New Observational Constraints for Atmospheric Hydroxyl on Global and Hemispheric Scales

See allHide authors and affiliations

Science  21 Apr 2000:
Vol. 288, Issue 5465, pp. 500-503
DOI: 10.1126/science.288.5465.500


Dramatic declines in emissions of methyl chloroform (1,1,1-trichloroethane) resulting from the Montreal Protocol provide an unprecedented opportunity to improve our understanding of the oxidizing power of Earth's atmosphere. Atmospheric observations of this industrial gas during the late 1990s yield new insights into the global burden and distribution of the hydroxyl radical. Our results set firm upper limits on the global and Southern Hemispheric lifetimes of methyl chloroform and confirm the predominance of hydroxyl in the tropics. Our analysis suggests a global lifetime for methyl chloroform of 5.2 (+0.2, −0.3) years, a Southern Hemispheric lifetime of 4.9 (+0.2, −0.3) years, and mean annual concentrations of OH that are 15 ± 10% higher south of the intertropical convergence zone than those north of this natural mixing boundary between the hemispheres.

The hydroxyl radical plays a central role in the chemistry of Earth's atmosphere (1). This powerful oxidant initiates photochemical reactions that cleanse the atmosphere of many gases affecting stratospheric ozone and global climate, and it is intimately involved in the production of tropospheric ozone. We show here how atmospheric measurements of CH3CCl3 during periods of rapidly declining and small emissions provide a unique opportunity to place constraints on estimates of hydroxyl radical concentrations on global and hemispheric scales.

To that end, we have measured CH3CCl3 in air from flask samples collected routinely since 1992 at 7 to 10 remote locations across the globe (Fig. 1) (2). In contrast to observations from previous years (3–6), our recent data reveal only small differences between mean mixing ratios of CH3CCl3 in the Northern and Southern Hemispheres and consistently lower mixing ratios in the tropics of both hemispheres than at higher latitudes (7) (Figs. 2 and 3). Also, intrahemispheric gradients were similar within both hemispheres during 1998–1999 (Fig. 3), which suggests that the distribution of methyl chloroform is now controlled primarily by atmospheric removal rather than by anthropogenic releases. These recent observations support theoretical predictions of the change in gradient expected with large reductions in emissions (8) and confirm the predominance of OH in the tropics. Furthermore, the data provide clear evidence that long-lived reduced gases such as CH4 and hydrochlorofluorocarbons are destroyed primarily in the tropics.

Figure 1

The measured atmospheric burden of CH3CCl3 in recent years. Monthly hemispheric mixing ratios for the Northern (NH, triangles) and Southern (SH, circles) Hemispheres at Earth's surface were calculated as weighted averages of observations from three to six sampling stations in each hemisphere. Since 1997, an average of two pairs per month was collected at each of the sampling sites. For more information, see the supplementary material (2). All mixing ratios are reported as parts per trillion by mole (pmol mol−1 or ppt).

Figure 2

The changing atmospheric distribution of CH3CCl3 at Earth's surface. Relative differences between monthly mean mixing ratios at individual sites (X i) and global monthly means (X g) are shown over the period of measurements. Global means were estimated as a simple average of the hemispheric means. Sampling sites and latitudes are indicated in the insets at right. See (2) for further details regarding these sampling sites and methods for estimating hemispheric means from these data.

Figure 3

Measured intrahemispheric gradients in annual mean mixing ratios of CH3CCl3. (A) Annual means for individual sites (X i) are shown normalized to respective hemispheric annual means for the years 1998 and 1999. In the Northern Hemisphere (triangles), only data from low-altitude sites (<1000 m) are shown with similar symbols and are connected with lines to allow for appropriate comparisons to similar sites in the Southern Hemisphere (circles). Lower mixing ratios are observed at higher altitude sites (crosses) in the Northern Hemisphere, and this may reflect the presence of continued emissions. Model calculations [(8) not shown in the figure], however, predict that similar vertical gradients are expected at mid-latitudes even in the absence of emissions, owing to downward transport of CH3CCl3-depleted air from the tropics. (B) Same as (A) but for the years 1994 and 1995, when Northern Hemispheric emissions were substantial (19). Because of a lack of measurements from all sites in this earlier period, data from both low- and high-altitude locations are shown with similar symbols in this panel. Error bars in both panels represent 1 SD of the annual mean ratios estimated for the two years.

To date, constraints on the global concentration of OH stem from trend and budget analyses of trace gases that are oxidized primarily by OH, such as CH3CCl3, 14CO, and CHClF2 (3, 4,8–12). The mean global burden of OH calculated in this way, however, depends strongly on the accuracy of calibration and source strength estimates for these gases (2–4). Because emissions of CH3CCl3 have declined rapidly and are now small, we can obtain an estimate of its atmospheric removal rate that is less sensitive to uncertainties in these quantities (13). If we consider the globe as a single box, the mean, global, first-order rate constant for removal of a gas from the atmosphere can be expressed asEmbedded Image(1)where G is the number of moles of a trace gas in the global atmosphere, E g is the number of moles emitted to the global atmosphere per year, and t is time. After emissions decrease rapidly or cease [that is, whenE g/G is small compared to (dG/dt)/G], the relative rate of change approaches k g as a limit. The inverse ofk g is the mean atmospheric lifetime of the gas. For methyl chloroform, k g is determined predominantly by OH oxidation in the troposphere but also includes slower losses in the stratosphere (14) and in warm waters of the ocean (15, 16).

Our measurements show that the exponential rate of decline for CH3CCl3 in the global troposphere was 0.182 year−1 during 1998 and 1999, which corresponds to ane-fold decay time and a firm upper limit to the global CH3CCl3 lifetime of 5.5 ± 0.1 years (1 SD) (17) (Fig. 4). Global industrial production and consumption data suggest thatE g/G was only 6% of the observed rate of decay over this period (18). In contrast,E g/G was comparable to or even larger than the atmospheric rate of change during the 1980s and early 1990s (19, 20). On the basis of the observed decay and these emissions, we calculate a global lifetime for CH3CCl3 of 5.2 (+0.2, −0.3) years over 1998–1999 (21). BecauseE g/G was small relative to the observed atmospheric decay over this period [(dG/dt)/G], this global lifetime estimate is much less sensitive to calibration accuracy and uncertainties in industrial emissions than are previous lifetime estimates.

Figure 4

Global and hemispheric rates of decay expressed as e-fold times (17). e-fold times appear at the end of each yearly period for whiche-fold times were computed [for example, τe , 1999.2 = (lnX 1999.2 − lnX 1998.2)−1 ; globale-fold times appear as plus signs, Southern Hemisphere times as circles, and Northern Hemisphere times as triangles]. The mean global decay time constant since the beginning of 1998 is estimated at 5.5 ± 0.1 years (17).

The global tropospheric concentration of OH implied by this lifetime estimate is 1.1 (±0.2) × 106 rad cm−3(22), or slightly lower than can be derived from trend and budget analyses of CH3CCl3 and CHClF2 in the presence of substantial emissions (4, 8, 12). Our results would be consistent with CH3CCl3 lifetimes of 4.8 years (4) or less (8, 12) and with higher OH concentrations only if emissions of CH3CCl3 have been underestimated by about a factor of 3 or more in recent years (23). A small portion of this discrepancy could stem from changes in the global lifetime of CH3CCl3 as emissions have declined, either as a result of interannual changes in mean annual OH or because the distribution of this gas in the atmosphere has changed relative to the distribution of sinks in recent years. Such lifetime changes, however, are expected to be small (2).

Knowledge of OH concentrations on hemispheric scales is critical for an improved understanding of reduced gases having poorly defined sources—gases such as CH4, CH3Br, and CH3Cl. In the past, our understanding of OH on these spatial scales has contained substantial uncertainty. Estimates of the relative hemispheric concentration of OH based on past observations of CH3CCl3 suffered from uncertainties in air exchange rates between the hemispheres and from difficulties in estimating hemispheric burdens of CH3CCl3 in the presence of substantial emissions (8). Significant uncertainties also remain for inferring the global and relative hemispheric abundance of OH from measurements of 14CO (8, 24). Uncertainties in distributions of precursors and sinks for OH preclude a definitive calculation of the hemispheric partitioning of this oxidant; different models can calculate more OH in either hemisphere (25–28) or similar amounts in both (8, 29).

Because the net flux of methyl chloroform across the intertropical convergence zone (ITCZ) is now reduced and because Southern Hemispheric emissions continue to account for only a small fraction of global emissions (18, 20), our recent data allow us to place tight constraints on the loss frequency of CH3CCl3 in the Southern Hemisphere for the first time. If we consider a two-box model in which the boxes represent the Northern and Southern Hemispheres and are separated at the ITCZ (a natural mixing boundary), mass balance considerations allow us to express the first-order rate constant for loss of methyl chloroform in the Southern Hemisphere asEmbedded Image(2)where E s represents Southern Hemispheric emissions, S and N are mean hemispheric mixing ratios (2), andk ex is a rate constant for air exchange between the hemispheres. With this approach, the mean decay observed south of the ITCZ and the measured hemispheric difference allow us to estimate a Southern Hemispheric lifetime (1/k s) for CH3CCl3 of 4.9 (+0.2, −0.3) years during 1998–1999 (2, 21). Emissions south of the ITCZ influence this estimate for 1/k s insignificantly because they accounted for only a small fraction of diminished global emissions over this period (2, 18);E s/S is estimated to have been only 1% of the observed relative rate of decay [(dS/dt)/S] in 1998 and 1999. Northern Hemispheric emissions need not be known in this approach. The accuracy of this hemispheric lifetime estimate does depend on our ability to constrain net hemispheric exchange of CH3CCl3 from the measured hemispheric difference and the interhemispheric exchange time 1/k ex. However, in the absence of the large concentration gradients and variability that arose in the past from substantial emissions, hemispheric means and differences can now be estimated with greater certainty. Furthermore, because the burden of CH3CCl3 is similar in the two hemispheres (Fig. 5), uncertainties in k ex now have a small influence on the value calculated for k s[we estimate that k ex(N/S– 1) was 15 ± 5% of the measured decay during 1998–1999].

Figure 5

Measured inter- and intrahemispheric gradients for atmospheric CH3CCl3. (A) The measured difference between the Northern (X NH) and Southern (X SH) Hemisphere mixing ratios relative to the global mean (X Globe) on a monthly basis. Monthly differences (solid circles) are connected with a thin line; the heavy line represents a 12-month running mean. Our measurements show that the mean annual difference between the two hemispheres was 2.9 ± 0.4% during 1998–1999 (2). (B) The relative difference in annual mean mixing ratio between each sampling site and Cape Grim (CGO, 41°S), normalized by the annual global mean (X Globe).

The global and hemispheric lifetimes derived from this analysis suggest that the loss frequency of CH3CCl3south of the ITCZ is 15 ± 10% higher than the loss frequency north of this natural boundary between the hemispheres (k n/k s = 0.88 ± 0.09). A similar asymmetry in the hemispheric loss frequency of CH3CCl3 can also be derived from the hemispheric difference we have measured for this gas during 1998–1999 (Figs. 5 and 6) (2). In the presence of significant emissions, hemispheric differences for longer-lived gases are determined predominantly by the magnitude and distribution of these emissions and by air exchange rates between the hemispheres. As emissions become small, however, this difference becomes sensitive to any asymmetry in loss frequencies as well (2, 8). Because losses of methyl chloroform to the ocean and stratosphere are relatively small and symmetric between the hemispheres (14–16), any significant asymmetry in its loss frequency reflects an asymmetry in the mean hemispheric concentrations of the hydroxyl radical.

Figure 6

The ratio of hemispheric loss frequencies (k n/k s) for CH3CCl3 implied by recent measurements. From the global and Southern Hemispheric lifetimes derived in the text, we estimate k n/k s = 0.88 ± 0.09 (open circle) (2). This estimate is based on atmospheric measurements and industrial emissions over 1998–1999 and on rates of air exchange between the hemispheres [see text and (2, 18)]. Because sinks other than OH oxidation are relatively small and symmetric,k n/k s is a good approximation for 〈OH〉n/〈OH〉s(2). Also shown are relative hemispheric loss frequencies implied by the measured hemispheric difference for a range of assumed global emissions and global lifetimes [solid line; see also (2)]. Mean concentrations of OH that are 10 to 30% higher in the Northern Hemisphere (25, 26) could be reconciled with our data only if industrial emissions were underestimated by more than a factor of 3 in 1998–1999 and if the global lifetime of CH3CCl3 were less than 4.8 years.

Our results suggest that mean annual concentrations of OH south of the ITCZ are 15 ± 10 % higher than they are north of this natural mixing boundary (Fig. 6) (2). Because the ITCZ is positioned slightly north of the equator on average, the hemispheric asymmetry in OH would be somewhat smaller if estimated for hemispheres divided at the equator (30). Despite this, our findings are not consistent with the significantly higher concentrations in the Northern Hemisphere that are calculated in some models (25,26) or implied from an analysis of CH2Cl2 measurements (8), and they suggest that higher concentrations of nitrogen oxides, ozone, and other OH precursors in the Northern Hemisphere are more than balanced by higher loss rates for this important oxidant as well.


View Abstract

Navigate This Article