ANALYSIS OF SOURCES AND SINKS OF ATMOSPHERIC NITROUS-OXIDE (N2O)

There are major unanswered questions about the sources of atmospheric nitrous oxide. Recent assessments of N2 O sources are diverging rather than converging. Here I use a simple model to place constraints on the relative sizes of N2 O sources in the northern and southern hemispheres (NH and SH). Using measurements of the rate of temporal increase of atmospheric nitrous oxide (N20) concentrations and observations of their cross-equator differences, I calculate the total sources of N20 for the (NH and SH). The NH source (cid:127)N must exceed the SH source q)$; their ratio R (-- c(cid:127)N/c(cid:127)$) is between 1.7 and 2.4 for the case where the NH concentration is I ppb higher than that in the SH. When the NH concentration is 0.75 ppb higher, R is between 1.5 and 1.9. These values of R increase by about 50% when more rapid loss to the NH stratosphere is introduced. Slightly wider ranges of R are calculated for certain choices of parameters. For these calculations I have used a two-box model which permits an array of assumptions to be tested; a range of values is employed for interhemispheric exchange times and for removal times due to stratospheric destruction of N9. O. In addition to the common case of no uptake of atmospheric N2 O by soils, I have tested the model sensitivity to two hypothesized nonzero soil sinks. Relatively small soil sinks would decrease the atmospheric residence time of N2 O to values below those that are calculated from stratospheric removal alone. for of N(cid:127)O; of (I)N, (I)$ and (I)N-$, moles per year. N(cid:127)O data for the distribution of N(cid:127)O between for N(cid:127)O, that is œN, œS, DN, and Ds,


INTRODUCTION
Nitrous oxide is an important trace gas in the atmosphere because it is an effective greenhouse gas and because it is the major source of stratospheric NO, which is important in ozone-layer chemistry. Its discovery and early views on its behavior are reviewed by Bates and Haps [1967]. Reactions that produce NO from N20 in the stratosphere and other reactions that destroy N20 and their rates are discussed by Johnston et al. [1979].
Evidence that N20 does release far more NO into the stratosphere than any other source was provided by Jackman et al. [1980]. As a greenhouse gas, N20 functions through several absorption bands in the wavelength region 7.7 /•m to 17 /•m [Ramanathan et al., 1985]. In the past century the atmospheric concentrations of several greenhouse gases have increased, including CO9., CH4, chlorofluorocarbons, N20, and tropospheric ozone. Of the enhanced greenhouse effect (radiative forcing) due to these increases to date, N20 increases have contributed 2 to 3%; depending on future concentrations, the contribution from N20 could increase to about 10% [Dickinson and Cicerone, 1986].

Measurements show that N•O is increasing in the global
atmosphere. Weiss [1981] and Weiss et al. [1981] showed that the 1978 concentration in the northern hemisphere was 300 ppb and that the average annual increase between 1976 and 1980 was about 0.6 ppb (parts per billion by volume in dry air) per year, or about 0.2% per year. These data are from several stations spanning the globe and from several extended oceanic cruises. Weiss [1981] also showed that an annum increase of 0.65 ppb could be inferred for the period 1961-1978 from other data. More recent data (R. F. Weiss, private communication, 1988) shows that through 1987 the average annual increase 1Now at Geosciences Department, University of California, Irvine. Copyright 1989 by the American Geophysical Union.
Paper number 89JD02757. 014a-0227/89/89 JD-02757505.00 was about 0.7 ppb and that the average excess in the northern hemisphere relative to the southern hemisphere of 0.• ppb reported by Weiss [1981], persi. sts. Analysis of an independent data set from Oregon and Tasmania (1979)(1980)(1981)(1982) showed possibly larger annual increases of 0.2 to 0.4% per year, an interhemispheric difference similar to that of Weiss [1981], and evidence for seasonal cycles [Khalil and Rasmussen, 1983]. Data from an extended ship cruise (50øN to 40øS) in 1987 have shown an excess of N20 in the northern hemisphere of about 1.0 ppb [Butler et al., 1989].
Current knowledge indicates that N20 is a very longlived gas once in the atmosphere. It is thought to be inert in the troposphere; photolysis (producing O(ID) and Nj) and attack by O(1D) (producing NO) in the stratosphere are the removal mechanisms. If these processes are the only removal mechanisms, then the atmospheric residence time rR for N20 is over 100 years. Are human perturbations to the cycling of nitrogen and within each hemisphere N:•O is relatively well mixed, through Earth's soils, oceans and atmosphere (see, for ex-at least at the surface. Consequently, a two-box model ample, Delwiche [1981]) responsible for increasing atmo-can be used to obtain estimates of differences between spheric N20? This important question has not yet been northern and southern sources and of interhemispheric answered, although human causes are likely. Attempts to transport; shortcomings to this approach are discussed construct an accurate ledger of N20 sources have not yet later.
succeeded; the proposed N20 sources do not sum to the Let N equal the total amount of N20 in the northern total of the atmospheric sink rates plus the observed an-hemisphere as a function of time, t, and let $ equal the nual increase. Recent attempts to construct balanced N20 total amount of N20 in the southern hemisphere. Then budgets [Keller et al., 1986;Gammon et al., 1986;HaG et al., 1987] have been undercut by newer measurements that indicate a much smaller (if any) N20 source from combustion processes. Specifically, it has been shown that artifact N20 is produced in sample flasks when air samples are rich in moisture, NO or NO2 and SO2 [Muzio and Krarnlich, 1988  north-south differences in concentrations; see section 4. x 10 • mol. For this study the exact N20 concentrations, Analysis by Levy et al. [1979Levy et al. [ , 1982 with three-dimensional for example, 305 ppb in the northern hemisphere, are not global transport models that assumed mostly continental of concern; the difference between hemispheres is the key sources (and hence, mostly NH) calculated slightly larger factor. Again, the exact number of moles for N and S is • tion to these cases where the atmospheric residence time For an atmospheric species like N20 whose atmospheric due to stratospheric removal, 1/oe, is set at 100, 125, 150, residence time rR >> re (the time for exchange between and 180 years and oeN = oe$, I have also modeled the case hemispheres), we expect significant interhemispheric ex-of these same four values of 1//• but with oeN = 1.6 oeS.

There is now a 15-or 20-year history of uncertainty for oeN and oe$. For many of the cases I take oeN --oeS, in N•O source sizes, identities, and r R [Liu et al., 1977]. but because there is some evidence from models that oeN It is this continuing uncertainty that prompts me to seek exceeds oe$ [Levy et al., 1982] I also study this case. I simple and useful constraints on some of these quantities. use the following units: for N and S, moles of N•O; for In this study, I examine the implications of atmospheric (I)N, (I)$ and (I)N-$, moles per year. Destruction rates N•O data for the distribution of N•O sources between for N•O, that is oeN
In principle, there could be faster loss of N20 from the would increase this ratio to about 2.5). Here I assume for NH troposphere to the NH stratosphere (followed by de-simplicity that soil sink activity is the same for all land struction) than occurs in the SH because NH topography area. With DN --2.1 DS and DN --(1/780 years), the leads to stronger planetary wave activity and because of a total global sink is 1.5 x 1012 g N/year or 5.4 x 10 lø stronger polar vortex in the SH. Stated differently, subsi-mol N20 per year, For the case DN = (1/390 years), dence from the NH stratosphere delivers more N20-poor the total global sink is 3 x 1012 g N/year or 1.07 x 10 TM air on an annum basis than in the SH. Three-dimensional mol N20/year. Note that an N20 consumption rate  Table 1. The upper third of Table I is  fluxes to the soil of order 109/cm 2 s were observed 10% The next column to the right in Table I is  In all these Ds twice as large as the previous case. cases, microbial reduction of N20 to N2 is suspected Several patterns are seen in the results of Table 1. to occur. Thus while the net contribution of soils to First, in all cases, R is greater than 1.4. It is only for the hemispherically, annually averaged N20 sources, (I) N the case of slow interhemispheric exchange, re --1.5 years, and (I)s, is positive I include soil sinks, DN and DS, to that R is less than 1.7. For re -1.0 year and 0.8 year, account for hemispheric differences.

< R < 3.1. For slow interhemispheric exchange the I have included nonzero soil sinks in the present effect of increasing the hypothesized soil sink is to increase calculations because sinks have been observed occasionally the ratio •N/•$.
But for re = 0.8 years, R actually in field measurements but also to illustrate how relatively decreases from 2.66 for Di -0 to 2.55 for the largest smaJl surface sinks could affect the atmospheric residence DN and Ds when 1/oe --150 years, and R decreases time of N20, given that the only other N20 sinks are in from 3.14 to 2.89 for 1/oe -180 years. This is due to  Table 2 presents the results for 36 cases where the northern hemisphere concentration of N20 is only 0.75 ppb greater than that in the southern hemisphere. The source ratio R that is required to maintain this smaller gradient is less for each combination in Table 2 than for the corresponding case in Table 1. In Figure 2, graphs of these results appear for the case DN = DS = 0. In Table 2 Figures 1 and  2 which are for oeN = oe$, the R values for oeN = 1.6oe$ are more than 50% higher. This should not be surprising; to maintain the same concentration gradient across the equator requires a higher ratio of northern hemispheric to southern hemispheric sources when there are larger sinks in the NH than the SH.

CONCLUSION
I have solved simplified equations for mass conservation of atmospheric N20 subject to the constraints of measured temporal increases and cross-equator differences, for a range of other related parameters including the time required for interhemispheric exchange, stratospheric destruction rates, the rates of consumption of atmospheric N20 by soils, and possible north-south differences in N20 loss rates. A two-box model was used; the total N20 sources for each hemisphere, northern and southern were computed. For all choices of parameters, NH sources dominate: •N/•$ lies between 1.4 and 5.2 for an N20 concentration that is 1 ppb higher in the NH than in the SH (see Table 1), and between 1.3 and 3.5 for an N20 concentration that is 0.75 ppb higher in the NH than in the SH (see Table 2 When an hypothesized soil sink is introduced, •N/•$ ratios increase somewhat, and atmospheric residence times for N20 become shorter. Soil sinks of the type assumed here could reduce rR by as much as 20% from the value with only stratospheric destruction. The present approach illustrates how the conclusions depend on several parameters, and how two processes, faster N20 loss to the NH stratosphere and assumed soil sink activity, each increase calculated R values, but it cannot resolve spatial differences within hemispheres without adding more boxes or temporal variations on seasonal (or shorter) time scales. More sophisticated analyses are needed with larger data sets. For example, Prinn and colleagues are introducing a multibox approach that deduces large tropical sources of N20 [Prinn et al., Atmospheric trends and emissions of nitrous oxide deduced from ten years of ALE-GAGE data, submitted to Journal of Geophysical Research, 1989], which if correct is an important finding. To reach such conclusions will require accurate differences between N20 concentrations at different latitudes; the present calculations are sensitive only to hemispheric differences. In deducing concentration gradients with latitude, seasonal cycles may have to be recognized [Khalil and Rasmussen, 1983]. Vertical gradients also contain information, even though they are known to be very small in the troposphere. Threedimensional transport models such as those by Mahlman et al. [1986] will be increasingly valuable as the data base grows. As they studies progress, it may become clear that there was as much or more N20 in the SH than the NH before there were significant anthropogenic sources, as Weiss [1981] suggested. Stable isotopes of nitrogen (15N) and oxygen (170 and 180), when measured in ratio to the major isotopes (14N and •60), could provide good indicators of which N20 sources are dominant, as well as evidence of soil uptake.
Such studies should be undertaken.
Field studies to quantify the sizes of soil sinks for atmospheric N20 are certainly needed, but they are difficult. First, small concentration changes must be measured. For a typical closed chamber (volume = 7 oe, height = 9 cm), a flux to the soil of 3 x 108/cm 2 s will decrease the N20 concentration by only 3 ppb after 30 min. Another particular problem is that leaks can mask N20 losses to soils in closed-chamber experiments.
When ambient air (with N20 present at 305 to 310 ppb) enters a typical closed chamber at a rate of 3 cm3/s an N20 loss of the order of 3 x 108/cm 2 s can be masked. The removal of, say, 50 cm 3 of air from such chambers in syringe sampling also forces external air into the chamber but is less of a problem. It is possible that in some field experiments to date, when N20 sources near zero were reported, small sinks may have been operating in reality.