Stratospheric ozone destruction: The importance of bromine

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 14, NO. DI9, PAGES 23,871-23,88, OCTOBER 2, 1999 Stratospheric ozone destruction: The importance of bromine relative to chlorine J. S. Daniel, S. Solomon, and R. W. Portmann NOAA Aeronomy Laboratory, Boulder, Colorado R. R. Garcia National Center for Atmospheric Research, Boulder, Colorado Abstract. The relative effectiveness of bromine compared to chlorine for destroying stratospheric ozone is explored. Two definitions previously used to quantify this relative effectiveness, typically referred to as ct, are compared and a definition is presented to calculate ct values applicable for column and global ozone loss. Calculations of ct are made with a twodimensional radiative/chemical/dynamical model and suggest that bromine is roughly 45 times more effective than chlorine for global ozone destruction. The physical processes underlying this result are probed, and sensitivity studies are presented that show that relatively large changes invoked in the modeled transport and heterogeneous chemistry lead to changes in this value of <15%. 1. Introduction Since 1974, when it was first suggested that emissions of chlorofluorocarbons (CFCs) could lead to stratospheric ozone destruction [Molina and Rowland, 1974], it has become apparent that anthropogenic halocarbon emissions have led to detectable ozone losses both at midlatitudes and high latitudes [World Meteorological Organization (WMO), 1995, 1999]. Observations and calculations demonstrate that anthropogenic halocarbons are the primary cause of polar ozone depletion, and the weight of evidence suggests that they are also the cause of significant midlatitude depletion [WMO, 1999]. There are several catalytic cycles that permit chlorine and bromine to destroy effectively stratospheric ozone. During polar springtime the primary cycles are [WMO, 1995]: Cycle 1 Cycle 2 2(C > C1 + 2) C1 + C1 + M -->(C1) 2 + M (C1) 2 + hv --> C1OO + C1 C1OO --> C1 + 2 net > 3 2 Br > BrO + 2 C > C1+O 2 C1 + BrO --> Br + C12 C12 + M --> C M net > 32 Copyfight 1999 by the American Geophysical Union. Paper number 1999JD /99/1999JD9381 $9. At midlatitudes, cycle 1 becomes much less important, but significant ozone losses in the lower stratosphere are caused by cycle 2 and by the catalytic cycles: Cycles 3 and 4 Cycles 5 and 6 OH > HO2 + 2 X > XO-I- O 2 XO + HO 2 --> HOX + 2 HOX + hv --> X + OH net > 32 X=C1, Br NO+O 3 ---> NO 2 +2 X > XO+O 2 XO + NO 2 + M --> XONO 2 + M XONO 2 + hv --> X + NO 3 NO 3 + hv --> NO+O 2 net > 32 X =C1, Br and in the middle and upper stratosphere by Cycles 7 and 8 X +3 --> XO+O 2 XO + O--> X +2 net O ---> 22 X=C1, Br Observations confirm that there have been significant increases in the tropospheric abundances of chlorine- and bromine-containing compounds over the past 25 years [Montzka et al., 1996; WMO, 1995] and that these increases are primarily due to the anthropogenically emitted, longlived, organic compounds that representhe primary source of 23,871

2 23,872 DANIEL ET AL.: STRATOSPI-[ERIC OZONE DESTRUCTION stratospheri chlorine and bromine. These halocarbon source gas increases can then lead to decreases in stratospheric ozone because of the previous chemical catalytic cycles. In Figure 1 we show the model-calculated monthly averaged relative rates of the primary catalytic cycles involving C1 x and BrOx for 49øN in March. The catalytic cycles 2, 3, 5, and 7 (defined above) represent most of the C1 -induced ozone loss, while cycles 2, 4, 6, and 8 represent most of the BrOw-induced ozone loss. Quantitative information concerning the relative ozone destructive ability of the various halogenated source gases can be useful both for science and policy analyses. Indices that have been used to accomplish this in the past include ozone depletion potentials (ODP) [Wuebbles, 1983; Solomon et al., 1992] and equivalent effective stratospherichlorine (EESC) [Daniel et al., 1995; WMO, 1995, 1999]. Each requires knowledge of the atmospheric lifetimes of the source gases as well as of the rates at which chlorine and bromine are released from the source gases in the stratosphere. For the brominated source gases it is also necessary to know the relative ozonedestroying ability of a bromine atom compared to a chlorine atom that enters the stratosphere. This quantity is often referred to as tx and remains one of the significant uncertainties in the implementation of ODPs and EESC for bromocarbons. Previous estimates of tx for polar springtime, supported by observations and calculations, range from 4 to [WMO, 1995], while estimates for midlatitudes have been more uncertain with values suggested to possibly be above 1 [Garcia and Solomon, 1994]. The primary difficulty in estimating a global or regional value for et outside the polar regions has been determining the best way to convert local et values as a function of altitude and latitude to values that are reflective of column-integrated ozone loss, either at specific latitudes or globally. Since the ODP and EESC indices are typically used to characterize global ozone destructive capabilities of source gases, it is critical that an appropriate value of et be used in these formulations. In past ozone assessments the adopted value of et for global averaged ozone loss has been between 4 and 6 [WMO, 1995, 1999]. If a value of 1 were more appropriate, then the relative impact of bromocarbons like the halons and CH3Br on stratospheric ozone could be significantly greater, implying that reduced future emissions of these compounds could have a larger effect on future ozone loss than previously estimated. In this paper we focus on two points concerning the relative impact of bromine- and chlorine-induced stratospheric ozone destruction. First, we compare and contrasthe past definitions used to quantify the value of ix. Second, we provide a general definition for et that we use to calculate values for the relative ozone destructive capability of bromine compared to chlorine for column ozone loss as a function of latitude and for globally averaged total ozone loss. In addition to providing Chlorine b Bromine 4 C1, O Cycle 7 4 BrO, O Cycle 8 3, 3 C122 2 Cycle 1.o 2 BrO, C1 Cycle 2 BrONO 2 Cycle 6 1 Tropopause Fractional Loss Rate Fractional Loss Rate Figure 1. The relative cumulative importance of (a) C1 and (b) BrO x catalytic cycles for ozone destruction as a function of altitude. The relative rates represent averages over the month of March at 49øN. The cycle numbers correspond to the catalytic cycles displayed in the text. 1

3 DANIEL ET AL.: STRATOSPHERIC OZONE DESTRUCTION 23,873 values for tx we explore the sensitivity of the calculated values to changes in aerosol amount, the presence of cirrus clouds, and temperature and dynamical model assumptions. Recently, Ko et al. [1998] discussed the sensitivity of the bromine efficiency factor (BEF) for methyl bromide to a variety of key chemical processes, where the bromine efficiency factor for CH3Br differs from tx by including in its value the fractional halogen release rate [Solomon et al., 1992] of CH3Br relative to CFC-11 in the stratosphere. Here our emphasis is not on chemical kinetics but rather on the variations of tx with latitude, altitude, transport, and heterogeneous chemistry. The kinetic rates assumed in this work important to bromine- and chlorine-catalyzed ozone depletion are taken from DeMore et al. [1997] and are thus consistent with the work of Ko et al. [1998]. In section 2 we begin with a discussion and comparison of two definitions of ix. We conclude this section by presenting a general definition of tx that can be applied to column and global ozone destruction. Our model calculations of tx are presented in section 3, which is divided into two parts. In the first part we discuss the variations in tx as a function of altitude, with continued emphasis on a comparison of the two tx definitions. In the second part we discuss the implications of these "local" tx calculations on tx quantities that are representative of column and global ozone loss. In section 4 we discuss the sensitivities of our tx calculations to several uncertainty factors, and our conclusions are presented in section The ct Fundamentals When an inorganic halogenated molecule containing chlorine, bromine, or iodine, is added to the stratosphere, the local ozone loss rates will generally increase, resulting in more ozone destruction. The amount of increased ozone destruction will depend on many factors, including the particular halogen that is added [Solomon et al., 1994b]. If an inorganic molecule containing, for example, fluorine is added to the stratosphere, it will have a negligible impact on ozone because of rapid conversion to tightly bound fluorine reservoir species. Almost all of the fluorine will reside in HF, leaving little to be in active forms (F and FO) that can destroy ozone. In contrast, inorganic stratospheric species containing iodine are quite potent for ozone destruction because iodine reservoir species are extremely weakly bound [Solomon et al., 1994a,b]; the majority of stratospheric iodine is likely found in the ozone destructive forms of I and IO. In this work we concentrate on the ozone destructive ability of bromine relative to chlorine. This quantity, referred to as ix, has been defined previously in two ways. It has been defined using ozone loss rates [e.g., Solomon et al., 1992] by ozone loss rate due to bromine catalytic cycles per inorganic bromine atom ozone loss rate due to chlorine catalytic cycles per inorganic chlorine atom and it has been defined by Danilin et al. [1996] using the change in ozone abundance by o = o O ly (1) (2) where the numerator represents the change in ozone at a particular stratospheric location due to a change in Bry divided by that change in Bry and the denominator represents the same quantity due to a change in Cly. Definition (1), which we will refer to here as the "rates" definition, reflects the relative impact on ozone chemical loss rates of the entire amount of inorganic bromine compared to the entire amount of inorganic chlorine on a per atom basis. In contrast, (2), which we will refer to as the "ozone change" definition, reflects the impact on ozone abundances of an incremental change in Bry compared to an incremental change in Cly from some initial state. With either definition it is expected that tx will be >1 throughout the lower stratosphere because of the relative stability of the chlorine and bromine inorganic reservoir species: HC1 and C1ONO 2 are more tightly bound than are HBr and BrONO2. The implication of this bonding is that a bromine atom added to the stratosphere is much more likely to be found in an ozone destructive form than is a chlorine atom. In spite of these definition differences the two approaches should yield the same value for tx in the absence of chemical feedbacks and when the ozone tendency is controlled by linear local photochemistry of C1,, and BrO x rather than by transport and other nonlocal effects such as self-healing. Selfhealing refers to the increases in ozone at certain altitudes that "replace" some of the ozone lost above because of increased UV flux and the resulting increase in 2 photolysis rates caused by the loss above. As a simplified example, assume that in some stratospheric location the local ozone loss is controlled by chemistry and is linear in C1 and in BrO, that ozone production is represented by P, and that ozone loss terms are given by A(C1), B(BrO), C(C1)(BrO), and D. A and B represent proportionality constants for the linear loss in C1 and BrO, respectively, while C represents the rate constant for the reaction of C1 and BrO, which is the rate limiting step for cycle 2. D represents all other losses that are assumed to be independent of C1 and BrO. The steady state ozone abundance can be written as P (3) 3 A(C1) + B(BrO) + C(C1)(BrO) + D If it is also assumed that C1/Cly=a and BrO/Bry=b, it is straightforward to write down the 2 definitions of tx using (1), (2), and (3) [ Bb(Bry ) + Cab(C1 y )(Bry )]/Bry Bb + Cab(C1 y ) (4) 1- [Aa(Cly)+Cab(Cly)(Bry)]/Cly Aa+Cab(Bry) AO3Bry / ABry a2 = AO3Cly / ACly den P 2 [Bb+Cab(Cly)] = P [Aa + Cab(Bry)] = al (5) den 2 where "den" represents the denominator of (3), AO3Bry is the ozone change due to a change in Bry and AO3cly is the ozone change due to a change in Cly. Thus, if only local linear C1,, and BrO x chemical processes are significant in determining the ozone abundance, the two tx definitions are identical. The two definitions do not necessarily produce the same tx value in regions where ozone loss is nonlinear with Cly or Bry.

4 23,874 DANIEL ET AL.: STRATOSPHERIC OZONE DESTRUCTION In polar springtime, for example, where the C1 dimer cycle (1) leads to a significant and nonlinear ozone loss, the rates only substantial changes implemented in the model for this study are that National Centers for Environmental Prediction definition and the ozone change definition lead to different (NCEP) temperatures, which are used in determining values, with the magnitude of the differences depending on heterogeneous rates, and aerosol surface areas are fixed at 1979 the importance of the C1 dimer cycle. values unless otherwise stated. This eliminates annual Both (1) and (2) are straightforward to use in quantifying the value of ct at a particular location; (1), however, is difficult to evaluate when estimating column and global ct values. While the chemical loss terms are readily defined in a local sense, it is difficult to determine a procedure to weight the local ct values with altitude that correctly accounts for the interplay of chemical and dynamical production and loss terms that govern the column ozone production and loss rates. variations in the heterogeneous processes due to these quantities. As discussed by Ko et al. [1998], changes in certain rates and absorption cross sections involving bromine chemistry since DeMore et al. [1994] have led to substantial differences in the calculated bromine efficiency value for CH3Br. In this work, we have therefore assumed values for the key rates and cross sections from DeMore et al. [1997] as was done by Ko et al. [1998]. The ct value from (1) is calculated For example, just as the ozone profile in the lower stratosphere using monthly averaged loss rate quantities. To calculate ct depends not only on local chemistry but also on transport, so too does the ozone depletion profile reflect transport from, for example, higher altitudes. Indeed, downward transport below -2 km can "freeze in" ozone losses from higher altitudes as shown later, so that the local depletion below 2 km does not only reflect local chemistry. A column ct definition, which avoids these ambiguities by considering the calculated from (2), three cases are run: (1) a baseline case with 1991 levels of chlorine and bromine source gases, (2) a chlorine perturbed case in which CFC-11 is chosen as the perturbation gas, and (3) an artificial bromine-perturbed case in which an artificial bromocarbon is adopted whose loss rate is assumed to be the same as that of CFC-11 but whose photolysis and O( D) reaction products are inorganic bromine rather than chlorine. column ozone change due to both dynamics and chemistry, The bromine case has thus been devised so that the amount of can be made analogously to the local ozone change ct definition: additional inorganic bromine released throughout the stratosphere will have the same spatial pattern as the additional inorganic chlorine in the chlorine perturbed case; in other AO3columnBry / AFlry words, ACly/ABry is a constanthroughouthe atmosphere. O column = CFC-11 is a natural choice to use as the perturbation source gas AO 3- column- CI y / AC1 y because it is the gas to which other halocarbons are compared in most ODP calculations. However, calculated ct values are AC1 I AO3-Bry (z)dz y dependent on the choice of source gas. If a gas with a shorter stratospheric lifetime were used, halon 1211, for example, ABry AO3cly (z)dz more inorganic halogen molecules would be released lower in the stratosphere and would affect the calculated ct values. Nevertheless, even using halon 1211, which has a 24-year I alocalao3-cly (z)dz stratospheric lifetime compared to 45 years for CFC-11 [WMO, (6) 1999], the global and column ct values that are presented in this work would increase by <1%. The results of our ct calculations for 49øN and 49øS in I AO3-cIy (z)dz March using both the rates definition (equation (1)) and the ozone change definition (equation(2)) are shown in Figure 2 where z is the altitude, 1ocal is defined with the ozone change as the dashed line and solid line, respectively. These local ct definition, and ACly /ABry is assumed to remain constant profiles illustrate two primary points. First, ct calculated from throughout the stratosphere. Hence the column t quantity (1) shows the behavior with altitude as given by Garcia and can be written as the average of the local txs (defined by (2)) Solomon [1994]. The value is quite large in the lower weighted by the ozone change profile arising from the Cly stratosphere because of the decreased stability of BrONO2 and perturbation. It is this relationship between local and column that HBr relative to C1ONO2 and HC1. The shorter chemical makes the ozone change ct definition particularly useful. lifetimes of BrONO2 and HBr make it much more likely that a By using (6) as the definition for the column ct values we bromine molecule in the lower stratosphere will be in an ozone include the chlorine- and bromine-induced response of ozone destructive form like BrO than that a chlorine atom would be in the troposphere as well as in the stratosphere. However, the in C1. With increasing altitude, C1 increases faster than tropospheric response, which is more difficult to calculate BrO, and the catalytic cycles 7 and 8 become important. The accurately than the stratospheric response, affects the global value of ct therefore decreases with increasing altitude in the cos presented in this work by <5%. middle stratosphere. In the upper stratosphere the expected value of ct from the rates definition can be written as 3. Calculations of c 2k2(BrO)(O)/Bry k2 (BrO)/Bry 3.1. Variations With Altitude (7) 2k (C1)(O)/Cly k (C1)/Cly In this section we compare and contrast the ct values calculated from (1) and (2) by examining the variation of the values with altitude. We perform the calculations with the where k is the rate constant for the reaction of C1 with O and two-dimensional interactive radiative/chemical/dynamical k2 is the rate constant for the reaction of BrO with O. During model, most recently described by Solomon et al. [1998]. The March at 49øN and an altitude of 4 km the value for ct is ~3.

5 ß ß DANIEL ET AL.' STRATOSPHERIC OZONE DESTRUCTION 23,875 a 4- ;,,, ß [, ß [ ß [, ].,. i 49øN i. '"' :... Rates Definition O Coe n hange Definition /' '... : 7, 7, 7;..- -"x,-, ii'rop. opa, use ' b 4- E - - -,-'u 3 :::J ß,, < 2-1, -2 i Rates Definition i O,.zon Change,Definition <x,, r, 7røP, øpa, use ' Figure 2. Calculated average a values during March at (a) 49øN and (b) 49øS as a function of altitude using the rates (equation(1), dashed line) and the ozone change (equation(2), solid line) a definitions. The second point apparent from Figure 2 is the relatively poor agreement between the as calculated from the rates definition compared to those calculated from the ozone change definition in the lower stratosphere. These discrepancies imply that the in situ chemistry of C1 x and BrOx are not the only factors determining the response of ozone to a Cly or a Bry perturbation. An estimate of the importance of in situ C1 and BrO chemistry can be made by calculating the ozone lifetime as a function of altitude with respect to C1 losses and with respect to BrO x losses. These lifetimes are shown in Figure 3 for 49øN in March. Below 3 km the ozone lifetime with respect to BrO losses is never <4 years, 5O I [ I I I I I I [ ' [ I [ I I I I I [ I I I I I I I I [ I I I I [ [ I I,, -, ß 4O ' BrOx 2 lo O.Ol Years Figure 3. Ozone loss lifetimes as a function of altitude calculated from monthly averaged losses during March at 49øN due to chlorine (solid line) and bromine (dashed line) chemistry.

6 ß ß 23,876 DANIEL ET AL' STRATOSPHERIC OZONE DESTRUCTION suggesting that other non-local processe such as transport and self-healing could play an important role in governing the specific amount of ozone loss. This implies that the calculated tx quantities from the rates definition can be significantly different than those calculated from the ozone change definition. Above -25 km the two values track each other more closely, but the lifetimes are still fairly long (greater than a month for BrOx), and the ozone change definition becomes noisy because the bromine-induced ozone depletion is small Column and Global We now use (6) to convert the local tx values discussed in the previous section to tx values that reflect the impact of bromine relative to chlorine for column and global ozone depletion. Column and global tx quantities are of more interest than local values because of their application in indices like ODPs and EESC. Because of the significant differences in local tx values in the lower stratosphere calculated using the rates and the ozone change definitions (Figure 2), it is important to understand the relationship between these local quantities and tx values representative of column and global ozone loss. The differences in the lower stratosphere (<25 km) are particularly important because of the large amount of observed and calculated ozone depletion at these altitudes [e.g., WMO, 1995]. The shape of the ozone change profile calculated because of a Cly increase is shown in Figure 4 for the tropics and midlatitudes in March. In both cases the values are normalized to the maximum value to reveal the shape. Figure 4 shows that the altitude regions contributing most to the column a values are roughly between 15 and 25 km at midlatitudes, with losses at higher altitudes in the tropics. These profiles can be used to convert the local ozone change tx quantities (such as those shown in Figure 2) to an tx value representative of total column ozone loss. Applying (6) to the curves of Figures 2 and 4 (dashed line) leads to the conclusion that bromine is almost times more important than chlorine for the column ozone loss in this midlatitude case. The tropical ozone loss profile peaks at higher altitudes where the local value of tx is smaller, so that the column integrated tx at low latitudes is -25. Figure 4 shows that the calculated column tx value is sensitive to the altitude profile of the ozone loss due to increasing Cly as well as to the local tx profile. These factors are important because they aid in an understanding of the significance of various model uncertainties as discussed later. Ozone change and x profiles are used for all latitudes to produce tx values reflective of column ozone loss in March shown in Figure 5. The significant decrease in the values of column c s in the tropics compared to higher latitudes discussed above is evident. While this difference is important in understanding the role of bromine in the tropics, it is of limited importance to the global value of tx because of the smaller amount of ozone destruction in tropical regions compared to midlatitudes and high latitudes. The second important feature of Figure 5 is the relatively constant column tx value of 45- in the midlatitudes and high latitudes of the Northern Hemisphere and a gradual increase from -4 to 65 from 3 ø to the pole in the Southern Hemisphere. While the altitude of peak ozone loss in March decreases from midlatitudes to the pole in both hemispheres, the behavior of the local tx profiles is different in the two hemispheres. In the spring hemisphere the local tx values decrease going toward the pole, counteracting the effect of the decreasing altitude of ozone loss, while in the autumn hemisphere the local tx values increase slightly, resulting in increased column txs with higher latitude. The decrease in local tx values with increasing latitude in the spring hemisphere occurs because of the effects of the decreasing temperatures with higher latitude on heterogeneous sulfate chemistry, which enhances C1/Cly. Conversely, in the autumn hemisphere the decrease in OH with increasing latitude leads to decreased conversion of HC1 to C1Ox and thus larger local tx values. These midlatitude column values of tx place constraints on the global value because of the significant amount of observed total ozone loss at Tropical, 3o... < Midlatitude ,,, I,,, I,,, I,,, I,, Normalized Ozone Change Figure 4. Normalized ozone loss profiles due to an increase in CFC-11 in the tropics and at midlatitudes.

7 DANIEL ET AL.' STRATOSPHERIC OZONE DESTRUCTION 23, ,, I,, I, I,, I,, I t Latitude (degrees) Figure 5. Average column c values as a function of latitude during March. midlatitudes. They also show that estimates of a midlatitude ct value >1 are too high. Although we have just discussed a straightforward method of converting local ct values defined with the ozone change definition to column ct quantities, there is no such easy method to use the rates definition of ct to compute a column or global ct value. This is primarily because the values of ct in the lower stratosphere defined with (1) do not account for transport and other processes not related to C1 x or BrOx ozone destruction that can play significant roles in the response of ozone to a chlorine or bromine perturbation. As a simple illustration, consider a hypothetical situation in which the chlorine-catalyzed ozone destruction rate per chlorine atom in the lower stratosphere at some latitude becomes so small that local {x values defined by the rates definition exceed 1. Further, assume that the lower stratospheric ozone losses are characterized by an ½x defined by the ozone change definition of because of being frozen in from higher altitudes. This illustrates a situation in which the relative effect of additional chlorine and bromine on the lower stratosphere, and thus much of the column, is not determined by the rates {x definition but by dynamics instead. If the rates ½x definition were used in combination with the ozone change profile, one would infer that bromine would lead to ~1 times more ozone destruction in the column than chlorine does, when actually i t would cause closer to times the destruction. Although not J F M A M J J A S O N Month Figure 6. Monthly averaged global {x values. D

8 23,878 DANIEL ET AL.' STRATOSPHERIC OZONE DESTRUCTION as dramatic as the hypothetical example, if the rates txs in Figure 2 (dashed line) were used in (6) instead of the ozone change txs, a column value of-9 would be calculated rather than the appropriate tx value of -. Because of these difficulties in converting local rates tx quantities to column or global tx values, we will use the ozone change definition in our calculations throughout the rest of this work. Using the same formalism of (6), monthly averaged global values of tx can be calculated by averaging local txs weighted by the calculated ozone change due to the chlorine perturbation. Figure 6 shows these global tx values calculated from our two-dimensional model. The values show little seasonal cycle and suggesthat bromine is roughly 45 times as effective as chlorine in its destruction of ozone on a global average basis. These results support the choice of tx values in past ozone assessments. When we calculate a BEF for CH3Br using this same formalism, we estimate a value of 53, which is ~1% lower than the value of 58 calculated by Ko et al. [1998] (recall that the BEF includes an additional term sometimes called fractional release). Knowledge of the abundances of the compounds in the ratelimiting reactions of the important catalytic cycles for ozone depletion can allow for an observational estimate of local tx values from the rates definition where chemical ozone losses are rapid and locally dominant. This approach, in fact, has been taken in order to observationally estimate the value of tx above Antarctica in the polar springtime [Solomon et al., 1992]. However, such an approach cannot be applied to the slower midlatitude losses, as shown above. This lack of observational support for the ozone change tx definition places additional demands on the accuracy of atmospheric models in order to obtain accurate tx quantities. Because of the role of dynamics in the ozone loss, the sensitivity to global transport must be examined. It is also necessary that models accurately calculate the chemical loss terms of ozone. This requires, for example, an accurate parameterization of chemistry on surfaces in addition to the accurate calculation of gas phase chemistry. 4. Sensitivities In this section we examine the sensitivity of calculated tx values to several model variables, including the strength of the dynamical forcing and mixing, the size of temperature fluctuations invoked, and the amount of surface area assumed for heterogeneous reactions. Changes in these modeled processes can result in two types of effects that can alter the column tx values. First, they have the potential to change the local tx profiles. This can result from either changes in the transport of ozone or from chemical changes that affect the C1/Cly and BrO/Bry ratios in the lower stratosphere. Second, they can cause a difference in the ozone vertical profile response to the addition of Cly. As is evident from (6), a total ozone loss change does not necessarily affect the column tx calculation, while a change in the ozone profile response does. If the peak oz6ne response to a change in chlorine were to occur at a lower altitude, for example, where local tx values are higher, this would tend to increase the column tx quantity. We conclude this section with a brief discussion of the sensitivity of tx values to the total chlorine and bromine abundance in the atmosphere. Three dynamical cases are examined. The column tx results of these three cases and our baseline case (solid line) are shown in Figure 7. In case A (dotted line) the temperature waves used to represent observed wave activity (which are particularly important to the accurate calculation of Northern Hemisphere midlatitude ozone depletion) [Solomon et al., 1998], are neglected from the calculation. In case B (dashed line) the background stratospheric horizontal diffusion coefficient (Kyy) is increased from lx15 to 3x15 cm 2 s 'l, and in case C (dasheddotted line) the planetary wave forcing is arbitrarily increased 71 4 u 3 " x f/ Baseline ', x '/... Case A -,,,,x,,,,//,: - - Case B - -,' ----CaseC Latitude (degrees) Figure 7. Column tx values for the baseline scenario compared to values for dynamic perturbation cases A (no temperature waves), B (increased background stratospheric Kyy), and C (increased planetary wave forcing).

9 stratosphere as the increase in the wave forcing drives the ozone change o value, which as shown in Figure 2 is less than the rates value below -24 km, even lower. Two surface area cases were performed, case D in which cirrus clouds are considered in the lower stratosphere and upper troposphere [Solomon et al., 1997] and case E in which 5 times the inferred Stratospheric Aerosol and Gas Experiment (SAGE) surface areas for the historically low 1979 observations is assumed. For these cases the March column o DANIEL ET AL.: STRATOSPHERIC OZONE DESTRUCTION 23,879 by 25%, leading to changes in the mean circulation. As shown column o values are converted to monthly averaged global o s in Figure 7, the changes in the calculated c for these cases at and compared to the baseline case, the values decrease by 8- particular latitudes vary fr m the baseline case by up to -2%. 13% in case D and increase by <5% in case E. Global o values exhibit a much smaller sensitivity, differing from the baseline case by <6% throughout the year. In case A These sensitivity cases demonstrate modest changes in the calculated column and global o values and show that o the impact of the temperature waves on heterogeneous calculations can be affected by various model assumptions. chemistry is most important in the Northern Hemisphere middle and high springtime latitudes. Here o values show substantial increases compared to the baseline case because of the decrease in C1/Cly caused by the reduced frequency of very cold temperatures. The increase in Kyy leads to more However, a true estimate of the calculated o uncertainty is more difficult to assess than calculating changes in o after varying certain model assumptions. The complex interactions between the transport and chemistry of stratospheric ozone in governing the value of o suggests that it would be particularly horizontal mixing and reduces the latitude gradient in the valuable for o calculations to be made with other twocolumn o values. Increases in the planetary wave forcing lead dimensional and three-dimensional interactive models and to decreases in midlatitude springtime column o values with more minor changes elsewhere. These springtime decreases compared to the o results discussed here. The chlorine abundance in the stratosphere is another factor result from decreases in the local o values in the lower that can affect calculated o values. Danilin et al. [1996] have shown that the abundance of chlorine and bromine can affect values calculated with the ozone change definition are shown in Figure 8. Both cases D and E result in column o values at all latitudes within 15% of the baseline case, in which we reduced, the potency of a bromine molecule for destroying ozone is reduced, and the value of o declines. A reduction of total stratospherichlorine from-3.5 to 1.5 ppbv leads to a assume 1979 aerosols and no cirrus clouds. In the springtime reduction in globally averaged o values of-4%. The midlatitudes, where the differences in o are most important, lower values of case D are generally due to the decrease in the local o values between 1 and 2 km caused by increased sensitivity of o to changes in total bromine is much less. A decrease in BrO does lead to a decreased potency of chlorine to destroy ozone through cycle 2, but this cycle plays a lesser C1/C1 r In both cases D and E the increase heterogeneous relative role in the chlorine-induced ozone destruction rate processing in the lower stratosphere leads to lower altitudes of peak ozone destruction, thus tending to increase the column o value; however, this effect is countered by the decrease in local than in the bromine-induced rate. This dependence of o on the atmospheric abundance of halogens, particularly chlorine, suggests that further studies should be performed that can improve estimates of o for use in future halocarbon emission o values, which is caused by increased C1/C1 r When the the calculated o values inside the polar vortex. Danilin et al. [1996] suggested that changes in the amount of bromine lead to alterations in the partitioning among the chlorinated compounds and thus lead to changes in. The values calculated in this work demonstrate significant sensitivity to changing chlorine abundances. This is expected because of the importance of the (C1)(BrO) catalytic cycle (cycle 2) in determining the total loss of ozone due to bromine. As C1 is 7 6 4O //... "' -.'X,\ /('? ---- Baseline ',,x,,,x,z... Case D -,,,,x,.' Case E Latitude (degrees) Figure 8. Column o values for the baseline scenario compared to values for chemical perturbation cases D (with cirrus clouds) and E (increased sulfate aerosols).

10 23,88 DANIEL ET AL.: STRATOSPHERIC OZONE DESTRUCTION scenarios that extend over the coming decades, when the halogen abundance is expected to decline significantly [WMO, 1999]. 5. Conclusions We have discussed two definitions of {x that have been used in the past. Care must be exercised in attempting to relate the {x value calculated from the rates definition (1) to a column or global {x value. If dynamics or any other ozone-destroying process not involving local chlorine or bromine chemistry plays a significant role to the ozone tendency, simply weighting a rates {x profile with altitude by the ozone loss profile can provide a misleading result. Use of the ozone change definition as shown in this work leads to straightforward relationships between {x values appropriate for column and global ozone destruction and local {x values. We have used a two-dimensional interactive radiative/ chemical/dynamical model to calculate values of {x and have probed the sensitivity to a number of parameters. These calculations suggest that the appropriate value characteristic of global ozone depletion at today's chlorine and bromine abundances is -45 with very little seasonality. The results are slightly sensitive to substantial changes in chemical and transport parameters in the present model. It would be beneficial to compare calculated {x values from various two-dimensional and three-dimensional models that are able to reproduce the observed ozone record of the past but which differ in their implementation of the important physical and chemical processes. Such a comparison would provide a more complete estimate of the accuracy of {x estimates. References Daniel, J. S., S. Solomon, and D. L. Albritton, On the evaluation of halocarbon radiative forcing and global warming potentials, J. Geophys. Res., 1, , Danilin, M. Y., N.-D. Sze, M. K. W. Ko, J. M. Rodrigues, and M. J. Prather, Bromine-chlorine coupling in the Antarctic ozone hole, Geophys. Res. Lett., 23, , DeMore, W. B., S. P. Sander, D. M. Golden, R. F. Hampson, M. J. Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb, and M. J. Molina, Evaluation number 11, in Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling, JPL Publ., 94-26, 273, DeMore, W. B., S. P. Sander, D. M. Golden, R. F. Hampson, M. J. Kurylo, C. J. Howard, A. R. Ravishankara, C. E. Kolb, and M. J. Molina, Evaluation number 12, in Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling, JPL Publ., 97-4, 266, Garcia, R. R., and S. Solomon, A new numerical model of the middle atmosphere: 2, Ozone and related species, J. Geophys. Res., 99, 12,937-12,951, Ko, M. K. W., N. D. Sze, C. Scott, J. M. Rodriguez, D. K. Weisenstein, and S. P. Sander, Ozone depletion potential of CH3Br, J. Geophys. Res., 13, 28,187-28,195, Molina, M. J., and F. S. Rowland, Stratospheric sink for chlorofluoromethanes: Chlorine atom catalyzed destruction of ozone, Nature, 249, , Montzka, S. A., J. H. Butler, R. C. Myers, T. M. Thompson, T. H. Swanson, A.D. Clarke, L. T. Lock, and J. W. Elkins, Decline in the tropospheric abundance of halogen from halocarbons: Implications for stratospheric ozone depletion, Science, 272, , Solomon, S., M. J. Mills, L. E. Heidt, W. H. Pollock, and A. F. Tuck, On the evaluation of ozone depletion potentials, J. Geophys. Res., 97, , Solomon, S., J. B. Burkholder, A. R. Ravishankara, and R. R. Garcia, Ozone depletion and global warming potentials of CF3I, J. Geophys. Res., 99, 2,929-2,935, 1994a. Solomon, S., R. R. Garcia, and A. R. Ravishankara, On the role of iodine in ozone depletion, J. Geophys. Res., 99, 2,491-2,499, 1994b. Solomon, S., S. Borrmann, R. R. Garcia, R. Portmann, L. Thomason, L. R. Poole, D. Winker, and M.P. McCormick, Heterogeneous chlorine chemistry in the tropopause region, J. Geophys. Res., 12, 21,411-21,429, Solomon, S., et al., Ozone depletion at midlatitudes: Coupling of volcanic aerosols and temperature variability to anthropogenic chlorine, Geophys. Res. Lett., 25, , World Meteorological Organization (WMO), Scientific Assessment of Ozone Depletion: 1994, Geneva, World Meteorological Organization (WMO), Scientific Assessment of Ozone Depletion: 1998, Geneva, Wuebbles, D. J., Chlorocarbon emission scenarios: Potential impact on stratospheric ozone, J. Geophys. Res., 88, , J. S. Daniel, R. W. Portmann, and S. Solomon, NOAA Aeronomy Laboratory, 325 Broadway, Boulder, CO 833 (jdaniel@al.noaa.gov) R. R. Garcia, National Center for Atmospheric Research, Boulder, CO (Received February 15, 1999; revised May 27, 1999; accepted May 28, 1999.)