Changing Climate: Report of the Carbon Dioxide Assessment Committee (1983)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Past and Future Atmospheric 3 Concentrations of Carbon Dioxide 3.1 INTRODUCTION Peter G. Brewer This chapter on how the carbon content of the atmosphere and other reservoirs may change over time has been written in several sections by individual authors. In reviewing the material it is clear that all controversy in this area has not been resolved. There are three principal goals that we seek in our evaluation of the carbon cycle among atmosphere, oceans, and biota. First, the climate change that we anticipate is based on the atmospheric C02 rise. We only have good measurements of atmospheric C02 from the time of the International Geophysical Year in l958 to the present. We need to know the preindus- trial value, the time course of its change in the decades prior to l958, and the factors causing this change. Second, we need to know as accurately as possible the fluxes of C02 among atmosphere, ocean, and biota today so as to be able to evaluate contemporary measurements; to separate natural, anthropogenic, and climatically modified effects; and to identify the role of other greenhouse gases. Third, if our projec- tions of the future are to have credibility, we must understand the sensitivity of our carbon reservoirs to change and the linkages and feedbacks that exist between them. We do have reasonable knowledge of the consumption of fossil fuels in the early part of this century. If we assume that the airborne C02 fraction has remained constant over this time, then simple backextrapo- lation yields an atmospheric C02 level of about 290 ppm at the turn of the century. Direct measurements of CO2 in air at that time yield equivocal values, having a mean of about 290 ppm and a low value of about 270 ppm. Recent measurements of the CO2 content of air trapped in glacial ice indicate a value of about 265 ppm for the middle of the last century. If a preindustrial value of 265 ppm is accepted, then two things are apparent. First, the discrepancy between the extrapo- lated 290 ppm and the observed 265 ppm must be accounted for; the difference of 25 ppm added to the atmosphere most likely would come from the terrestrial biosphere. There is little evidence for an oceanic source, although modest degassing of the ocean would occur on warming. Second, the discrepancy is not an insignificant fraction of the atmospheric C02 rise but accounts for some 30% of the C02 signal that we see today. The warming due to CO2 is complex but may be l86

l87 approximated by the logarithmic relationship (see Chapter 5) 3.0 • ln 2 ln V [CO2 where [CO2] is the present CO2 value, [CO2]0 is the preindustrial value, and 3.0° represents the mean increase in temperature estimated for a doubling of atmospheric CO2. The overall warming today then could be as high as l.l°C, and the initial CO2 difference between the low and high estimates is 0.4°C. In the long run the difference will be insignificant; for the present it is uncertain whether we have observed a CO2-induced warming, and it is of great interest to know the theoretical size and shape of our signal. In evaluating the fluxes of CO2 among our atmospheric, oceanic and biotic reservoirs today we note that we have direct measurements only of the atmospheric reservoir. Measurements of carbon stocks in the global biosphere are complex and are inferred from local measurements, patterns of land use, soil changes, and deforestation. In the ocean we feel keenly the lack of a high-quality time series of measurements. Measurements of the contemporary ocean reveal the large natural CO2 cycle and only hint at the anthropogenic signal. Our principal infor- mation comes from the observed fractionation of C between air and sea from which we calculate oceanic CO2 uptake. The result is an averaged signal, and resolution is poor on time scales less than a decade. The principal causes of the annual fluctuations in atmospheric CO2 are the seasonal growth and decay of the terrestrial biota. It is as yet difficult to ascribe annual atmospheric fluctuations to oceanic changes. One exception to this has been the correlation by Bacastow and Keeling of atmospheric CO2 trends correlated with the Southern Oscillation Index. The amplitude of the annual CO2 fluctuations revealed at the Mauna Loa observatory shows a tendency to increase with time, suggesting increasing terrestrial photosynthetic and respiratory activity. Ocean-atmosphere carbon models calculate the partioning of fossil fuel CO2 released over the last two decades to be about 40% oceanic uptake and 60% atmospheric fraction, with minor net transfers from the terrestrial biota. If we express the global carbon balance as A = F - S + B (Woodwell, this volume, Section 3.3), where A is the increase in the carbon content of the atmosphere over any period, F is the release of carbon to the atmosphere from combustion of fossil fuels in the same period, S is the net transfer to the oceans in the same period, and B is the absorption or release of carbon by the biota in the same period, then the terms S and B are the least well determined. The "airborne fraction" is not directly determined from either oceanic or biosphere experiments or models. Estimates of carbon changes in the terrestrial biota vary widely from small net increases to large net decreases. In this chapter the current net release of carbon from the biosphere is estimated as about 2 gigatons of carbon (Gt of C) per year. This is at the upper limit that can be accommodated by atmosphere-ocean models. Two things could confound our current

l88 thinking: If the rate of release of CO2 from the biota was growing at an identical rate to the CO2 release from fossil fuels, we would find it hard to detect; and if the biotic release was matched by some unknown CO2 sink, the releases could indeed be large. So far we cannot unequivocally support either of these scenarios. In summary, the recent estimates of an atmospheric CO2 concentra- tion of about 265 ppm around l850 lead to a predicted warming greater than that yet observed today if we use the upper range of climate model results, and point to a net CO2 source from the terrestrial biosphere contributing about 25 ppm to the atmospheric levels. This flux from the biota most likely occurred in the late nineteenth century and early decades of this century. The net release from the biosphere today could be about 2 Gt of C per year, although lesser or negative fluxes would not be inconsistent with oceanic and atmospheric models. Finally, we must keep in mind, as Revelle's analysis (Section 3.5) of methane hydrates in continental slope sediments suggests, that climate change may bring about surprising changes in fluxes of carbon. 3.2 CARBON DIOXIDE AND THE OCEANS Peter G. Brewer The effect of solution of the gas by the sea water was next considered, because the sea acts as a giant regulator of carbon dioxide and holds some sixty times as much as the atmosphere. The rate at which the sea water could correct an excess of atmospheric carbon dioxide depends mainly upon the fresh volume of water exposed to the air each year, because equilibrium with the atmospheric gases is only established to a depth of about 200 m during such a period. The vertical circulation of the oceans is not well understood, but several factors point to an equilibrium time, in which the whole sea volume is exposed to the atmosphere of between two and five thousand years. —Callendar (l938) 3.2.l Introduction The quotation above is still, after more than 40 years of progress, as succinct a statement of the problem as one could desire. The rising atmospheric C02 level has been carefully measured since l958. Cal- culations made on the amount of fossil fuel CO2 released during this time, based on good records of oil, coal, and gas combustion, show that the observed increase in atmospheric CO2 is a little more than half the amount of fossil fuel CO2 input.

l89 The CO2 not present in the atmosphere must have been transferred to some other reservoir, and all investigators who have examined the problem over the last four decades have concluded that the ocean is, and will remain, the primary sink for fossil fuel CO2. The ocean holds about 53 times the total atmospheric carbon dioxide content, or about 3.7 gigatons of carbon (Gt of C) as CO2. The depth of the ocean mixed layer that establishes annual contact with the atmosphere is about 75 m, and the mean circulation time for the deep oceans is about 500 years (Stuiver et al., l983). The ocean acts as a "giant regulator" not only of CO2 but also of climate and thus occupies a central role in the debate over the effects of increasing atmospheric CO2 levels on our society. The capacity of the ocean for CO2 uptake is a function of its chemistry; the rate at which this capacity can be brought into play is a function of ocean physics. In addition to these direct and present contributions, the deep ocean carbonate sediments provide, on a longer time scale, a vast buffer against chemical change. The natural vertical gradient of CO2 with depth in the oceans is driven by the biological flux of particulate matter. There have been many recent papers and reviews on these topics (e.g., Broecker et al., l979; Takahashi and Azevedo, l982), and models designed to reproduce their effects (e.g., Oeschger et al., l975; Killough and Emanuel, l98l). The Scientific Committee on Problems of the Environment (SCOPE) reports by Bolin et al. (l979) and Bolin (l98l) provide excel- lent assessments of the carbon cycle. The scene is one of constant research and evaluation; some basic facts, however, hold constant, and some uncertainties are widely recognized. These form the basis of this review. In attempting to model future atmospheric CO2 levels, the largest uncertainty of course surrounds the economic and energy resource decisions facing mankind. The CO2 content of the future atmosphere will largely reflect how much CO2 we choose to put in. The key word is choose, for however hard those decisions may be, they represent choices distinct from the natural laws that will inevitably be obeyed as the CO2 level rises. 3.2.2 The Cycle of Carbon Dioxide within the Oceans Measurements of the carbon dioxide system in seawater plainly reveal the natural cycle. If we wish to detect changes in the ocean resulting from anthropogenic CO2, then a prerequisite is that the natural cycle be well understood. This cycle is intimately linked to that of oxygen, and the nutrient elements nitrogen and phosphorus, and to ocean circulation. 3.2.3 The Deep Circulation The oceans are stably stratified, capped by a warm, less dense surface layer and increasing in density with depth. The deep ocean waters have a salinity of about 34.9 parts per thousand and a temperature of about 2°C. Most of the deep waters of the world's oceans are formed in wintertime in the Norwegian and Greenland Seas and in the Weddell Sea.

l90 Here winter cooling increases the density of surface waters until the stratification of the water column breaks down and, by a poorly under- stood process, the deep source regions are renewed. Once formed, the bottom waters of the basins exit via various sills and proceed on their grand tour. From the Norwegian and Greenland Seas the flow is to the south; the residence time of deep water in the Atlantic Ocean has recently been estimated as 275 years (Stuiver et al., l983). In the southern ocean these deep waters become entrained in the great clock- wise circulation of the Antarctic Circumpolar Ocean. Here they branch, after a residence time of some 40 years, into either the Pacific or the Indian Oceans. The residence time of deep water in the Pacific Ocean is about 600 years, and in the Indian Ocean about 335 years. Within these ocean basins the deep waters are gradually entrained into the shallower flows of the intermediate waters and are eventually returned to the surface. The flows are not simply advective, and large-scale turbulent processes along and across density surfaces predominate. Reid and Lynn (l97l) have elegantly shown that the salinity maximum of the North Atlantic deep waters can be traced on their trajectory around the globe. Stuiver et al. (l983) have followed the decay of radiocarbon within these waters and calculated their ages. 3.2.4 Biological Activity Photosynthetic activity by phytoplankton in ocean surface waters fixes CO2 into organic tissues. The global oceanic annual primary produc- tivity is uncertain but is approximately one half that on land (Peterson, l980). The amount, then, is large. Oxygen is produced in this process, and ocean surface waters typically show slight (l-2%) supersaturation with respect to equilibrium with atmospheric 02. In contrast to primary productivity on land, where large standing stocks of carbon are formed in woody tissues, oceanic production by phytoplankton is very rapidly consumed by grazing organisms. Some 90% of the organic matter formed is consumed within the euphotic zone. The remainder falls through the ocean-water column, as excreted fecal pellets and discrete cells, and is subject to oxidative decomposition by microbes. Production at the ocean surface is limited largely by the availability of the nutrient elements nitrogen and phosphorus. These combine with carbon in proportions generally represented by the reaction l06 C02 (g) + l6 N05 + H2PO4 + l7 H+ + l22 H20 - Cl06H263°ll0 Nl6P + l3802 (g). where the subscript g denotes the gaseous state. Although local variations in these ratios are found, the mean oceanic signal is remarkably constant. The removal of C02 from surface waters raises the pH; the removal of nitrate and phosphate raises the pH and the alkalinity (Brewer and Goldman, l976). The pattern of oceanic primary productivity is such that large areas in the center of the great oceanic gyres are nutrient-impoverished oligotrophic regions, where little production occurs. In upwelling regions, such as those at the eastern sides of oceanic basins where

l9l nutrients are brought to the surface, intense biological activity occurs. 3.2.5 Deep Decomposition of Organic Matter Below the euphotic zone reaction (l) proceeds in the reverse direction. Oxygen is consumed, and carbon dioxide and the nutrient elements are released to the dissolved state. The rate of this reverse reaction is quite variable. It is generally found to decrease quasi-exponentially with depth, with oxygen consumption rates ranging from about 0.5 ml of O2/L/yr at shallow depths to less than 0.0l ml of O2/L/yr in the abyss - (Jenkins, l980) . Carbon dioxide is released in proportion to this. The result is that the ocean water column is characterized everywhere by an oxygen minimum below the euphotic zone, where decomposition is rapid and the water column is poorly ventilated. Combining these processes with the scheme of the deep circulation given earlier, we see a progressive change in the chemistry of the deep water during its 500-year deep ocean tour. In traveling from the North Atlantic to the Antarctic to the North Pacific, the water becomes sys- tematically depleted in oxygen and enriched in CO2 and the nutrients. The GEOSECS series of atlases beautifully reveal these trends as the rain of material from above inexorably shifts the oceanic CO2 chemistry. 3.2.6 Calcium Carbonate The increase in deep ocean CO2 due to oxidation of organic matter lowers the pH of seawater, increasing its corrosiveness to calcium carbonate. Surface seawater is supersaturated with respect to calcium carbonate, which is secreted by many marine organisms to form their shells. These rain down through the ocean to form calcareous sediments. Calcium carbonate solubility increases with increasing pressure and decreasing temperature, and at some point a horizon occurs below which calcium carbonate dissolves. In the North Atlantic Ocean this horizon occurs at great depth, approximately 5000 m. Progressing along the path of the deep circulation, oxygen and pH are lowered with increasing (X>2 levels, and progressively more calcium carbonate is dissolved. The dissolution horizon shoals markedly along this trajectory. The calcium carbonate thus dissolved from the sediments raises the CO2 concentration of the deep water further and increases the alkalinity. Of the increase in CO2 experienced by deep ocean waters, some 70% is attributable to decomposition of organic matter and 30% to the dis- solution of calcium carbonate. Baes (l982) has carefully reviewed this topic. Takahashi et al. (l98l) have compiled the result of the GEOSECS expedition to illustrate the progressive change in these properties. In Figure 3.l is shown the depth distribution of CO2 in seawater for the various ocean basins from the North Atlantic to the North Pacific. In Figure 3.2, the equivalent change in alkalinity is shown. With this background in mind we now examine some details of the chemistry of seawater.

l92

l93 CT LU o to *• OJ O O <fr CM ^ o V^ U") -J ^ < w O O ro CM o tf> V ^ OJO c\J ro u •H M (0 c 41 8 n o T k l 00 > H •H • H u 10 4I 03 41 (0 »H 41 X! W (0 O (0 • >-4 td 41 3 O. W 2 O 4J •H ig •y N 9 -H A H •p n o c • Is- « Hld3Q «J 8 I

l94 3.2.7 The Chemistry of CO2 in Seawater The reaction of gaseous CO2 with water produces hydrated CO2 and carbonic acid as in co2g + H2o *- H2co3. (2) The carbonic acid may dissociate by losing hydrogen ions as in H2CO3 *- H+ + HCO5 (3) and HCO^ *- H+ + COf. (4) with the relative proportions of these species at any time being set by the pH of the system. This representation is somewhat crude, and a great many minor species also contribute to the acid-base balance of seawater, in particular the boric acid equilibrium B(OH)3 + H2O +- B(OH)4 + H+. (5) The addition of CO2 to seawater changes its chemistry in accordance with established laws as ocean and atmosphere strive to attain equilibrium. Any large body of water will tend toward equilibrium with atmo- spheric CO2; the unique feature of the oceans, in addition to their enormous size, lies in their alkalinity. The alkalinity of seawater arises from the dissolution of basic minerals in seawater, principally calcium carbonate. Alkalinity is operationally defined as the amount of acid required to titrate l kg of seawater to a constant pH value corresponding to conversion of bicarbonate and carbonate ions to car- bonic acid. In practice very high precision in measurement is required for useful work and a precise equivalence of reactions (2) through (5) above is sought, not simply a pH value, so that the acid present exactly balances the bases as in [H+] = [HCO^] + 2[CO3] + [B(OH)4] + [OH-]. (6) The alkalinity of the present-day oceans is reasonably well known. It has been measured as a basic component of the GEOSECS (Takahashi et al., l980a) and TTO expeditions (PCODF, l98l). The alkalinity of ocean surface waters is quite well correlated with salinity; at a salinity of 35°/oo it is approximately 2300 equivalents/kg. The review by Skirrow (l975) provides a comprehensive and scholarly account of ocean CO2 chemistry, and the paper by Bradshaw et al. (l98l) illuminates the complexity of ocean CO2 system measurement. The principal effect of adding CO2 to ocean surface water is to consume carbonate ion: + H2O »2HCO3. (7)

l95 The reaction does not proceed entirely to the right, and considerable resistance to change occurs. This resistance is accurately reflected in the thermodynamics of the CO2 system. The buffer factor, or Revelle factor as it is widely known, appropriate for this reaction may be presented by (dpCO2/pCO2) TA,T,S R, (8) (dTCO2/TCO2) TA,T,S where TCO2 is the total concentration of carbon dioxide in all its forms, pCO2 is the partial pressure of carbon dioxide gas, TA is the total alkalinity, T is the temperature, and S is the salinity. Sundquist et al. (l979) have pointed out that this property is quite well known. It varies with temperature and has a numerical value of about l0. In essence a l0% change in pCO2 produces only a l% change in CO2. Takahashi et al. (l980b) have described the change in this factor that will inexorably occur as ocean CO2 levels rise. Figure 3.3 shows the change as function of CO2 for alkalinities of 2.2 and 2.4 milliequivalents/kg, taken from their paper. As the CO2 content of the atmosphere and therefore the surface ocean increases, we move to the right on this figure encountering sharply rising values of R. The resistance to change increases, the ocean absorbs proportionately less CO2, and the airborne fraction rises. This is a complex system, sensitive to the alkalinity/total CO2 ratio, and hence pH. The sharp maximum that occurs in Figure 3.l is readily understandable in terms of carbonate chemistry equilibria (Takahashi et al., l980b) and occurs when the concentration of CO3 becomes equal to that of H2CO3; thereafter a decrease in R will take place with R asymptotically approaching l. How accurately is the curve in Figure 3.3 defined, what processes are likely to alter it, and how will we know if the ocean does indeed proceed along the thermodynamic course that we have charted? The curve is a theoretical construct, based on sound principles of solution chemistry. We would like to have a series of field observations of the varying concentration of CO2 in the ocean with time so as to follow these changes; however, there are no adequate measurements for this purpose. In practice, the buffer factor is not a measured variable but a calculated property. The accuracy of these calculations depends on our knowledge of the solubility of CO2 gas in seawater and on the thermodynamic constants describing the dissociation of carbonic and boric acids in seawater. Although these have long been investigated, it is only relatively recently that results of sufficient accuracy have been obtained. The solubility of CO2 gas has been determined by Murray and Riley (l97l) and by Weiss (l974). The results of these experiments are in excellent agreement and have been fitted by Weiss (l974) to the equation ln a1 = -60.2409 + 93.45l7 (l00/T) + 23.3585 ln (T/l00) + [0.0235l7 - 0.023656 (T/l00) + 0.0047036 (T/l00)2]S, (9)

l96 E O " < LU <r 844tiatm 18 17 16 16 14 13 12 11 10 9 8 S = 35 °/oo SB - 0.41 mM/kg 1.8 2.0 2C0210 2.2 -3 2.4 M/kg 0°C FIGURE 3.3 Variation of the buffer factor, or Revelle factor, (R) of seawater with changing total CO2. The calculation is for seawater of 35% salinity and a total boron content of 0.4l mM/kg. Curves for waters of two different alkalinities are shown. Increasing CO2 levels raise the buffer factor and diminish the oceans tendency to absorb CO2. (From Takahashi et al., l980b.) where a' is the solubility in mol/kg of seawater/atm, T is the absolute temperature, and S is the salinity. This solubility equation has been used in virtually all recent models of ocean CO2 uptake. The solubility of CO2 is much greater than that of O2 or N2; the relative proportions of N2:O2:CO2 in the atmosphere are about 2400:630:l, whereas in seawater the corresponding ratios are 28:l9:l depending on the salinity and temperature (Skirrow, l975). There is little uncertainty in our knowledge of CO2 solubility. The dissociation constants (K1 and K2) of carbonic and boric acids in seawater have had a rich investigative history, and a complex literature attests to this. The dissociation constants are formally defined as

l97 H+[HCO;j] (l0) [H2CO3] and K2 - - » (ll) [HCOj] where the square brackets denote the concentration of the species in seawater. These are apparent, not true, thermodynamic constants combining the activity of the hydrogen in with the concentrations of the CO2 species. Much of the difficulty has surrounded the definition of the pH and ionic medium scales used by the various experimentalists who have determined these constants. Also, the lack of any convention for fitting the data obtained to mathematical functions has resulted in an arcane set of equations. Millero (l979) has reviewed this situation. He finds that the data of Lyman (l956), Hansson (l973), and Mehrbach et al. (l973) all yield very similar results when the apparent constants (K^) for the equilibria at various salinities (S) are fitted to equations of the form ln Ki = ln Kiw + AiS + BiS, (l2) where Ki is the constant for pure water, and A^ and B^ are temperature- dependent adjustable parameters. The discrepancies in calculated values of [HCO^] and [CO3] for water of fixed alkalinity and total CO2 are about +10 y mol/kg among the various constants. These differences have almost no effect on our ability to model ocean CO2 uptake but are a considerable irritant to researchers attempting to make and verify accurate CO2 measurements under often trying field conditions. In one area there is a degree of uncertainty: the dissolved organic matter in natural seawater is not represented in any way in these for- mulations. Typical dissolved organic matter concentrations in ocean water are l mg of C/kg (83 y mol/kg) . The acid-base characteristics of this material, and thus its contribution to the alkalinity, are poorly known. Huizenga and Kester (l979) report about ll ymol of sites per mg of C on this material with a dissociation constant of about l03'5. The effect then is not likely to be large. The preceding paragraphs show that for a seawater of constant alkalinity and chemical composition we can calculate quite well the effects of adding CO2. However, changing the alkalinity of seawater will have a marked effect. Adding CO2 gas to seawater [Equation (7)] does not change the alkalinity since charge balance is not altered; the dissolution or precipitation of CaCO3 [Equation (6) ] does. The principal forms of CaCO3 in the ocean are the polymorphs calcite and aragonite, and these are secreted by calcareous organisms to form their shells. Surface seawater is greatly supersaturated with respect to both calcite and aragonite, spontaneous precipitation being ki- netically inhibited. The various chapters in the volume edited by

l98 Andersen and Malahoff (l977) testify to the complexity surrounding CaCO3 formation and dissolution in the oceans. The solubility of CaCO3 in seawater increases with increasing pressure, decreasing temperature, and decreasing pH; thus, the deep oceans are undersat- urated with respect to CaCO3, and dissolution occurs. As we add CO2 to the surface ocean we decrease the pH and increase the tendency for CaCO3 dissolution. If this dissolution occurs, then both the alkalinity and the total CO2 increase; although this process generates an increase in total CO2, the net effect of the alkalinity increase would be to enhance the ocean's capacity for CO2 uptake by maintaining constant the factor R (Figure 3.3) and providing CO| ions. Model calculations tend to assume a constant alkalinity scenario, and there is no evidence that the alkalinity of the ocean has increased in recent times. A skeptic could however point out that there is precious little evidence that it has not. Ambiguities in definition of alkalinity [e.g., the inclusion of minor species such as HPOj- and SiO(OH)3], imprecision in measurement, and lack of a historical time series leave us with a poor temporal record of the alkalinity and total CO2 content of the ocean. Several things make this problem complex. First, the concept of the solubility of pure CaCO3 in seawater is moot; Morse et al. (l980) show that the surface undergoing dissolution rapidly becomes transformed into a Mg-Ca-CO3 interfacial layer with complex kinetic and solubility controls. Second, biogenically produced magnesian calcites (such as in some algae and the spines of sea urchins), containing l5 mol % magnesium or more, commonly occur in ocean surface waters. The stability of this material is poorly understood, and its dissolution would change alka- linity. Garrels and Mackenzie (l98l) recently reviewed the susceptibil- ity of magnesian calcite phases to CO2-induced dissolution. Their conclusion was that insufficient magnesian calcites existed globally to have major impact, if dissolved, on ocean CO2 uptake. Dissolution of this material, however, would certainly be noticed on a local scale. Finally, since the surface ocean is so strongly supersaturated with respect to calcite and aragonite and is likely to remain so, it is widely assumed that no dissolution of these minerals takes place there. Aller (l982) has pointed out that this is not so. Calcareous shells in nearshore muds are exposed to interstitial waters rich in respiratory CO2 and low in pH. The shells are dissolved quite rapidly, resulting in a diffusive flux of Ca2+ and COf ions to the overlying waters. As we change the CO2 content of surface waters, we change the upper boundary condition controlling this flux. The effects of this process are still being explored. It is quite possible to incorporate the effects of a postulated increase in ocean alkalinity into an atmosphere-ocean CO2 model, such as has been done by Bacastow and Keeling (l979), although the accuracy of the conclusions is uncertain. In this calculation, the dissolution of deep calcium carbonate was found to have little immediate effect on rising atmospheric CO2 levels, since the affected seawater would be sequestered in the deep ocean. As this water is brought into contact with the atmosphere, it will slowly draw down the atmospheric CO2

l99 levels some hundreds of years in the future. If we were to dissolve an average depth of 3 cm of pure calcium carbonate from the ocean floor, then in l500 years the atmospheric C02 level would be some 30% lower than with no change in alkalinity. If shallow-water calcium carbonate dissolves, the effects are more dramatic. If we were to dissolve an average depth of 40 cm of pure calcium carbonate from shallow-water sediments, then the peak atmospheric C02 level would only be some 60% of that with no dissolution occurring, and the drawdown in the future would be more rapid. This, as emphasized by Bacastow and Keeling (l979), is of course unrealistic. There are kinetic limits and controls on carbonate dissolution extrinsic to the model considered here (e.g., Sayles, l98l; Emerson and Bender, l98l), and the disappearance of such massive amounts of carbonate shells and corals and sediments from our shores and shallow seas would present a crisis for man arousing far greater concern than any incremental effect on C02 levels. In the future it appears inevitable that fossil fuel C02-indueed dissolution of calcium carbonate will take place in the ocean. The most sensitive site appears to be in the deep North Atlantic Ocean where waters enriched in industrial CO2 begin their deep ocean tour (Broecker and Takahashi, l977) and conceptually, since these waters have demonstrably "seen" fossil fuel CO2, some dissolution has likely already occurred. However, there is no time series of measurements in the oceans adequate to confirm or deny these statements, and some thought must be given to this if progress is to occur. 3.2.8 Measurements of Ocean C02 Ocean surface water today contains about 2000 ymol/kg of C02. The amount varies with temperature, location, and season. We do not know the concentration in the past. The atmospheric C02 content in the last century appears to have been about 270 ppm, although uncertainty exists. Assuming the maintenance of overall equilibrium, we can calculate that modern-day surface ocean waters must contain about 35 pmol/kg more CO2 than in the past, an approximate l.8% increase. Although this appears to be so, there is no time series of ocean C02 measurements adequate to support this claim. The ocean surface pH is similarly variable in the range 8.0-8.3; we can calculate that ocean surface pH has been reduced by about 0.06 pH units. The carbonate ion content of surface seawater depends on the complex equilibria estab- lished between the species H2CO3, HCO^ and CO|; it is typically l5% or so of the total CO2 concentration. We calculate that ocean surface water today contains about l0% less carbonate ion than in the past. How well can we measure these changes? The most problematic and least satisfactory measurement is pH; a measurement precision of +0.0l pH unit is possible, but the long-term change would be hard to detect over seasonal fluctuations. Alkalinity can be measured with a precision and accuracy of about 3 vequivalents/kg. The total C02 content can be determined to +4 ymol/kg

200 by potentiometric titration (Bradshaw et al, l98l); recently Keeling (l983) has presented ocean CO2 data based on gas extraction and manometric measurement accurate to +0.5 ymol/kg. The current rate of increase of the total CO2 concentration in ocean surface waters is calculated to be approximately l pmol/kg/yr. Plainly such an increase would be observable from time series measurements made today within the space of a few years. The most sensitive measurement is pCO2, which may be determined to within a few tenths of a part per million. By pCO2 we mean the pressure of CO2 gas that would be found in a small volume of air that had been allowed to reach gaseous equilibrium with a large volume of seawater. Surface seawater strives to maintain a pCO2 globally in equilibrium with the atmosphere, lagging behind by some small value due to the finite time required for gas exchange to take place. There is a large natural variability in ocean surface pCO2 (Keeling l968; Miyake et al., l974; Takahashi, l979) as shown in Figure 3.4. The data shown in Figure 3.4 are in fact the deviations of ocean surface pCO2 from equilibrium with the atmosphere. It is not a synoptic data set but an ensemble of results from many expeditions, quasi-normalized in time and containing considerable seasonal noise. The essential features are strongly negative values at high latitudes, where rapid cooling and biological activity have markedly lowered pCO2, and high values at the equator, where upwelling of CO2-rich water and warming have raised pCO2. Gaseous exchange of CO2 is sufficiently slow that equilibrium with the atmosphere is not achieved locally, and these patterns persist. Negative values imply invasion of CO2 from the atmosphere to the ocean; positive values indicate evasion of CO2 from the ocean to the air. This distribution results in a net flux of CO2 between the equa- torial and polar oceans (Bolin and Keeling, l963). Peatman et al. (l983) have recently calculated that the net release of CO2 by the equatorial oceans is currently about l.3 Gt of C/yr; the net uptake by the high-latitude oceans is about 4.4 Gt of C/yr. The presence of fossil fuel CO2 has enhanced the polar uptake and suppressed the equatorial source. There is now evidence for a systematic change in ocean surface pCO2 with time. Takahashi et al. (l983) have compiled their measurements of pCO2 in Sargasso Sea surface waters from the IGY (l957), GEOSECS (l972), and TTO (l98l) expeditions. The results are shown in Figure 3.5, together with the atmospheric trend. It is clear that ocean pCO2 is rising; however, the interpretation of this signal requires caution. The problem is not with the precision of measurement but with compensating for the natural noise due to fluctuating ocean surface conditions. The atmospheric and oceanic lines are not parallel in Figure 3.5, and some explanation of this is needed. The Sargasso Sea exhibits low surface pCO2 with respect to the atmosphere (Figure 3.4). This arises from the marked negative heat flux observed there (Bunker and Worthington, l976), so that surface seawater is cooled during its residence time in the Sargasso Sea. The slope of the oceanic line in Figure 3.5 thus represents not only the anthropogenic

20l d id E 11) -H M iM 3 k l H tM 41 id 3 O *J Q1 0 * -HtM Q4 4J o o1 id -o cn c 41 O id z -H n 09 (0 OJ • > ,o S* 01 O 41 • k i 4J a 3 id oi o o n n -H 41 C -H k i 41 4J n> 41 3 0) 10 41 -p > a1 id -H o •o *J CO 41 iH .. c A « -H g ai 41 CO to to id c oi Cn oi .c M C Q id O CO 41 O n £ g 41 EH 5 — W 10 . Qi oo • 41 VD « Oi aj a .a o . &Q o1 oi O « o -H g O m iD c Cti J3 O -•' M Ol-H « W CO D O id • o e > k l M 5 fi -H n id -H id

202 350 >• a: (/> z z — - ! 300 E & i O o u 04 o u 250 (^ I ' • ATMOSPHERIC C02 TTO/NA IGY m en O <D cn in U> 0> B YEAR ID r^ cn o 00 0> FIGURE 3.5 The change in the mean pCO2 of surface Sargasso Sea water from the IGY expedition (l957) to the TTO-North Atlantic expedition (l98l). (From Takahashi et al., l983.) CO2 trend but climatic variations as well, suggesting perhaps a modest warming during this period. The increase in ocean surface pCO2 is readily measurable from time-series measurements made today. There appears to be little reason to maintain in the future the lack of knowledge of trends in ocean surface CO2 properties that has characterized the past. 3.2.9 Models of Ocean CO? Uptake Mathematical models describing the transfer of CO2 among atmosphere, biosphere, and ocean are an essential tool to scientists working in this field. The features of ocean CO2 chemistry reviewed above have long been recognized and are adequately represented in recent models (Oeschger et al., l975; Bacastow and Keeling, l979). Future refinement is probable; however, the essential chemical concepts are in place.

203 How do we model ocean C02 uptake, what data are available for model testing, and how do we know if the models are correct? The physical basis for the modeling of ocean C02 uptake is reasonably well established. The first criterion is that the gas exchange rate be known. Broecker and Peng (l974) have examined the problem of determining gas exchange rates, and Broecker et al. (l979) have evaluated the effects of uncertainty in this property on cal- culations of ocean CQ2 uptake. Most field data have been obtained by the radon technique by which the deficiency of naturally produced radon-222, due to gas exchange at the sea surface, is measured. A simple stagnant film model is used to calculate the gas exchange parameter and to relate the observations to C02 exchange. The two-way C02 exchange rate obtained in this way is l6 mol/m2/yr. A much larger-scale estimate is available from the natural balance of 14C by radioactive decay in the interior of the ocean must be balanced by a fresh influx of CO2. The result from this calculation yields a rate of l9 + 6 mol/m2/yr. C02 exchange in a wind-wave tunnel has been examined by Broecker et al. (l979), yielding results in substantial agreement with the above. The calculations of gas exchange rates appear to rest on sound principles. The characteristic exchange time for CO2 is about l0 times longer than for the nonreactive gases (N2 and 02, for example) and is about l year (Broecker and Peng, l974). The time scale for 14C02 exchange is about l0 years, owing to the time required for complete isotopic equilibration with the carbon pool. Gas exchange for C02, though quite slow, does not appear to be the rate-limiting step in models of ocean CO2 uptake. The rate of vertical mixing in the sea is widely viewed as the critical parameter. The annual cycle over most of the ocean is such that in the summer the sea becomes capped by a warm surface layer, which undergoes continuous gaseous exchange with the atmosphere. Wintertime cooling and late winter storms increase the density of this upper layer so that it finally becomes unstable and undergoes turbulent exchange with waters of equivalent density below. This scheme of wintertime surface turnover, followed by lateral penetration along surfaces of constant density, is what is parameterized in various ways in ocean models. Most water mass formation takes place in a location and at a time when we cannot observe it. Some integrated measure of the effect is required, and this has been provided by the tracer approach. The principal tracers used have been 14C and 3H; the transcendental virtue of l4C is that it identically mimics CO2, taking part in both the gaseous and biological exchange cycles. Revelle and Suess (l957) and Craig (l957) used C box models to evaluate carbon dioxide exchange between atmosphere and ocean and within the ocean. As these models grew in complexity (Keeling and Bolin, l967, l968), the neces- sity for multiple tracers became apparent, as did the difficulty of assigning realistic transfer rates between boxes based on objective experimental evidence. The injection into the atmosphere of massive amounts of radionuclides in the nuclear bomb tests of l958-l962 prior to the signing of the nuclear test-ban treaty provided such a tracer suite, and a decade later ocean scientists organized the GEOSECS

204 WESTERN NORTH ATLANTIC 0' 60* GEOSECS TRITIUM CTU81N) 60* 70* 80'N IOOO 2000 0. UJ o 3000 4000 5000 6000 FIGURE 3.6 Penetration of tritium into the North Atlantic Ocean in l972 along the GEOSECS cruise track. (From Ostlund et al., l974.) experiment to observe the oceanic distribution of these species. The tritium and radiocarbon results from this program (Ostlund et al., l974) provided a snapshot of ocean tracer penetration on a l0-year time scale; Figure 3.6 shows tritium penetration into the Atlantic Ocean in l972. The results shown in Figure 3.6 represent a vertical slice through the western basin of the North Atlantic. They plainly reveal the formation of new deep waters; however, such a representation understates the great horizontal extent of the oceans. C02 is taken up by surface seawater globally, and between 50-70% of the ocean's anthropogenic C02 burden is stored in the surface and thermocline waters of the great oceanic gyres (Stuiver, l978; Siegenthaler, l983). Although the penetration of tritium (Figure 3.6) provides a vivid pictorial representation of the extent of the ocean labeled by an invading tracer in the l0-year period l962-l972, the conversion of this signal to that of C02 is complex. First, tritium was injected as a pulse early in the decade, and largely in the northern hemisphere. Second, the mean age of the fossil fuel C02 increase is about 28

205 WESTERN NORTH ATLANTlC O° 1O° 2.O° 30 4O° 5.O° TTO TRlTIUM (TU81N) 6.O° 7O° 8O* N 1OOO 2OOO i X £3OOO 1II 4OOO- 5OOO- 6OOO FIGURE 3.7 The re-occupation of the GEOSECS stations in l98l on the TTO Expedition. Tritium was principally injected into the atmosphere of the northern hemisphere by nuclear bomb tests in l962. The figure illustrates the labeling of the ocean by the invasion of a passive tracer in a l9-year period. The mean age of the fossil fuel CO2 signal is about 28 years. (From Ostlund, l983.) years (Broecker et al., l979). What fraction of the ocean will be labeled on the 28-year time scale appropriate for CO2? Fortunately, the experiment shown in Figure 3.6 has been repeated. In Figure 3.7, results are shown for a reoccupation of this section in l98l, l9 years after the principal tritium pulse (Ostlund, l983). The progression of the deep-water front is plainly to be seen. These data provide a critical test for models of ocean CO2 uptake, for unless the models can match these observations, they are unlikely to be correct. It should be noted, however, that the converse is not necessarily true. A model that simply matches data at one point in time and contains unrealistic physical principles or dynamic characteristics cannot be said to be "true" or "correct" in any satisfactory sense.

206 Oeschger et al. (l975) gave major impetus to this field with their successful implementation of a box diffusion model. The model con- sisted of an atmosphere, a biosphere, an ocean mixed layer, and a diffusive deep ocean. Their detailed evaluation of the transfer equations and careful selection of numerical properties (mixed layer depth » 75 m, eddy-diffusion coefficient » l.3 cm2/sec) yielded a good match to the atmospheric signal. The model successfully coped with the different dynamic responses required for penetration of bomb 14C and excess fossil fuel CO2, and they concluded that the model would be valid for predictions of the atmospheric C02 response to the various possible future C02 input time functions. Killough and Emanuel (l98l) recently compared several models of ocean CO2 uptake, the models differing principally in the size and number of reservoirs and their sequence of interconnection so that the dynamic response characteristics varied. Each of the models was calibrated from observations of natural l4C activity and tuned to match the observed response to the penetration of bomb C. Their evaluation gave results in substantial agreement with those of Oeschger et al. (l975), Stuiver (l978), and Broecker et al. (l979). At this point it is clear that virtually all successful models of ocean C02 uptake have relied on the tracer approach, particularly l4C and tritium. The models are tuned to simulate the natural radiocarbon distribution and their dynamic response tested against the bomb transient. The tracer, rather than direct, approach has been necessary for several reasons. First, there is no time series of ocean CO2 measurements of high accuracy that would match the Mauna Loa and other atmospheric records. Early oceanic CO2 measurements quite simply lack accuracy and precision and rest on an unsatisfactory thermodynamic basis. Second, models depending purely on physical principles and measurements (T,S) are frequently undetermined. Without the constraints supplied by independent tracers of differing source functions, satisfac- tory solutions are not easily achieved. Finally, measurements of ocean CO2 made today are inadequate on their own to solve the fossil fuel CO2 problem. The models described above, which currently serve to calculate ocean CO2 uptake, while highly ingenious, are nonetheless viewed with considerable skepticism by physical oceanographers. The concern is not that the overall estimate of ocean CO2 uptake is substantially in error but that the parameterization of all of ocean physics into a single vertical eddy-diffusion coefficient is unacceptable. Garrett (l979) has carefully reviewed the evidence for vertical diffusion in the ocean and finds no physical basis for a vertical diffusion coef- ficient of^l cm2 sec"l as required by Oeschger et al. (l975), and indeed by all one-dimensional vertical models. A value of one tenth that number appears realistic. Jenkins (l980) has followed the time history of the penetration of tritium, and its stable, gaseous daughter product helium-3, into the Sargasso Sea. He shows unequivo- cally that a one-dimensional model with high vertical diffusivity cannot explain the results and that a scheme of wintertime surface- water turnover followed by lateral penetration along density surfaces is required.

207 There are of course many models of ocean circulation that give a more realistic portrayal of ocean physics (Holland, l97l, l978), and similar models are now being applied to study the time-dependent penetration of tracers into the interior of the ocean (Sarmiento, l982). The application of these models to the O>2 problem will be complex but seems to be possible. Rapid progress is to be expected in this field. However, the principal criticism of ocean CO2 uptake models has come not from their representation of ocean physics but from their failure to include some components of the natural, or perturbed, CO2 cycle. The basic assumption for models such as those of Oeschger et al. (l975) or Killough and Emanuel (l98l) is that the natural cycle of CO2 within the ocean has been unchanged by the activities of man—i.e., primary production remains at past levels, and the vertical flux of particulate carbon has not been altered. The initial ocean co2 profile thus appears in the models as a constant value. The true ocean CO2 cycle is enormously complex, so that the simplification introduced by this procedure is most attractive. The recent report by Bolin et al. (l982), who have incorporated realistic profiles of oceanic phosphate, oxygen, a>2, alkalinity, and l4C in a highly developed l2-box ocean model, illustrates this point well. The carbon dioxide concentration in the North Atlantic has been measured in parallel with the tritium section in Figure 3.7 (Brewer, l983). The results (Figure 3.8) show the large natural variability of the oceanic CO2 system. Hidden within these data lies the increase in CO2 caused by the activities of man. Kempe (l982) has reviewed the long-term trends in river fluxes to the ocean of carbon and nutrients in great detail and has shown that fertilization, land use, and industrial activities have altered many of these fluxes markedly. Walsh (l98l) has suggested that overfishing off Peru and other perturbations (Walsh et al., l98l) have changed the storage of carbon on continental shelves. It is clear that the facets of carbon research are so many and varied that to attempt to satisfy all claims for omission would result in models of endless and labyrinthine complexity. We would like to know if these effects are offset by competing processes and if the net effect is of sufficient size to alter our conclusions regarding the fossil fuel signal. Keeling (l983) has recently examined this problem. He has compared measurements of atmospheric CO2 from a time series of samples obtained on board ships from north-south sections in the Pacific Ocean. Recent seasonal coverage has greatly increased the utility of these data. By assuming that the seasonal oscillations observed today hold true for the recent past, we can greatly improve the interpretation of previously obtained shipboard data. Figure 3.9 shows plots of such data grouped from the periods near l962, l968, and l980 from Keeling (l983). The data are normalized relative to a constant value at the South Pole, shown as zero on this figure. The peak, centered at the equator, is attributed to degassing of CO2 from the high pCO2 zone in the equatorial Pacific Ocean (Keeling, l968). This is a natural phenomenon. In the northern hemisphere a steady increase has occurred that is consistent with rising fossil fuel usage, over 90% of which

208 WESTERN TTO MAM 1000 5000 6000 EON 40N 50N 60N 70N FIGURE 3.8 The distribution of total carbon dioxide in seawater along the section shown in Figure 3.7. The large natural variability seen here must have been perturbed by the invasion of fossil fuel CO2, although we have no time series of measurements to document this change. (From Brewer, l983.) takes place in the northern hemisphere. By assuming that the change in atmospheric CO2 is due solely to fossil fuel usage, with a constant oceanic uptake over this period, then model profiles can be developed that may be compared with the observed data. Figure 3.l0, again from Keeling (l983), shows the result of subtraction of the model profiles from the observations. The residual signal reflects the expected constant equatorial degassing but does not indicate any significant perturbation of atmospheric CO2 from other than the fossil fuel source. These data, obtained over approximately a 20-year period, appear to be the best available. They support, but do not prove, the idea of a constant, or very slowly changing, ocean and terrestrial biosphere. The atmosphere integrates the effects of changes in these systems, so that if a large change is postulated in one of them, then there appears to have been a most fortunate compensation in the other. Pearman et al. (l983) have also examined this problem. They calculate that the dominantly northern hemisphere input of C&2 to the atmosphere has changed the interhemispheric difference in atmospheric

209 3.0 a -S 2.0 8 1.0 1 TT 1980 1968 1962 1 1 1 1 I I I i I M] 90° S 40° S 20° S 0 20 N LATlTUDE (deg) 40° N 90° N FIGURE 3.9 Atmospheric CO2 levels along a Pacific Ocean transect, normalized to zero at the South Pole. (From Keeling, l983.) concentration (north-south) from -l ppm in the last century to +4-5 ppm today. Their atmosphere-ocean model simulation points to a net upper limit of 2 Gt of C per year from a terrestrial carbon source. The conclusion of Oeschger et al. (l975) was that the proportioning of fossil fuel CO2 between atmosphere and ocean was 60 to 40%. Stuiver (l978) concluded that some 47% of fossil fuel CO2 was stored in the ocean. Broecker et al. (l979) calculated that 37 + 4% of the fossil fuel CO2 generated between l958 and the present has been taken up by the sea. Bolin et al. (l982) have recently concluded that the oceanic uptake falls in the range 30-38%. Since these estimates all fall in a narrow range, depend on accepted physical principles, and are verifiable by several different means, there does not appear to be the chance of significant error. 3.2.l0 Future Studies and Problems Although present-day models appear satisfactorily to answer the question of how much fossil fuel CO2 the sea takes up, there are many things that we do not know.

2l0 8 , -2 Mil I 90 60 50 40 30 20 10 0 10 20 30 40 50 60 90 °N LATITUDE FIGURE 3.l0 The results shown in Figure 3.9 corrected for seasonal and fossil fuel C02 effects alone. No perturbation of the signal from other than these effects is apparent. (From Keeling, l983.) We do not know the atmospheric or oceanic CO2 levels in the last century. Recent ice core data (Neftel et al., l982) promise to place new constraints on this. Calculations based on oceanic measurements (Brewer l978, l983) yield a preindustrial value of about 265 ppm of C02. However, this is likely a lower limit since oceanic deep waters are formed in areas of marked negative disequilibrium with the atmosphere. As our ocean data base grows, the one-dimensional models will become increasingly inadequate, and incorporation of the C02 and tracer data into new models will be required. The warming accompanying the atmospheric CO2 rise will affect the ocean. Storage of heat in the upper layers will mitigate, but not prevent, climate change (Bryan et al., l982). Models of this heat storage currently treat it as passive uptake, not affecting water mass formation and vertical circulation. If such changes, however, did occur, it would affect ocean CO2 uptake in unknown ways. The warming of the ocean will reduce CO2 solubility and expel further C02 to the atmosphere. This effect can be calculated reasonably well (Bacastow and Keeling, l979) and is incorporated into current models (Killough and Bmanuel, l98l) of ocean C02 uptake.

2ll Whether we will detect this effect and separate it in our budgeting of CO2 between atmosphere, ocean, and biosphere, or confuse it with some other signal, is problematic. 3.2.ll Summary On average, each year the ocean currently takes up an amount of CO2 approximately equal to 40% of the fossil fuel CO2 added to the atmosphere by man. Calculations based on sound principles of solution chemistry and an estimated atmospheric CO2 content of about 270 ppm in the last century show that ocean surface pH must have decreased by about 0.06 pH unit during this time; the total CO2 concentration of surface seawater must have increased by about 35 yinol/kg (about l.8%). The proportion of carbonate ion must have decreased by about l0%, increasing the tendency for calcium carbonate dissolution. The alkalinity is believed not to have changed. We do not know this on the basis of a direct time series of measurements, and until quite recently, oceanic CO2 system measurements contained substantial inaccuracies. Models of ocean CO2 uptake have depended greatly on tracer data, particularly natural and bomb-produced 14C. These models represent some features of ocean chemistry quite well, but they represent ocean physics by a simple vertical diffusion coefficient. The models treat only the CO2 perturbation and do not yet attempt to mimic the natural and complex CO2-oxygen-nutrient cycles within the ocean. Rapid progress in incorporating these features into more advanced ocean circulation models is anticipated. It is quite possible to measure the changing CO2 properties of the ocean with time using modern techniques, although no ongoing program yet exists to do so. There are some significant uncertainties awaiting. We do not know when and where calcium carbonate dissolution will occur. We do not know how future warming will affect ocean circulation and whether we will detect this warming in the CO2 signal. We do not know adequately the time history of CO2 releases from the biosphere so as firmly to close the atmosphere-ocean-biosphere triangle represented in our models. Finally, we should not be too complacent. Nature has vast resources with which to fool us; the last glaciation was apparently accompanied by massive CO2 transfers to and from the ocean, the causes, conse- quences, and explanations of which are poorly understood today (Broecker, l982). References Aller, R. C. (l982). Carbonate dissolution in nearshore terrigenous muds: the role of physical and biological reworking. J. Geol. 20:79-95. Andersen, N. R., and A. Malahoff, eds. (l977). The Fate of Fossil Fuel COi in the Oceans. Plenum, New York, 749 pp.

2l2 Bacastow, R. B., and C. D. Keeling (l979). Models to predict future atmospheric CO2 concentrations. In Workshop on the Global Effects of Carbon Dioxide from Fossil Fuels. U.S. Dept. of Energy Report CONF-770385, pp. 72-90. Baes, C. F. (l982). Ocean Chemistry and Biology. In Carbon Dioxide Review: l982, W. C. Clark, ed. Oxford U. Press, New York, pp. l87-2ll. Bolin, B. (l98l). Carbon Cycle Modelling. SCOPE Report l6. Wiley, New York. Bolin, B., and C. D. Keeling (l963). Large-scale atmospheric mixing as deduced from the seasonal and meridional variations of carbon dioxide. J. Geophys. Res. 68:3899-3920. Bolin, B., E. T. Degens, S. Kempe, and P. Ketner (l979). The Global Carbon Cycle. SCOPE Report l3. Wiley, New York, pp. l-49l. Bolin, B., A. Bjorkstrom, K. Holmen, and B. Moore (l982). The simultaneous use of tracers for ocean circulation studies. Report CM-58, Dept. of Meteorology, U. of Stockholm, p. 70. Bradshaw, A. L., P. G. Brewer, D. K. Shafer, and R. T. Williams (l98l). Measurements of total carbon dioxide and alkalinity by potentiometric titration in the GEOSECS program. Earth Planet. Sci. Lett. 55:99-ll5. Brewer, P. G. (l978). Direct observation of the oceanic CO2 increase. Geophys. Res. Lett. 5:997-l000. Brewer, P. G. (l983). The T.T.O. North Atlantic Study—A Progress Report. In Proceedings: Carbon Dioxide Research Conference: Carbon Dioxide, Science and Consensus. U.S. Dept. of Energy Report CONF-820970, Part II, pp. 9l-l22. Brewer, P. G., and J. C. Goldman (l976). Alkalinity changes generated by phytoplankton growth. Limnol. Oceanog. 2l:l08-ll7. Broecker, H. C., J. Peterman, and W. Siems (l978). The influence of wind on CO2-exchange in a wind-wave tunnel, including the effects of monolayers. J. Mar. Res. 36:595-6l0. Broecker, W. S. (l982). Ocean chemistry during glacial time. Geochim. Cosmochim. Acta 46:l689-l705. Broecker, W. S., and T. H. Peng (l974). Gas exchange rates between air and sea. Tellus 26:2l-35. Broecker, W. S., and T. Takahashi (l977). Neutralization of fossil fuel CO2 by marine calcium carbonate. In The Fate of Fossil Fuel CO2 in the Oceans, N. E. Andersen and A. Malahoff, eds. Plenum, New York, pp. 2l3-24l. Broecker, W. S., T Takahashi, H. J. Simpson, and T. H. Peng (l979). Fate of fossil fuel carbon dioxide and the global carbon budget. Science 206:409-4l8. Bryan, K., F. G. Komro, S. Manabe, and M. J. Spelman (l982). Transient climate response to increasing atmospheric carbon dioxide. Science .2l5:56-58. Bunker, A. F., and L. V. Worthington (l976). Energy exchange charts of the North Atlantic Ocean. Bull. Am. Meteorol. Soc. 57:670-678. Callendar, G. S. (l938). The artificial production of carbon dioxide and its influence on temperature. Q. J. Roy. Meteorol. Soc. 64:223-240.

2l3 Craig, H. (l957). The natural distribution of radiocarbon and the exchange time of carbon dioxide between atmosphere and sea. Tellus 9.: l-l7. Emerson, S. R., and M. L. Bender, l98l. Carbon fluxes at the sediment-water interface of the deep sea: calcium carbonate preservation. J. Mar. Res. 39:l39-l62. Garrels, R. M., and F. T. Mackenzie, l98l. Some aspects of the role of the shallow ocean in global carbon dioxide uptake. U.S. Dept. of Energy Report CONF-8003ll5. Garrett, C. (l979). Mixing in the ocean interior. Dynamics Atmos. Oceans 3:239-265. GEOSECS Atlases Vols. l-6. U.S. Government Printing Office, Washington, D.C. Hansson, I. (l973). A new set of acidity constants for carbonic acid and boric acid in sea water. Deep-Sea Res. 20:46l-l78. Holland, W. R. (l97l). Ocean tracer distributions. Tellus 23:37l-392. Holland, W. R. (l978). The role of mesoscale eddies in the general circulation of the ocean—numerical experiments using a wind-driven quasi-geostrophic model. J. Phys. Oceanog. 8:363-392. Huizenga, D. L., and D. R. Kester (l979). Protonation equilibria of marine dissolved organic matter. Limnol. Oceanog. 24:l45-l50. Jenkins, W. J. (l980). Tritium and -*He in the Sargasso Sea, J. Mar. Res. 38:533-569. Keeling, C. D. (l968). Carbon dioxide in surface oceans waters, 4. Global distribution. J. Geophys. Res. 73:4543-4553. Keeling, C. D. (l983). The Global Carbon Cycle: What we know and could know from atmospheric, biospheric and oceanic observations. In Proceedings: Carbon Dioxide Research Conference: Carbon Dioxide, Science and Consensus. U.S. Dept. of Energy Report CONF-820970, Part II, pp. l-62. Keeling, C. D., and B. Bolin (l967). The simultaneous use of chemical tracers in oceanic studies, I. General theory of reservoir models. Tellus l9:566-58l. Keeling, C. D., and B. Bolin (l968). The simultaneous use of chemical tracers in oceanic studies, II. A three reservoir model of the North and South Pacific Oceans. Tellus 20:l7-54. Kempe, S. (l982). Long-term records of C02 pressure fluctuations in fresh waters. In Transport of Carbon and Minerals in Major World Rivers. Mitt. Geol.-Palaont. Inst. Univ. Hamburg, Federal Republic of Germany. SCOPE Report 52, pp. 9l-332. Killough, G. G., and W. R. Emanuel (l98l). A comparison of several models of carbon turnover in the ocean with respect to their distributions of transit time and age responses to atmospheric CO2 and l4C. Tellus 33:274-290. Lyman, J. (l956). Buffer mechanism of sea water. Ph.D. Thesis, University of California, Los Angeles, l96 pp. Mehrbach, C., C. H. Culberson, J. E. Hawley, and R. M. Pytkowicz (l973) . Measurement of the apparent dissociation constants of carbonic acid in sea water at atmospheric pressure. Limnol. Oceanog. l8:897-907.

2l4 Millero, F. J. (l979). The thermodynamics of the carbonate system in sea water. Geochim. Cosmochim. Acta 43:l65l-l66l. Miyake, Y., Y. Sugimura, and K. Saruhashi (l974). The carbon dioxide content in the surface waters of the Pacific Ocean. Rec. Oceanog. Works in Jpn. l2:45-52. Morse, J. W., A. Mucci, and F. J. Millero (l980). The solubility of aragonite and calcite in sea water of 35°/oo salinity at 25°C and atmospheric pressure. Geochim. Cosmochim. Acta 44:85-94. Murray, C. N., and J. P. Riley (l97l). The solubility of gases in distilled water and sea water, IV. Carbon dioxide. Deep-Sea Res. l8-: 533-54l. Neftel, A., H. Oeschger, J. Schwander, B. Stauffer, and R. Zumbrunn (l982). New measurements on ice core samples to determine the CO2 content of the atmosphere during the last 40,000 years. Nature 295:220-223. Oeschger, H., U. Siegenthaler, U. Schotterer, and A. Gugelmann (l975). A box diffusion model to study the carbon dioxide exchange in nature. Tellus 27:l68-l92. Ostlund, H. G. (l983). Tritium and Radiocarbon: TTO Western North Atlantic Section GEOSBCS Re-occupation. Tritium Laboratory Data Release 83-07, Rosenstiel School of Marine and Atmospheric Sciences, Miami, Fla., unpublished data. Ostlund, H. G., H. G. Dorsey, and C. G. Rooth (l974). GEOSECS North Atlantic radiocarbon and tritium results. Earth Planet. Sci. Lett. .23:69-86. PCODF (l98l). TTO Preliminary Hydrographic Data Reports, Vol. I-IV. Scripps Institution of Oceanography Reports, La Jolla, Calif. Pearman, G. I., P. Hyson, and P. J. Fraser (l983). The global distribution of atmospheric carbon dioxide: l. Aspects of observations and modelling. J. Geophys. Res. 88:358l-3590. Peterson, B. J. (l980). Aquatic primary productivity and the l4C-CO2 method: a history of the productivity problem. Ann. Rev. Ecol. Syst. ll:359-385. Reid, J. L., and R. J. Lynn (l97l). On the influence of the Norwegian- Greenland and Heddell seas upon the bottom waters of the Indian and Pacific Oceans. Deep-Sea Res. l8:l063-l088. Revelle, R., and H. E. Suess (l957). Carbon dioxide exchange between atmosphere and ocean and the question of an increase of atmospheric CO2 during past decades. Tellus 9:l8-27. Sarmiento, J. L. (l982). A simulation of bomb tritium entry into the Atlantic Ocean. J. Phys. Oceanog., in press. Sayles, F. L. (l98l). The composition and diagenesis of interstitial solutions, II. Fluxes and diagenesis at the water-sediment interface in the high latitude North and South Atlantic. Geochim. Cosmochim. Acta 45:l06l-l086. Siegenthaler, U. (l983). Uptake of excess CO2 by an outcrop-diffusion model of the ocean. J. Geophys. Res. 88:3599-3608. Skirrow, G. (l975). The dissolved gases-carbon dioxde. In Chemical Oceanography, Vol. 2, 2nd ed., J. P. Riley and G. Skirrow, eds. Academic, New York, pp. l-l92.

2l5 Stuiver, M. (l978). Atmospheric carbon dioxide and carbon reservoir changes. Science l99:253-258. Stuiver, M., P. D. Quay, and H. G. Ostlund (l983). Abyssal water carbon-l4 distribution and the age of the world oceans. Science 2l9:849-85l. Sundquist, E. T., L. N. Plummer, and T. M. L. Wigley (l979). Carbon dioxide in the ocean surface: the homogeneous buffer factor. Science 204:l203-l205. Takahashi, T. (l979). Carbon dioxide chemistry in ocean water. In Workshop on the Global Effects of Carbon Dioxide from Fossil Fuels. U.S. Dept. of Energy Report CONF-770385, pp. 63-7l. Takahashi, T., and A. G. E. Azevedo (l982). The oceans as a CO2 reservoir. In Interpretation of Climate and Photochemical Models, Ozone and Temperature Measurements, R. A. Beck and J. R. Hummel, eds. American Institute of Physics Conf. Proc. No. 82, pp. 83-l09. Takahashi, T., W. S. Broecker, A. E. Bainbridge, and R. F. Weiss (l980a). Carbonate chemistry of the Atlantic, Pacific and Indian Oceans: The results of the GEOSECS Expeditions, l972-l978. Report l, cv-l-80. Lamont-Doherty Geological Observatory. Takahashi, T., W. S. Broecker, A. E. Bainbridge, and R. P. Weiss (l980b). Carbonate chemistry of the surface waters of the world oceans. In Isotope Marine Chemistry, E. Goldberg, Y. Horibe, and K. Saruhashi, eds. Uchida Rokakuho, Tokyo, pp. l47-l82. Takahashi, T., W. S. Broecker, and A. E. Bainbridge (l98l). The alkalinity and total carbon dioxide concentration in the world oceans. In Carbon Cycle Modelling, B. Bolin, ed. SCOPE Report l6. Wiley, New York, pp. l59-l99. Takahashi, T., D. Chipman, and T. Volk (l983). Geographical, seasonal, and secular variations of the partial pressure of CO2 in surface waters of the North Atlantic Ocean: The results of the North Atlantic TTO Program. In Proceedings: Carbon Dioxide Research Conference: Carbon Dioxide, Science and Consensus. U.S. Dept. of Energy Report CONF-820970, Part II, pp. l23-l45. Walsh, J. J. (l98l). A carbon budget for over-fishing off Peru. Nature 290:300-304. Walsh, J. J., G. T. Rowe, R. L. Iverson, and C. P. McRoy (l98l). Biological export of shelf carbon is a sink of the global CO2 cycle. Nature 29l:l96-20l. Weiss, R. F. (l974). Carbon dioxide in water and sea water: the solubility of a non-ideal gas. Mar. Chem. 2:203-2l5.

2l6 3.3 BIOTIC EFFECTS ON THE CONCENTRATION OF ATMOSPHERIC CARBON DIOXIDE: A REVIEW AND PROJECTION George M. Wbodwell 3.3.l Introduction The composition of the atmosphere is changing. It has changed greatly, of course, throughout the period of the Earth's evolution. Few doubt the dominant role of the biota in the evolution of the atmosphere: the oxygen is residual from storage in the crust of reduced carbon com- pounds fixed by plants over hundreds of millions of years. The current changes are due in part to mobilization of these fossil reserves as fuel, as well as to changes in the amount of carbon retained in the biota and soils globally. The immediate question is how large the recent and future influence of the biota may be. The primary evidence for current change is the record of observations of the C02 content of the atmosphere. Modern records show not only a year-by-year increase of between about 0.5 and 2 ppm in CO2 annually but also a seasonal fluctuation. The amplitude of the oscillation varies with latitude, altitude, and probably with other factors (see Machta, Section 3.4). The amplitude approaches 20 ppm in central Long Island (Woodwell et al., l973a) and in Barrow, Alaska (Kelley, l969). It is about 5 ppm at Mauna Loa and l ppm at the South Pole (Keeling et al., l976a,b). The peak CO2 concentration occurs in late winter? the minimum occurs in early fall. The oscillation is reversed in the southern hemisphere to coincide with the southern seasons. These observations are evidence that biotic factors are large enough to influence the C02 content of the atmosphere in the short term over large regions, possibly the Earth as a whole. We know, in addition, that the terrestrial biota and soils contain two to three times as much carbon as the atmosphere. A small change in the size of these reser- voirs globally has the potential for storing or releasing a quantity of CO2 sufficient to affect the amount in the atmosphere appreciably. Despite the strength of these observations, answers to specific questions about the role of the biota have proven elusive. Can we use the history of the terrestrial biota over the past century to help determine the preindustrial atmospheric C02 concentration (and thus the sensitivity of climate to increasing C02)? What fraction of the increase in C02 observed since l958 at Mauna Loa is due to oxidation of carbon compounds in plants and soils as opposed to combustion of fossil fuels? What proportion of future C02 emissions will be absorbed into biotic reservoirs? If current trends in land use con- tinue, what effect will they have on future atmospheric concentrations? To what extent could the future C02 content of the atmosphere be controlled by management of land and forests? In the sections that follow we discuss the factors that affect the role of the biota in determining atmospheric C02 concentrations. These factors include the size and location of major reservoirs of carbon, the various transitions in metabolism that affect the reservoirs, and direct human effects, such as the clearing of forested land for agriculture.

2l7 3.3.2 How Much Carbon is Held in the Biota and Soils? 3.3.2.l The Biota Over the past decade, the most widely used appraisal of the global carbon content of biotic systems has been that of Whittaker and Likens (l973). Whittaker and Likens estimated that the world of l950 held in the biota a total of 829 gigatons of carbon (Gt of C), almost entirely on land. By far the largest biotic reservoir, 743 Gt of C, was estimated to be in forests. Forests have been reduced in area in the more than 30 years since l950. More recent estimates (e.g., Ajtay et al., l979) of terrestrial biomass have suggested additionally that the average standing crop of organic matter per unit area is lower than estimated by Whittaker and Likens (l973). The most comprehensive recent analysis is that of Olson (l982), who has suggested a total biomass for l980 of 560 Gt of C. The differences among these estimates are probably due largely to differences in interpretation and in assumptions; they may be due in part to reduction in the area of forests between l950 and l980. There is no easy resolution. To the extent that areas in forests are in question, satellite imagery will offer greatly improved estimates. Better data on biomass will require additional field studies. 3.3.2.2 The Soils The total carbon retained in soils has been estimated globally as between about l400 and 3000 Gt (Table 3.l). Schlesinger (l977, l983) has reviewed the estimates and has suggested that there is a total of about l500 Gt of readily mobilized carbon available. This conclusion has been supported recently by Post et al. (l982), who summarized data from more than 2000 samples of soils from around the world. 3.3.2.3 Total Carbon Pool under Biotic Influences The total amount of carbon readily transformable into CO2 by metabolic processes is probably in the range of 2000-2500 Gt, about 3 times the 725 Gt of C currently held in the atmosphere. Most of this inventory is associated with forests. The fraction of the biotic pool that will actually be transformed to atmospheric CO2 in future decades is speculative. In principle, a large portion could be. Evidence is that when forests are disturbed or replaced by agriculture there is a substantial loss of carbon. Loss from the original standing stock of plants usually exceeds 90%, and loss from soils can also be substantial. Plough horizons of soils long in agriculture commonly contain l-3% carbon or less on a total dry weight basis. Original soils of the forests from which the agricultural soils were developed may have contained as much as 40-50% carbon in the surface horizon. Total loss over the entire profile is probably 20-50% of the original amount in the soil.

2l8 TABLE 3.l Amount of Carbon Retained in Soils Globally According to Various Recent Estimates** Total Carbon in Soils of the Earth (Gt) Source 3000 Bohn, l978 2070 Ajtay et al., l979 l456 Schlesinger, l977 l395 Post et al., l982 l477 Buringh (in press, l983) *.The data of Post et al. (l982) are the most comprehensive and recent estimates and are based on more than 2000 samples from around the world. 3.3.3 Metabolism and the Storage of Carbon in Terrestrial and Aquatic Ecosystems 3.3.3.l The Production Equation i The net flux of carbon between the atmosphere and any ecosystem is determined by the balance between gross photosynthesis and total ; respiration. The relationship is shown by the production equation for an ecosystem (Woodwell and Whittaker, l968): ( NEP « GP - (RA + %) , where NEP is the net ecosystem production, the net flux of carbon into or from an ecosystem; GP is the gross production, total photosynthesis of the ecosystem; RA is the respiration of the autotrophs, the green plants; Rg is the respiration of the heterotrophs, including all animals and organisms of decay; and RA + Rg is the total . respiration of the ecosystem. The potential range of values for NEP is from a large negative number, indicating a loss of stored carbon, to a positive value that approaches the net amount of carbon available from green plants after their own needs for respiration (RA) have been met. This excess above respiration is commonly called net primary production (NP). Its relationship to gross production is given by the equation NP = GP - RA, where GP is the total photosynthesis of the ecosystem, as above; R& is the respiration of the autotrophs; and NEP is NP - RH. The production equations have been especially useful in analyses of the metabolism and carbon flux of forests, where the terms can be

2l9 Biomass and soil carbon (gC/m ) YEARS FIGURE 3.ll Relationship between net ecosystem production (NEP) and the total accumulation of carbon in a forest. evaluated conveniently (for example, see Woodwell and Whittaker, l968; Whittaker and Woodwell, l969; Woodwell and Botkin, l970; Reichle et al., l973). The equations are applicable in aquatic systems as well (Woodwell et al., l973b, l979). Net ecosystem production varies from zero at the start of the successional development of a forest to a maximum at midsuccession and back to zero at climax (Figure 3.ll). The relationship emphasizes that undisturbed forests approach an equilibrium (climax) in which gross photosynthesis is equaled by total respiration. Any change in the relationship between gross production and either segment of the total respiration shifts the ratio and causes a positive or negative net ecosystem production. In general, photosynthesis is more vulnerable to disruption than total respiration. This generalization holds for both the individual plant and for the ecosystem as a whole. The reason is that photosynthesis is limited to certain plants and is dependent on many factors, all of which must be favorable; respiration is a general characteristic of all life and occurs in some form under a wide range of conditions. Virtually any disturbance (e.g., forest fire, land clearing, air pollution) favors respiration over photosynthesis, at least initially (see discussion below) and tends to result in transfer of carbon from the biota into the atmosphere. 3.3.3.2 A Basis in the Metabolism of Forests for the Oscillation in Atmospheric CO2 Concentration I Our understanding of the metabolism of terrestrial ecosystems, including forests, agricultural systems, grasslands, and other communities, supports the hypothesis that the annual oscillation in CO2 is due largely to the metabolism of temperate zone forests. The forests dominate because of their size, both in area and in magnitude of their metabolism. The annual course of metabolism of a temperate zone forest, taken in toto on a unit of land, expressed as gross photosynthesis and total respiration in separate curves, appears in Figure 3.l2(a). Net ecosystem production at any time is the algebraic

220 FACTORS AFFECTING CO2 IN AIR 7 o CD oc < o IL. O M ir O O i 330 8 J310 11/76 M M FIGURE 3.l2 (a) The course of total respiration and gross photo- synthesis of an oak-pine forest in central Long Island, New York. Integration of these two curves produced the prediction of the annual change in atmospheric CO2 shown in the curve below. (b) The amplitude predicted in this way was considerably greater than observed (Woodwell et al., l973a) , apparently because of mixing with air from over the oceans.

22l sum of one point on each of the two curves. Integrating over the entire year produces a curve for net ecosystem production [Figure 3.l2(b)]. This new curve follows closely the pattern of oscillation observed in the CO2 concentration at Mauna Loa and elsewhere, although the amplitude observed around the world differs greatly from that calculated for the forest of central Long Island shown in Figure 3.l2. 3.3.3.3 Factors Affecting Global Net Ecosystem Production 3.3.3.3.l Succession and the Equilibrium Hypothesis The equilibrium between gross photosynthesis and total respiration is achieved in forests after a period of successional development that may last decades to a century or so. It probably occurs in the pelagic segment of aquatic systems in days to weeks after disturbance. It is approximate: there is a residuum of the annual increment of fixed carbon that is stored in sediments both on land and in the sea, but the storage is a small fraction of annual production (Broecker, l974). To illustrate, one of the largest accumulations of stored carbon on land is in the peat of the tundra and boreal forest, which we might estimate at s6o Gt of C. This mass is believed to have accumulated over the l0,000 years since the retreat of glacial ice, an annual rate of accumulation 0.05 Gt of C. The assumption of an equilibrium seems appropriate for this first approximation of the biotic flux. Aquatic ecosystems move more rapidly than terrestrial systems toward equilibrium between gross photosynthesis and total respiration because the species are small-bodied and reproduce rapidly. Mechanical dis- turbance is common and has little effect on the ratio of gross produc- tion to total respiration under most circumstances. Chemical distur- bance in the form of enrichment or pollution may simply speed the rate of turnover of carbon molecules, not increase the rate of sedimentation, at least through a wide range (Peterson, l982), although Walsh et al. (l98l) have suggested that enrichment of the coastal zone with nitrogen and phosphorus is causing increased accumulation of sediments on the shelf. 3.3.3.3.2 Gross Photosynthesis Gross photosynthesis is awkward to measure directly in nature. It is normally measured as net photosynthesis of leaves plus total respiration of the ecosystem. Data are usually taken on a small scale in controlled environments or in carefully monitored natural ecosystems. Techniques are available now to provide more and better data on patterns and trends in the metabolism of ecosystems and should be exploited. Several factors affect gross photosynthesis. The most important are light, moisture, and availability of nutrients, especially nitrogen, phosphorus, and CO2. While it is common to think of one factor at a time as limiting the rate of any process such as photosynthesis, experi- ence indicates that, throughout wide ranges, a change in the avail-

222 ability of any factor will produce a response. The dominant question for our consideration is whether the increase in atmospheric CO2 is causing an increase in net ecosystem production globally. Waggoner (this volume, Chapter 6), Strain (l978), and a recent symposium (AAAS, in press) in Athens, Georgia, have reviewed studies in which efforts have been made to measure the effect of increased CO2 on the growth of crop plants. While there is probably an important effect on growth of well-watered, fertilized plants, there is question as to whether these effects extend to natural communities. The following tentative generalizations may be offered (Kramer, l98l): • Species differ greatly in response to enhanced CO2. • The response is greater in plants with indeterminate growth (cotton) than in plants with determinate growth (corn). • The response is greater in C3 plants such as soybeans than in C4 plants such as corn. • The largest response is in seedlings; in older plants the response decreases or ceases. One of the most important factors is that any increase in growth observed is not always due to an increase in the rate of photosynthesis per unit of leaf area. Enhanced CO2 concentrations cause changes in the morphology of growing plants, including an increase in branching of both woody and herbaceous plants, greater stem elongation, and an increase in the ratio between roots and shoots. One of the most per- sistent effects is an increase in the area of leaves, a result observed by Wong (l979) in a series of studies of effects of nitrogen nutrition and CO2 on cotton and by others in several studies summarized by Kramer (l98l) and Strain (l978). While Waggoner (this volume, Chapter 6) reports generally beneficial effects on growth of crop plants, a positive response to increased CO2 is not universal. Wong (l979) showed that corn plants assimilated less carbon under high CO2 and speculated that contrary observations in other experiments may have been due to lower light intensities. Responses to enhanced CO2 concentrations are particularly strong in younger plants, a factor that affects many of the experiments reported, such as that by Gifford (l979), in which a stimulation of growth was shown in wheat grown under moisture stress with enhanced CO2 concen- tration. In longer-term experiments and in older plants response is diminished, disappears, or may involve a reduction in growth rate. These considerations should lead to caution in projecting effects of enhanced concentrations of CO2 on photosynthesis in forests (Strain, l978; Kramer, l98l). Despite the expansion of agriculture, natural (unmanaged) forests still dominate the biotic segment of the global carbon cycle. Plants in natural forests live in conditions of extreme competition for light, water, nutrients, space, and, probably, C02 during daylight. None of the research reported applies directly to this circumstance, a fact that suggests great caution in predicting enhanced storage of carbon in natural systems due to increased atmospheric CO2. In fact, Kramer (l98l) concluded that in general, increase in CO2 concentration will

223 probably have the least effect on growth of plants in closed stands where light, water, and mineral nutrition, separately or collectively, already limit the rate of photosynthesis. Kramer's conclusion is supported by observations that the rate of photosynthesis per unit leaf area is not always increased by an increase in CO2 concentration and that effects on plants often involve a change in the morphology in young plants. Forests are not modified rapidly in the latter respect. Effects of temperature on gross photosynthesis are through effects on respiration, or other processes apart from the photochemical process, which is nearly independent of temperature (Larcher, l980, p. lll). 3.3..3.3.3 Total Respiration The data on respiration that are of significance in detecting a change in net ecosystem production are those that define the rate per unit of land area. They include the respiration of the community of higher plants, the community of animals, and the various communities of lichens, mosses, and organisms of decay. Such data have rarely been taken for terrestrial ecosystems; analysis must be based largely on inference from first principles or from data obtained for other purposes. Kates of respiration are also affected by many factors, including availability of water, nutrients, especially nitrogen and phosphorus, and temperature. As in all chemical reactions, rates are affected by the availability of substrates and the accumulation of products. The greatest sensitivity is probably to temperature. A l0°C increase in temperature through the middle range of the response curve for a whole plant commonly produces a twofold to threefold increase in the rate of respiration (Table 3.2). Experimental evidence from tundra communities confirms the effect of warming (Billings et al., l982). By comparison, the effects of other factors appear small. Direct effects of CO2 concentrations in the range of 300-600 ppm on total respiration of an ecosystem are so small as to remain unmeasured and probably unmeasure- able. The observation that rates of respiration in photosynthetic tissues differ in the light and in the dark (Zelitch, l97l) has little bearing on the total respiration, so large is this total relative to the fraction that occurs in photosynthetic tissues. 3.3.3.3.4 Net Ecosystem Production and the 6-Factor £ In an effort to resolve a discrepancy between the amount of carbon reportedly released through combustion of fossil fuels and the amount apparently transferred to the oceans, Bacastow and Keeling (l973) introduced a factor into their analysis that allowed an expansion of the biotic pool of carbon as a function of the increase in CO2 in air. It was assumed that this so-called "B-factor" was the only important biotic consideration in the global carbon cycle. B was estimated to be a constant with a value of 0.26. The B-factor as formulated by Bacastow and Keeling was limited to the putatively positive effect of the increase in CO2 on net ecosystem production.

224 TABLE 3.2 Respiratory Quotients (QiQ) for Plants and Plant Communities3. Respiratory Temperature Zone/Species QlO Range (°C) Reference Pea seedlings 3.0 0-l0 Giese, l968 2.4 l0-20 l.8 20-30 l.4 30-40 Plants 2.l-2.6 0-30 Fitter and Hay, l98l Tundra/Taiga 2 Low temperature Miller, l98l 2 High temperature Tropics 3 l0 Larcher, l980 Clover 2.4 3-l3 Woledge and Dennis, l982 Greenland plants 2.0-2.7 0-25 Eckhardt et al. , l982 -The Ql0 is the factor by which respiration is increased by a l0°C increase in temperature. No consideration was given the possibility that processes other than CO2 enrichment might affect the amount of carbon retained in the biota and soils or the possibility that the area of forests might be changing globally. The use of the 6-factor was a pragmatic solution to a complex and puzzling issue that arose from attempts to analyze the global carbon cycle through a simple model. Its use should now be replaced by separate analyses of the effects of (a) changes in the area of forests and (b) potential changes in net ecosystem production caused by both increased atmospheric CO2 and changes in climate. The latter will require modeling based on processes in terrestrial ecosystems. The C02 content of the atmosphere has increased since l860 to its current 340 ppm from a concentration now estimated at 260-280 ppm (see Machta, Section 3.4). The increase may be approaching 30%. Such a change in C02 content alone has probably had no effect on respiration. It may have affected gross photosynthesis, but if so, the change is detectable neither directly as a measurement of net or gross photo- synthesis nor indirectly as a measurement of some segment of net ecosystem production such as the annual increment of wood in trees of seasonal forests. A widespread increase in annual tree growth of as little as l0% should be detectable as a universal or very common change in width of tree rings. No such stimulation is conspicuous. Rebello and Wagener (l976) found evidence in Europe of an increase in diameter growth, but Whittaker et al. (l974) found the opposite in North America. A smaller increase might remain undetected at present. Waggoner (this volume, Chapter 6) suggests a small (~5%) increase for crop plants in a 400-ppm atmosphere. While some atmospheric changes may have favored growth in the total carbon held in the biota and soils globally, others may not. In par-

225 ticular, there is a need to consider the possible role of a long-term global warming of about 0.5°C since the lows of the late l880s (see Weller et al., this volume, Chapter 5). If a l0°C increase in tem- perature increases rates of respiration twofold to threefold (Table 3.2), the 0.5°C warming observed may have increased total respiration of terrestrial ecosystems by l0-l5%. Such a change would appear as a reduction in net ecosystem production; it might, of course, simply offset an increase in gross photosynthesis. The topic of changes in the biota as a result of enhanced CO2 and climatic change requires detailed study through descriptive surveys and careful field experi- mentation under controlled circumstances. At the moment there is no direct evidence that net ecosystem production has changed per unit area of existing forests regionally or globally over the past century. 3.3.4 Changes in Area of Forests of the World There have been many changes in the area of forests in postglacial time. The transitions have been caused by climatic changes such as those that accompanied the retreat of glacial ice, by shifts in patterns of distribution of rainfall, and by activities of man. The Levant, for instance, was probably largely deforested 5000 or more years ago through harvest of wood followed by intensive and prolonged grazing by goats. Other sections of the Mediterranean Basin were deforested more recently, but still l000-5000 years ago. The British Isles were largely forested until the seventeenth century. Other areas, including much of Europe and northeastern North America, were cleared for agriculture and grazing two to three centuries ago. Agri- culture was subsequently abandoned in some areas, and these have been partially reforested. Overall, the expansion of human population throughout history has been accompanied by an almost continuous decline in the area of forests globally. The rate at which deforestation occurred in the past was slow by comparison with recent rates. The changes that affected areas as large as the Mediterranean Basin, the Levant, or the British Isles took place over centuries to millennia. Rates of CO2 releases from biotic sources probably varied considerably and may have affected the CO2 content of the atmosphere significantly, but there is no basis for measurements either of the biotic releases or of the CO2 accumulation that followed them. During the past century, higher rates of deforesta- tion may have been resulting in annual releases of CO2 of as much as several billions of tons of carbon, in some years increasing the total atmospheric burden by as much as 0.5%. The amount of the biotic contribution to the atmospheric increase is in question. The challenge has been measurement: how can information on rates of deforestation, and reforestation whenever it occurs, be summed for the Earth as a whole? The greatest uncertainty is in the rates of deforestation in the tropics, but there is uncertainty about the size of the contribution from loss of temperate zone and boreal forests as well. The largest areas of forest remaining in the world are,in the tropics, especially the Amazon, and in the northern tem-

226 TABLE 3.3 Estimates of Annual Net Carbon Flux between Terrestrial Ecosystems and the Atmosphere in or about l980*. Author l0l5 g of C/yr Adams et al., l977 0.4 to 4 Bolin, l977 0.4 to l.6 Revelle and Munk, l977 l.6 Wong, l978 l.9 Woodwell et al., l978 4 to 8 Hampicke, l979 l.5 to 4.5 Seiler and Crutzen, l980 -2.0 to 2.0 Brown and Lugo, l98l -l.0 to 0.5 Moore et al., l98l 2.2 to 4.7 Olson, l982 0.5 to 2.0 Houghton et al., l983 l.8 to 4.7 ^Positive values indicate net release to the atmosphere. Full citations in references. perate and boreal zones. There are significant areas of forests remaining, however, in tropical Africa and in Southeast Asia. There have been numerous attempts over the past decade to estimate the current annual net carbon flux between terrestrial ecosystems and the atmosphere. Estimates, summarized by Clark et al. (l982), range from a net absorption of 2.0 Gt of C to a net release of 20 Gt of C per year. Table 3.3 presents a selection of recent estimates. All the estimates suffer from a lack of persuasive detail as to rates of defor- estation in key areas. The estimates differ in large part because they have treated different segments of the problem. When corrected to a common basis, the recent estimates converge considerably (Woodwell et al., l982). The most important advances in these analyses have come through recognition that sufficient information is available to allow prediction of the details of changes in forested areas if the time of (a) harvest or (b) transformation to agricultural or grazing land is known. A forest that is harvested by clearcutting and allowed to recover, for example, follows a predictable pattern of successional development. The observation that disturbance occurred is the critical point: the sequel is predictable. Similarly, a forest transformed to pasture or to row-crop agriculture loses its carbon stock predictably. If agriculture is abandoned, the forest recovers, again at predictable rates. Precision in the total inventory of carbon is less important than the evidence of change in the inventory. The evidence of change is abrupt and discontinuous, namely, the harvest of a forest or the abandonment of agriculture. Evidence on changes in land use can be accumulated, tabulated, and summed in a model to offer an estimate at any time of the trends in

227 carbon storage in forests, locally or globally. Such a model, con- structed around the central principle of ecological succession, has been developed and used; results are reported below. Details of the construction of the model, including the data and assumptions used and tests of sensitivity, have been presented elsewhere (Moore et al., l98l; Houghton et al., l983; Woodwell et al., l983a). The model accommodates l2 geographic regions and l0 different types of vegetation in each region. Transitions in soils are also included. A few comments about quality and sources of data are necessary. Data appraising rates of change in forested areas are surprisingly difficult to obtain and verify. The major factors to be measured are (l) change in the amount of carbon per unit area of forests and (2) change in the area of forests. Neither can be measured unequivocally for the globe at present. Several factors contribute to the difficulties in obtaining sound data. One is that no nation is proud of the destruction of a valued resource, and national statistics are often unreliable. In addition, many nations lack equipment and personnel needed for gathering data on area of forests remaining and for evaluating such data. There is always question as to what is forest. Are the successional stands that replace moist tropical forests following harvest to be considered equivalent to the forests they replaced? Are impoverished, heavily grazed woodlands forests? Various economic considerations tend to bias reporting first one way, then another (Persson l974, l977). Recent studies of remote sensing using the LANDSAT and NOAA systems show that satellite imagery offers great promise for improving data on areas of forests globally (Woodwell, l980; Woodwell et al., l983b; Woodwell et al., in press). Three sources for estimating rates of deforestation have been used extensively in recent analyses. One source is the series of Production Yearbooks of the UN Food and Agriculture Organization (FAO), published since l949. They rely on data reported by governments. A second source is the work of Myers (l980), who has compiled detailed estimates of the rate of conversion of tropical forests using a variety of sources. The emphasis on the tropics is appropriate because of the extraordinary growth in the human population in the tropics and the surge of economic development that has affected the tropical regions since the Second World War. A third source of estimation is based on the assumption that there is a simple correlation between growth in human population and rate of deforestation. The basis of the assumption is that most of the loss of forests is due to conversion for agriculture. Revelle and Munk (l977) developed this approach initially.* *Richards et al. (l983) have recently used historical records to determine land area converted from unmanaged ecosystems to regularly planted cropland. They estimate net conversion of 85l million hectares between l860-l978; Revelle and Munk estimated clearing between l860-l970 of 853 million hectares.

228 rH 10 C H O tH •H a CM O rH 0*0*CM 0 rH m O* 0900 Is 4J co r- w * rH » CM CMrOH CMCM CM CM r-|rH •^ c i1 . O 10 •H 4J •H k i CQ k l k i 3 s *, m m m m m m m m m a) 5i 4t u 0} CM CO CM CMCMCM CMCM CM CM CMCM rH O 10 c g1 3« C OOO OOO OO O O OO k l •rl 41 4J O1 10 41 1 -S (0 a - _ 41 0* G* 0* cnc*o* 0*0* en o* oo* CO k l m m m mm m m meo r- c2 r >5 O OOO ooo oo o o oo a* C 0 u ooo ooo oo o o oo M 10 'o (C^ 1 1 1 III II 1 1 II a 4J ti 41 c (0 w i s 41 Q u o C rH *J »$ O 3 VO U5 ^C CMtOtO to rH tO m to rrj •O o soi6i0 ovo •** to at to i C -H 10 k l OOO OOO CMO O O 0O 3^3 ooo ooo oo o o oo III II 1 1 II U 41 O * S k l O *5 *9 — . 0 >t O > i j* 1 r* « « OD <JD m oo * v o oo in crt mr^ro vom ifl r- in^ „ •» r- m 0j ^J UH a o 3 ooo ooo oo o o oo 41 a o J3 * a u «H EH 2 41 m u " m >t> m in f),r*ifO Of) PO o^ u>ro S 0< 5* o 10 w O to o to 2 V *- 09 m o hi 10 « > (O **H 0* H O rH O ooo oo o o oo u 41 D -u C H Ij e fH S1 1 10 41 w 41 •H 3 «-4 J^ £ a *O <U Q1 a a o S V 10 -H O*OO *nr»cM OttD rH o OPO I Ml •jj 41 k i k i in r- ro ^ ot ^ *^ o fi roo rH O CT O • OS U ii < O CM rH HrHO i-HrH .H rH HM •H B rH CO 4-' d1 O 41 rCC Tl v 00 3 0* a . (D X rH CM o r- <MtOO> ovo rH in r- in S* 41 00 m r- in 0} -<H c 41 0* m CO rH a rH -H k l 10 41 H X3 e « rH „ O ^^ ^J 41 CJ °0 01 C rH A O* UH rH ^ in o o co in «JD m o in CM o4 • rH O 1 00 00 00 •Hrjro GOP- in r- mm — — Q rd (0 O *"* -Q -H 4J *J *O 09 — O u oo en •H JS — O O t o . § UH k i 3 „ 3 « « 5 S 5 8 § ° ° H k i 4J 10 ll O w „ a"°J 41 1-1 4> U U 41 U (y c_^ QQ CO Q1 § 3 ni oi rHOioCiou-1 a 0 a c o-woiji.*: 41 4J 41 -H • H 1 O • > q u u noicjc to u u 10 <u 3' io -H 3 OJ 11 rH oQ ai 0 Q to es 4J u O Ul O to — 3 C0 CM 41 41 COOTJTJiorH j-j TJ O iT 1 41 g rH •5 * a H n o o k l c rH tr 41 u-i (U C 41 >, §§ 2 * u ^ «j in 3 Q 1 3 > enoOtojbr/i41 41 i04141 3S -H •* w • O t? O «C 41 . 10 H ^irHDuZOilotoU rH 4i>O 4j §io H k i 3 UH Oi (T) 4J -H CO fc X Z P *J rH e " Ml, 3& >o an« it „ • • fl HklO 4^ * * " C • M 0 V g rH csr, „ rH rH f-9 jQ 4-1 |0 -H •p a < a IA

229 Analyses using the three sources have been reported in detail by Houghton et al. (l983) , and comparisons have been made with other, less comprehensive analyses that appeared to reach different conclusions, such as those of Detwiler et al. (l98l) and Brown and Lugo (l98l). In each instance, when appropriate adjustments were made to include all major factors, such as decay of soil organic matter, the analyses converged to a narrower range. Olson (l982), on the other hand, estimated a net loss of slightly less than +l Gt of C from the terrestrial sources in l980; the estimate was based on lower appraisals of the mass of carbon per unit area than those used by Houghton et al. (l983) and interpretations of changes in the area of forests that are difficult to document. The results of the application of the model using the three sources appear in Table 3.4. The range of the total release of carbon from the biota and soils into the atmosphere since l860 varied little, between l80 and l85 Gt. Most of the difference occurred in recent decades: the range of estimates between l958 and l980 was 52-70 Gt of C. The annual net release in l980 was estimated as between l.8 and 4.7 Gt of C. The model was used to test assumptions about the data and the significance of various biotic processes. These tests showed that lower total releases over l20 years could have occurred if agricultural land were taken only from nonforested areas (5b in Table 3.4), if there were reduced decay of soil carbon (8 in Table 3.4), if there were very much more rapid recovery of disturbed forests than experience dictated (l0 in Table 3.4), and if there were substantially less carbon in the vegetation and soils than sources such as Whittaker and Likens (l973) had indicated (ll in Table 3.4). These changes were all applied to the population-based estimate. None of these modifications turned the biotic pools into a net sink for atmospheric carbon at any time in the l20-year period. The annual release in l980 reached a minimum of l.7 x l0 Gt of C when the assumption was made that all agricultural land was taken from nonforested areas. That assumption must be considered unrealistic; it was included in the analysis to provide a limit for comparison. 3.3.5 The Biota in the Context of the Global Carbon Balance The effort to this point has been to summarize the most probable transitions in the biotic pools of carbon globally over the past century. The results are not consistent with current estimates of other segments of the global carbon cycle. The global carbon balance can be expressed as A » F - S + B, where A is the increase in the carbon content of the atmosphere over any period, F is the release of carbon to the atmosphere from combustion of fossil fuels in the same period, S is the net transfer to the oceans in the same period, and B is the absorption or release of carbon by the biota in the same period.

230 This equation can be evaluated to estimate the role of the biota. For example, for l980 Houghton et al. (l983) reported (in Gt of C per year) 2.5 (+0.2) / 5.2 (+0.7) - 2.0 (+0.5) - 0.7(+l.4). The sources of the quantities in this case were A - 2.5 for l980 from Bacastow and Keeling (l98l), F = 5.2 for l980 from Rotty (l98l), and S * 2.0 calculated as 39% of the fossil fuel release following Broecker et al. (l979). If the estimates of A, F, and 8 are accepted, the biota absorbed 0.7 Gt of C in l980. If A, F, and S are stretched to their proposed limits of uncertainty, the biota may have absorbed 2.l Gt of C or released 0.7 Gt of C. The method is indirect, but it has the advantage of presenting simply the relationships among the various major fluxes of carbon. Alternatively, methods based on isotopic dilution may be used to estimate total biotic effects over a long period of time. Results suggest a net release from the biota of between 70 and l95 Gt of C over a century or more (Table 3.5). These studies cannot provide precise estimates for a given year. As we have seen, most direct analyses of recent changes in the biotic reservoirs of carbon, when they include releases of carbon from decay of organic residues from the plants, organic matter in soils, and the decay of wood and forest products removed from the site, show a net release of carbon from destruction of forests. The release is estimated currently here on the basis of extensive experience as between l.8 and 4.7 Gt of C per year (Houghton et al., l983; Woodwell et al., l983a). This range is well outside the estimates above, including the ranges of uncertainty. Nonetheless, if the actual release is at the lower end of this range, there may be some basis for argument that the global carbon equation is balanced within the range of uncertainty associated with the other terms. If the actual value is near the upper end of the range, the equation is unreconcilable. A recent reinterpretation of TABLE 3.5 Estimates of the Release of Carbon from the Terrestrial Biota and Soils to the Atmosphere during the Past Century Based on Studies of Isotopes of Carbon in Tree Rings§. Reference Period Gt of C Stuiver, l978 l850-l950 l20 Wagener, l978 l800-l935 l70 Freyer, l978 l860-l974 70 Siegenthaler and Oeschger, l978 l860-l974 l35-l95 Tans, l978 l850-l950 l50 ^Adapted from Houghton et al. (l982).

23l the data from Myers (l980) on forest conversion in the tropics suggests that the upper limit of this range may be as low as 3.0 Gt of C. One hypothesis frequently proposed to balance the carbon equation is that net ecosystem production on land has increased in response to the increase in CO2 in the atmosphere. Such an increase would have to have been substantial, as much as l00-200 Gt of C over the past l20 years, to overcome the biotic losses to the atmosphere described above. It would probably be detected as a universal increase in diameter growth of trees or storage of humus in soils. Any such stimulation of carbon storage would have been accelerating as deforestation has proceeded, reducing areas where storage could occur. Moreover, it would require an increased spread between gross production and total respiration globally. A warming trend would probably work counter to this by increasing rates of respiration of plants and soil organic matter without a corresponding increase in gross production. This relationship would persist unless other factors, too, were ameliorated, such as water supply and the availability of nutrient elements, especially N and P. Additional insight on the role of the biota has been sought through exploration of the oscillation in C02 concentration observed at Mauna Loa (Hall et al., l975; Bacastow et al., l98lb; Pearman and Hyson, l98l). The hypothesis is that a change in the area of forests or a change in net ecosystem production regionally should be reflected over the 25-year record in a systematic change in the amplitude of the oscillation. Analyses showed no identifiable trend over the first l5 years (Hall et al., l975). There are now indications that the ampli- tude has been increasing in recent years (Bacastow et al., l98la,b; Machta, this volume, Chapter 3.4). How is such a change in amplitude to be interpreted? There appears to be little question that the oscillation itself is caused by the metabolism of forests. If the amplitude can be assured to be free of potentially confounding effects, it may offer an appraisal of the status of net ecosystem production at any moment for a segment of the northern hemisphere (Figure 3.l2). Unfortunately, the coupling between the oscillation and metabolism of forests is too loose for the amplitude to be considered a unique measurement of metabolism. Many factors affect it. These include, for example, atmospheric mixing: the ampli- tude of the oscillation is reduced at higher elevations and at lower latitudes. Small changes in patterns of circulation of air can be expected to affect it, as well as variability in the temperature of seawater. The amplitude is also open to various direct effects of metabolism of forests. A warm winter in the northern hemisphere, for example, would increase the total respiration without affecting the photosynthetic withdrawal during summer appreciably; the excess CO2 would appear as an increased late-winter peak in the Mauna Loa record. That CO2 would be mixed into the rest of the atmosphere over the ensuing weeks and would have little or no effect on the late summer minimum. The amplitude of the oscillation would have been increased and the total biotic pool of carbon on land reduced. The converse, an increase in the storage of carbon during summer, whatever the cause, would also appear as an increased amplitude in the oscillation. Vari-

232 ations in fossil fuel use may also be a confusing factor. While the oscillation of the Mauna Loa record may be interpretable, reliance on the record to identify transitions in the biota must await considerably greater attention to detail and better data than have been available so far. The possibility remains that aquatic systems have been stimulated in some way into accelerated storage of fixed carbon in sediments or in the deeper waters of the oceans. Freshwater systems can be ignored: they are very small in comparison with the oceans, which cover two thirds of the surface of the Earth. Baes (l982) and others (Smith, l98l; Walsh et al., l98l) have suggested various mechanisms by which biotic activity might result in sedimentation of carbon. To be signifi- cant in the global cycle these mechanisms would have to account for l Gt of C or more annually and would have to respond in some way roughly proportional to the increase in CO2 in the atmosphere. While there is no question about the capacity of the oceanic biota, either in coastal areas (Walsh et al., l98l) or in the open oceans (Baes, l982) to fix sufficient carbon to be significant in the global balance, there is question as to whether enough fixed carbon is sequestered in these waters to affect the global cycle in ways measurable now on a year-by- year basis. In pelagic aquatic systems gross production and total respiration tend to be closely coupled. An increase in photosynthesis is quickly followed by an increase in respiration; storage in sediments is small. Peterson (l982) addressed the question raised by Baes (1982) as to the capacity of the marine biota for storing carbon in any form. His conclusion was that there is so much carbon in seawater as dissolved CO2 in equilibrium with the oceanic carbonate-bicarbonate system that errors in estimates of oceanic absorption of CO2 are most likely to involve rates of mixing of surface waters into intermediate or greater depths. The issue remains unresolved. If the terrestrial biota appear to be a substantial net source of CO2 for the atmosphere beyond the fossil fuel source, the oceans must be absorbing substantially more CO2 than has been measured. The discrepancies are emphasized further by consideration of the fraction of the CO2 released that remains in the atmosphere. The annual increase in CO2 in the atmosphere is caused by the accumulation of some fraction of the total CO2 released. Because the fossil fuel CO2 has been thought to be the major, and sometimes the only, source of additional CO2, a frequent practice has been to express the increase in atmospheric CO2 as a fraction of the fossil fuel release. The fraction calculated in this way has the advantage of being based on two numbers that are measured with considerable accuracy. The fraction has the further advantage of being expected to approach a constant in the simplified models commonly used (Bacastow and Keeling l979, l98l). The airborne fraction, calculated solely on the basis of the.Mauna Loa data and estimates of combustion of fossil fuels was 0.55 for the period l959-l978, according to Bacastow and Keeling (l98l). The range of airborne fractions consistent with current carbon-cycle models was explored by Oeschger and Heimann (l983). They suggest that a range

233 from about 0.4 to 0.7 is possible. If the fraction is less than 0.4, there are deficiencies in current descriptions of the carbon cycle. When the fossil fuel contribution alone is considered, the preindustrial atmospheric CC>2 concentration is usually thought to have been 290-300 ppm. Recognition that there has been a large additional release from the biota and soils is consistent with recent measurements and estimates of a lower preindustrial concentration (See Machta, Section 3.4). If the biotic release were larger during the latter part of the last century than now, the airborne fraction would have a curious time history. To illustrate, we can calculate the airborne fraction for two periods, l860-l958 and l959-l980, using the estimate of biotic releases based on population (see Table 3.4 above) and data from Rotty (l98l, l982) on releases from combustion of fossil fuels. Time Period Biotic Release Fossil Release Total Release Atmospheric Increase Airborne Fraction e! l860-l958 l959-l980 Total l23 57 l80 76 l99 l43 342 l06 49 l55 0.53 0.34 0.45 86 l62 The estimates are made on the assumption, taken arbitrarily, that the preindustrial CO2 concentration in the atmosphere was 265 ppm. During the earlier period, l860-l958, 53% of the total CO2 thought to have been released to the atmosphere appears to have remained there. During the latter period, about 34% of the total release seems to have accumulated. Over the entire l20 years about 45% of the total has remained in the atmosphere according to these estimates. A circumstance in which the fraction of atmospheric carbon trans- ferred into the oceans or other sinks increases as the concentration of CO2 in the atmosphere rises is difficult to envision. No persuasive explanation is available. Mass balance considerations do not appear to support the hypothesis that entrophication of coastal waters, for instance, is causing accelerated sedimentation of carbon currently (Peterson, l982). There do not appear to be mechanisms for sequestering sufficient carbon on land. The uncertainty of the basic numbers, especially the preindustrial CO2 concentration and the magnitude and timing of biotic releases, both used to estimate the airborne fraction, re-emphasize that little should be inferred at present from calcula- tions such as these. There is, nonetheless, ample basis for arguing that the total release of carbon into the atmosphere has been and remains larger than the release from fossil fuels alone. Such a release means that the total accumulation in the atmosphere has been a lower fraction of the total release than estimated solely on the basis of combustion of fossil fuels. Presumably there has been a greater transfer to the oceans than commonly recognized.

234 3.3.6 A Projection of Further Releases from Biotic Pools If the analyses above are correct, the most important biotic transition that affects the global carbon cycle is the destruction of forests. The rate of destruction and the rate of release of carbon to the atmo- sphere can be anticipated. If the rate is proportional to the increase in population as assumed in the intermediate analysis reported above from Houghton et al. (l983), and it is assumed further that population continues to grow through the year 2000 and that the rate of growth declines thenceforth to zero in the year 2l00, the release from biotic sources might be expected to follow the solid line of Figure 3.l3. Releases would range between the current estimate of approximately 2 Gt of C annually to a maximum of between 6 and 7 Gt of C in 2000. If the growth in population were to continue beyond 2000, the forests them- selves would limit the release by the year 2030 to a maximum of about l0 Gt of C annually. Forests would then cease to exist as closed stands; the impoverished stands would decline progressively in carbon content and productivity. Rapid oxidation of 230 Gt of C, roughly the current total carbon content of tropical forests according to Olson (l982), would lead to an appreciable increase of atmospheric C02 (see Machta, Section 3.6). 3.3.7 Summary and Conclusions The biota and soils of the Earth contain more than three times as much carbon as the atmosphere. The most powerful evidence for the importance of the biota in affecting the C02 content of the atmosphere is the annual oscillation in the C02 concentration observed at Mauna Loa and in virtually every other annual record of atmospheric CO2. The extent of the biotic influence and the factors that govern it remain uncertain. The influence, however, is due primarily to global changes in forest' biomass. Forests have two types of effects on atmospheric C02, a shorter- term effect that is apparent in the oscillation in the concentration of CO2 and a longer-term effect due to changes in the total amount of carbon stored in them. The largest change in the mass of carbon in forests appears to be a net global reduction due to deforestation to support the expansion of agriculture. The biotic release from all changes in area, expressed as a net for the world as a whole, is probably in the range of l.8-4.7 Gt of C annually. This conclusion is derived from tabulations of data on rates of deforestation and forest harvest evaluated with a model. The model provides for the recovery of forests following harvest or abandonment of agriculture. Most studies, when expressed globally with adjustments to include soils and suc- cessional recovery, produce results that fall within the range stated. The question remains as to whether the metabolism of forests is being affected to change the storage of carbon in otherwise untouched stands. Such a change would require an increase in the spread between gross photosynthesis and total respiration. The factors that are most likely to affect this spread are light, moisture, nutrients, and tem-

235 o O m cc < O LL O X 1860 1880 1900 1920 1940 1960 1980 2000 2020 2040 2060 2080 2100 YEAR FIGURE 3.l3 Predictions of the net release of carbon from the biota over the next several decades based on the assumption that deforestation is proportional to growth in population. The solid line is the net release on the assumption that the population continues to grow until the year 2000, when growth then declines exponentially to zero; the dashed line shows the release if the population continues to grow and forests simply disappear early in the next century. For comparison, Nordhaus and Yohe (this volume, Chapter 2, Section 2.l) offer a mean estimate of fossil fuel CO2 emissions of about 5 Gt of C in the year 2000, l0 Gt of C in 2025, l5 Gt of C in 2050, and 20 Gt of C in 2l00. Deforestation is not a simple function of growth in population; it is in part due to economic and industrial circumstances. Nonetheless, for this prediction the assumption as made here is adequate to show that releases of CO2 from deforestation could increase well into the next century and then decline. perature, in addition to the concentration of CO2. Of these, temperature would seem to have the greatest potential: a l°C change might be expected to produce a 20-30% change in rate of respiration. Any effect on total photosynthesis would be largely through an increase in the length of the growing season. An increase in the growing season with impact on photosynthesis comparable with that of temperature on respiration would not be expected from a l°C increase in temperature. Interpretations of the apparent increase in amplitude of the oscilla-

236 tion of C02 concentration at Mauna Loa as due to increased regrowth of forests or stimulation of carbon storage on land are premature; the amplitude is affected by several factors. The probability seems high that a global warming will, at least initially, stimulate respiration in existing forests, thereby releasing additional stored carbon to the atmosphere; offsetting this trend could be a longer-term expansion of the forested zone poleward, with additional storage of carbon in the expanded area of forests. Recognition that there has been an appreciable biotic release in addition to the release of carbon from combustion of fossil fuels has important implications both for estimating the seriousness of CO2 increase and for potential mitigation of CO2-induced changes in climate. If there has been an appreciably larger total release than generally accepted, then the average fraction of total anthropogenic emissions (fossil fuel and biotic) remaining airborne over recent decades has a value near 0.4 (Clark et al. in Clark, ed., l982; Woodwell et al., l983a). Equally important, a total release from the biota and soils of approximately l80 Gt of C over the past l20 years would be consistent with an atmospheric C02 concentration in the mid-nineteenth century in the lower part of the 250-290-ppm range. Independent evidence is growing for the lower part of the range. Finally, the potential of the biota, especially forests, to release or store carbon is large enough to affect the C02 content of the atmosphere significantly year by year. If the surge in use of fossil fuels continues, the relative importance of the biotic contribution will diminish. If fossil use increases at a moderate rate, management of the biotic pools of carbon can affect the time that any given atmospheric CO2 concentration is reached by several decades. Better prediction, even control, of future CO2 concentrations are possible but will require substantially strengthened research and understanding of the carbon cycle. References Adams, J. A. S., M. S. M. Mantovani, and L. L. Lundell (l977). Wood versus fossil fuel as a source of excess carbon dioxide in the atmosphere: a preliminary report. Science l96:54-56. Ajtay, G. L., P. Ketner, and P. Duvigneaud (l979). Terrestrial primary production and phytomass. In The Global Carbon Cycle, B. Bolin et al., eds. SCOPE Report l3. Wiley, New York, pp. l29-l8l. Bacastow, R., and C. D. Keeling (l973). Atmospheric carbon dioxide and radiocarbon in the natural carbon cycle. II: Changes from A.D. l700 to 2070 as deduced from a geochemical model. In Carbon in the Biosphere, G. M. Woodwell and E. V. Pecan, eds. USAEC Symposium Series No. 30., U.S. Dept. of Commerce. NTIS, Springfield, Va., pp. 86-l36. Bacastow, R. B., and C. D. Keeling (l979). Models to predict future atmospheric C02 concentrations. In Workshop on the Global Effects of Carbon Dioxide from Fossil Fuels, W. P. Elliott and L.

237 Machta, eds. U.S. Dept. of Energy, CONF-770385, NTIS, Springfield, Va., pp. 72-90. Bacastow, R., and C. D. Keeling (l98l). Atmospheric carbon dioxide concentration and the observed airborne fraction. In Carbon Cycle Modelling, B. Bolin, ed. SCOPE Report l6. Wiley, New York, pp. l03-ll2. Bacastow, R. B., C. D. Keeling, T. P. Whorf, and C. S. Wong (l98la). Seasonal amplitude in atmospheric CO2 concentration at Canadian weather station P, l970-l980. Papers presented at WMO/ICSU/UNEP Scientific Conference on Analysis and Interpretation of Atmospheric CO2 Data, Bern, l98l. World Climate Programme. Bacastow, R. B., C. D. Keeling, and T. P. Whorf (l98lb). Seasonal amplitude in atmospheric CO2 concentration at Mauna Loa, Hawaii, l959-l980. Papers presented at WMO/ICSU/UNEP Scientific Conference on Analysis and Interpretation of Atmospheric CO2 Data, Bern, l981. World Climate Programme. Baes, C. F. (l982). Effects of ocean chemistry and biology on atmospheric carbon dioxide. In Carbon Dioxide Review: l982, W. C. Clark, ed. Oxford U. Press, New York, pp. l87-204 Billings, W. D., J. O. Luken, D. A. Mortensen, and K. M. Peterson (l982). Arctic tundra: A source sink for atmospheric carbon dioxide in a changing environment? Oecologia 53:7-ll. Bohn, H. L. (l976). Estimate of organic carbon in world soils. Soil. Sci. Soc. Am. J. 40:468-470. Bolin, B. (l977). Changes of land biota and their importance for the carbon cycle. Science l96:6l3-6l5. Broecker, W. S. (l974). Chemical Oceanography. Harcourt Brace Jovanovich, New York. Broecker, W. S., T. Takahashi, H. J. Simpson, T.-H. Peng (l979). Fate of fossil fuel carbon dioxide and the global carbon budget. Science 206:409-4l8. Brown, C. W., and C. D. Keeling (l965). The concentration of atmospheric carbon dioxide in Antarctica. J. Geophys. Res. 20:6077-6085. Brown, S., and A. E. Lugo (l98l). The role of the terrestrial biota in the global CO2 cycle. Proceedings of a Symposium: A Review of the Carbon Dioxide Problem. Am. Chem. Soc., Div. Petrol. Chem. 26-: l0l9-l02 5. Buringh, P. (l983). Organic carbon soils of the world. In The Role of Terrestrial Vegetation in the Global Carbon Cycle: Measurement by Remote Sensing, G. M. Woodwell, ed. SCOPE. Wiley, New York (in press). Clark, W. C., ed. (l982). Carbon Dioxide Review: l982. Oxford U. Press, New York. Detwiler, R. P., C. A. S. Hall, and P. Bogdonoff (l98l). Simulating the impact of tropical land use changes on the exchange of carbon between vegetation and the atmosphere. In Global Dynamics of Atmospheric Carbon, S. Brown, ed. Proceedings of a symposium held at the annual meeting of the Ecological Society of America under the auspices of the American Institute of Biological Sciences, Bloomington, Indiana, l98l. U.S. Dept. of Energy, pp. l4l-l59

238 Eckhardt, F. E., L. Heerfordt, H. M. Jorgensen, and P. Vaag (l982). Photo-synthetic production in Greenland as related to climate, plant cover, and grazing pressure. Photosynthetics l6:7l-l00. Giese, A. C. (l968). Cell Physiology. 3rd ed. Saunders, Philadelphia, Pa. Gifford, R. M. (l979). CO2 and plant growth under water and light stress: implications for balancing the global carbon budget. Search l0:3l6-3l8. Hall, C. A. S., C. A. Ekdahl, and D. E. Wartenberg (l975). A fifteen year record of biotic metabolism in the northern hemisphere. Nature .255: l36-l38. Hampicke, U. (l979). Sources and sinks of carbon dioxide in ter- restrial ecosystems. Environ. Internat. 2:30l-3l6. Houghton, R. A., J. E. Bobbie, J. M. Melillo, B. Moore, B. J. Peterson, G. R. Shaver, and G. M. Woodwell (l983). Changes in the carbon content of terrestrial biota and soils between l860 and l980: A net release of CO2 to the atmosphere. Ecolog. Monogr. (in press). Keeling, C. D., J. A. Adams, Jr., C. A. Ekdahl, and P. R. Guenther (l976a). Atmospheric carbon dioxide variation at the South Pole. Tellus 28:552-564. Reeling, C. D., R. B. Bacastow, A. E. Bainbridge, C. A. Ekdahl, Jr., P. R. Guenther, L. S. Waterman, and J. F. S. Chin (l976b). Atmospheric carbon dioxide variations at Mauna Loa Observatory, Hawaii. Tellus 28: 538-55l. Kelley, J. J. (l969). An analysis of carbon dioxide in the arctic atmosphere near Barrow, Alaska, l96l to l967. University of Washington, Dept. of Atmospheric Sciences, Scientific Report, Naval Research Contract No. 6l4-67-A-0l03-0007. NR. 307-252, Seattle, Washington. Kramer, P. J. (l98l). Carbon dioxide concentration, photosynthesis and dry matter production. BioScience 3l:29-33. Larcher, W. (l980). Physiological Plant Ecology. Springer-Verlag, Berlin. Moore, B., R. D. Boone, J. E. Hobbie, R. A. Houghton, J. M. Melillo, B. J. Peterson, G. R. Shaver, C. J. Vorosmarty, and G. M. Woodwell (l98l). A simple model for analysis of the role of terrestrial ecosystems in the global carbon budget. In Carbon Cycle Modelling, B. Bolin, ed. SCOPE Report l6. Wiley, New York. Myers, N. (l980). Conversion of Tropical Moist Forests. Prepared for Committee on Research Priorities in Tropical Biology, National Research Council, National Academy of Sciences, Washington, D.C. Oeschger, H., and M. Heimann (l983). Uncertainties of predictions of future atmospheric CO2 concentrations. J. Geophys. Res. 88_: l258-l262. Olson, J. S. (l982). Earth's vegetation and atmospheric carbon dioxide. In Carbon Dioxide Review: l982, W. C. Clark, ed. Oxford U. Press, New York, pp. 388-398. Pearman, G. I., and P. Hyson (l98l). The annual variation of atmospheric CO2 concentration observed in the Northern Hemisphere. J. Geophys. Res. 86: 9836-9843.

239 Persson, R. (l974). Review of the world's forest resources in the early l970s. Dept. of Forest Survey Res. Notes No. l7. Royal College of Forestry, Stockholm. 265 pp. Persson, R. (l977). Scope and approach to world forest resource appraisals. Dept. of Forestry Survey Res. Notes No. 23. Royal College of Forestry, Stockholm. Peterson, B. J. (l982). In Carbon Dioxide Review: l982, W. C. Clark, ed. Oxford U. Press, New York, pp. 205-206. Post, W. M., W. R. Emanuel, P. J. Zinke, and A. G. Stangenberger (l982). Soil carbon pools and world life zones. Nature 298:l56-l59. Rebello, A., and K. Wagener (l976). Evaluation of l2C-and~r3C-"data on atmospheric CO2 on the basis of a diffusion model for oceanic mixing. In Environmental Biogeochemistry, Vol. lt Carbon, Nitrogen, Phosphorus, Sulfur, and Selenium, J. O. Nriagu, ed. Ann Arbor Science Publishers, Ann Arbor, Mich., pp. l3-23. Reichle, D. E., B. E. Dinger, N. T. Edwards, W. F. Harris, and P. Sollins (l973). Carbon flow and storage in a forest ecosystem. In Carbon and the Biosphere, G. M. Woodwell and E. V. Pecan, eds. U.S. Atomic Energy Commission, pp. 345-365. Revelle, R., and W. Munk (l977). The carbon dioxide cycle and the biosphere. Energy and Climate. National Research Council, Geophysics Study Committee, National Academy Press, Washington, D.C., pp. l40-l58. Richards, J. F., J. S. Olson, and R. M. Rotty (l983). Development of a data base for carbon dioxide releases resulting from conversion of land to agricultural uses. ORAU/IEA-82-l0(M). Institute for Energy Analysis, Oak Ridge, Tenn. Rotty, R. M. (l98l). Data for global CO2 production from fossil fuels and cement. In Carbon Cycle Modelling, B. Bolin, ed. SCOPE Report l6. Wiley, New York, pp. l2l-l25. Rotty, R. M. (l982). Distribution and changes in industrial carbon dioxide production. J. Geophys. Res. 88:l30l-l308. Schlesinger, W. H. (l977). Carbon balance in terrestrial detritus. Ann. Rev. Ecol. Syst. 8:5l-8l. Schlesinger, W. H. (l983). The world carbon pool in soil organic matter: a source of atmospheric CO2. In The Role of Terrestrial Vegetation in the Global Carbon Cycle: Measurement by Remote Sensing. SCOPE. Wiley, New York. Seiler, W., and P. J. Crutzen (l980). Estimates of gross and net fluxes of carbon between the biosphere and the atmosphere from biomass burning. Climatic Change 2:207-247. Siegenthaler, U., and H. Oeschger (l978). Predicting future atmospheric carbon dioxide levels. Science l99:388-394. Smith, S. V. (l98l). Marine macrophytes as a global carbon sink. Science 2ll:838-840. Strain, B. R., ed. (l978). Report of the workshop on anticipated plant responses to the global carbon dioxide enrichment, held August 4-5, l977. Dept. of Botany, Duke U., Durham, N.C. 9l pp. Stuiver, M. (l978). Atmospheric carbon dioxide and carbon reservoir changes. Science l99:253-270.

240 Tans, P. P. (l978). Carbon l3 and Carbon l4 in trees and the atmospheric C02 increase. Thesis. Rijsuniversiteit te Groningen, The Netherlands. Wagener, K. (l978). Total anthropogenic CO2 production during the period l800-l935 from carbon-l3 measurements in tree rings. Radiat. Environ. Biophys. l5:l0l-lll. Walsh, J. J., G. T. Rowe, R. L. Iverson, and C. P. McRoy (l98l). Biological export of shelf carbon is a sink of the global C02 cycle. Nature 29l: l96-20l. Whittaker, R. H., F. H.Bovmann, G. E. Likens, and T. G. Siccama (l974). The Hubbard Brook ecosystem study: forest biomass and production. Ecolog. Manag. 44:233-254. Whittaker, R. H., and G. E. Likens (l973). Carbon in the biota. In Carbon and the Biosphere, G. M. Woodwell and E. V. Pecan, eds. USAEC Symposium Series No. 30. NTIS, Springfield, Va., pp. 28l-302. Whittaker, R. H., and G. M. Woodwell (l969). Structure, production and diversity of the oak-pine forest at Brookhaven, New York. J. Ecol. .57: l55-l74. Woledge, J., and W. D. Dennis (l982). The effect of temperature on photosynthesis of ryegrass and whole clover leaves. Ann. Bot. .5J): 25-35. Wong, C. S. (l978). Atmospheric input of carbon dioxide from burning wood. Science 200:l97-l99. Wong, C. S. (l979). Elevated atmospheric partial pressure of CO2 and plant growth. Oecologia 44:68-74. Woodwell, G. M., ed. (l980). Measurement of changes in terrestrial carbon using remote sensing. U.S. Dept. of Energy CONF-7905l76, UC-ll. Available from NTIS, Springfield, Va. Woodwell, G. M., ed. (in press). The Role of Terrestrial Vegetation in the Global Carbon Cycle: Measurement by Remote Sensing. SCOPE Report 23. Wiley, New York. Woodwell, G. M., and D. B. Botkin (l970). Metabolism of terrestrial ecosystems by gas exchange techniques: the Brookhaven approach. In Ecological Studies, D. E. Reichle, ed. Analysis and Synthesis, Volume l. Springer-Verlag, Berlin, pp. 73-85. Woodwell, G. M., and R. A. Houghton (l977). Biotic influences on the world carbon budget. In Global Chemical Cycles and Their Alteration by Man, W. Stumm, ed. Report of the Dahlem Workshop, November l5-l9, l976. Dahlem Konferenzen, Berlin, pp. 6l-72. Woodwell, G. M., and R. H. Whittaker (l968). Primary production in terrestrial ecosystems. Am. Zool. 8:l9-30. Woodwell, G. M., R. A. Houghton, and N. R. Tempel (l973a). Atmospheric C02 at Brookhaven, Long Island, New York: patterns of variation up to l25 meters. J. Geophy. Res. 78:932-940. Woodwell, G. M., P. H. Rich, and C. A. S. Hall (l973b). Carbon in estuaries. In Carbon in the Biosphere, G. M. Woodwell and E. V. Pecan, eds. Proceedings of the Twenty-fourth Brookhaven Symposium in Biology, Upton, New York. USAEC, Office of Information Sources. NTIS, Springfield, Va., pp. 22l-240.

24l Woodwell, G. M., R. H. Whittaker, W. A. Reiners, G. E. Likens, C. C. Delwiche, and D. B. Botkin (l978). The biota and the world carbon budget. Science l99:l4l-l46. Woodwell, G. M., R. A. Houghton, C. A. S. Hall, D. E. Whitney, R. A. Moll, and D. W. Juers (l979). The Flax Pond ecosystem study: the annual metabolism and nutrient budgets of a salt marsh. In Ecological Processes in Coastal Environments, R. L. Jeffries and A. J. Davy, eds. The First European Ecological Symposium and the Nineteenth Symposium of the British Ecological Society, Norwich, September l977. Blackwell Scientific Publications, Boston, Mass., pp. 49l-5ll. Woodwell, G. M., J. E. Bobbie, R. A. Houghton, J. M. Melillo, B. Moore, C. A. Palm, B. J. Peterson, and G. R. Shaver (l982). Report of the Woods Hole Conference on the Biotic Contributions to the Global Carbon Cycle at the Ecosystems Center. Ecosystems Center, Marine Biological Laboratory, Woods Hole, Mass., March l982. Woodwell, G. M., J. E. Hobbie, R. A. Houghton, J. M. Melillo, B. Moore, B. J. Peterson, and G. R. Shaver (l983a). The contributions of global deforestation to atmospheric carbon dioxide. Photocopies available from the Ecosystem Center, Marine Biological Laboratory, Woods Hole, Mass. Woodwell, G. M., J. E. Hobbie, R. A. Houghton, J. M. Melillo, B. J. Peterson, G. R. Shaver, T. A. Stone, B. Moore, and A. B. Paru (l983b). Deforestation measured by LANDSAT: steps toward a method. DOE/EV/l0468-l. NTIS, Springfield, Va. Zelitch, I. (l97l). Photosynthesis, Photorespiration, and Plant Productivity. Academic, New York.

242 3.4 THE ATMOSPHERE Lester Machta 3.4.l Introduction It is the growing concentration of CO2 in the atmospheric reservoir that has attracted most attention to the CO2 issue. The pre- Industrial Revolution (e.g., l850) concentration probably lay in the range 250 to 295 ppmv (parts per million by volume or mole fraction). Measurements by chemical analysis (Callendar, l958; Keeling, l978) and extrapolations backward based only on records of fossil fuel emissions suggest a late-nineteenth-century concentration at the upper end. Measurements from ice cores (Neftel et al., l982; Oeschger, l983) and reconstructed ocean measurements (see Brewer, Section 3.2) suggest preindustrial concentrations at the lower end. A WMO-sponsored Meeting of Experts in June l983 concluded the most likely mid-nineteenth-century concentration was between 260 and 280 ppm, based on consideration of all the various estimates including carbon isotope data in tree rings (the meeting report will be issued at a later date). Concentrations significantly less than 290 ppm imply the existence of a large nonfossil fuel source of C02 and are thus consistent with a large early input from disturbances of the biosphere. By l980 the atmospheric C02 concentration had risen to about 340 ppmv. The behavior of C02 in air is simpler and better understood than in the other two major reservoirs—the land biosphere and the oceans. C02 is conservative in air, that is, it is not subject to chemical transformation at least up to an altitude of about 60 km. It moves with the other inert air molecules with which it is embedded. Most, if not all, of the known variations of CO2 in time and space in the air appear to follow known meteorological principles. Since interest often focuses on time scales of years to decades, as a first approximation, it is usually acceptable to treat the whole atmosphere as a single well-mixed box and apply first-order kinetics to the CO2 transfer to other reservoirs. The growth of C02 in the air can be demonstrated at almost any location on Earth over a period of several years. Modern-day measure- ments were begun by C. D. Keeling of Scripps Institution of Oceanography during the l957-l958 International Geophysical Year. Stations were established at the South Pole and at ll,l50 feet aside Mauna Loa in Hawaii. The latter, the better record, is reproduced in Figure 3.l4. Since the pioneering measurements of Keeling, other stations operated by many countries and organizations have been established. A map of the location of stations as of July l982 as supplied by the World Meteorological Organization (WMO) appears as Figure 3.l5. At most of the stations air is collected in containers for subsequent analysis J.n central laboratories. With few exceptions, both on-station and lab- oratory analyses are performed by nondispersive infrared analysis comparing the ambient samples with standard gases. Since the response of the analyzer depends on the carrier gas, it is now agreed that the carrier gas of the standard should duplicate air as closely as possible,

243 e.g., nitrogen, oxygen, and argon instead of the former nitrogen gas. A transition to standards of CO2 in air (or simulated air) now in progress around the world should proceed as quickly as possible and be consistent with ensuring long-term integrity of the standards. Carrier-gas corrections based on CO2-in-N2 standards are required to bring concentrations closer to their true values. 3.4.2 Changes in Atmospheric CO? Growth Rate with Time and Space There appear to be two kinds of changes in the year-to-year growth rate at Mauna Loa (Figure 3.l4): a shorter- and longer-term variation. 3.4.2.l Shorter-Term Variation and Its Possible Cause •* •. Here we follow the analysis of Machta et al. (l977); the same result is arrived at by the analysis of Bacastow (l976) and Newell and Weare (l977), although a different interpretation is offered by them. The monthly mean concentrations at a given station exhibit a seasonal oscillation and a long-term trend, both of which can be removed mathe- matically. The resulting monthly values of two stations, in this case o I- er LU O O o 340 335 330 325 320 315 310 Q Eit1maud Valua ( } Based on Est1mated Data 1958 1960 1965 1970 1975 19801981 (.66) .94 .70 .90 .50 (.69)(.581.67 .6 .71 .94 1.90 1.33 .87 1.222.22 .57 .601.06 1.57 1.571.351.81 ANNUAL CHANGE (ppmv/yr) FIGURE 3.l4 Mean monthly concentrations of atmospheric C02 at Hauna Loa.

244

245 RESIDUALS FROM FIT TO AQ + A,*bt 400 300 — _ 200 100 Annual Consumption of Fossil Fuel o Carbon Dioxide (x 1012gCO2) -100 -200 _— -300 f • :, .:;^:. . . .j,-' ^ 0 \. . • — • -j—f, -•- "• • - «—*-*• -v *- (ppm) *•.'• > -1 — • •— — • •— — « •— i El Nino 1 -A -• 1+ -^^ Xuw- A -"~^ ^ '"'.-V1' •" ^V -t/t- (ppm) • *,'»1• -1 — i i i i i i i i i i i *i i i i i i i 1958 1960 1962 1964 1966 1968 1970 1972 197419 YEAR FIGURE 3.l6 Time history of residuals of monthly carbon dioxide concentrations after removing the long-term upward trend and seasonal variability at Mauna Loa and the South Pole (lower section). The horizontal lines in the upper section represent the residuals of the annual consumption (actually production) of fossil fuels after removing the long-term upward trend. The periods of El Nino are also shown in the lower section. The horizontal lines in the lower section among the circles and crosses are the annual average residuals. Mauna Loa (small circles) and the South Pole (crosses), appear in Figure 3.l6. This figure shows a pattern of short-term variation with fluctuations reversing themselves every 2 to 6 years. The horizontal bars among the circles and crosses represent the annual values of the l2 monthly averages. The horizontal bars in the upper part of Figure 3.16 represent the departures from the best exponential fit of the annual fossil fuel emissions of CO2 from the mean value for the l7 years. Variation in annual emissions might be the reason for the year-to- year fluctuation in Figure 3.l6 of the atmospheric CO2 content; however, the correlation coefficient between concurrent anomalies of emissions and atmospheric content is only 0.42, not statistically significant. Allowing for a lag of up to 6 years between emissions and atmospheric C02 content simply reduces the correlation coefficient

246 below 0.42. While year-to-year fluctuations in fossil fuel combustion contribute to variability in atmospheric CO2 concentrations, other factors appear to dominate the variability. There seems to be a good correspondence between increasing anomalies, the open circles, and the periodic variations in the southern hemisphere oceans and atmosphere known as El Nino events, shown as the horizontal bars between the record at the two stations. A similar relationship can be found between the anomalies in atmospheric CO2 content and tempera- ture in the tropical eastern Pacific Ocean and with the Southern Oscil- lation. An analysis using a two-dimensional transport model (vertical and north-south directions) suggests that the lag of changing concentra- tions among stations (Mauna Loa, the South Pole and Australia) fits a ground- or sea-level source or sink of (X>2 (i.e., a forcing function) near 5 to l0° S, the region of the El Nino. But the cause of the forcing function in the tropical Pacific is less clear. The warmer sea-surface temperatures associated with the El Nino could produce a higher-than-normal partial pressure of CO2, enhancing the tropical oceanic source; the warmer temperatures also reflect lesser upwelling, which reduced the oceanic biological activity, which, in turn, can affect the air-sea CO2 difference in the same sense as the warmer water (e.g., a smaller biosphere will take up less atmospheric CO2); and finally, the changing wind speeds related to the Southern Oscillation can also alter the rate of air-sea exchange of CO2. Thus, the above analysis, while leaving questions about the cause of the forcing function unanswered, does indicate that the shorter-term (2 to 6 years) variations in atmospheric CO2 concentrations are empirically correlated to some phenomena in the eastern tropical Pacific Ocean (El Nino, Southern Oscillation, biota change). 3.4.2.2 Longer-Term Variations An inspection of year-to-year increases in concentration at Mauna Loa reveals that they are generally becoming larger with time. Figure 3.l4 shows this trend in the values of annual changes in ppmv/yr. Through l968 the annual increase was below l ppmv, while in recent years it has often been nearer to l.5 ppmv. The emissions of man-made CO2 (mainly fossil fuel combustion plus cement manufacture and flaring of natural gas*) are also becoming larger with time. Elliott (l983) estimates an overall annual growth rate of emissions from industrial activity of about 3.5% over the past l20 years, with wide variation due to economic fluctuations, and Nordhaus and Yohe (this volume, Chapter 2, Section 2.l) predict that emissions from fossil fuels will most likely grow at a rate of about l% or 2% per year over the next hundred years. But there are other likely sources for atmospheric CO2; in particular the CO2 produced as a result of deforestation. Elliott and Machta (l98l) *Hereafter, the term "fossil fuel C02" is understood to include the other two minor contributions as well.

247 have sought to determine whether the Mauna Loa and South Pole records of CO2 increase (the average of the two is taken as the average for the Earth in this analysis) are better fitted by the increasing fossil fuel combustion source of CO2 or whether the addition of another significant source, such as from deforestation, results in a better fit to the measurements. In principle, if there were an accurate model of the carbon cycle, one might enter alternate amounts of CO2 into it, and the best fit to the observed data would be the best size of the source to fit observed concentrations. Elliott and Machta (l98l) have tried to avoid the issue of defining the carbon cycle. They assume that each year roughly the same fraction of that year's CO2 emissions remains airborne. This airborne fraction is determined from the ratio of the C02 increase in the atmosphere (as found from the average annual increases at Mauna Loa and the South Pole) to the amount added to the air from all sources that one wishes to assume. The interval of this analysis is from l958 to l98l. The result indicates that the observed increases in atmospheric CO2 are best fitted by only a fossil fuel source, without any additional constant or random CO2 source. In fact, the analysis suggests that there could be a small loss of CO2 from the atmosphere, possibly through CO2 fertilization of photosynthesis because of the elevated atmospheric CO2 concentration. There is one caveat, however. This analysis could not distinguish between fossil fuel and nonfossil fuel emissions, such as deforestation, that are increasing at the same rate. It does, however, provide estimates of how closely the growth of the nonfossil source had to match the fossil fuel growth to be undetectable. For sources averaging more than 2 Gt of C per year, the growth rate would have to have been quite close to the fossil fuel growth rate to be undetected. 3.4.2.3 Change in Annual Cycle The annual cycle shown in Figure 3.l4 for Mauna Loa C02 concentrations is almost certainly the result of the warm season uptake of CO2 during land biosphere photosynthesis (see Woodwell, Section 3.3). There could be a contribution or diminution from the seasonality of fossil fuel com- bustion or the air-sea exchange of C02. The seasonality of the latter processes are believed to be small in determining the annual cycle of atmospheric C02 concentrations. The long, high-quality record at Mauna Loa has been analyzed to determine whether the amplitude of the annual cycle is changing with time. The result is shown in Figure 3.l7. The increase of amplitude suggested by a best-fit line, the dashed line, in large part depends on an increase that occurs only during the past 6 years. The finding of increasing amplitude at Mauna Loa is supported by analysis of a shorter and less convincing C02 record at the Canadian ship Papa located in the Gulf of Alaska, according to Keeling (l983). The most plausible explanation of the increasing amplitude is increased biological activ- ity, such as, but not necessarily, a larger temperate and high-latitude biosphere (e.g., larger forests) (c.f., Woodwell, Section 3.3).

248 120 110 olOO CE 90 80 L 110 r Relative Amplitude of the Seasonal Cycle STN 'P' 50° N SLOPE= 0.77 %/YR 100 o 90 80 \ SLOPED 0.66% /YR MAUNA LOA (19° N) 1958 '60 '62 '64 '66 '68 '70 '72 '74 '76 '78 '80 (b) YEAR FIGURE 3.l7 Seasonal amplitude in atmospheric CO2 concentration at (a) Weather Ship P at 50.0° N and (b) Mauna Loa Observatory at l9.5° N. Dots connected by solid lines represent an estimation of the amplitude for individual years as determined by a best fit of a four-harmonic seasonal cycle as described by Bacastow et al. (l98l). The dashed straight line is a least-squares fit of a linearly increasing amplitude over the entire period of record. (Source: Keeling, l982.) 3.4.2.4 Spatial Distribution Keeling (l982) has provided north-south profiles of ground-level air concentrations of CO2 relative to that at the South Pole (see Brewer, Section 3.2). Profiles for 3 years, all adjusted to a common value at the South Pole are shown in Figure 3.9 in Section 3.2. It is evident that the secular growth in the northern hemisphere exceeds that in the southern hemisphere, as might be expected from the location of fossil fuel CO2 sources mainly in the northern hemisphere. The concentration in each of the 3 years is also higher in the northern than southern hemisphere. Near the equator there is a secondary peak, which might be due to either the release of CO2 from the tropical oceans or defor- estation in the tropics. Keeling has estimated the transfer from the sea to air and compared this with the equatorial peak in Figure 3.l8. His conclusion is that CO2 from tropical deforestation during l962-l980 is unlikely to exceed l or 2 Gt of C per year or no more than

249 3.0 I I I i i ri| 1980 1968 1962 ll I I I I I I 90° S 40° S 20° S 0° 20° N LATlTUDE (deg) 40 N 90° N FIGURE 3.l8 North-south profile of ground-level air concentration relative to that at the South Pole for 3 years. 40% of the current CO2 released from fossil fuel combustion and likely much less. 3.4.2.5 Isotopic Content of Atmospheric CC>2 Isotopic (^C) analyses of atmospheric CO2 samples have been undertaken in a systematic fashion in only the past few years so that conclusions derived from these data must still be viewed with caution. Keeling (l983) contends that the seasonal cycle in 13C measurements are consistent with land plants being the primary source of the annual cycle of CO2 concentration for northern hemisphere stations. At the South Pole the isotopic data suggest an oceanic source for the cause of the much smaller annual cycle. Tentative results from Keeling from Fanning Island in the tropical Pacific Ocean and during oceanographic cruises in the tropics (the First Global Atmospheric Research Program Global Experiment expedition) support the contention that the peak COo concentration in the equatorial region in Figure 3.l8 is the result of air-sea exchange and not due to deforestation in the tropics. Leavitt

250 and Long (l983), while allowing for other interpretations, report that the shape of the best-fit reconstruction of 50 yr of l3C/12C measure- ments from tree rings suggests that the biosphere has acted as a CO2 source to about l965 but has become a sink afterward. 3.4.3 Conclusions Atmospheric C02 data provide information far beyond the single obser- vation that atmospheric CO2 is increasing. Through careful measure- ments, one is able to derive valuable information from the temporal and spatial variability. The pattern of results is highly suggestive of a minimal contribution of nonfossil fuel sources of CO2. Globally, during the past 20 years, most of the variations are more readily accounted for by the growing fossil fuel source alone than from any significant additional source from, say, deforestation. Both the limited quantity of data and the possibility of alternate explanations prevent any definitive statement today that excludes nonfossil fuel sources. The need to continue quality observations cannot be overemphasized. The spatial gradients of atmospheric CO2 are so small that the total minimum to maximum concentration at a clean air location for the entire globe is no more than about l% of the mean global concentration. Data collected with very high precision are needed to detect such small gradients. Historical and geologic data from past records such as from ice cores have proven to be very valuable, and an expanded effort to confirm the previous findings (about 200 ppmv found l8,000 years ago) should be undertaken. The isotopic studies in tree rings and from current air samples offer potential to elucidate further the carbon cycle and the contribution of the nonfossil fuel C02. For example, following the fate with time of the nuclear weapons test l4CO2 in the atmosphere can continue to provide new information on the atmospheric residence time of the fossil fuel C02. References Bacastow, R. B. (l976). Modulation of atmospheric carbon dioxide by the Southern Oscillation. Nature 26l:ll6. Bacastow, R. B., C. D. Keeling, and T. P. whorf (l98l). Seasonal amplitude in atmospheric CO2 concentration at Mauna Loa, Hawaii l959-l980. In papers presented at the WMO/ICSU/UNEP Scientific Conference on Analysis and Interpretation of Atmospheric CO2 Data. World Meteorological Organization, Geneva, Switzerland, pp. l69-l76. Callendar, G. S. (l958). On the amount of carbon dioxide in the atmosphere. Tellus l0:243-248. Elliott, W. P. (l983). A note on the historical industrial production of carbon dioxide. Climate Change 5:l4l-l44.

25l Elliott, W. P., and L. Machta (l98l). In papers presented at the WMO/ICSU/UNEP Scientific Conference on Analysis and Interpretation of Atmospheric CO2 Data. World Meteorological Organization, Geneva, Switzerland, p. l9l. Keeling, C. D. (l978) . Atmospheric carbon dioxide in the l9th century. Science 202:ll09. Keeling, C. D. (l983). The global carbon cycle: what we know and could know from atmospheric, biospheric, and oceanic observations. In Proceedings, CO? Research Conference: Carbon Dioxide, Science, and Consensus, Berkeley Springs, West Virginia. CONF-820970. NTIS, Springfield, Va. 22l6l. Leavitt, S. W., and A. Long (l983). An atmospheric l3C/l2C reconstruction generated through removal of climate effects from tree-ring l3C/l*C measurements. Tellus 35B:92-l02. Machta, L., K. Hanson, and C. D. Keeling (l977). Atmospheric carbon dioxide and some interpretations. In The Fate of Fossil Fuel CO? in the Oceans, N. R. Andersen and A. Malahoff, eds. Plenum, New York, pp. l3l-l44. Neftel, A., H. Oeschger, J. Schwander, B. Stauffer, and R. Zumbrunn (l982). New measurements on ice core samples to determine the CO2 content of the atmosphere during the last 40,000 years. Nature _295:220-223. Newell, R. E., and B. C. Weare (l977). A relation between atmospheric carbon dioxide and Pacific sea-surface temperature. Geophys. Res. Lett. 4:l-2. Oeschger, H., and M. Heimann (l983). Uncertainties of predictions of future atmospheric CO2 concentrations, J. Geophys. Res. 88:l258.

252 3.5 METHANE HYDRATES IN CONTINENTAL SLOPE SEDIMENTS AND INCREASING ATMOSPHERIC CARBON DIOXIDE Roger R. Revelle 3.5.l Methane in the Atmosphere About 4.8 Gt of methane (CH4) are present in the Earth's atmosphere, corresponding to l.7 ppm by volume (see Machta, this volume, Chapter 4, Section 4.4). Methane is a strong absorber of infrared radiation in the part of the atmospheric "window" centered around a wavelength of 7.66 um. According to Lacis et al. (l98l), a doubling of the atmospheric methane concentration would cause an increase in global average surface temperature of 0.4l°C. Chamberlain et al. (l982) estimate a larger value, 0.95°C for methane doubling and report lower and higher results by other groups as well. These calculations allow for positive feedbacks resulting from the increase in absolute humidity with rising temperatures and the consequent higher infrared absorption by water vapor, decreases of planetary albedo due to melting of snow and ice, and assumed cloud behavior. Chamberlain et al. estimate that methane is now being added to the atmosphere at a rate between 0.5 and l.0 Gt per year, primarily from anaerobic fermentation of organic material in rice paddies, swamps, and tundras, plus enteric fermentation in the digestive tracts of ruminant animals. Anaerobic fermentation in the guts of termites, which contain cellulose-digesting symbiotic bac- teria, is probably also a significant source. Some methane is being added to the ocean-atmosphere system from vents in the rift zones of the ocean floor (Welhan and Craig, l979) and perhaps of East Africa (Deuser et al., l973). As we shall see, ocean sediments on the con- tinental slopes may be a relatively small source now but an important source in the future (MacDonald, l982a). Methane is removed from the lower atmosphere by a reaction with hydroxyl (HO) and is eventually oxidized to CO2. With the above estimate of the rate of input of 0.5 to l.0 Gt per year, the residence time in the air should be between 5 and l0 years. Measurements indicate that the methane content of the air is increasing perhaps by 0.07 Gt per year, or about l.4% per year, doubling in 70 years (Rassmussen and Khalil, l98l; Craig and Chou, l982). Part of this increase may be the direct result of such human actions as expansion of the area of rice paddies to meet the food needs of growing populations. A small part may represent release of methane from methane hydrates in continental slope sediments as the ocean responds to atmospheric warming. MacDonald (l982a) defines methane hydrate as a "type of clathrate in which methane and smaller amounts of ethane and other higher hydro- carbons are trapped within a cage of water molecules in the form of ice." Though it is not a stoichiometric compound, about 6 mol of water are required for l mol of methane in the clathrate. Methane hydrates are stable at low temperatures and relatively high pressures (Claypool and Kaplan, l974; Miller, l974). They are found at depths between 200 and l000 m below the ground surface in permafrost (Kvenvolden and

253 McMenamin, l980; Chersky and Makogon, l970) and should be present near the surface of marine bottom sediments below water depths of 290 to more than 800 m, depending on bottom-water temperature. Within the sediments the thickness of the clathrate zone will be limited by the geothermal gradient of about 30°C km"l, which reflects heat conduc- tion from the interior of the Earth. As Bell (l982) has demonstrated, even with a CO2-induced rise in surface air temperatures of around l0°C, virtually none of the clathrate in permafrost would become unstable during the next several hundred years because the surface heating of the "frozen ground" would first have to penetrate and melt the permafrost in a 200-m-thick clathrate- free zone. The enthalpy (latent heat of fusion) of the ice in perma- frost would greatly slow the downward penetration of the heat wave. But with a rise in ocean-bottom temperatures, the uppermost layers of sediments would also become warmer and methane hydrates would become unstable in the upper limit of their depth range, that is, about 300 m in the Arctic and about 600 m at low latitudes. 3.5.2 Formation of Methane Clathrate in Continental Slope Sediments The quantity of clathrates that will be released from sediments under the seafloor as a result of ocean warming depends on the distribution of clathrates with depth and on their total abundance in the sediments. Estimates of total abundance by different authors differ by a factor of 500, from l03 to 5 x l05 Gt (MacDonald, l982b). If the methane locked up in clathrates were produced by anaerobic fermentation of organic matter in the sediments, one would expect that most oceanic clathrates would be found in deep semienclosed basins and on continental slopes, particularly on passive continental margins such as those on both sides of the Atlantic. The rate of sedimentation on continental slopes is relatively high (of the order of l0 to 20 cm/l000 years), and samples taken from near the surface of the deposits are high in organic matter—on the average about 2% of the dry weight (Trask, l932). This organic matter is some- times called "marine humus." It consists mainly of the partially decomposed tissues of marine plankton and nekton and to a lesser extent of the remains of terrestrial plants. On average, according to Trask, the carbon content of the organic matter is 56%; hence, organic carbon averages l.l2% of the dry weight of the uppermost layers of sediments.* The average density of the dry material is 2.6 g caT*. *J. G. Erdman and colleagues of the Phillips Petroleum Company measured the organic carbon content of many samples collected by the Deep Sea Drilling Project from outer continental margins. The results were published in the Initial Reports of the Deep Sea Drilling Project (Volumes XXIV, XXXI, XXXVIII, XL, XLI, XLII, XLIII, XLIV, XLVII, XLVIII, L, and LXVI, published between l975 and l98l, inclusive). Seventy-three samples of Quaternary to late Pliocene age from the (continued overleaf)

254 These deposits are very porous; cores of freshly collected mud usually contain two thirds water by volume. Thus, an average liter of mud from near the surface of the deposits will contain 650 g of water and 870 g of solids, including silicate mineral grains, fragments of calcareous and siliceous skeletons, and shells, and about l7 g of organic matter containing close to l0 g of carbon. Because of the relatively high rates of deposition and the abundance of decomposable organic material, free oxygen in the interstitial water of these sediments is rapidly depleted as they are buried, and "reduc- ing" conditions prevail a short distance below the seafloor. The principal living organisms under these conditions are anaerobic bac- teria, which are able to carry out their metabolic activities in the absence of free oxygen. Dissolved sulfate in the interstitial water will first be reduced to sulfide, and a small fraction (less than 0.5 g per liter) of organic carbon will be oxidized to C02. All the sulfate is usually depleted in the top meter of the sediments. Beneath this top layer, methane is produced (Claypool and Kaplan, l974). Below water depths of 300 to 600 m, depending on bottom-water temperature, methane in excess of the quantity that can be dissolved in the interstitial water will be converted to methane hydrate as soon as it is formed. There are no measurements of the actual methane concentration in deep-sea muds in situ. In several cores from areas of rapid deposition collected by the Deep Sea Drilling Project (DSDP), gas was observed escaping when the core liners were removed from the bore barrels on board Challenger. Some mud was ejected from the liners by the force of the escaping gas. Subsequent analysis showed this gas to be almost entirely methane (Mclver, l974). Presumably, most of the methane escaped from the samples while they were being raised from the seafloor and handled on deck, but the remaining amounts were surprisingly high—up to l5 mmol of methane per liter of interstitial water. The organic carbon content of DSDP sample with high remaining gas content after shipboard handling (presumably those in which the methane was originally present as a clathrate) ranged from 0.28 to l.l4%— averaging 0.62% by dry weight (Mclver, l974). Assuming that this organic carbon represents the residue of organic matter after methano- genesis has been completed, and that the proportion of residual carbon to carbon in methane produced is roughly the same (l:0.5l) as in the (continued from overleaf) northern Indian Ocean, eastern Pacific, north and south Atlantic, and the Japan, Mediterranean, and Black Seas have an average organic carbon content of l.4% of the dry weight of the sediments. The depths beneath the sediment surface ranged from less than l to l000 m, with most of the samples being from 30 to 300 m below the top of the sediment; the average depth of the overlying water was around 2000 m. Presumably the measured organic carbon represents the residue after sulfate reduction and methanogenesis. Erdman believes that Trask's analysis of surface sediment gave low results because the samples were poorly preserved.

255 "biogas" digesters described by Makhijani and Poole (l975), the cal- culated average methane content of these muds in situ is 0.42% by weight of dry sediment, or 3.6 g per liter of wet mud. The correspond- ing concentration of methane in the interstitial seawater would be about 330 mmol kg"l. The biochemical processes in the buried sediments are not well under- stood and may be quite different from those in methane-producing biogas digesters. J. G. Erdman (Bartlesville, Oklahoma, personal communica- tion) has pointed out that the microbial population in marine sediments drops rapidly with depth in the sediments and that the organic matter, unlike terrestrial biomass, is relatively lean in the hydrolyzable con- stituents of plant materials that ferment easily. Erdman is convinced that the mass of methane hydrate in marine sediments is larger than the amount calculated above. He believes that most of this methane was formed from sedimentary organic matter by thermolytic processes under heat and pressure at substantial depth in the sediments. The methane then migrated upward until it became trapped in the zone of methane hydrate stability in the upper sedimentary layers. This hypothesis has the significant advantage that it does not require the methane in the upper part of the hydrate zone to have formed since the last inter- glacial approximately l25,000 years ago, when subsurface ocean warming may have been as great as that expected with a doubling of atmospheric carbon dioxide. An estimate of the minimum concentration of methane can be made from the inferred existence of methane clathrate in muds from the Blake Plateau off the southeastern coast of the United States at a total depth (overlying water plus depth in the sediments) of about 4000 m (Stoll et al., l97l; Bryan, l974). In order for a clathrate to form, the concentration of dissolved methane must have been close to 64-69 mmol kg"l of interstitial water, which is the solubility of methane at a hydrostatic pressure of 400 atm (Claypool and Kaplan, l974). This concentration is about 20% of that calculated above by comparison with observed methane production in biogas digesters. For our present purposes, we may assume that the concentration of methane in continental slope muds is halfway between these two esti- mates, say 200 mmol kg"l of interstitial water, or 2.2 g of methane per liter of mud. If all of this methane is present as clathrates, about l200 mmol kg"l of water will also be in the same state or 2l.6 g kg"l of interstitial water. This is 3.2% of the water in an average mud. Miller (l974) shows that in seawater, the minimum hydrostatic pressure (P) at which methane hydrate is stable between temperatures (T) of 0° and l0°C is given by ^ogl0 P (atmospheres) = l.46l3 + 0.04l6T + 2.93 x l0-4 (T)2, where P (atmospheres) is the partial pressure of methane, which is equal to hydrostatic pressure when the water is saturated with methane. The water depths below which methane hydrate in the uppermost layers of marine bottom sediments will be stable at different temperatures can be calculated from this equation, with the simplifying assumption that

256 hydrostatic pressure in the ocean increases by approximately l atm for each l0 m of depth: Bottom-Water Minimum Depth of Temperature (°C) Clathrate Stability (m) _ _ 289 l 3l9 2 35l 3 388 4 429 5 475 6 528 7 588 8 650 9 724 l0 807 3.5.3 Effect of Carbon Dioxide-Induced Warming on Continental Slope Clathrates With carbon dioxide-induced warming of the atmosphere, ocean surface temperatures will rise by a nearly equal amount, and heat will be carried downward by advection and eddy diffusion into the subsurface water layers. For a doubling of atmospheric CO2 and the expected increase in other "greenhouse gases," with an assumed sensitivity of global average temperature of 3°C for a CO2 doubling, the temperature increase in different latitudes at the water depths below which methane hydrate is stable at present can be estimated (see this volume, Chapter 8, Section 8.3). The corresponding increases in clathrate stability depths are shown in Table 3.6. The depth of melting of the clathrate below the sediment-water interface at any time after the bottom-water temperature is raised will be much smaller than the depth at which warming will occur in the absence of clathrate. When bubbles of methane are formed, the latent heat (enthalpy) of vaporization of the methane must be added to that for melting of the water-ice in the clathrate. Miller (l974) has calculated that with 6 mol of water per mol of methane, the combined enthalpies are l20 cal g 1 of H20 in the clathrate. Because most of the water in the sediment remains liquid even after clathrate has formed, we can compute a "virtual" latent heat, L, required for the depth of wetting to advance downward by l cm. L = 0.032 x l20 + 0.968 x l = 4.8 cal g"l of The problem of the rate of downward advance of the depth of melting has been solved by W. H. Munk (La Jolla, California, personal communication). He shows that the depth h (cm) of the melted zone at time t (sec) after an instantaneous rise in water temperature at the sediment-water interface is given approximately by,

257 h - a (ict)l/2, where ic(cm2 sec"l) » K/pC; K is thermal conductivity of wet sediment, in cal sec"1 cm"l "C"l; p is density of wet sediments, g cm-^j C is specific heat, cal g"l °C- , and r- -i l/2 [2C 1 a(dimensionless) « — (T, - T ) L ra J ; L is "virtual" enthalpy of melting for clathrate in sediments, cal g"l of H2O in both clathrate and liquid fractions of interstitial water; Tl - Tm is temperature at the sediment-water interface minus temperature of melting of clathrate. With the following approximate numerical values: K = 4 x l0"^ (calculated from average geothermal heat flow through the seafloor and estimated temperature gradient in sediment: K = l.3 x l0-^ •=• 3 x l0-4),p = l.54, T! - Tm = l, C = l, L = 4.8, Ic = 2.6 X l(T3, and a = 0.65, then h = 0.033l(t)l/2 cm. After an instantaneous rise of bottom-water temperature of l°C, the depth of melting would advance downward l8.6 m beneath the sediment- water interface during l00 years, or an average of l8.6 cm yr"l. An average of 0.04l g of methane per cm of seafloor should be released each year in the area where clathrates have become unstable. The rate of exchange of dissolved materials between the sediments and the overlying water is very slow, and consequently methane will probably not diffuse significantly out of the sediments until its con- centration becomes high enough to form bubbles of the gas. According to Miller (l974), bubbles of methane will form at a hydrostatic pres- sure of 400 atm when the concentration exceeds 350 mmol liter"l. Provided the relation between the concentration required for bubble formation and hydrostatic pressure is approximately linear, the required concentration at about 500 m will be 44 mmol. With our assumed concen- tration of 200 mmol kg"l of interstitial water, close to 80% of the methane released from clathrate should escape from the mud in bubbles and should rise rapidly to the sea surface before it can be oxidized in the water. 3.5.4 Future Rate of Methane Release from Sedimentary Clathrates From Table 3.6 we see that the average depth interval on the continental slope over which the clathrates will become unstable with a C02 doub- ling is about l00 m. Using Kossina's l92l estimate (see also Menard and Smith, l966) that the area of seafloor between 200 and l000 m is l5.5 x lOl^ cm2, and assuming that the depths of continental slopes increase linearly in this depth range, the area in which clathrates will become unstable is l.94 x l0l(* cm2, and the total quantity of methane released annually as bubbles in the bottom muds should be 0.8 x 0.04l x l.94 x l0l6 g, or, 0.64 Gt (l09 metric tons). With an atmospheric residence time of 5 to l0 years, the quantity of methane in

258 C >i i •H SH •H 1 a JQ n 1 41 a 41 u *• c 0 W 0 H 41 Bam 41 «J -H -o i O1 V I 3 i0 JB JS *1 u 4-t H -H 1 41 10 CU *J — E > rH oi M a iTiinor-[--r-ooao «C o Q 3 ~- 01 01 rH rH rH 25 5 u 10 O rH iM 4I U k l iM 4-i 4J ° H * 5 2 S, i Ql M • «*« ssssssssss i 1 41 Q Ul EH ^-" ijo VD r^ r*- vo r^ ^ 1/1 ^ 8 « 0 3 -M 4J O «* * -H 1 SU u oi 4I n ! Qi A 41 ,9 H OirH v oi 01 ja \ 4J 4J OJu- m M n_i f* _ ocoooinmoor- i cor^otaicoatoor^e*> i oi 1 § * a ! 4J u 4J 0 & °* 8|l t 4J EH W C -H » ) * « " oooommooo ! 0< S « — • *mr>«oi-«oc-^M > > i 4J 41 a •" S y >i i *J 3 -rl -C -tJ C « rH 41 >H -H • U 4 • o 41 10 *- U ij ^J J-J S 3«Sw«8S«M u O. O 01 -- 1 k l 4I 41 r* ! CU 9 zzzzzncacnca | 3 •I eeeeeeeee 3 ooooooooo O o

259 the air could rise by 3.2 to 6.4 Gt, two thirds to four thirds of the present amount. This quantity would be in addition to most of the present rate of increase of about 7 Gt/century. The corresponding increase in global average su'rface temperature from the methane "green- house" effect toward the latter part of the twenty-first century could be 0.65 to 0.8°C if we accept the estimate of Lacis et al. (l98l) for the equilibrium warming resulting from a doubling of methane, and l.5 to l.8°C, using the estimate of Chamberlain et al. (l982). As the C02 produced by oxidation of the methane accumulates in the air, there will be a slight further rise in temperature, of the order of 0.l to 0.2°C in a hundred years. Bell (l982) assumed that a CO2-induced warming of the Arctic Ocean would release methane hydrates during l00 years from the top 40 m of the bottom sediments over half the area between water depths of 280 and 370 m. The estimated bottom area between these depths in the Arctic Ocean and Arctic Mediterranean is 0.240 million knr or about l.5% of the area between 200 and l000 m in the world ocean (Menard and Smith, l966). Bell assumes that l0,000 Gt of C as methane hydrate are uniformly distributed in the top 250 m of the sediments over the world ocean area between 200 and l000 m. The total release from the Arctic Ocean regions in l00 years would then be l2 Gt of C or 0.l2 Gt/yr. (Bell's figure of 8 Gt of C/yr is an obvious computational error.) Obviously, the uncertainties in calculations of release of methane from continental slope sediments are so great that the results cannot be thought of as a projection for the future. But it is equally obvious that an extensive sediment sampling program on continental ONSHORE OFFSHORE FIGURE 3.l9 Reported occurrences of natural gas hydrates. (Updated from Kvenvolden and McMenamin, l980.)

260 slopes throughout the world should be undertaken to determine the depth, thickness, and distribution of methane hydrate clathrates, especially where oceanfloor temperatures and depths are such that methane release is likely from ocean warming during the next century. A small release of methane clathrates may already be taking place as a consequence of the estimated 0.5°C increase of global air temperature (Hansen et al., l982; Jones et al., l982; Weller et al., this volume, Chapter 5, Section 5.2) during the last century, and the probable increase of ocean-bottom temperatures by 0.l-0.2°C from eddy diffusion and advection down to 500 m. Some indications that clathrates may be widespread and abundant in ocean sediments are given in Figure 3.l9 (from MacDonald, l982b). This shows the locations in the continental slopes of the ocean floor and in the Black and Caspian Seas, in which the existence of methane clathrates has been inferred from high gas contents in cores of the Deep Sea Drilling Project or in which their presence is suspected from acoustic reflections from a layer below the sediment surface that parallels the ocean-bottom topography (Bryan, l974; Stoll et al., l97l). It is believed that these are reflections from the bottom of the methane clathrate zone, although other explanations are possible. The likelihood of the widespread occurrence of clathrates in continental slope sediments gives force to our argument that a systematic survey should be made in an attempt to determine their abundance and distribution. References Bell, P. R. (l982). Methane hydrate and the carbon dioxide question. In Carbon Dioxide Review: l982, W. C. Clark, ed. Oxford U. Press, New York, pp. 40l-406. Bryan, G. M. (l974). In situ indications of gas hydrate. In Natural Gases in Marine Sediments, I. R. Kaplan, ed. Plenum, New York, pp. l5l-l70. Chamberlain, J. W., H. M. Foley, G. J. MacDonald, and M. A. Ruderman (l982). Climate effects of minor atmospheric constituents. In Carbon Dioxide Review: l982, W. C. Clark, ed. Oxford U. Press, New York, pp. 255-277. Chersky, N., and Y. Makogon (l970). Solid gas—world reserves are enormous. Oil and Gas Internat. l0:8. Claypool, G. E., and I. R. Kaplan (l974). The origin and distribution of methane in marine sediments. In Natural Gases in Marine Sediments, I. R. Kaplan, ed. Plenum, New York, pp. 99-l40. Craig, H., and C. C. Chou (l982). Methane, the record in polar ice cores. Geophys. Res. Lett. 9:l22l-l224. Deuser, W., E. Degens, G. Harvey, and M. Rubin (l973). Methane in Lake Kiva: new data bearing on its origin. Science l8l:5l-53. Hansen, J., D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind, and G. Russell (l98l). Climatic impact of increasing atmospheric carbon dioxide. Science 2l3:957-966.

26l Jones, P. D., R. M. L. Wigley, and P. M. Kelly (l982). Mon. Weather Rev. ll0:59-70. Kossina, E. (l92l). Die Tiefen des Weltmeeres. Berlin U. Inst. fur Meereskunde. Veroff. N.F., A Geogr. Naturwiss. Heft 9, 70 pp. Kvenvolden, K. A., and M. A. McMenamin (l980). Hydrates of natural gas: a review of their geologic occurrence. U.S. Geological Survey Circ. 825. U.S. Dept. of the Interior, Washington, D.C. Lacis, A., J. Hansen, P. Lee, T. Mitchell, and S. Lebedeff (l98l). Greenhouse effect of trace gases, l970-l980. Geophys. Res. Lett. 8.: l035-l038. MacDonald, G. J., ed. (l982a). The Long-Term Impacts of Increasing Atmospheric Carbon Dioxide Levels. Ballinger, Cambridge, Mass., 273 pp. MacDonald, G. J. (1982b). The many origins of natural gas. Paper presented May l982 at Deep Source Gas Workshop sponsored by Morgantown Energy Technology Center. U.S. Dept. of Energy. Mclver, R. D. (l974). Hydrocarbon gas (methane) in canned Deep Sea Drilling Project core samples. In Natural Gases in Marine Sediments, I. R. Kaplan, ed. Plenum, New York, pp. 63-70. Makhijani, A., and A. Poole (l975). Energy and Agriculture in the Third World, Appendix B, Biogasification. Ballinger, Cambridge, Mass., pp. l43-l60. Menard, H. W., and S. M. Smith (l966). Hypsometry of ocean basin provinces. J. Geophys. Res. 7l:4305-432l. Miller, S. R. (l974). The nature and occurrence of clathrate hydrates. In Natural Gases in Marine Sediments, I. R. Kaplan, ed. Plenum, New York, pp. l5l-l70. Rasmussen, R. A., and M. A. K. Khalil (l98l). Atmospheric methane (CH4): trends and seasonal cycles. J. Geophys. Res. 86:9826-9832. Stoll, R. D., J. Ewing, and G. Bryan (l97l). Anomalous wave velocities in sediments containing gas hydrates. J. Geophys. Res. 76(8):2090. Trask, P. D. (l932). Origin and environment of source sediments of petroleum. American Petroleum Institute, Gulf Publ. Co., Houston, Tex., reprinted l982, 332 pp. Quoted in H. Sverdrup et al. (l942), The Oceans. Prentice-Hall, New York, pp. l009-l0l5. Welhan, J., and H. Craig (l979). Methane and hydrogen in East Pacific Rise hydrothermal fluids. Geophys. Res. Lett. 6:829-83l.

262 3.6 SENSITIVITY STUDIES USING CARBON CYCLE MODELS Lester Machta Sensitivity studies using carbon cycle models provide a way of estimating uncertainties in model predictions and aid in distinguishing between those factors in the model that require improvement and those whose uncertainty makes little difference in a final answer. On the other hand, sensitivity studies cannot identify defects that are incor- porated in the model and for which no sensitivity analysis is possible. Furthermore, a model based on the present physical and biological world usually assumes that future behavior of that world can be derived from its past and current status. One should remember that all models of the real world limit their treatment to one or at most a very few forcing functions. In reality many other factors may also produce relevant changes. There are a wide variety of carbon cycle models currently available into which one may enter values for fossil fuel or other O>2 sources. These models have in common three reservoirs: the atmosphere, the oceans, and the biosphere. They also limit themselves to exchange processes that act on a time scale less than several thousand years; that is, many geologic processes are omitted. Each model subdivides the reservoirs and transports carbon and its isotopes in different ways. All models inject the past releases of fossil fuel CO2 and, in a few cases other sources of CO2f into the atmospheric reservoir and try to reconstruct the growth of C02 concentration in air as found at, say, Mauna Loa Observatory after l958. In some instances, other observations, not involved in model development, may also be used to validate the model. Few, if any, models of the carbon cycle would likely be published unless there were some agreement with the Mauna Loa C02 record. Thus, using predictions from the preindustrialized period to the l958-l982 interval may not be useful for sensitivity studies. Rather, predictions into the future, as long as they use the same projected C02 releases, are a better approach to sensitivity. Note that this in no way implies that the future emissions are known; merely that they represent convenient and realistic numbers to use for sensitivity studies. 3.6.l Comparison among Different Models Killough and Emanuel (l98l) have compared the projections made from two simple ocean box models and three other more complex ocean layer models. In each case, plausible parameters for the reservoirs, which include an atmosphere and biosphere as well as the oceans, are used to determine the size and exchange between reservoirs. Figure 3.20 illustrates the projection to year 2275. The topmost curve labeled "cumulative release +290 ppmv" represents the assumed curve of the input of CO2 into the atmosphere expressed in units of the atmospheric concentration as though all the CO2 remained airborne and were uniformly mixed in the air. The input scenario and the models suggest a peak concentration in the air about six times the preindustrial concentration of 290 ppmv.

FIGURE 3.20 Atmospheric response levels of CO2 based on cumulative release as shown. Net exchange with terrestrial biota was assumed to be zero after an asymptotic transition period within the decade after l975. (From Killough and Emmanuel, l98l.) I 1 I ! I I I1 ! II Cumulative release + 290 ppmv Atmospheric response curves for the five models t_J_ I 1 J At this peak, the range between the minimum and maximum concentrations from the five models was 350 ppmv or about l8% of the peak concentra- tion. Were one to select the average among five models, the maximum deviation of the lowest and highest concentration from the five models would differ from the mean by less than l0%. Laurmann and Spreiter (l983) have compared three simple box-type models (2, 3, and 4 boxes or carbon reservoirs). They conclude that not only are predictions of future concentrations from these three models sufficiently similar, but the use of a single airborne fraction (they use the term "retention fraction") yields similar results as the box models. Two exceptions are noted. First, the models do diverge if the growth rate of CO2 emissions is small, an exponential growth rate of less than l.5% per year. Second, if the transfer to the deep oceans is much faster than used in the models then all bets are off (i.e., the models do not represent nature). It is noted that problems will arise if there has been significant deforestation CO2 during the past several decades; the three models assume none. 3.6.2 Comparison of Parameters within a Single Model As part of the analysis of their carbon cycle model, Enting and Pearman (l982) have undertaken a sensitivity study. Although the model does not allow for geographical variation in any of the three major reser- voirs (atmosphere, oceans, and biosphere), it contains many features beyond those of the other models described in this review of sensitiv- ity studies. Virtually all of the parameters (or observations leading to the choice of a value for a parameter) were studied. The conclu- sions drawn by Enting and Pearman are: "The sensitivity analysis . . .

264 TABLE 3.7 Sensitivity Study Using a Box Model of Keeling and Bacastow (l977) and the Nordhaus and Yohe 50th Percentile CO2 Emissions Scenario Range Range Variation in Parameter (ppmv)3. (%) — Rate of exchange between air and sea 2x and 0.5x standard rate of exchange 2 0.3 Rate of exchange between mixed layer of the ocean and the deep ocean 2x and 0.5x standard rate of exchange 70 9 Both of above taken together 74 l0 Biospheric uptake due to enhanced atmospheric CO2 No uptake and a standard value of 0.266 229 29 Buffer factor Constant (l0) and variable according to predicted oceanic chemistry change 6l 8 JLThe range is the higher minus the lower predicted by the changes in arithmetic number used for the parameter in the year 2l00. -i&ange divided by 784 ppmv, the predicted value for the year 2l00, times l00. indicated that for practical purposes the best-fit parameter set is not unique but that there is a range of values within which the model agrees with the data. . . . These uncertainties have only a limited effect on the model predictions for the year 2000, the main uncertainty being the future rate of fossil fuel release." Using the Keeling and Bacastow (l977) box model, we have varied some of the parameters one at a time to determine the effect of such a change on the prediction of atmospheric CO2 concentration in the year 2l00. Table 3.7 shows some of the results based mainly on doubling or halving the most probable value of the parameter and expressing the uncertainty in the prediction as a range in predicted concentration. For all cases, the Nordhaus and Yohe (see this volume, Chapter 2, Section 2.l) 50 percentile scenario of CO2 emissions was used to the year A.D. 2l00. Again, if a departure from a mean value is preferred to the range, the extremes would be at most l5% from the mean CO2 concentration, 784 ppmv. 3.6.3 Deforestation as a Source of CO? We consider deforestation CO2 as a possible real and important source of atmospheric CO2. If some reasonable amounts of future CO2 from deforestation are added to CO2 from future fossil fuel combustion, the error that would be introduced by the omission of the future deforestation CO2 would still be small in the year 2l00 assuming, say, a 2% per year growth rate in the emission of fossil fuel CO2 after l980. For example, oxidizing half of the living biosphere (or

265 half of about 600 x l0l5 g of C) would result in an increase of perhaps 75 ppmv in a predicted value of about l,000 ppmv in the year 2l00. On the other hand, if significant deforestation is currently in progress, and has been for the past several decades, irrespective of future deforestation, the issue is different from that described in the above paragraph. If, for example, the deforestation CO2 were about as large as that from fossil fuel sources, the current models would fail to reproduce the observed atmospheric CO2 growth after l958. The models would likely have to be modified since no reasonable adjust- ment of the parameters will allow a good fit of predictions to observa- tions after l958. The airborne fraction, the ratio of atmospheric increase in a year to the net amount added to the atmosphere, would be calculated to drop to about 0.3 from a value of almost 0.6. Instead of an increase of predicted concentration from 340 to l000 ppmv, the increase might be only from 340 to 670 ppmv to the year 2l00. 3.6.4 Conclusion It may be worth noting what the limited survey of sensitivity studies does and does not reveal. The studies do suggest that if the carbon cycle models are accepted as valid representations of reality, reason- able variations in the numerical values of the parameters do not appear to affect significantly the predictions of future concentrations of atmospheric CC^. Research that accepts the physics, chemistry, and biology of the existing models but tries simply to refine the parameters may not be so effective as research in other aspects of the carbon dioxide issue. On the other hand, the sensitivity studies reveal nothing about the ability of the models to represent nature either today or in the future except possibly indirectly. The guidance provided by sensitivity studies suggests the need for research in those aspects of the carbon cycle likely to make a difference for predicting future atmospheric CO2 concentrations. References Enting, I. G., and G. I. Pearman (l982). Description of a one-dimensional global carbon cycle model. Paper No. 42. Div. of Atmos. Phys., CSIRO, Australia. Keeling, C. D., and R. B. Bacastow (l977). Impact of industrial gases on climate. Energy and Climate. Geophysics Research Board, National Research Council, National Academy Press, Washington, D.C. Killough, G. G., and W. R. Emanuel (l98l). A comparison of several models of carbon turnover in the ocean with respect to their distributions of transit time and age, and responses to atmospheric CO2 and l4C. Tellus 33:274-290. Laurmann, J. A., and J. R. Spreiter (l983). The effects of carbon cycle model error in calculating future atmospheric carbon dioxide levels. Climatic Change 5:l45-l75.