Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Carbon sinks and responsibility indices

The amount of gases remaining in (and thus warming) the atmosphere is determined by the balance between emissions from natural and anthropogenic sources and the amount taken up in natural and human-engendered processes called sinks. The ocean is a major carbon dioxide sink, mainly through and chemical absorption at the surface but also through phytoplankton that may fall from the surface to carry carbon into ocean sediments. Indeed, any place photosynthesis results in a net increase in biomass will be classified as a sink until the increase stops. In addition, bogs where carbon is being buried, forests where fire causes carbon to be sequestered in the soil as charcoal, places where soil carbon is growing, and many other widespread phenomena ad as sinks.

There have been thoughtful suggestions that the distribution of natural sinks for atmospheric carbon be considered in allocating emissions rights (for example, Epstein and Gupta 1990; Agarwal and Narain 1991; Mitchell 1992). Since the operation of sinks is just as important as the magnitude of emissions in determining atmospheric carbon levels, this approach has some appeal.

Unfortunately, unlike important emissions sources, the sizes of important sinks change as the concentrations of the two most important carbon-based greenhouse gases, carbon dioxide and methane, change. When atmospheric concentrations rise, some sinks increase because the processes that drive most of them are functions of concentrations. On the other hand, some other sinks may decline. Unfortunately, therefore, each sink type reacts differently to changes in the atmospheric concentrations. Sinks, thus, are moving forgets as well as being substantially different for the different gases.

To avoid this problem, one might take 'pre-human' CO2 concentrations as the baseline from which to determine the 'natural' size of sinks. The 'natural' sinks could then be allocated on a per capita basis or whatever. By definition, however, 'natural' sinks and sources were essentially in balance. More precisely, the prehuman carbon balance was maintained by a net terrestrial flow that was slightly (some tenths of a GT out of about a 200 GT annual flow) positive (a source), but countered by an equivalent slightly negative flow for the oceans (a sink) (Schmidbauer et al. 1991). Since most people agree that the ocean sink ought to be part of the common human heritage, that is, allocated equally to each person, it is not clear how useful (or practical) it is to attempt to allocate the prehuman natural terrestrial net source. As a whole, both are extremely small in any case.

Although small as a whole, it might be countered that there are large net terrestrial sinks and sources in individual localities that ought to be considered in allocations. This approach is frustrated, however, by our inability to define the 'natural' state of the landscape in any one place. A common assumption is that the landscape/atmosphere relationship was in a 'natural' condition before humanity began to burn significant amounts of fossil fuel, that is, before the early 1800s. Since this seems also to correspond roughly to the date at which atmospheric levels of the major carbon greenhouse gases (CH4 and CO2) began to rise, the changes are attributed to the combination of fossil carbon input and land-use changes that have occurred since. The IPCC, for example, subsequent to stating that 'concentrations of CO2 and CH4, after remaining relatively constant up to the 1 8th century, have risen sharply since then due to man's activities' (IPCC 1990: xvi), confines its examination of the impact of land use changes to the period after 1850 (Section

There is a questionable assumption hidden here, however, that apparently relatively stable atmospheric concentrations previous to 1800 meant that there were no major land-use changes. In fad, humanity has caused rather dramatic land-use changes ever since learning how to use fire. Much of the landscape that may seem to us 'natural' is actually the result of pre-industrial human management by fire and other means. No continent has been spared this management. Starting 12,000 years back, for example, an estimated 25 per cent of subsequent human deforestation had actually occurred by 1700, 50 per cent by 1850, and 75 per cent by 1915 (Turner et al. 1990). Furthermore, human land management had started long before 10,000 B.C. The deforestation going on now, while vitally important as a fraction of the remaining forest and perhaps occurring at a greater rate than in the past, is not excessively large compared to what our ancestors accomplished.

Most of this land-use change, then as now, was undoubtedly related to food production and security, but some must have been due to the harvesting of biomass as fuel, which remains even today an important component of total biomass combustion. Reliance on biomass fuel was, of course, even more widespread before use of fossil fuels and may have resulted in significant net emissions of carbon (Kammen and Marino 1993).

Unfortunately, our knowledge of historical population distribution and its impact on the biosphere is not sufficient to allow adequate estimates of land-use changes and the corresponding gross emissions of carbon, let alone the net result for the atmosphere taking into account changing carbon sinks (Andreae 1991). That there do not seem to be any major historical (human history) changes in atmospheric carbon levels previous to 1800 does not prove, however, that there were no major land-use changes and corresponding gross carbon emissions, only that over time they must have been balanced by appropriate sinks.

It does seem clear, however, that the idea of ever being able to define a 'natural' baseline in every locality is questionable. Indeed, it may well be that the same climate, soil, and other physical conditions can support entirely distinct 'climax' landscapes depending on the pathway taken. Even short-term temporary intervention by human management can completely change all subsequent history in an area by instituting positive feedback mechanisms. Alternatively, removal of human management may not cause reversion to a unique "natural' landscape (Woodcock 1 992).

We are greatly hampered by lack of data and knowledge not only about past sinks, but also about those in operation today. As shown in Figure 4.3, out of the 7 GT of carbon thought to be released annually by human activities (5.4 from fossil fuels, 1.6 from land-use changes), less than 50 per cent (3.4 GT) stays in the atmosphere and only 2 GT goes into the ocean (IPCC 1990, 1992). The remaining 1.6 GT goes to the infamous 'missing carbon' sink. As indicated on the figure, there is some evidence that the carbon is going into terrestrial sinks in the northern hemisphere, but much uncertainty remains. This uncertainty, alone, handicaps any effort to incorporate sinks into allocation schemes.

Assume, for the moment, that the missing carbon is shown conclusively to be absorbed by sinks in temperate and boreal forests, as many today believe. What would be the implications of this for global negotiations? Should the tropical, mostly developing, countries now undergoing deforestation be charged with emissions from land-use changes on top of those from fossil fuel, as is the case in many proposed allocation schemes (such as in WRI 1990)? If so, should not the high-latitude, mostly developed, countries be accorded a net decrease in their fossil-fuel totals to account for the sinks in their forests! If so, the result would be a great levelling of international accountability.

Figure 4.3 The global carbon cycle approximate current flows (GT carbon)

On the other hand, whereas the forests of the currently developed countries may be absorbing carbon today, they are generally still far from their 'original' (that is, pre-human) size. Should these countries then be held responsible for the carbon released in the great deforestations of the past? If so, since we do not know much about the 'natural' condition before these deforestations, how would this be done? Should some currently developing countries be handicapped because they are deforesting relatively late in human history! It is certainly not clear that they are, or will be, deforesting more than others did in the past. Other developing countries, such as China, may have completed much of their deforestation even before North America and Europe. Mesopotamia may have done so even earlier.

The current global flow of biospheric carbon from low to high latitudes is likely just the reverse of what went on previously when the high-latitude forests were being cleared. Some, and perhaps a significant amount, of the gross carbon released by high-latitude clearing was once taken up by sinks in the tropics, through CO2 fertilization, for example. This balance is consistent with the (apparent) fact that it did not result in significant increases in atmospheric carbon. In effect, people have been using each other's carbon sinks for a long time. Today, this is called the 'missing carbon' phenomenon.

These ecological/biophysical factors affect how responsibility can be attributed in negotiations today. Someday, perhaps, we may be able to determine which

local parts of the natural pre-human landscape were sources and which were sinks. We would then be able to judge what is occurring in these localities today relative to a meaningful baseline. The value of doing so for purposes of allocation is questionable, however. Just as we would not want to bias the allocation of sink rights against people who do not happen to live near the ocean, so too there is no reason to punish those who happen to live near swamps or reward the people who happen to live near forests (or wherever the natural sources and sinks are). This logic implies that the natural terrestrial sources and sinks also ought to be allocated equally to all people (we certainly do not have the data at present to do otherwise).

Although conceptually tricky, this analysis seems to argue that in addition to sinks, changes in the biosphere itself should be left out of responsibility indices. This conclusion follows because

1 We cannot and may never be able to define in any meaningful way the natural baselines from which we have departed, either on a total or local basis.

2 Even if we could quantify them, departures from these baselines are much too ancient for us to link them rationally to particular human populations today.

3 We cannot define the baseline as today's biospheric situation, because we still understand it quite poorly. (Some 1.6 GT of carbon of the approximately 7 GT total emitted is going into sinks, probably biospheric, that we have not identified.)

For fossil fuels, on the other hand, the situation is quite different

1 We do know accurately the natural baseline (near zero) and the geographic and temporal distributions of emissions (Ebert and Karmali 1992).

2 The bulk of the emissions has occurred within a period such that they can be rationally allocated to countries today.

3 Fossil fuel emissions are larger, growing more rapidly, and less intrinsically limited than net biospheric emissions.

As a result, in this book, we do not include biospheric changes as part of the responsibility index.

To optimize present management of the carbon sphere, however, humanity clearly must address biospheric sinks. Nations should be encouraged to maintain and enhance whatever sinks they control and to develop others. The techniques to estimate the marginal costs of doing so and to compare these costs with other ways of limiting carbon emissions are the subjects of Chapters 5 and beyond. Of necessity, however, is choosing a baseline, that is, the mixture of biospheric sinks and sources from which departures will be accounted henceforth. This baseline is taken to be 1990, an arbitrary, but practical, choice, one also selected in the Framework Convention on Climate Change discussed at the June 1992 UNCED meeting in Rio.

Conceptually intermediate between fossil fuel emissions and human-engendered changes in the 'natural' biosphere are the atmospheric consequences of managed agroecosystems. Of most importance for global warming is the methane from ruminants, rice paddies, and biomass burning. Although obviously 'managed,' given the scale of total human intervention in the biosphere, agroecosystems are merely at one end of a spectrum of management. We thus classify them with the rest of the biosphere and do not include them in the responsibility index, but do take account of them in evaluating mitigation proposals. For calculating marginal costs of emissions reduction, we again take 1990 as the base year.


Agarwal, A, and S Narain, 1991. Global Warming in an Unequal World A Case of
Environmental Colonialism, Centre for Science and Environment, New Delhi, India

Ahuja, D R. 1992. 'Estimating National Contributions to Greenhouse Gas Emissions,' in J K Mitchell, ed. Global Environmental Change 2(2) 83-87

Andreae, M O. 1991. 'Biomass Burning Its History, Use, and Distribution and Its Impact on

Environmental Quality and Global Climate,' in J S Levine, ea., Global Biomass Burning,
MIT Press, Cambridge, MA, USA

Ebert, C, and A Karmali, 1992. 'Uncertainties in Estimating Greenhouse Gas Emissions,'

Environment Department Working Paper No. 52, World Bank, Washington, DC, USA

Epstein, F. and R Gupta, 1 990. Controlling the Greenhouse Effect: Five Global Regimes

Compared Brookings Institution, Washington DC, USA

IPCC, 1990, ;992. Scientific Assessment, Supplement to the Scientific Assessment, Cam bridge University Press, Cambridge, UK

Jaworski, 2, T V Segalstad and N Ono, 1992. 'Do Glaciers Tell a True Atmospheric CO2

Story!' Science of the Total Environment 114 (April) 227-284 Kammen, D M, and B D Marino, 1993. 'On the Origin and Magnitude of Pre-industrial Anthropogenic CO2 and CH4 Emissions,' Chemosphere, 26(1-4): 69-86

Mitchell, J K, ed., 1992. 'Greenhouse Equity' (six authors comment on Agarwal and Narain 1991) Global Environmental Change 2(2) 82-100

Pyne, S J. 1982. fire in America, Princeton University Press, NJ, USA

Pyne, S J. 1991. Burning Bush, A Fire History of Australia, Henry Holt, NY, USA

Rambo, A T. 1984. 'No Free Lunch A Reexamination of the Energetic Efficiency of Swidden Agriculture,' in A T Rambo and P E Sajise, eds, An Introduction to Human Ecology
Research on Agricultural Systems in Southeast Asia, University of the Philippines, Los Banos, pp. 154-163

Schmidbauer, B. et al., 1991. Vol. 1, Protecting the Earth, German Bundestag, Bonner

Universitats-Buchdruckerei, Bonn

Turner, B L, and K W Butzer, 1992. 'The Columbian Encounter and Land-Use Change,'

Environment 34(8) 16-20, 37-44

Turner, B L, W C Clark, R W Kates, et al., 1990. The Earth as Transformed by Human Action: Global and Regional Changes in the Biosphere over the Past 300 Years, Cambridge
University Press, Cambridge, UK

Woodcock, D W. 1992. 'The Rain on the Plain Are There Vegetation-Climate Feedbacks?' Paleogeography, Paleoclimatology, Paleoecology (GIobal and Planetary Change Section)
97 191 -201

WRI World Resources Institute), 1990. World Resources 1990-97, Oxford University Press, NY, USA

The way out of this paradox is to recognize the need for either a non-linear weighting of emissions in something like a progressive income-tax schedule, a threshold, or both. This is parallel to the way the ability-to-pay index was constructed, which recognized that a dollar of GNP above the threshold is to be counted differently to one below. Just so with natural debt per capita. This natural debt threshold is analogous to the already accepted 300 g/capita threshold under the Montreal Protocol (Table 4.1).

Using global warming potentials in project evaluations

Even though comprehensive indices that aggregate a range of greenhouse gases may yet be impractical agreements because of data uncertainties and verification difficulties, they can still find a place in project evaluation (Levander 1990; Wilson 1990; Ellington et al. 1992; Pitstick et al. 1992). When choosing among possible greenhouse gas mitigation measures, for example, there is a need to compare costs per unit greenhouse reduction. In many cases, in project evaluation it will be possible to estimate and verify the changes in emissions of the other greenhouse gases as well as of CO2. In some cases, inclusion of these non-CO2 greenhouse gases can be crucial to making the right choice.

To illustrate the importance of non-CO2 gases, consider the following question: 'What are the greenhouse-gas implications of moving up the household energy ladder from wood to kerosene or LPG for cooking?' Given that something like half the world's households still use biomass fuels for cooking, this question is important.

Essentially all the products of incomplete combustion (PIC) that are produced during biomass and fossil-fuel combustion are also effectively greenhouse gases. These include methane, carbon monoxide, and non-methane hydrocarbons, the last two mainly exert their effects indirectly. Indeed, as shown in Table 2.1 of Chapter 2, considering indirect warming effects, if carbon from fuel combustion is to go into the air, carbon dioxide is actually about the least damaging form.

Unfortunately, the combustion efficiency of traditional biomass burning in simple stoves is usually much less than 100 per cent. It is not uncommon, for example, for well over 10 per cent of the carbon to be released as PIC rather than carbon dioxide, which would be the only carbon product if combustion was complete. Because PIC on average have higher global warming potentials (GWP) than carbon dioxide, the total impact can be substantially higher than would be indicated by an evaluation based an carbon dioxide alone. A recent pilot study in the Philippines, for example, has shown that the total GWP of woodstoves can, depending on circumstances, be more than double that of the carbon dioxide (i.e., the GWP of the PIC can rival that of the carbon dioxide alone) (Smith et al. 1993).

This paint is illustrated in Figure 4.4, which straws the flow of carbon through one of these stoves. Note that applying the 20-year GWP to the PIC gives a total CO2-equivalent even greater than that of the CO2 itself, even though the latter has more than seven times more of the original fuel carbon. By comparison, a 100-year time horizon gives a much lower, but still significant GWP (40 per cent of CO2).

For comparison, the carbon balance of one possible alternative, a kerosene stove cooking the same meal, is shown in Figure 4.5. Since kerosene has 1.5 times more energy per carbon atom than wood, and kerosene stoves are 2.5 times more energy efficient, the total fuel carbon needed is much less 27 per cent. In addition, in a 100-year horizon, kerosene stoves seem to produce less than 10 per cent additional GWP over the CO2 (20 per cent for a 20-year GWP).

There are several tentative but potentially important implications of these findings. First the overall greenhouse gas benefit of improved biomass cookstove programmes may be much larger than previously estimated. This outcome depends, however, on the degree to which the stoves actually improve combustion efficiency, rather than just overall efficiency, which is also affected by the efficiency of heat transferring to the cooking pots. Assuming that the wood is not being harvested renewably, for example, the cost of improved stoves to reduce carbon emissions (CO2-equivalents) might be something like $40/tonne without considering the PIC, but only $25/tonne considering the PIC at a 100year horizon. At a 20-year horizon, it would only be $14/tonne.

Figure 4.4 Flow of carbon through a wood-fuelled cooking stove

Second, surprisingly, in some cases, there may be substantial greenhouse benefits in switching from some kinds of biomass stoves to modern fuels such as kerosene and LPG. Although these fossil fuels produce significant GWP because they are non-renewable, the overall GWP impact may be lower because of the high emissions of PIC from some biomass stoves. This result is shown in Table 4.3 (case A) where, at all time horizons, kerosene and LPG have much lower total GWP than a stove fuelled with non-renewably harvested wood. Less obvious, however, are the results shown in case B. based on a completely renewable wood harvesting system. Because the wood stove produces so much PIC, there is an advantage of moving to the fossil fuels in all but the 500-year time horizon, even though burning the modern fuels releases fossil carbon (net CO2 emissions from the stove are considered to be zero). This conclusion may provide the extra incentive needed in some areas to shift policy toward encouraging fuel substitution rather than improved woodstoves.

Figure 4.5 Flow of carbon through a kerosene-fired stove

Table 4.3 Benefits GWP from fuel switching in cookstoves

Case A Where wood is harvested on a completely non-renewable basis, i.e., woodstove CO2 emissions are included as net atmospheric additions.

Relative GWP* in
going from wood to
Time horizon
Instantaneous 20-year 100-year 500-year
Kerosene 0.22 0.19 0.23 0.25
LPG 0.14 0.12 0.16 0.17

Case B Where wood is harvested on a completely renewable oasis, i.e., woodstove CO2 emissions are not included for the woodstove because the CO2 is completely recycled, but CO? from the fossil fuel stoves is included.

Relative GWP in going from wood to Time horizon
Instantaneous 20-year 700-year 500-year
Kerosene 0.69 0.44 0.96 1.5
LPG 0.46 0.27 0.63 1.0

* This does not take into account GGs emitted in other parts of these fuel cycles, e.g., decay of wood residues and emissions at oil refineries.

This GWP includes the production of CO2 as well as the non-CO2 GGs CO, CH4, and NMHC and is dependent on relative stove fuel efficiencies (wood = 20%; kerosene = 50%; LPG = 70%). Energy per carbon atom is considered to be equal for the fossil fuels and two-thirds as much for wood (see Smith et al., 1993).

Third although charcoal stoves ore more efficient and produce less GWP (and health-damaging) pollution than do woodstoves, the charcoal manufacturing process is quite a different matter. Another impact of the generally quite inefficient charcoal kilns in developing countries, besides excessive wood demand, is that a good fraction of the original carbon in the wood ends up as PIC released at the kiln. The resulting total GWP of the charcoal fuel cycle apparently con be many times that of a comparable wood fuel cycle. This effect gives additional impetus to improve charcoal kiln efficiencies and may also argue for more efforts to encourage some charcoal-using populations to use other fuels (Delmas et al. 1991).

More research is being undertaken to verify these findings. The general impact, however, would seem to be that the PIC of biomass stoves is on even greater enemy than has been recognized. We have known that they rob households of some of the energy contained in the fuel and impose health problems on the householders (Smith 1987). Now we have still another incentive to reduce PIC as much as possible (Smith and Thomeloe 1992).

This example illustrates how project evaluations can be affected by consideration of non-CO2 gases, as well as using GWPs that include estimates of indirect effects. We have not, however, taken such a comprehensive approach in the cost calculations in later chapters of this book. This is partly because many of the projects (for example, improved lighting) reduce electricity demand. When generated by fossil fuels, electricity is generally produced in large-scale units that emit small amounts of non-CO2 greenhouse gases relative to the CO2 they emit (although if powered by natural gas or coal, there could be significant methane emissions elsewhere in the fuel cycle). This assumption bears examination, however, in future evaluations for large-scale devices and is clearly wrong for some small-scale combustion devices like simple stoves.


Delmas, R A, A Marenco, J P Tathy et al., 1991. 'Sources and Sinks of Methane in the African
Savanna,' J of Geophysical Res. 96(D4): 7287-7299

Ellington, R T. M Meo, and D W Bough, 1992. 'The Total Greenhouse Warming Potential of
Technical Systems Analysis for Decision Making,' J Air and Waste Management
Association 42(4): 422-428

Levander, T. 1990. 'The Relative Contributions to the Greenhouse Effect from the Use of
Different Fuels,' Atmospheric Environment 24A(11) 2707-2714

Pitstick, M E, D J Santini and H Chauhan, 1992. 'Reduction in Global Warming Due to Fuel
Economy improvements and Emissions Controls New US Light-DutyVehicles,' Proceedings of the Intersociety Energy Conversion Engineering Conference, San Diego, August 3-7

Smith, K R. 1987. Biofuels, Air Pollution, and Health, Plenum, NY, USA

Smith K R. M A K Khalil, R A Rasmussen, S A Thorneloe F Manegdeg and M Apte, 1993.
'Greenhouse Gases from Biomass and Fossil Fuel Stoves in Developing Countries: A
Manila Pilot Study,' Chemosphere, 26(1-4): 479-505

Smith, K R. and S A Thorneloe, 1992. 'Household Fuels in Developing Countries: Global
Warming, Health, and Energy Implications,' Proceedings from the 7 992 Greenhouse Gas
Emissions and Mitigation Research Symposium, USEPA, RTP, NC, USA

Wilson D, 1990. 'Quantifying and Comparing Fuel-Cycle Greenhouse Gas Emissions,'
Energy Policy 18(6): 550-562

The most direct, unambiguous, and accurate reflection of a country's contribution to the greenhouse problem is then given by the total emissions of all the relevant gases (on a CO2-equivalent basis) over the period of interest. As discussed in Chapter 2, deciding on which gases and time periods to incorporate will be a matter for negotiations, but 100 years of warming into the future seems like a good starting point for discussions.

The responsibility for emissions needs to be tempered by the number of people for which these emissions were released. Thus responsibility is proportional to the natural debt, which we define as the amount of GGs remaining divided by the current population.

Before determining responsibility to pay, it is necessary to choose a threshold natural debt. A number of ways to do this exist: for example, based on PQLI, as with ATP; based on the natural debt needed for stable global GG concentration of some specified level; or based on global averages.

The third column of Table 4.2 compares RESP for fossil-fuel CO2 emissions since 1950 using natural debt at three different thresholds: 5, 10, and 20 tonnes per capita. Note that lower thresholds tend to decrease the relative responsibilities of developed countries, such as the United States, Germany, and the UK, but increase those of developing countries. Only with the lowest threshold does China have any responsibility, and India's responsibility remains at zero for all the thresholds shown.

Obligation to pay

It now remains to combine the practical (resources - ATP) and the ethical (responsibility - RESP) components into an overall obligation-to-pay (OTP) index that is transparent in formulation so that it can be easily manipulated in the course of negotiations. In parallel with Wirth and Lashof (1990), the combination could be a simple sum or the two could be multiplied together.

The second combination (product) implies that if either ATP or RESP is zero, payments are zero. It could be argued, however, that responsibility for a debt does not disappear just because the borrower is unable to pay now. This argument suggests the first formulation where OTP will be greater than zero for a larger group of countries.

Columns 4 and 5 of Table 4.2 compare the additive version of this obligation-to-pay (OTP) index using equal weights for both ATP and RESP and showing also the effect of varying natural debt threshold and income measure (GNP or PPP). The result is more stable than either the ATP or RESP index by itself. The US obligation remains between 35 and 39 per cent, for example. China retains a modest obligation in two scenarios, but India's remains at zero for all combinations.

Smith et al. (1991) compare the multiplicative version of the OTP using equal weights for the two indicators, for both types of income indicators and two different natural debt thresholds. The US contribution is above 69 per cent in all cases, undoubtedly a politically unacceptable result. Most developing countries, on the other hand, have zero OTP.

The present constraint on the maximum UN assessment for any one nation is 25 per cent, substantially less than the 35-39 per cent range determined here for the OTP of the United States. It might be noted, however, that the original limit and US assessment set in 1946 was 40 per cent (UN 1989).

Contents - Previous - Next