The oceans: A source or a sink of methyl bromide?

. The global ocean/atmosphere flux of methyl bromide. extrapolated globally on basis of constant saturation anomaly, the net methyl bromide The same data can also be extrapolated on the basis of steady-state production rate of methyl bromide in the water column, allowing regional and seasonal variations in temperature to affect the saturation anomaly. We have carried out this type of extrapolation, and we found that the oceans are a strong net source of methyl bromide to the atmosphere. The difference arises mainly due to slow degradation rates in water of higher latitudes. A reduction of the applied production rate by more than 35% is needed in order to switch the ocean from a source to a sink of methyl bromide. These results demonstrate the sensitivity of current estimates of oceanic flux to assumptions about methyl bromide production and destruction in the water column.


Introduction
Methyl bromide (CH3Br) is considered to be a major source of bromine to the stratosphere and an important contributor to the destruction of stratospheric ozone [WMO, 1994]. This issue has led to attempts to regulate the antlu:opogenic use of methyl bromide through the Montreal Protocol and the Clean Air Act [USEPA, 1993]. The implicit assumption is that reducing anthropogenic emissions of methyl bromide will lead to a reduction in its concentration in the at•nosphere, in a mmmer similar to that of chlorofluorocarbons. Unlike clfiorofluorocarbons, methyl bromide has both natural and anthropogenic sources. It is both produced and destroyed in the oceans, and the rates of these processes can exert a major control on atmospheric methyl bromide concentrations. Recently, Lobert et al. [1995] observed that eastern Pacific Ocean surface waters were undersaturated in most areas exc. ept for coastal and upwelling regions. They extrapolated these observations globally and concluded that the oceans act as a net sink for methyl bromide. Lobert et al. [1995] also used a simple model of methyl bromide cycling in the surface ocean to estimate the production rate needed to support the observed saturatiøn state in the water colunto along their cruise track. Those calculations assumed that methyl bromide concentrations reflect a steadystate between production (presumably biological), chemical losses, vertical mixing and air-sea exchange [Butler, 1994]. Some of the parameters used in such calculations are strongly temperature dependent, notably the solubility and the chemical loss rate of methyl bromide in seawater [Elliott and Rowland, 1993]. For this reason, it can be argued that the Lobert et al.
[1995] data set should be extrapolated globally on the basis of Copyright 1996 by the American Geophysical Union.

Paper number 96GL00424
0094-8534/96/96G L-00424505.00 the calculated production rates, rather than the observed saturation anomalies. In fact, there are unsupported asstunptions associated with either method of extrapolation and these contribute to uncertainty in our ability to estimate the global ocean/atmosphere flux. In this paper we estimate the global flux from production rate-based extrapolations in order to estimate this uncertainty.

Model Description
In this effort we have used the approach of Butler [1994]  of the COADS data set used in this work is that described by Yvon and Butler [1996]. For the duration of each month, since the meteorological parameters remain unchanged and the gas phase concentrations of methyl bromide are constant, Equation (1) can be solved analytically to give the concentration of methyl bromide at each grid, as a function of time C2 ICo C2)e-Cl(t-t*) where C1 = ka+(Kw+kea)/z, C2=po+Kwpg/Hz and t*, t are the initial and final times, respectively. As a result, the net flux, F, of methyl bromide, between t* and t, is given by the equation:

Model Application
The major task associated with the oceanic component of the methyl bromide budget is parameterizing the regional and temporal variability in the production rate of methyl bromide in seawater. As a "best guess" scenario we used the saturation anomaly data of Lobert et al. error in the slope is 7.3x10 '9, while the standard error in the intercept is 3.8x10 '9. Squares correspond to 33 high latitude winter production rates. The standard error in the slope is 5.9x10 '9, while the standard error in the intercept is 1.4x10 '9.
Production rates have been extracted from the measurements of Lobert et al. [1995].
fluorescence data to derive a relationship between chlorophyll and methyl bromide production. We subdivided the data set into two regimes: tropical/subtropical latitudes and high latitude wintertime (Figure 1). Due to the lack of measurements, we assumed that high latitude summertime has the same behavior as the tropical/subtropical latitudes. We then used the Coastal Zone Color Scanner (CZCS) monthly color images of phytoplankton pigment concentrations to develop monthly gridded concentrations of chlorophyll in the oceans and from those obtained a gridded data set of methyl bromide production compatible with the model. The estimated global production rate of methyl bromide was 183.9 Gg yr 'l in reasonable agreement with the 151.2 Gg yr '• estimated by Lobert et al. [1995]. The total uncertainty in the estimated production rate is about 20%. Major components of this uncertainty are the oceanic loss rate, the airsea exchange coefficient, the solubility, as well as the linear relationship between production and chlorophyll. The loss rate is the largest component of this uncertainty, and the contribution would become greater if the uncertainty in the magnitude of biological removal had been included. However, since insufficient data are currently available to extrapolate the biological removal rate globally, the loss rate was only calculated in ternis of the chemical removal.
Although methyl bromide production and chlorophyll are adequately correlated, as shown in Figure 1, it should be viewed skeptically in terms of causality, as there are many characteristics of seawater which vary in concert with chlorophyll in a given oceanic regime. Moore et al. [1995], for example, have reported that preliminary results of their experiments indicate that marine phytoplankton do produce methyl bromide, but with rates that account for only a fraction of the global production rates estimated by Lobert et al. [1995]. In order to examine the dependence of the oceanic behavior on this assumption, we performed a second set of simulations, with the same global methyl bromide production rate, but uniformly distributed across the oceans. Figure 2 illustrates the saturation anomaly, •, of the surface ocean for the month of January for both simulations. The saturation anomaly, in percent, is defined as the departure of the partial pressure in seawater from equilibrium, Ag =100•vw-ps)/ps. The saturation anomaly is the quantity which is directly determined in field studies and used to derive the air/sea flux . This quantity is highly dependent on the air and seawater concentrations and on the temperature of the waters. As shown in Figure 2a sensitive to the assumption used to estimate the production rate in our base case scenario, but in both cases the ocean is predicted to be a net source of methyl bromide. The above finding is of major importance, because it indicates that there may be a strong positive natural net flux of methyl bromide from the oceans, that cannot be eliminated with any control of anthropogenic emissions. If the above results are correct, even within a thctor of two, it is expected that the oceanic flux will increase substantially, as the anthropogenic emissions are reduced. If, for example, we assume that the average concentration of methyl bronfide in the northern hemisphere is reduced to 10 ppt, our model predicts that the global flux froln the oceans will increase by about 6 Gg yr '•. The prediction that the ocean is a net source of methyl bromide may also help in the effort of reconciling the methyl bromide budget [Butler, 1995]. Balancing the sources and sinks of methyl bromide, while accounting for the observed interhemispheric gradient is one of the biggest issues facing scientists in the field. For example, if the ocean is a net sink of methyl bromide, what are the sources of methyl bromide in the southern hemisphere that result in an atmospheric methyl bromide concentration as high as 9 ppt? We have carried out simulations with a full three dimensional Chemical Transport Model (GRANTOUR) including current best estimates of anthropogenic emissions and biomass burning. Preliminary results indicate that, if the oceans are a net sink, the methyl bromide concentrations in the southern hemisphere should be much lower than 9 ppt [Pilinis et al., 1995]. On the other hand, if the recent estimate of a global annual soil sink of 42 Gg [Shorter et al., 1995] is correct, most of which exists in the northern hemisphere, it is hard to explain 9 ppt of methyl bromide in the southern hemisphere due to oceanic sources alone, while reconciling the N/S gradient of 1.3. Another possible source is biomass burning, but it has no significant emphasis in either hemisphere [Butler, 1995].
In order to examine the sensitivity of the saturation state of the ocean to the production rate used, we performed two additional sets of simulations. In the first set the total production was reduced by 18% to 151.2 Gg yr 4, the global production rate estimated by Lobeft et al. [1995]. The model predicted a net oceanic flux of 13 Gg yr 'l, a reduction of about 42% froin the base-case silmdation. Therefore, despite the sensitivity of the net flux to the production rate, the oceans are still predicted to be a net source of methyl bromide. hi the second set of simulations, the total production was reduced by 35%, to 119.8 Gg yr 'l. The model predicted that, for this reduced production rate, the net oceanic flux is 0.0 Gg yr", i.e. the ocean is neither a net source nor sink of methyl bromide. Reducing the total production rate of methyl bromide even further causes the ocean to behave as a net sink of methyl bromide. If methyl bromide is also removed from the oceans biologically, the result would be equivalent to a reduction in the net production rate in the oceans. Since neither the mechanisms nor the magnitude of this removal are known, the simulations with reduced production rates presented above provide an estimate of the effects of an increase in the loss rate due to biological activity.
An additional set of runs were perfornied to examine the sensitivity of the oceanic flux to the choice of the depth of the mixed layer. An increase of the mixed layer depth by 20% resulted in a decrease of the net oceanic flux by about 30%, while a decrease of the mixed layer depth by 20% resulted in an increase of the net flux by 40%. Therefore, in all cases, the ocean is still predicted to be a net source of methyl bromide.

Conclusions
Production-based extrapolations yield a positive global net flux of methyl bromide from the ocean to the atmosphere of 22.5 Gg yr 4, while the saturation state-based extrapolation performed by Lobert et al. [1995] yields a global negative flux of approximately 13 Gg yr '•. Both approaches suffer from the relatively limited geographic coverage of field measurements and from our limited understanding of the biological processes involved in production and destruction of methyl bromide in the water column. The saturation state-based extrapolation involves the implicit assumption that regional and seasonal variability in production and destruction rates are tightly coupled. This is perhaps possible if the water coltunn losses of methyl bromide are biologically controlled, rather than chemically, as asstuned in the model. The production-based extrapolations assume that there is no coupling between production and destruction, and in this approach the temperature dependence of the chemical loss largely drives the variability in the saturation state. At this time we have no basis with which to assess the validity of the assumptions inherent in the two approaches. Given the importance of the global oceanic flux of methyl bromide and its sensitivity to assumptions about the production and destruction rates, further research into the processes controlling the methyl bromide distribution is warranted.
Acimowledgments. This work was partially supported by NOAA grant NA46GP0310. We would like to thank Drs. J. H. Butler, J. M. Lobert and S. A. Yvon for providing us the COADS and BLAST94 databases and for valuable suggestions, as well as Dr. Francisco Chavez for the fluorescence data.