Global atmospheric distributions and source strengths of light hydrocarbons and tetrachloroethene

. The atmospheric distributions of CH4, C2H6, C3Hs, C2H2, and C2Ci 4 and their annual chemical removal rates in steady state are determined versus latitude using a modified version of the Oslo two-dimensional global tropospheric photochemical model. A photochemically calculated hydroxyl radical distribution, which has been validated with methylchloroform data, and seasonally varying surface measurements of the title species are used to compute their respective global annual surface source strengths and steady state lifetimes. Computed annual surface source strengths of CH4, C2H6, C3H8, C2H2, and The calculated annual chemical removal rates of these compounds show distinct latitudinal distributions. Because their steady state global lifetimes are less than the model interhemispheric exchange time (about the calculated north to south ratios of the deduced surface emission strengths of C2H6, C3H8, C2H2, and CxC14 probably reflect the locations of their sources. Within the limits of previously estimated industrial emissions of CxCi 4 (3-4 kT) for the southern hemisphere, our calculations indicate that about 47 kT of additional southern hemispheric source of C2Ci 4 is required for 1989-1990 to attain steady state mass balance in this region. There are two possibilities for this needed source: either other industrial sources are missing, or there are unidentified natural sources of C2C14. So far, oceans have been suggested as a natural source. Normalization of monthly varying ratios of hemispherically averaged calculated surface mixing ratios of C2H6, C3H8, and C2H 2 and their respective observed mixing ratios with respect to those for C2Ci 4 indicates that the sources of these hydrocarbons are seasonal in nature. It is also shown that convective transport effectively redistributes these short-lived species but their calculated surface source strengths are relatively independent of this transport process. only source of chlorine radicals in the MBL is due to reaction of HC1 with OH radicals. The concentrations of C1 radicals in the surface emissions of calculated using the nonconvective scheme


Introduction
The chemical composition of the Earth's atmosphere and its spatial and temporal variabilities are controlled by coupled photochemical and dynamical processes. This coupling becomes even more complicated by changing biological phenomena [Prinn, 1994] and by active exchanges between the Earth's surface and atmosphere. The Earth's atmospheric composition has been continually modified by natural and anthropogenic forces, particularly during the industrial era. Observations of trace gases in the atmosphere have revealed the extent to which the Earth's atmosphere and its oxidative capacity have changed [Isaksen, 1988;Thompson, 1992 in this paper, we calculate the annual global surface source strengths of relatively long-lived C2H6, C2H2, and C3H 8 and their corresponding latitudinal distributions. In addition to these light parent NMHCs, CH 4 and tetrachloroethene (C2C14) , a mainly industrially produced chemical, have also been included in the present study to assess their tropospheric budgets. In steady state, in the absence of in situ atmospheric sources, for all these species the global surface source strengths can be equated to their corresponding total atmospheric photochemical losses. The fact that OH radicals are the main sink of surface-emitted light NMHCs and C2C14 (with more than 80% of their sources originating from the northern hemisphere (NH), as concluded in this paper) and because industrial sources of C2C14 are seasonally independent [McCulloch and Midgley, 1996], C2C14 is used to examine the seasonal changes in the surface source strengths of C2H6, C2H2, and C3H 8. To approach these objectives, we have used the seasonally varying surface concentration measurements of these hydrocarbons and C2C14 (described in section 2) as their lower boundary conditions in a two-dimensional global photochemical model. A brief description of this model is given in section 3. For this purpose, we have developed two photochemical schemes PC1 and PC3 which contain CH 4 and CH4, C2H6, C2H2, C2H 4, C3H8, and C3H 6 as parent hydrocarbons, respectively. Details of these schemes are given in section 4. Features and dependence of OH radical distributions calculated using these schemes are described in section 5. Results and discussions of computed global source strengths for C2H6, C2H2, C3H8, and C2C14, including CH4, and seasonal variations in surface sources of light NMHCs are given in sections 6 and 7, respectively. Finally, section 8 summarizes the major findings of this paper.

Surface Measurements of Light Hydrocarbons and C2C14
One of the goals of this paper is to estimate surface source strengths of light hydrocarbons and C2C14. Surface emission estimates of CH 4 from various sources are reasonably well constrained within a certain range [Cicerone and Oreroland, 1988;Crutzen, 1991], but for NMHCs, as stated earlier, there still exist many uncertainties in the identities, and hence in the magnitudes of their source strengths. With the exception of CH 4 which also has soil sinks (about 6% of its total global sink [Cicerone and Oremland, 1988;WMO, 1994]), all the listed hydrocarbons and C2CI 4 included in this study have their surface emissions balanced by corresponding atmospheric chemical losses primarily by OH radicals. For some species, reactions with 03, NO3, and C1 radicals are other minor sinks. The characteristic that in steady state, global atmospheric chemical removals of hydrocarbons and C2C14 are equal to their total surface sources has made it possible to interactively calculate their surface emissions using a photochemical model and their corresponding surface measurements.
To accomplish this, we used surface observation data for the listed hydrocarbons and C2C14 collected during University of California, Irvine's (UCI's) trace gas program [Smith, 1993;Wang et al., 1995]. Under this program, air samples were collected each season along the Pacific rim over a wide latitude range from 71øN to 47øS. To avoid the possible calibration differences among various working groups around the world [Apel et al., 1994], no effort is made to combine their data with UCI's observations. The use of data from a single investigator is required to allow the self-consistent deduction of latitude gradients of surface sources. For CH4, seasonally varying data averaged over four successive years (1988-1991) and for NMHCs and C2C14, data for (1989-1991) and (1989-1990), respectively, were used. Before treating the raw data over these periods, data corresponding to contamination and pollution events were removed. All observation data collected at the same latitude for different days of the same season were averaged. For each model grid location, these averaged data were spatially interpolated using the running average technique. For the model latitudes that lie outside the domain 71øN-47øS for which no observations exist, we assumed the same concentration as for the nearest available latitude. After averaging over the same seasons (which is applicable only for hydrocarbons), all these points were linearly interpolated on 1 day time resolution. Figure 1 shows the resultant seasonally and latitudinally varying treated surface mixing ratios of CH4, C2H6, C2H2, C3H8, and C2C14. Annually averaged observed mixing ratios at the surface for these species are 1.75 ppm (CH4), 1150 ppt (C2H6), 330 ppt (C3H8), 230 ppt (C2H2), and 12 ppt (C2C14) for the NH and 1.67 ppm (CH4) , 310 ppt (C2H6), 50 ppt (C3H8), 40 ppt (C2H2), and 3 ppt (C2C14) for the southern hemisphere (SH). In general, all species show significant seasonal and latitudinal variations (see Figure 1) that are governed by their diverse spatially and temporally varying source distributions combined with their variable atmospheric lifetimes and interhemispheric exchange transport. For further discussion of these measurements and corresponding analysis and interpretation, refer to the references cited in this section.

Model Description
For the present study, we used a modified version of the Oslo two-dimensional (altitude-latitude) global tropospheric photochemical model [Gupta, 1996]. The original version of this model was developed by Isaksen and Rodhe [1978]. In the modified model the vertical domain has been increased by 8 km. In the vertical direction the model extends from the surface to 24.5 km with a total of 49 vertical layers and uniform resolution of 0.5 km. In latitude the model resolution is 10 ø.
The main purpose of extending the model domain is to isolate the dependence of atmospheric distributions of light NMHCs from their upper boundary condition and to better simulate their phc•tc•c'h•rniqtry in th• •nn•r trcmc•qnh•r• gxz rnnkino thiq change the rigidity of the upper boundary condition can be relaxed because of the combined fast removal of these NMHCs by C1 and OH radicals in the upper troposphere and lower stratosphere [Chameides and Cicerone, 1978;Singh and Kasting, 1988].
The advective and diffusive transport coefficients for the modified model are those derived from the Geophysical Fluid Dynamics Laboratory general circulation model (GFDL GCM) [Plumb and Mahlman, 1987]. The transport features of this model have been tested with observed distributions and trends of CFC-11, CFC-12 [Cunnold et al., 1994], and 85Kr [Weiss et al., 1983] tracers yielding an interhemispheric exchange time of about 1 year [Gupta, 1996]. This model has been used to simulate the global distribution of •lSCH4 and its dependence on kinetic isotopic fractionations associated with various sinks of CH 4 [Gupta et al., , 1997].
Another modification made to the Oslo model is the inclusion of parameterized treatment for convective or subgrid scale vertical transport. This transport process is very important for the short-lived species which have chemical lifetimes of the order of a few days to weeks. Through this process, a polluted air mass containing a mixture of nitrogen oxides and hydrocarbons and CO, in addition to other pollutants, can be transported upward from the planetary boundary layer (PBL) to the free troposphere where the chemistry of these pollutants very efficiently forms 03 molecules and hence significantly affects its distribution [Lelieveld and Crutzen, 1994]. Changing the distribution of 03 in the upper troposphere and lower stratosphere can be very important radiatively [Lacis et al., 1990]. In this model the treatment for fast vertical transport by convective clouds and frontal circulation due to cyclones at midlatitudes is based on the parameterization scheme developed by Langner et al. [1990]. For both of these fast vertical transports the model draws air out of the PBL (considered to be the first two model layers) at a specified rate depending upon the lat-itude and season. The boundary layer air is then detrained into the remaining part of the troposphere above the boundary layer according to a specified detrainment vertical profile which is also a function of latitude and season. For each latitude the mass continuity is maintained by compensating subsidence in the respective vertical column. For convective cloud transport this formulation is similar to the one developed by Chatfield and Crutzen [1984]. This approach is believed to capture the most important features of the convective transport mechanism for atmospheric species with a source in the boundary layer and has been applied to simulate the distributions for SO2 and 222Rn [Langner et al., 1990]. For the modified model the seasonal and latitudinal distributions of mass fluxes due to cloud convection and midlatitude frontal cyclones and their respective vertical detrainment profiles are adopted from Langner et al. [1990].

Development of Photochemical Modules
In addition to the transport processes, sources and sinks of various chemical species and their mutual chemical interactions define the behavior of atmospheric pollutants. These pollutants may originate either from their surface emissions (natural and/or anthropogenic) or in situ production, or a combination of both. Various gas-and liquid-phase reactions and heterogeneous losses due to deposition at the Earth's surface, on aerosols, and rain droplets account for the ultimate fate of atmospheric pollutants. The following subsections briefly describe these points including the details about the photochemical schemes PC1 and PC3.

Emissions and Boundary Conditions
For the parent hydrocarbons, depending upon the purpose of the model simulation, two different types of lower boundary conditions were employed. For the deduction of surface emission strengths of light hydrocarbons and C2C14, their observed surface concentrations were used as the lower boundary condition. To investigate the seasonality in their sources, the calculated annual latitudinal surface emissions were uniformly applied on per time step basis at the lower boundary. For acetone, also a byproduct of C.• or higher hydrocarbon oxidation, a surface emission strength of 25 Tg/yr due to biomass burning and biogenic emissions, as concluded by Singh et al. [1994], was equally distributed on per time step basis primarily at tropical latitudes. For C2H 4 and C3H 6 their seasonally varying observed surface mixing ratios averaged for (1989-1991), as described in section 2, were used as the lower boundary condition. At the upper boundary, CH 4 was allowed to diffuse through the top of the model according to the one-dimensional vertical steady state scale height approximation using its losses against OH and C1 radicals in the stratosphere. For NMHCs and C2C14, mixing ratios just above the model domain were fixed as 0.25 times that at the topmost layer; that is, X5o = f x X49, where X is the mixing ratio andf is equal to 0.25. This upper boundary condition for NMHCs is relatively rigid but is appropriate because by 25 km, their concentrations decrease very rapidly with altitude [Rudolph et al., 1981[Rudolph et al., , 1984, thereby making their atmospheric distributions completely independent of upper boundary conditions. In fact, simulations performed with variable value of f ranging between 0 and 1 showed no difference in the vertical profiles of these species [Gupta, 1996].
Annual source strengths of NOx, HNO3, 03, and CO used in  Fuglestvedt et al. [1993]. For NOx and CO, north to south surface emission ratios are 4.2 and 2.6, respectively. For both of these species the surface emissions were applied equally on per time step basis. Stratospheric subsidence rates of 03, NO, and HNO 3 and their corresponding latitudinal and seasonal distributions were taken from the Oslo two-dimensional tropospheric-stratospheric model [Stordal et al., 1985;Isaksen and Stordal, 1986]. For all simulations, full interactive photochemical calculations with transport were performed for the model region between surface and 16.5 km. The extended region of the model between 16.5 and 24.5 km was only used for the transport of light hydrocarbons and C2C14 and for their chemical losses due to prescribed distributions of OH and C1 radicals. Monthly averaged distributions of C1 and OH radicals for the extended region were taken from the Oslo two-dimensional tropospheric-stratospheric model. Also, for the marine boundary layer (MBL), monthly varying distributions of C1 radicals were adopted from the Oslo model. In this model the only source of chlorine radicals in the MBL is due to reaction of HC1 with OH radicals. The concentrations of C1 radicals in the MBL and atmospheric region of (16.5-24.5) km are of the order of (5 x 102-4 x 103) and (7 x 103-4 x 10 4) molecules cm -3, respectively. Monthly varying two-dimensional water vapor distributions were adopted from Barnett and Corney [1985]. Dry deposition loss frequencies for 03, NO2, H202, HNO3, and PAN were calculated for the lowest layer in terms of their corresponding deposition velocity given at 1 m [Isaksen and Rodhe, 1978]. These loss frequencies were weighted by the land and oceanic areas per 10 ø latitude band. Also, the seasonal dependence of land surface deposition velocity was also incorporated as mentioned by Isaksen et al. [1985]. Table 2 lists the deposition velocities of various species used in the model calculations. To examine the effects of the CO soil sink, the deposition velocity corresponding to this loss is also given in Table 2.

Photochemical Schemes
Two photochemical schemes, PC1 and PC3, were developed for the present study which focus on the fate of six light parent hydrocarbons, namely CH4, (C2H6, C2H4, C2H2) and (C3Hs, C3H6) and their oxidation byproducts. Photochemical scheme PC1 contains only CH4, while in addition to CH4, scheme PC3 also includes C2 and C3 compounds. Tetrachloroethene is included in both of these schemes. Initially, we adopted the PC1 chemical scheme from Isaksen and Hov [1987] and modified it to include some reactions of oxidation byproducts such as methyl alcohol, formic acid, nitrous acid, etc. The rate constants of this scheme were updated using the recommendations ofAtkinson et al. [1992] and DeMote et al. [1992]. Presently, the chemical scheme PC1 has about 150 thermal and photolytic reactions involving 45 species. Inclusion of complex and extensively branched chemistry of C2 and C3 parent hydrocarbons and their byproducts increased the total number of reactions to about 750 involving more than 200 species [Gupta, 1996]. Some reactions of parent hydrocarbons and their byproducts involving C1 radicals in the MBL, upper troposphere, and lower stratosphere were also included. For both schemes, rate constants of primary reactions of parent hydrocarbons and C2CI 4 were updated from DeMote et al. [1997], Kaiser and Wellington [1996], and Atkinson [1994].
Photodissociation rates for all species of both photochemical schemes were calculated according to the two-stream approximation method developed by Isaksen et al. [1977] and modified by Jonson and Isaksen [1991] to account for diffuse radiation by Rayleigh and Mie scattering due to cloud droplets and aerosol particles. The data for actinic solar flux and absorption cross sections and quantum yields of all the species undergoing photodissociation were taken from DeMote et al. [1985,1992]. For the middle of each month, all photodissociation rates were calculated off-line at 15 min time and 10 ø latitude intervals, with and without cloud cover. Both sets of these photodissociation rates (with and without cloud cover) were averaged over 1 hour time intervals for their later use in diurnal photochemical calculations.
The chemical part of the two-dimensional model equation was solved using the time-flux operator splitting method and "quasi steady state approximation (QSSA)" [Hesstvedt et al., 1978]. To maximize mass conservation and to use longer time steps for integration, the chemical family technique was used in addition to constraining the sum of concentrations of rapidly cycling species such as OH/HO2, HO2/HO2NO2, and NO/NO2 [Berntsen and Isaksen, 1994]. For all the simulations, typically the model was allowed to run for at least 4 decades to ensure the attainment of steady state concentrations of CH 4 in the entire model domain. More details about photochemical schemes, calculation of photorates, and diurnal averaging and their use for the solution of long-lived species are given by Gupta [1996].  Table 1 were used to perform the experiment named "delta-CH4." This experiment was designed by IPCC working group I to evaluate the indirect effects of a 20% increase in CH 4 concentration on the tropospheric 03 and NOx distributions and on OH concentrations. Following the protocol for this experiment (M. J. Prather, personal communication, 1996), we also determined the ratio of adjustment time to steady state lifetime [Gupta, 1996]. For both photochemical schemes PC1 and PC3 the calculated ratios were 1.43 and 1.39, respectively, which are in good agreement with estimate of 1.45 _+ 0.25 reported by IPCC [1994]. The value of this ratio was also verified by direct adjustment time calculation for perturbation simulation. The adjustment time for a CH 4 pulse perturbation to scheme PC1 was found to be 14 years, which when divided by its steady state lifetime of 9.72 years yielded 1.44 as the ratio of adjustment time to steady state lifetime.  Figure 3 shows the diurnally averaged OH distributions for the months of July and January. During the summer of both hemispheres the maximum concentration is observed around 25 ø which is consistent with other model studies [Crutzen, 1987;Spivakovsky et al., 1990]. This seasonal variation in OH distribution is mainly due to seasonality in the solar UV flux reaching the troposphere which is responsible for the OH production through the photodissociation of 03 molecules followed by the reaction O(•D) + H20. Generally, OH radical number densities decrease with altitudes, but in the upper troposphere and lower stratosphere a slight increase in OH density is calculated. This pattern in vertical profile of OH radicals is consistent with that of Tie et al. [1992] but differs from that of Spivakovsky et al. [1990]. and SH, respectively) and the minimum (6.6 x 105 cm -3 and 7.4 x 105 cm -3 for NH and SH, respectively) in winter. Peakto-peak ratio of hemispherically averaged OH concentration over a year is found to be higher in the SH (2.2) as compared to that in the NH (2.1). Globally averaged OH radical concen-  [1990]. For the termolecular reaction of OH and NO 2 leading to formation of HNO3, we used the rate constant recommended by Atkinson et al. [1992]. New rate measurements for this reaction reported by Donahue et al. [1997] indicate that at 300 K and for pressure region 20-600 torr, the rate constant may be higher by 10-30%. It remains, however, to determine whether this correction applies to all tropospheric temperatures. If so, the revised rate constant will lead to an increase in OH concentration in comparison with that calculated here. There are two primary reasons for this speculation: lowering the OH radical loss rate due to the stated reaction and increase in 03 formation due to more availability of NO 2 radicals to photolyze.

Effect of Convective Transport
As mentioned before, vertical convection draws polluted air from the boundary layer which has enhanced concentrations of CO, hydrocarbons, NO x, and HNO 3 and returns the upper tropospheric air which is rich in 03. Simulating this fast exchange of air makes tropospheric chemistry more complex and augments the oxidizing capacity of the troposphere by increasing overall OH concentration, as shown in Figure 4. This figure illustrates the percent change in OH concentration for the month of July from convection. In the lower troposphere this increase is as much as 5%, and for the upper tropical troposphere it ranges from 15% to more than 45%. On an annual basis, compared to the nonconvective PC1 case, the global average OH concentration increased by 3%. In the lower troposphere this increase is due to more 03, whose photolysis forms O(•D) and hence more OH radicals. Another important effect of convective transport on tropospheric chemistry is to increase the 03 formation efficiency per NO molecule consumed. For given NO emissions the increase in 03 formation efficiency per NO molecule in the PBL is due to decreased background NO concentrations caused by convective dilution [Liu et al., 1987]. Decreases in concentrations of CO and CH 4 resulting from dilution of the PBL layer also increased OH concentrations. In the upper troposphere the relatively enhanced oxidation of CO and CH 4 in the NO-rich environment increased the production of OH radicals.

Effect of Addition of C2 and C3 Parent NMHCs
Addition of C2 and C3 parent hydrocarbon oxidation has significant effects on regional OH concentrations, but little effect on global annually averaged OH concentration. In the lower troposphere at higher latitudes, diurnally averaged OH concentrations derived from scheme PC3 are decreased by as much as 10% relative to that from scheme PC1, as shown for July in Figure 5. However, in the northern hemispheric upper troposphere during summer, OH concentration is increased by more than 8% due to higher NOx concentrations (from oxidation of organic nitrates) which caused increased oxidation of hydrocarbons and hence enhanced production of 03 and OH radicals [Crutzen, 1987]. Also for this region, the OH removal rate is small due to the negligible rainout probability of HNO3, a major sink of OH in the lower troposphere. Overall, the global averaged OH concentration is decreased by 2%. Also, for scheme PC3 the calculated interhemispheric north to south OH concentration ratio decreased slightly (to 0.888 as compared to 0.893) from scheme PC1 because of higher northern hemispheric concentrations of C2 and C3 parent hydrocarbons and their oxidation byproducts.

Effect of Soil Sinks of CO and CH 4
Inclusion of soil sinks of CH 4 and CO to photochemical scheme PC1 has a significant effect on global and hemispheric OH distributions. For the CO soil sink a surface deposition velocity of 0.03 cm/s was used [Seiler and Conrad, 1987;Hough, 1991] which is higher than the 0.02 cm/s used by Crutzen and Zimmermann [1991]. In steady state the calculated magnitude of this soil sink for CO is 278 Tg/yr which is comparable to the lower limit (250 Tg/yr) recommended by WMO [1994]. Pinto et al. [1983] used 0.04 cm/s for the CO deposition velocity and calculated a corresponding sink of 480 Tg/yr. For CH 4 we used a constant soil sink of 30 Tg/yr [WMO, 1994], which is distributed proportionally to the land surface area. For July, inclusion of both sinks increased diurnally averaged OH concentrations by as much as 8% at higher northern hemispheric altitudes. At lower altitudes this increase is relatively smaller as shown in Figure 6. On the global average basis the hydroxyl radical concentration increased by 2%, and the interhemispheric ratio increased to 0.91 as compared to the corresponding values of 0.89, respectively, derived without the soil sink and scheme PC1.

Test of Validation Using Methyl Chloroform
The hydroxyl radical distribution calculated interactively with photochemistry was validated by simulating the latest reported observed trend and surface distributions of methyl chloroform (MCF) [Prinn et al., 1995]. Like CFC-11 and CFC-12, MCF is also solely of anthropogenic origin. It is mainly destroyed by OH radicals in the troposphere, while oceans and stratospheric photolysis are two minor sinks. Thus MCF serves as an indicator for tropospheric OH concentrations.   Table 4). Because of the lack of understanding of types and strengths of various sources of individual hydrocarbons, we have calculated the total surface source strengths of C2H•,, C2H 2, and C3H s using their seasonally and latitudinally varying observed surface mixing ratios, as described in section 2, and OH radical distribution derived from scheme PC3 and validated by methyl chloroform simulations. In addition, we have also calculated annual surface source strengths of CH 4 and C2C14.

Global Distributions and Source Strengths of
At the end of each 1 day time step our calculated surface mixing ratios of the stated species at all latitudes were replaced by the corresponding interpolated observed values of the next day. In steady state the cumulative annual difference between the calculated and updated observed surface mixing ratios at all latitudes for each species was translated into a global source strength. For these initial calculations the convective transport process was not included. Computed steady state annual surface emissions of these hydrocarbons, corresponding north to south ratios of their annual source strengths, and global lifetimes are given in Table 3. Among these NMHCs, C2H•, is the longest-lived with a global lifetime of 83 days. Figures 7-9 show the latitude-time cross section of monthly surface removal rates (which are monthly integrated combined

The latitude belt (30øN-60øN) accounts for more than 54% of the total annual global emissions of these NMHCs, and the latitude belt (20øN-20øS) accounts for more than 27% of their total annual global emissions. With lifetimes short compared to the model interhemispheric exchange time of about 1 year,
the north to south surface source ratios of 4.0, 4.7, and 4.0 for C2H•, , C3H8, and C2H2, respectively, indicate that 80% or more of their emissions occur in the NH. In steady state the total northern hemispheric annual sources are 8.3, 6.9, and 2.5 Tg for C2H½, , C3H8, and C2H2, respectively, of which 1.1, 0.2, and 0.1 Tg, respectively, are transported to the SH where they are chemically destroyed. The maximum north to south interhemispheric flux for January of 0.2, 0.06, and 0.03 Tg for C2H6, C3Hs, and C2H2, respectively, was calculated. Above 16.5 km, removal of C2H 6 by chlorine dominates the total losses and causes a rapid decrease in its vertical mixing ratio; however, this loss of C2H 6 by chlorine radicals is only 2% of its total global loss.
The calculated surface source strengths of these NMHCs and their comparison with other estimates are given in Table 4 Isaksen et al. [1985] but are lower than other reported estimates (see Table 4  C3H8, and C2H 2 for the month of January. Significant presence of all these species is calculated in the model region above 10 km. In the southern hemispheric tropics, vertical profiles of these species show some inversions, that is, increase in concentration with altitude at about 5-15 km altitude that we attribute to the interhemispheric transport mechanism in the model. The extent and frequency of occurrence of these inversions were found to be completely independent of upper boundary conditions. Computed vertical profiles of C2H6, C2H2, and C3H8 compare quite closely to those measured by Goldman Figure 11 shows the distribution of percent change in mixing ratios of C2H6, C3H8, and C2H 2 for the month of January due to the inclusion of convective transport which is intense in the NH summer. For all three species a small decrease (more than 2%) was calculated in the lowest 2 km in the southern hemispheric tropics due to upward movement of air with high mixing ratios of these compounds and downward movement of air from the free troposphere, which is less concentrated in these species. Because of convective transport, throughout most of the free troposphere, C2H 6 mixing ratios increased by 5-40%, whereas for C3H8 and C2H 2 the calculated increase ranges between 25-125% and 25-75%, respectively. The difference in the extent of change in mixing ratio distributions of C2H6, C3H8, and All these source strengths are given in Tg/yr. "Using scheme PC1 with chlorine removal between (16.5-24.5) km and without convection.
•' Using scheme PC1 with chlorine removal between ( Figure 10) for the month of January due to inclusion of convective transport to the photochemical scheme PC3.

C2H2 due to convection is entirely governed by the differences in their lifetimes
and their concentrations in the PBL. For C:Hr•, because of its longer chemical lifetime, diffusion tends to smooth out the excessive gradient distribution caused by convection. Because our convective transport scheme employs seasonally averaged mass fluxes to be drawn out of the PBL and detrainment profiles, the model does not predict the increase in the mixing ratio of these species with altitude in the NH observed by N.J. Blake et al. [1996] which was likely due to episodic convective phenomena. We have also calculated the steady state global source strength of CH 4 using its surface measurements. For OH fields derived from scheme PC1, the computed emission is 498 Tg/yr with steady state global lifetime of 9.74 years. The ratio of northern hemispheric to southern hemispheric surface emissions for CH 4 is 2.1. Addition of C2 and C3 hydrocarbon photochemistry decreased the total surface emission to 490 Tg/yr. Inclusion of convection to scheme PC3 increased this source strength to 500 Tg/yr. Inclusion of soil sinks of CH 4 and CO to scheme PC1 increased the globally averaged OH concentration and resulted in total CH 4 surface emission of 538 Tg/yr, of which 10 Tg/yr is due to chemical feedback between OH and CH 4. These estimates of CH 4 emission are within the range reported by Cicerone and Oreroland [1988], Crutzen [1991], WMO [1994], and IPCC [1994]. Because of its long lifetime and well-mixed distribution, only a very small increase (<1.0%) in the CH 4 mixing ratio was calculated in the free troposphere due to convection.

Tetrachloroethene
Tetrachloroethene, C2C14, is a gas whose only known sources are anthropogenic, although it was suggested that it may have some natural sources from seawater algae Inclusion of chlorine oxidation of C2C14, which at 288 K is 270 times faster than that due to its reaction with OH radicals, in the MBL (assumed to be the lowest model layer) increased the calculated source strength by 17 kT/yr. Adding convective transport to schemes PC1 and PC3 resulted in total calculated annual strengths of 457 and 448 kT, respectively. Owing to its primary emissions in the NH and relatively short lifetime, C2C14 is most concentrated there, as is evident from its surface measurements (Figure 1) and the simulated distribution using scheme PC3 shown in Figure 13 for the month of July. The calculated mixing ratios of C2C14 for most of the NH at 13 km range between 3 and 4 ppt, whereas, for the entire SH, C2C14 mixing ratios are always lower than 3 ppt. Similar to light NMHCs, inclusion of convective transport decreased the C2C14 mixing ratios by 1% in the lowest 2 km around 20øN and increased its mixing ratios by 10-20% in most of the free northern hemispheric troposphere, as shown in Figure 13.

7.
Seasonal Nature of Surface

Sources of Light NMHCs
Wang et al. [1995] suggested that C2C14 could be a sensitive chemical tracer to examine the seasonal amplitude in the hemi-  that the calculated minor oceanic source of C2C14 is also seasonally independent), we have attempted to estimate seasonality in the source strengths of relatively short-lived species C2H6, C3H8, and C2H 2. Of these species, C2H 6 is the longestlived, and its steady state global lifetime is about one fourth of the model interhemispheric exchange time.
To achieve this objective, we simulated the steady state atmospheric distributions of light NMHCs and C2C14 using photochemical scheme PC3 (without convection) and their deduced annual source strengths (applied equally per time step) and latitudinal distributions, as described in previous sections, as their lower boundary conditions. For all 12 months the observed and simulated surface mixing ratios of C2H6, C3H8, C2H2, and C2CI 4 were hemispherically averaged. For each species the ratios of hemispherically averaged simulated and observed mixing ratios, denoted as S/O, were calculated. For all four species this ratio was found to be different from 1.0, as shown in Table 5 Figure 14. Calculated monthly variations in the hemispheric emissions of C2Hr,, C3H s, and C2H 2 for case PC3A shown in Table 4. and MCF. Because both of these factors, that is, seasonal variations in OH concentration in both hemispheres and interhemispheric exchange, are equally applicable to C2C14 and light NMHCs, the calculated deviation in S/O for C2C14 from 1.0, which will have most contribution due to seasonal variations in OH because of its short lifetime (more than 3 times smaller) as compared to the model interhemispheric exchange time, can be used for each month to normalize the S/O ratios for light NMHCs. This normalization of S/O for light NMHCs with respect to S/O of C2C14 will cancel the effects of both the factors mentioned before. Therefore any deviation in these normalized ratios for light NMHCs from 1.0 will reflect the seasonality in their sources. This approach should be most accurate for the NMHCs whose atmospheric lifetime is closest to that of C2C14, that is, C2H6, and semiquantitative for other NMHC species. Figure 14 shows the monthly variations in surface source strengths of light NMHCs obtained after multiplying their monthly source strengths with their respective normalized S/O ratios. The calculation shows that emissions of C2Hr, should be higher than average for the months of August-January in the NH and for July-November in the SH, with maximum amounts during September (0.9 Tg for NH and 0.2 Tg for SH). Emissions of C3H 8 should be higher than average for the months of September-March in the NH and for August-January in the SH, with maximum amounts during November (0.9 Tg) for the NH and during September (0.2 Tg) for the SH. Similarly, C2H 2 emissions should be higher than average for the months of September-January in the NH and for August-February in the SH, with maximum amounts during November (0.3 Tg) for the NH and during December (0.1 Tg) for the SH. These conclusions were unaffected by uniformly changing the surface mixing ratio of C2C14 by +25% and introducing the convective transport and repeating the same exercise. To some extent, these conclusions may be affected by the problems associated with background air sampling, the ways that the raw observed data were treated and the hemispheric averaging was performed, and by chromatographic artifacts.
Over a year, northern hemispheric emissions for C2H6, C3Hs, and C2H 2 for the months of August-January, September-March, and September-January, respectively, are more than the monthly average emissions.
In future studies we will use these calculated emissions of hydrocarbons to examine the effects of perturbations due to anthropogenic and natural sources of various gases on tropospheric chemical nonlinearity and on the distribution and budgets of key chemical species such as 03, CO, NOx, HOx, and CH 4.

Summary
In the present study, we have calculated the global surface removal rates and corresponding latitudinal distributions for CH4, C2H6, C3H8, C2H2, and C2C14 using surface measurements from all seasons. The distribution of OH radicals, the primary sink of these hydrocarbons, is photochemically computed using detailed photochemical schemes. For the methane-only model (PC1) the global and annual average OH concentration is 1.0 x 10 6 molecules cm -3. Annually averaged OH concentrations are found to be higher in the SH which is consistent with some previous studies. Addition of soil sinks for CO and CH 4 and inclusion of NMHC chemistry to photochemical scheme PC1 have significant effects on the global and regional distributions of OH radical and its interhemispheric concentration ratios. Nighttime OH radical concentrations in the lower troposphere were found to be only 40-80 times lower than the daytime maximum value. In the upper troposphere, nighttime OH concentrations drop to almost zero. This difference in diurnal variations in the OH concentration at different altitudes is due to the concentration of HO2 radicals and their rapid recycling with 03 and HO2NO 2. The annual surface emissions of CH4, C2H6, C3H8, C2H2, and C2C14 calculated using the nonconvective scheme PC3 are 490 Tg, 10.4 Tg, 8.4 Tg, 3.1 Tg, and 432 kT, respectively. Inclusion of the convective scheme increased these emission estimates by 2.0, 4.0, 3.9, 4.1, and 3.7%, respectively. For NMHCs and C2C14 these emissions show distinct latitudinal variations. Because of their relatively short lifetimes compared to the I year interhemispheric exchange time, the calculated higher north to south ratios of total emissions of all three NMHCs (which range between 4.0 and 4.7) and C2C14 (7.5) clearly indicate that more than 80% of their global annual sources originated from the NH. Given the industrial emissions of C2C14 for the southern hemisphere, estimated by Mc-Culloch and Midgley [1996], our calculations indicate that as much as 47 kT of additional southern hemispheric source of C2CI 4 is needed for 1989-1990 to attain its steady state mass balance in this region. Because of relatively long lifetimes, these NMHCs and C2C14 show significant presence in the upper troposphere and lower stratosphere. Modeled increases in southern tropical mixing ratios of these species with altitude in the upper troposphere are explicitly due to the interhemispheric transport feature and not from effects of the upper boundary condition. The calculated atmospheric distributions of NMHCs and C2C14 are shown to be strongly affected by convective transport.
With the use of C2C14, which has sources independent of season and also has similar atmospheric chemical sinks as those for NMHCs, the comparison of observed surface mixing ratios of NMHCs with the corresponding modeled surface mixing ratios (calculated using uniform surface emissions throughout the year as the lower boundary condition) indicates that the sources of these species may be seasonal in nature.