Potential impact of iodinated replacement compounds CF 3 I and CH 3 I on atmospheric ozone : a three-dimensional modeling study

The concept of Ozone Depletion Potentials (ODPs) is extensively used in policy considerations related to concerns about the effects of various halocarbons and other gases on stratospheric ozone. Many of the recent candidate replacement compounds have atmospheric lifetimes shorter than one year in order to limit their environmental effects, especially on stratospheric ozone. Using a three-dimensional global chemistry-transport model (CTM) of the troposphere and the stratosphere, the purpose of this study is to evaluate the potential effects of several very short-lived iodinated substances, namely iodotrifluoromethane (CF 3I) and methyl iodide (CH3I), on atmospheric ozone. Like other chemicals with extremely short lifetimes, the stratospheric halogen loading and resulting ozone effects from these compounds are strongly dependent on the location of emissions. For CF3I, a possible replacement candidate for bromotrifluoromethane (CF 3Br), ODPs derived by the three-dimensional model are 0.008 with chemical lifetime of 5.03 days and 0.016 with a lifetime of 1.13 days for emissions assumed to be evenly distributed over land surfaces at mid-latitudes and the tropics, respectively. While this is the first time the ODPs have been evaluated with a three-dimensional model, these values are in good agreement with those derived previously. The model calculations suggest that tropical convection could deliver a larger portion of the gas and their breakdown products to the upper troposphere and lower stratosphere if emission source is located in the tropics. The resulting ODP for CH3I, emitted from mid-latitudes, is 0.017 with lifetime of 13.59 days. Valid simulations of convective transport, vertical mixing and degradation chemistry of Correspondence to: D. J. Wuebbles (wuebbles@atmos.uiuc.edu) CH3I are shown that have good qualitative agreement between the model derived distribution of background CH 3I, based on global source emission fluxes from previous studi s, and available observations especially in vertical profiles.


Introduction
A number of analyses indicate that human-related emissions of chlorine-containing and bromine-containing halocarbons, such as chlorofluorocarbons (CFCs), carbon tetrachloride (CCl 4 ), methyl chloroform (CH 3 CCl 3 ) and bromochlorofluorocarbons (Halons), have been largely responsible for the observed decrease in stratospheric ozone concentrations over recent decades (WMO, 2007).Those ozone-depleting substances (ODSs) are primarily comprised of the halocarbons that have very long lifetimes in the atmosphere.Without strong policy actions, these human produced ODSs would have continued to have a significant impact on future levels of stratospheric ozone.Thus, it is not surprising that the development of our understanding of their effects on ozone led to international regulations of the ODSs under the landmark Montreal Protocol in 1987, and the amendments and adjustment to the Protocol that followed to further strengthen the Protocol (Velders et al., 2007).Because of the Protocol, a number of replacement compounds have arisen to replace the ODSs being controlled.
The concept of Ozone Depletion Potentials (ODPs) (Wuebbles, 1981(Wuebbles, , 1983) ) arose as a means of determining the relative ability of a chemical to destroy stratospheric ozone.ODPs are an integral part of national and international considerations on ozone-protection policy, including the Montreal Protocol and its Amendments and the U.S. Clean Air D. Youn et al.: Potential impact of iodinated replacement compounds on atmospheric ozone Act.ODPs provide an important and relatively straightforward way of analyzing the potential for a new chemical to affect ozone relative to the CFCs, Halons, and other replacement compounds.ODPs are currently determined by two different means: calculations from chemistry-transport models (CTMs) of the global atmosphere (e.g., WMO, 1995WMO, , 1999WMO, , 2003WMO, , 2007)), and calculations using a semi-empirical approach (Solomon et al., 1992;WMO, 1999).Traditionally, zonally averaged two-dimensional (2-D) CTMs have been the accepted tools for calculating the ozone response in the stratosphere.The past evaluations of ODP were conducted for chemicals with atmospheric lifetime sufficiently long (more than approximately one year).These long-lived gases are well mixed throughout the troposphere after surface release and large portion of surface emissions can reach the stratosphere.
However, many of the compounds being considered either for new applications or as replacements for substances controlled under the Montreal Protocol are now designed to be very or even extremely short-lived, on the order of days to a few months, so as to reduce the impacts on ozone and climate.Many of these very short-lived (VSL) compounds still contain halogens including chlorine, bromine and iodine.The VSL replacement gases still can be vertically transported into the lower stratosphere by intense convection in the tropical troposphere, which is relevant to possible effects on ozone from these compounds.Thus the ODP of a VSL species depends upon its distribution in the atmosphere and the location of its source (Wuebbles and Ko, 1999;Wuebbles et al., 2001;Ko and Poulet, 2003).Unfortunately, the determination of ODPs for these gases is not straightforward because their short atmospheric lifetimes leave them poorly mixed in the troposphere (Wuebbles and Ko, 1999;Wuebbles et al., 2001;Ko and Poulet, 2003;Butler et al., 2007).The traditional 2-D model analysis of ODPs also is not sufficiently accurate for calculating the integrated amount of the halogenated VSL source and reaction product gases in the troposphere that enter the stratosphere (Wuebbles et al., 2001).
Three-dimensional (3-D) CTMs fully representing the complete troposphere and stratosphere are necessary for evaluating the halogen loading and ozone depletion in the stratosphere for VSL species.Thus the definition of ODPs has been revised for VSL compounds (Wuebbles et al., 2001;WMO, 2003).The ODP definition for the VSL compounds thus now accounts for the variation that can occur in the ODP as a function of where and when the compound is used and emitted.The most important factor in evaluating the ODP of VLS compounds is shown to be geographical distribution, or latitude, of the surface emissions because gases emitted at higher latitudes take longer to reach the stratosphere than gases emitted in the tropics (Bridgeman et al., 2000;Olsen et al., 2000;Wuebbles et al., 2001).Seasonal variations in the emissions can also be important if that is a relevant criterion in the atmospheric emissions of the compound.
The only VSL compound currently evaluated and published using 3-D models of the chemistry and physics of the troposphere and stratosphere is n-propyl bromide (nPB, 1-C 3 H 7 Br); at this time, only limited studies and a range of estimated ODPs are available for it (Wuebbles et al., 2001;WMO, 2003WMO, , 2007)).The past studies of the ODPs for nPB have been only partially based on 3-D model results (see Ko and Poulet (2003) for a more complete summary).3-D CTMs, which include a much more comprehensive treatment of transport than the zonally-averaged 2-D models, are now much more preferable for calculations of ODPs both for the longer-lived gases and for the very short-lived candidate replacement compounds.These models cannot only determine the amount of the substance reaching the stratosphere directly, without the simplifying assumptions of the 2-D models, but can also more directly follow the processes affecting the reaction products.Studies suggest that the vast majority of chlorine or bromine reaching the stratosphere from VSL substances is transported there in reaction products (Wuebbles et al., 2001;WMO, 2003WMO, , 2007)).
Iodotrifluoromethane (CF 3 I) is a gaseous fire suppression flooding agent for in-flight aircraft and electronic equipment fires and a candidate replacement for bromotrifluoromethane (Halon-1301, CF 3 Br).Methyl iodide (CH 3 I), a potential replacement fumigant for methyl bromide (CH 3 Br), has been of interest as a natural source of atmospheric iodine that is primarily emitted from the oceans and as a potentially useful tracer of marine convection and the subsequent role of iodine in atmospheric photochemistry (Bell et al., 2002;Cohan et al., 2003;Vogt et al., 1999).Since the C-I bond is easily cleaved, CF 3 I can be rapidly photo-dissociated even at low UV intensity and visible light under daylight conditions in the planetary boundary layer or the troposphere.CH 3 I also readily photolyzes under day light conditions as well as reacting with the hydroxyl radical (OH).For both species, the initial reaction results rapidly in the availability of atmospheric iodine for chemical reactions with ozone and other gases.Although current atmospheric concentrations of iodine-containing gases are very small, CF 3 I and CH 3 I are both a potentially significant source of atmospheric iodine despite being very short-lived and are of interest in the upper troposphere and lower stratosphere as a source of iodine radicals for ozone destruction (WMO, 2007).Ozone destruction in the lower stratosphere due to catalytic cycles involving iodine was not well understood, but recent laboratory studies have improved the understanding of atmospheric chemical processes of iodine species including their photochemistry (WMO, 2007;Li et al., 2006;Bösch et al., 2003).
As the next stage for further potential impact analysis of the compounds on ozone beyond the previous studies using 2-D models (Li et al. 2006;Solomon et al., 1994), ODPs of the two iodinated VSL replacement compounds CF 3 I and CH 3 I are evaluated using the National Center for Atmospheric Research (NCAR) Model for OZone And Related Tracers version 3 (MOZART-3) 3-D chemistry-transport model (we also update the ODP using the latest version of the University of Illinois at Urbana-Champaign (UIUC) 2-D chemical-radiative-transport (CRT) model).A thorough understanding of the potential effects of the iodine compounds on stratospheric ozone builds upon the previous twodimensional studies (Li et al., 2006).Using the available analyses of global source emissions for CH 3 I (Lee-Taylor and Redeker, 2005;Bell et al., 2002), background CH 3 I concentrations in the atmosphere are also evaluated and compared with observational data towards determining the capabilities of 3-D models towards accounting for CH 3 I chemistry.
This paper is organized as follows.In Sect.2, the methodologies to obtain ODPs for very short-lived substances (VSLS) are described and discussed.A brief description of models used for model-derived ODPs is given in Sect. 3 and simulated CH 3 I concentrations are discussed and compared with observational datasets in Sect. 4. Results of ODPs are presented in Sect. 5. Discussion and conclusions of the findings are in Sect.6.

Methodology
The concept of ODPs provides a relative cumulative measure of the expected effects, on stratospheric ozone, of the emissions of a gas relative to one of the gases of most concern to ozone change, namely CFC-11 (CFCl 3 ) (Wuebbles et al., 1981(Wuebbles et al., , 1983;;WMO, 1995WMO, , 1999WMO, , 2003WMO, , 2007)).The ODP of a gas is defined as the integrated change in total ozone per unit mass emission of the gas, relative to the change in total ozone per unit mass emission of CFC-11.As a relative measure, ODPs are subject to fewer uncertainties than estimates of the absolute percentage of ozone depletion caused by different gases.ODPs provide an important and relatively straightforward way of analyzing the potential for a new chemical to affect ozone relative to CFCs, Halons, and other replacement compounds.However, a chemical compound with extremely short lifetime, on the order of several to tens of days, is not expected to be well mixed in the atmosphere and thus has ODPs that should be dependent on location of the emissions (Wuebbles and Ko, 1999;Wuebbles et al., 2001;Ko and Poulet, 2003).
The ODP of an ozone-destroying species χ is defined by: where the percent change in O 3 burden ( O 3 (χ)) for a surface emission of the compound χ is divided by the same factors for CFC-11.The flux of the compound χ introduced into the atmospheric boundary layer is represented by φ(χ ), and the CFC-11 flux in the perturbation run minus that in the reference run is represented by φ(CFC-11).
Therefore, the derivation of the ODP for a given compound for assumed emissions at given locations and/or seasons requires two steady-state perturbation model runs with the CTM, for an assumed flux of the gases χ and CFC-11 relative to a steady-state reference atmosphere.A CFC-11 perturbed atmosphere is obtained from the CFC-11 perturbation run by increasing the surface mixing ratio of CFC-11 from its value in the reference atmosphere run.When the reference and CFC-11 perturbed steady-state atmospheres are produced, the atmospheric burdens of O 3 and CFC-11 as well as the total chemical removal rate of CFC-11 from the atmosphere are calculated to provide a denominator for the ODP as given in Eq. (1).In a separate χ perturbation run, χ is introduced as a surface flux emission for specific latitude ranges to provide the ODP numerator.The model-derived ODPs are steady-state ODPs.
VSL compounds (assumed to be those with atmospheric lifetimes less than 0.5 year) generally do not mix evenly in the troposphere, and most of the surface fluxes are oxidized within the troposphere rather than reaching the stratosphere.It is important to note that through the degradation process, halogen-containing product gases can be transported to the stratosphere.Thus in analyzing such shortlived gases, chemical degradation in the troposphere, removal of degradation products by incorporation into clouds and rainout, and transport of the gases and their degradation products to the stratosphere must be treated accurately.Because tropospheric transport processes cannot be accurately treated in a zonally-averaged two-dimensional model, a three-dimensional model is required to investigate the impact of halogenated replacement compounds on ozone depletion as a function of location (Wuebbles et al., 2001).However, at this time, ODP studies have not been done with a fully coupled CTM of the troposphere and stratosphere.
Since VSL compounds are likely to be used primarily at northern mid-latitudes, the emissions at the surface are assumed to occur for latitudes from 30 • N to 60 • N (see Fig. 1).The surface emissions in the MOZART-3 perturbation run are considered only over land grid cells while those in the 2-D model are assumed equally distributed over the latitude range.The emissions are assumed to be the same over model land grids and assigned with fractional land surface emissions at boundary grids between land and ocean/lake surfaces considering the actual land fraction.Total annual masses emitted into the atmosphere in the 3-D model were adjusted to match those in the 2-D model for comparison purpose.Model ODPs of the iodinated compounds have been derived assuming the mid-latitude emissions.Model analysis of a tropical emissions case has also been carried out for CF 3 I.Global atmospheric lifetimes are calculated based on model atmospheric loading and chemical loss rate of the compounds.

Modeling tools
The 3-D model better represents some of the physical processes affecting ozone as well as transport processes, especially in the troposphere and the polar stratosphere, than the 2-D model does.Therefore, 3-D CTMs are now the tool of choice and much more preferable for evaluating ODPs both for longer-lived gases and for the very short-lived candidate replacement compounds.These models can not only determine the amount of the substance reaching the stratosphere directly but can also follow the processes affecting the reaction products.The drawback in using 3-D models is that they are very computationally expensive and require much additional human resources for analysis.
In this study, the National Center for Atmospheric Research (NCAR) Model for OZone And Related Tracers version 3 (MOZART-3) (Kinnison et al., 2007) has been used to explicitly calculate the impact of VSL iodinated compounds on ozone.The MOZART-3 CTM of the global atmosphere is one of the leading tools for the study and prediction of current and future atmospheric composition and of possible effects from anthropogenic emissions into the atmosphere.The model was updated to incorporate the complete set of iodine chemistry used in the previous 2-D modeling study by Li et al. (2006).Major chemistry reactions and rates that could impact effects of CF 3 I and CH 3 I on atmospheric ozone concentrations relative to earlier studies (e.g., Solomon et al., 1994) include: CF 3 I, CH 3 I, IO, HOI, INO, INO 2 , IONO 2 , OIO, ICl, IBr, and I 2 absorption cross-sections for photochemical reactions (Sander et al., 2003(Sander et al., , 2007)), thermal reaction-rate coefficients of CH 3 I + OH → I + H 2 O + CH 2 O, and catalytic cycles involving reactions of IO with ClO or BrO and I with O 3 , HO 2 or O (see Li et al., 2006).
The MOZART-3 model includes a complete representation of stratospheric processes and extends to altitudes greater than 80 km as described and evaluated in Kinnison et al. (2007).It incorporates a full stratosphere, including the chemistry of chlorine species (Cl y ) and bromine species (Br y ) important to stratospheric ozone as well as updated hydrogen, nitrogen, and hydrocarbon oxidation chemistry relevant to stratospheric and tropospheric chemistry included in its predecessor model MOZART version 2 (Horowitz et al., 2003;Lin et al., 2008).State-of-the-art representations of relevant heterogeneous and physical processes for winter/spring polar vortex related to ozone destruction are also fully included in the model.This model has been evaluated extensively via comparisons with measurements of atmospheric trace gases from satellite data and a large number of aircraft field campaigns (e.g., Kinnison et al., 2007;Pan et al., 2007;Horowitz et al., 2003;Wei et al., 2003).
Chemical reaction-rate constants and photochemical data follow the recommendations of Sander et al. (2003).The MOZART-3 CTM used in this study is driven by a meteorology field derived from the Whole Atmosphere Community Climate Model (WACCM) version 1b.The MOZART-3 thus has a 2.8 • resolution in latitude and longitude and a hybrid sigma-pressure vertical coordinate including 66 layers from the surface to 5.1×10 −6 mbar (approximately 140 km).
The UIUC 2-D CRT model (see Youn et al., 2006Youn et al., , 2009;;Guillas et al., 2006) updated from the version used in Li et al. (2006) is also used.The UIUC 2-D CRT model includes a 3-D representation of the propagation of low wave-number planetary waves, and thus is called as "2.5-D model" in the sense that transport and diffusion effects due to the breaking of the waves are self-consistently handled with the zonal wind and the residual mean meridional circulation (mass transport circulation).Wave propagation properties and the residual circulation, calculated with climatological data from United Kingdom Meteorological Office (UKMO) data assimilation system for the specific years 1992-2002, are computed self-consistently with the zonal wind, and tracers are transported in the zonal-mean plane by the residual circulation and the diffusion due to the wave breaking (Choi and Youn, 2001;Youn et al., 2009).
The UIUC 2-D CRT model has been an adequate and efficient tool for the evaluation of ODPs although 2-D models have limitations in adequately representing the zonally asymmetric features such as tropospheric transport processes and the edge and evolution of the ozone hole.The 2-D model applied to traditional ODP studies, especially for long-lived compounds, has been performed in a daily-average mode with a time step of one day outside the operators.However, the 2-D model runs in this study of evaluating the impact of VSL compounds on atmospheric ozone were conducted in more detailed integration with a time step of 20 min due to extremely short lifetimes of the source and product gases although the above-mentioned limitations exist.The 2-D model still has the advantage of being much more computa-tionally efficient compared to 3-D models, and thus can provide a quick reference to the slower 3-D simulations (Youn et al., 2009).

Simulation of CH 3 I background concentration
CH 3 I, an effective carrier of iodine from the biosphere into the atmosphere, is emitted into the atmosphere mainly as a result of biological methylation of iodine (Moore and Groszko, 1999;Amachi et al., 2001) along with photochemical production in open sea (Happell and Wallace, 1996;Li et al., 2001;Yokouchi et al., 2001).After the emissions to the atmosphere, CH 3 I can be photolyzed or oxidized by the OH radical, to produce iodine radicals.Photolysis of CH 3 I is a main pathway of its removal from the troposphere, so as to produce iodine atoms even under visible light conditions (Chameides andDavis, 1980: Davis et al., 1996).Iodine produced from CH 3 I plays many important roles in marine boundary layer chemistry, influencing ozone destruction, the oxidizing capacity of the troposphere, denoxification, and particle formation (Vogt et al., 1999;Carpenter, 2003;O'Dowd et al., 2002).The ultimate fate of gas phase iodine is abstraction to aerosol particles as iodide (I − ) and iodate (IO3 − ) in the troposphere.Unfortunately, most of the existing measurements of iodine in marine aerosols are based on bulk measurement of total iodine, so the individual chemical iodine species contributing to particulate iodine have rarely been measured (O'Dowd and Hoffmann, 2005).Although recent modeling work tried to include aerosol phase reactions with organic substances (Pechtl et al., 2007), any aerosol chemistry was not included in our current study.Therefore, our derived ODPs of iodine species are the upper limit.In this Sect.4, a simulation of background CH 3 I concentrations in the atmosphere is evaluated to determine the capabilities of 3-D models towards accounting for CH 3 I chemistry.
As described in detail by Bell et al. (2002), the known or extrapolated global budget of oceanic CH 3 I emissions ranges from 130 to 1300 Gg yr −1 .However, land-based sources may be important regionally.CH 3 I fluxes from growing rice are significant compared to other land-based sources such as inland and coastal wetlands, peatlands, and biomass burning (Lee-Taylor and Redeker, 2005 and references therein).Flux rates from inland and coastal wetlands, peatlands, and biomass burning all yield global source terms below 10 Gg yr −1 (Lee-Taylor and Redeker, 2005).
Table 1 shows the annual emission flux of CH 3 I to the atmosphere used in this modeling study.The CH 3 I emission sources include ocean, rice paddies, biomass burning, wood fuel, and wetlands.Except for emissions from rice paddies, all of the emissions in Table 1 are computed following Bell et al. (2002), using the GEOS-Chem global 3-D model of tropospheric chemistry (Park et al., 2006;Bey et al., 2001) driven by assimilated meteorological data from the Goddard  (Colman et al., 2001;Blake et al., 2001).Simulated atmospheric burden of CH 3 I amounting to 8.0 Gg and photolysis loss rate of 264.0 Gg yr −1 in the model yield lifetimes of 11.1 days.The lifetime of 11.1 days is about twice as large as that in Bell et al. (2002).It should be noted that if total gaseous inorganic iodine is defined as I y =I+IO+HOI+INO+INO 2 +IONO 2 +OIO+ICl+IBr+I 2 +CH 3 I, simulated total iodine in the stratosphere using the CH 3 I emissions in Table 1 constitutes about 2.6% of total atmospheric I y .
Besides uncertainty ranges of source emissions, a horizontal resolution of approximately 250 km in the MOZART 3-D CTM is coarse for comparison with aircraft measurements.Satellite observations of CH 3 I, if any, could be used for the better comparison.However, since our current ultimate target is evaluating ODPs of two iodinated compounds, the following comparisons of 3-D model results with aircraft measurements are used to check if the model chemistry of atmospheric iodine described by Li et al. (2006) is reasonable enough to evaluate ODPs of iodine species.Figure 3 shows [pptv]  4 show enhanced concentration in the middle and upper troposphere (UT) associated with deep convection, as also found in Bell et al. (2002).The simulated and observed profiles are in good agreement; the model reproduces the vertical gradient (decrease) with altitude in the lower atmosphere.However, over tropical central Pacific regions, the simulated concentrations are over 4 times lower than observed.The simulation of MBL CH 3 I concentrations has errors possibly due to the lack in understanding of emission fluxes over oceans.The MOZART-3 CTM in the present study is driven by the WACCM meteorology.Since the climate model meteorology compared to assimilated data will not provide a good simulation of deep marine convective mass fluxes over the Pacific, it is not surprising that the MOZART-3 results show a less defined "C-Shaped" profile than the GEOS-Chem does in Bell et al. (2002).
Discrepancy in spatial patterns of the simulated and observed CH 3 I concentrations in the MBL is associated with uncertain source emissions that need to be better understood.However, the current model results qualitatively agree with aircraft measurement data in vertical profiles.The good comparisons between measured and modeled CH 3 I concentrations do not fully justify the set of iodine chemistry used in our model simulations.However, those comparisons do denote valid simulation of convective transport, vertical mixing and degradation chemistry of CH 3 I.
In simulating the effect of fast convective updraft at global scale, such as large longitudinal gradients at a latitude circle and the maximum tropospheric trace gas concentrations with the overshooting features over land tropical convective systems, Park et al. (2004) showed overall good agreement between satellite observations and MOZART-3 simulations for methane and water vapor in the upper troposphere and lower stratosphere (UTLS).Therefore, the MOZART-3 simulation in the UTLS region is overall success although there are significant differences in the observed concentrations of the long-lived species between the continents.We note that it is beyond our research scope to quantify the uncertainties in global-scale model calculations in that the current horizontal resolution could misrepresent subgrid convective processes along with the uncertain convection parameterization producing overshootings significantly high, rapid and intense enough for penetrating the lower stratosphere.However, the substantial uncertainties in calculations of large-scale vertical transport (and the contribution of overshooting convection) are topics that require more investigation.

Ozone Depletion Potentials for CF 3 I and CH 3 I
Several earlier studies have evaluated potential effects of CF 3 I on ozone in terms of ODPs.Assuming CF 3 I emissions from the Earth surface only, early studies obtained extremely small ODPs of CF 3 I, which were <0.008 by Solomon et al. (1994) and 0.006 by Connell et al. (1996).Since the  1 and 2 and Fig. 1 in Li et al., 2006) relative to the earlier studies of the ODP of CF 3 I (Solomon et al., 1994).
In this study, the ODPs for CF 3 I and CH 3 I have been evaluated using the state-of-the-art MOZART-3 3-D CTM incorporated with the chemical processes relative to the iodinecontaining compounds as in Li et al. (2006).We also use the current UIUC 2-D CRT model for direct comparison to the 3-D model results.Three-dimensional model case studies for mid-latitude emissions of CF 3 I and CH 3 I and for tropical emissions of CF 3 I has been conducted to evaluate the potential effects of these postulated anthropogenic surface flux emissions of halogenated VSL compounds on atmospheric ozone based on the geographic distributions of emissions in Fig. 1.Land-based emissions of CF 3 I and CH 3 I over the latitude range of 30 • N to 60 • N used in the model correspond to a total emission rate of 0.74 and 0.536 Tg yr −1 , respectively.An additional simulation for CF 3 I land emission (total 0.73 Tg yr −1 ) for 20 • S to 20 • N was also conducted.
The atmospheric burden changes of CF 3 I (CH 3 I), I y , and O 3 per Tg yr −1 of CF 3 I (CH 3 I) surface flux introduced to the UIUC 2-D CTM and MOZART are evaluated for the troposphere and the stratosphere.The burden is obviously higher in the troposphere where the species is being emitted or and its by-products are produced via degradation.Stratospheric burdens are smaller because removal processes including photo-dissociation of CF 3 I and wet scavenging of its reaction products prohibit those iodine species from being transported into the stratosphere.
Latitude-height distributions of zonal mean CF 3 I determined by the UIUC 2-D and the MOZART-3 3-D model are shown in Fig. 5.The contour lines are isolines of the base-10 logarithms of CF 3 I volume mixing ratios where the mixing ratios are scaled as per teragram of one year CF 3 I loading.3-D model-determined distributions of annually averaged CF 3 I including topography and the tropopause variations along longitudes show more detailed features than those from the two-dimensional model.In general, CF 3 I distributions from the different model simulations compare relatively well, including the annual variation from January to December (not shown).
The annual average percent change in total inorganic iodine (I y ) mixing ratios per Tg yr −1 of CF 3 I surface flux in Fig. 6 shows agreement in pattern with that in CF 3 I in Fig. 5.As expected, the Northern Hemisphere has a higher mixing ratio of I y because of the direct influence of CF 3 I.However, since the values on contour lines in Fig. 6 are the base 10 logarithm of "I y mixing ratio in ppt", contour values of about −1.5 to −2.5 in the stratosphere amount to about 0.03 to 0.003 ppt which is not significant.Despite the very small concentrations, total inorganic iodine from mid-latitude emission scenarios of iodine species in the stratosphere in Fig. 6a (also in Fig. 8b) is noted because the distributions shows asymmetry about equator, differently from that from tropical emission scenario in Fig. 6b.It would partly indicate dependency of short-lived ODS's ODP on emission location and support the existence of Northern Hemisphere (NH) transport pathway (the NH summer monsoons as regions for transport) to the lower stratosphere as shown with satellite data and MOZART-3 output by Park et al. (2004).The UIUC 2-D and MOZART 3-D models show a general agreement in the transport and distribution of the I y .However, 3-D simulated I y for mid-latitude emissions shows smaller mixing ratio in the Southern Hemispheric stratosphere than 2-D simulated one does.Larger amounts of 3-D I y for tropics emissions are elevated up to the tropopause level.2).
ODPs are a vital part of the current U.S. Clean Air Act, in which 0.2 is the upper bound of ODP values of any chemical that is not controlled.However, much smaller ODPs are considered to still be of concern in recent policy considerations.Global atmospheric chemical lifetimes and ODPs of iodine compounds for the surface emission scenarios in Fig. 1 are summarized in Table 2.The 3-D model-based ODPs for mid-latitude emission cases of CF 3 I and CH 3 I are 0.008 and 0.017, respectively, and that for tropical emission case of CF 3 I is 0.016.For comparison, the 2-D model-based ODPs obtained are 0.006 and 0.016, respectively, for CF 3 I and CH 3 I emissions evenly distributed over the mid-latitudes of 30 • N to 60 • N, while the ODP for CF 3 I emissions distributed over 20 • S to 20 • N is 0.013.
The 2-D and 3-D model-derived lifetimes of CF 3 I and CH 3 I are comparable to the values reported in Chapter 2 of WMO (2003WMO ( , 2007)).Consistently with ozone depletion distributions in Fig. 7, three-dimensional ODPs for CF 3 I and CH 3 I are higher but are in good agreement with 2-D ODPs.Both models show the dependence of the ODPs of CF 3 I on the location of emissions.The CF 3 I ODP for tropical emissions is twice that found for mid-latitude emissions.The consistent results obtained with different models for the two iodine compounds is in part attributable to the consistent set of chemistry incorporated into the models.The computationally efficient zonally-averaged 2-D model still appears to perform relatively well in the ODP studies of those VSL species.This is in part because the residual circulation used in the 2-D advection well represents mass transport in latitudinal and vertical (meridional) directions.The global lifetimes of CF 3 I are smaller than those for CH 3 I.This accounts for the smaller ODP values of CF 3 I than that of CH 3 I.The shorter lifetimes of the iodinated VSL compounds, when compared with most brominated VSL compounds, suggest less efficient transport of total inorganic iodine to the stratosphere and thus smaller ODPs.

Key findings and conclusions
The MOZART-3 3-D CTM, fully representing the troposphere and stratosphere, as well as the current version of the UIUC 2-D model, has been used to evaluate the potential effects of CF 3 I and CH 3 I on atmospheric ozone.Through the model analyses, we have calculated ODPs for the two iodinated VSL compounds to evaluate the relative effects on global atmospheric ozone from surface emissions of such questionable gases.
The fraction of VSL compounds involved in ozone depletion depends strongly on the location of the surface sources.The atmospheric distribution of the source gases is determined by competitive atmospheric processes including vertical transport and chemical destruction or removal via washout.3-D model-derived ODPs of CF 3 I and CH 3 I for mid-latitude emissions case are 0.008 and 0.017, respectively.The 2-D model gives similar ODP values, 0.006 for CF 3 I and 0.016 for CH 3 I, despite of the limited 2-D representations of atmospheric processes.The 3-D and 2-D modelderived CF 3 I ODP for tropical emissions, 0.016 and 0.013, respectively, also agree well.The chemical lifetime of CF 3 I from tropical emissions is shorter than that for mid-latitude emissions due to fast photolysis but the model-derived ODP values for CF 3 I tropical emissions are larger (see Table 2).The larger CF 3 I ODPs for tropical emissions can be explained by the rapid vertical transport due to convective activities in equatorial regions that is more effective in the vertical transport of total halogens since the transport timescale in convection events (hours) is shorter than chemical loss or washout removal timescale.Youn et al. (2009) showed that the differences between time variations of Halon perturbations and ozone changes, simulated by both the 2-D and 3-D models, were found not to give significantly different values for both the direct and indirect GWPs.Trace gas changes in perturbed model atmosphere for the calculation of ODP are determined with two steady-state (perturbed and baseline) model atmospheres for each scenario so that the limitation of the 2-D model mostly related to zonally-asymmetric features such as tropospheric circulation and stratospheric ozone hole does not signify in this study of evaluating potential impacts on global ozone www.atmos-chem-phys.net/10/10129/2010/Atmos.Chem.Phys., 10, 10129-10144, 2010 using the two simulations.Also, the 2-D model is expected to yield similar mass transport to the 3-D model as long as both large-scale transport features and eddy effects in the atmosphere are reasonably consistent in the 2-D models.Therefore, despite known limitations of the 2-D model, the 2-D model with shorter integration time step for simulating the very-short lived compounds in this study is proven to be useful.
We simulate background concentrations of CH 3 I using the MOZART-3 3-D CTM to evaluate the model results with aircraft measurement data.Using CH 3 I source emission fluxes reported by the previous studies and calculated from the oceanic mixed layer model coupled to the GEOS-Chem 3-D CTM, there were discrepancies in spatial patterns of simulated and observed CH 3 I concentrations in MBL.However, the current model results qualitatively agree with aircraft measurement data in vertical profiles of CH 3 I in the atmosphere.Therefore, the MOZART-3 model can be used to evaluate ODPs of iodine species as long as postulated emissions are used.From the simulation of background CH 3 I concentration, we found that about 2.6% of total inorganic iodine as the upper limit is transported across the tropopause into the stratosphere.
The extremely small ODPs for CF 3 I and CH 3 I values indicate that little of the CF 3 I or CH 3 I, and their reaction products containing iodine atoms, would reach the stratosphere.Most of the iodine would be removed rapidly from the atmosphere by rainout processes.These studies also emphasize the importance of using state-of-the-art three-dimensional numerical models of the chemistry and physics of the troposphere and stratosphere in evaluating the ODPs of replacement compounds for short-lived compounds.
Fig. 1.Land-based surface emission fluxes of CF 3 I over (a) 30 • N to 60 • N and (b) 20 • S to 20 • N in the unit of molecules cm −1 s −1 .These postulated emissions are inputs to model perturbation runs.The surface flux is adjusted at boundary grids considering actual land fractions.

Fig. 2 .
Fig. 2. Annual average surface emission fluxes of CH 3 I in the unit of molecules cm −1 s −1 .The emissions include fluxes from ocean, rice paddies, biomass burning, wetlands, and wood burning to the atmosphere.

Fig. 3 .
Fig. 3. Atmospheric concentrations of CH 3 I in the marine boundary layer.Circles are aircraft measurement data from PEM-Tropics A in August-October (left) and PEM-Tropics B in March-April (right).The aircraft and model data are averaged between 0 and 1 km for boundary layer concentrations.

Fig. 4 .
Fig. 4. Vertical profiles of CH 3 I concentrations measured from aircraft and simulated by model.Horizontal bars are standard deviations computed from the individual measurements.Solid black lines denote the corresponding model profiles.

Fig. 5 .Fig. 6 .
Fig. 5. Annually averaged CF 3 I mixing ratios introduced into the atmosphere by one teragram of total CF 3 I surface emissions in ranges of (a) mid-latitudes and (b) tropics.The base 10 logarithm was taken of the CF 3 I mixing ratios determined by the two scenarios using two different models after being scaled as per teragram CF 3 I loading.The MOZART-3 3-D (UIUC 2-D) model results are on the left-handside (right-hand-side).The tropopause levels are denoted by thick line and the 3-D model results only includes detailed topography and tropopause variations.

Figure 8
Figure8shows changes in annually-averaged CH 3 I, I y , and O 3 that are made by CH 3 I surface emissions over midlatitude range of 30 • N to 60 • N. Spatial patterns of changes in I y and O 3 due to the CF 3 I and CH 3 I mid-latitude emissions in Figs.6a and 7aand Fig.8care very similar.However, the loading of CH 3 I into the atmosphere yields larger depletion of atmospheric ozone than that of CF 3 I does because CH 3 I has longer lifetime (see Table2).ODPs are a vital part of the current U.S. Clean Air Act, in which 0.2 is the upper bound of ODP values of any chemical that is not controlled.However, much smaller ODPs are considered to still be of concern in recent policy considerations.Global atmospheric chemical lifetimes and ODPs of iodine compounds for the surface emission scenarios in Fig.1are summarized in Table2.The 3-D model-based ODPs for mid-latitude emission cases of CF 3 I and CH 3 I are 0.008 and 0.017, respectively, and that for tropical emission case of CF 3 I is 0.016.For comparison, the 2-D model-based ODPs obtained are 0.006 and 0.016, respectively, for CF 3 I

DFig. 7 .
Fig. 7. Annually averaged ozone percent changes for (a) mid-latitude and (b) tropical emission scenarios relative to background atmospheric conditions without CF 3 I emissions in the unit of % Tg −1 yr.

Fig. 8 .
Fig. 8. Annually averaged changes in (a) CH 3 I and (b) I y , (c) ozone for CH 3 I mid-latitude emission scenario relative to background atmospheric conditions without CH 3 I emissions.

Table 2 .
Li et al. (2006)es of CF 3 I for mid-latitudes and tropics emission scenarios and CH 3 I for mid-latitudes emission scenarios using the UIUC 2-D and the MOZART-3 3-D models.Derived lifetimes are global chemical lifetimes calculated using atmospheric total burden (loading) and chemical loss-rates in the models.earlierstudies,thelaboratory results on iodine reactions keep being updated.Using the 2004 version UIUC 2-D model,Li et al. (2006)re-evaluated the ODPs for surface emissions of CF 3 I and determined ODPs of 0.013 for emissions evenly distributed over the Northern Hemisphere, 0.011 for emissions distributed over 30 • N to 60 • N, and 0.018 for emissions distributed over 5 • S to 30 • N. The model used an updated representation of iodine chemistry (see Tables