissions of long-lived trace gases can be inferred from their observed mixing ratio differences between the Northern Hemisphere (NH) and Southern Hemisphere (SH), given realistic estimates of their NH-SH exchange time, the emission partitioning between the two hemispheres, and the NH versus SH OH abundance ratio. Using the observed long-term trend and emissions derived from the measured hemispheric gradient, the combination of HFC-32 (CH₂F₂), HFC-134a (CH₂FCF₃, HFC-152a (CH₃CHF₂), and HCFC-22 (CHClF₂), instead of a single gas, will be useful as a MCF alternative to infer global and hemispheric OH abundance and trace gas lifetimes. The primary assumption on which this multispecies approach relies is that the OH lifetimes can be estimated by scaling the thermal reaction rates of a reference gas at 272 K on global and hemispheric scales. Thus, the derived hemispheric and global OH estimates are forced to reconcile the observed trends and gradient for all four compounds simultaneously. However, currently, observations of these gases from the surface networks do not provide more accurate OH abundance estimate than that from MCF.

Figure 1.

Latitude-pressure cross sections of zonally integrated annual loss rates of MCF (CH₃CCl₃), CH₄, and HCFC-22 from the WACCM model using the 30 year averaged model output from the 2000 time-slice run, with warm colors indicating larger loss rates. The black contours outline the regions where more than 50%, 75%, and 90% of the globally integrated loss occur.

Figure 2.

(a) Modeled steady state atmospheric lifetimes (years) from the TS2000 simulation for CH₃CCl₃ versus mass-weighted tropospheric OH abundance. (b) Comparison of tropical (30°S–30°N) annual mean OH profiles from Spivakovsky et al. (2000) (thick black line with gray shading indicates $1\sigma$ spatial and temporal variance in that model) and the models. The global mean OH abundances from Spivakovsky et al. (2000) and those derived from the models (${[\text{OH}]}_{\text{GM}}$; molecules cm⁻³) are given in the legend. All CCM results are 30 year annual average from the 2000 time-slice simulations.

Figure 3.

Comparison of 30 year averaged model zonal annual mean OH (×10⁶ molecules cm⁻³) from the 2000 time-slice simulation for the GSFC2D, GEOSCCM, SOCOL, ULAQ, UMUKCA, and WACCM models.

Figure 4.

Modeled ${\tau }_{\text{OH}}$ (years) for (a) CH₄, (b) HCFC-22, (c) CH₃Cl, and (d) CH₃Br versus modeled ${\tau }_{\text{OH}}$ of MCF. Results shown are the steady state ${\tau }_{\text{OH}}$ from the TS2000 simulations (squares) and the transient ${\tau }_{\text{OH}}$ in 2000 from the TRANS simulations (asterisks). The ${\tau }_{\text{OH}}$ of the FBC tracers from individual models, when available, is also shown (diamonds). The 12-box IM-inferred ${\tau }_{\text{OH}}$ for CH₄ and HCFC-22 are shown as well (black filled squares).

Figure 5.

(a) Comparison of industrial-based bottom-up (black), interhemispheric concentration difference-based (red), and global trend and MCF lifetime-based (gray) emissions during 1980–2012. (b) The contribution of terms A–D in equation (7) to the MCF trend over the same time period. (c) The AGAGE observed (black) and two-box model calculated (red) global mean CH₃CCl₃ mixing ratios during 1980–2012 using the gradient-based emissions derived in Figure 5a.

Figure 6.

Correlation of annually averaged model interhemispheric mixing ratio difference, $\text{Δ}{C}_{n-s}$, for HCFC-22 and CH₃CCl₃ with the annual global emissions used in the model simulation. The linear fit regression lines are shown as dashed lines. Numbers in each circle mark the year of the model results. Different NH emission fraction ($F_n$) values were used in the GEOSCCM simulation for HCFC-22 for 1980–1995 and 1996–2012. Since the regression slope (af/b in equation (6)) is a function of $F_n$, the 1980–1995 results fall on a different regression line than the 1996–2012 results.

Figure 7.

(left) The two-box model calculated trends for HFC-32, HFC-134a, HCFC-22, and HFC-152a using a MCF-based $X_{ns}$ of 1.3 years (gray lines) and the optimal $X_{ns}$ (dark red dashed lines) that best match the observations (black lines). (right) Comparison of the gradient-based emissions (dark red dashed lines), the trend lifetime-based emissions (blue lines), and the inventory bottom-up emissions from the nonarticle five countries (black lines). Annual mean $C$, $\text{Δ}{C}_{n-s}$, $\partial C/\partial t$, and $\partial \text{Δ}{C}_{n-s}/\partial t$ are calculated using atmospheric observations from the AGAGE network. Annually varying $F_n$ are calculated based on emission inventory reported to UNFCCC (section 4.2). Sensitivity of the simulated surface mixing ratios and gradient-based emissions are also shown to reflect emission uncertainties in (1) the fraction of global emissions emitted in the NH, $F_n$ (0.05 to +0.05 or unity of 1 whichever is the smallest; light green shading) and (2) NH-SH OH difference (NH/SH ratio of 1 ± 0.1; light red shading).

Figure 8.

The modeled annual mean interhemispheric mixing ratio difference ($\text{Δ}{C}_{n-s}$) versus emissions and its dependence on (a) trace gas lifetime, from 1.6 years to 14.1 years, with a 5%/yr increase in emissions and (b) rate of change in emissions, from an increase of 1%/yr to 100%/yr for a HCFC-22 like tracer with a lifetime of 12 years, based on idealized tracer sensitivity simulations conducted using the GSFC2D model (see text for details). The symbols represent the values for individual years, and the lines show the regression slopes for each sensitivity simulation. The $X_{ns}$ values derived from these GSFC2D model results for each lifetime/emission scenario are also included.

Figure 9.

The model probability distribution function of the (left) SH and (right) NH detrended mixing ratio anomalies (with respect to global mean values) for CH₃CCl₃ and HCFC-22 during 1980–2014. $\text{Δ}{C}_{n-s}$ calculated using all global surface grids ($\text{Δ}{C}_{n-s,\text{gsurf}}$; gray shading for distribution probability and thin gray dashed lines for area-weighted hemispheric mean anomaly) differ from those calculated using the modeled values sampled at the AGAGE stations (thick red lines for distribution probability and red dashed lines for hemispheric mean) and the NOAA-GMD stations (thick blue lines for distribution probability and blue dashed lines for hemispheric mean). Station-based values are interpolated to all latitudes, from 90°S to 90°N, for full global representation. The modeled distribution probability (thick black lines), hemispheric means (black dashed lines), and $\text{Δ}{C}_{n-s,\text{trop}}$ calculated using all grid cell results below 500 hPa are also shown.