Working Group I: The Scientific Basis

Other reports in this collection

Executive Summary

CO2 concentration trends and budgets

Before the Industrial Era, circa 1750, atmospheric carbon dioxide (CO2) concentration was 280 ± 10 ppm for several thousand years. It has risen continuously since then, reaching 367 ppm in 1999.

The present atmospheric CO2 concentration has not been exceeded during the past 420,000 years, and likely not during the past 20 million years. The rate of increase over the past century is unprecedented, at least during the past 20,000 years.

The present atmospheric CO2 increase is caused by anthropogenic emissions of CO2. About three-quarters of these emissions are due to fossil fuel burning. Fossil fuel burning (plus a small contribution from cement production) released on average 5.4 ± 0.3 PgC/yr during 1980 to 1989, and 6.3 ± 0.4 PgC/yr during 1990 to 1999. Land use change is responsible for the rest of the emissions.

The rate of increase of atmospheric CO2 content was 3.3 ± 0.1 PgC/yr during 1980 to 1989 and 3.2 ± 0.1 PgC/yr during 1990 to 1999. These rates are less than the emissions, because some of the emitted CO2 dissolves in the oceans, and some is taken up by terrestrial ecosystems. Individual years show different rates of increase. For example, 1992 was low (1.9 PgC/yr), and 1998 was the highest (6.0 PgC/yr) since direct measurements began in 1957. This variability is mainly caused by variations in land and ocean uptake.

Statistically, high rates of increase in atmospheric CO2 have occurred in most El Niño years, although low rates occurred during the extended El Niño of 1991 to 1994. Surface water CO2 measurements from the equatorial Pacific show that the natural source of CO2 from this region is reduced by between 0.2 and 1.0 PgC/yr during El Niño events, counter to the atmospheric increase. It is likely that the high rates of CO2 increase during most El Niño events are explained by reductions in land uptake, caused in part by the effects of high temperatures, drought and fire on terrestrial ecosystems in the tropics.

Land and ocean uptake of CO2 can now be separated using atmospheric measurements (CO2, oxygen (O2) and 13CO2). For 1980 to 1989, the ocean-atmosphere flux is estimated as -1.9 ± 0.6 PgC/yr and the land-atmosphere flux as -0.2 ± 0.7 PgC/yr based on CO2 and O2 measurements (negative signs denote net uptake). For 1990 to 1999, the ocean-atmosphere flux is estimated as -1.7 ± 0.5 PgC/yr and the land-atmosphere flux as -1.4 ± 0.7 PgC/yr. These figures are consistent with alternative budgets based on CO2 and 13CO2 measurements, and with independent estimates based on measurements of CO2 and 13CO2 in sea water. The new 1980s estimates are also consistent with the ocean-model based carbon budget of the IPCC WGI Second Assessment Report (IPCC, 1996a) (hereafter SAR). The new 1990s estimates update the budget derived using SAR methodologies for the IPCC Special Report on Land Use, Land Use Change and Forestry (IPCC, 2000a).

The net CO2 release due to land-use change during the 1980s has been estimated as 0.6 to 2.5 PgC/yr (central estimate 1.7 PgC/yr). This net CO2 release is mainly due to deforestation in the tropics. Uncertainties about land-use changes limit the accuracy of these estimates. Comparable data for the 1990s are not yet available.

The land-atmosphere flux estimated from atmospheric observations comprises the balance of net CO2 release due to land-use changes and CO2 uptake by terrestrial systems (the “residual terrestrial sink”). The residual terrestrial sink is estimated as -1.9 PgC/yr (range -3.8 to +0.3 PgC/yr) during the 1980s. It has several likely causes, including changes in land management practices and fertilisation effects of increased atmospheric CO2 and nitrogen (N) deposition, leading to increased vegetation and soil carbon.

Modelling based on atmospheric observations (inverse modelling) enables the land-atmosphere and ocean-atmosphere fluxes to be partitioned between broad latitudinal bands. The sites of anthropogenic CO2 uptake in the ocean are not resolved by inverse modelling because of the large, natural background air-sea fluxes (outgassing in the tropics and uptake in high latitudes). Estimates of the land-atmosphere flux north of 30°N during 1980 to 1989 range from -2.3 to -0.6 PgC/yr; for the tropics, -1.0 to +1.5 PgC/yr. These results imply substantial terrestrial sinks for anthropogenic CO2 in the northern extra-tropics, and in the tropics (to balance deforestation). The pattern for the 1980s persisted into the 1990s.

Terrestrial carbon inventory data indicate carbon sinks in northern and tropical forests, consistent with the results of inverse modelling.

East-west gradients of atmospheric CO2 concentration are an order of magnitude smaller than north-south gradients. Estimates of continental-scale CO2 balance are possible in principle but are poorly constrained because there are too few well-calibrated CO2 monitoring sites, especially in the interior of continents, and insufficient data on air-sea fluxes and vertical transport in the atmosphere.

The global carbon cycle and anthropogenic CO2

The global carbon cycle operates through a variety of response and feedback mechanisms. The most relevant for decade to century time-scales are listed here.

Responses of the carbon cycle to changing CO2 concentrations

  • Uptake of anthropogenic CO2 by the ocean is primarily governed by ocean circulation and carbonate chemistry. So long as atmospheric CO2 concentration is increasing there is net uptake of carbon by the ocean, driven by the atmosphere-ocean difference in partial pressure of CO2. The fraction of anthropogenic CO2 that is taken up by the ocean declines with increasing CO2 concentration, due to reduced buffer capacity of the carbonate system. The fraction taken up by the ocean also declines with the rate of increase of atmospheric CO2, because the rate of mixing between deep water and surface water limits CO2 uptake.
  • Increasing atmospheric CO2 has no significant fertilisation effect on marine biological productivity, but it decreases pH. Over a century, changes in marine biology brought about by changes in calcification at low pH could increase the ocean uptake of CO2 by a few percentage points.
  • Terrestrial uptake of CO2 is governed by net biome production (NBP), which is the balance of net primary production (NPP) and carbon losses due to heterotrophic respiration (decomposition and herbivory) and fire, including the fate of harvested biomass. NPP increases when atmospheric CO2 concentration is increased above present levels (the “fertilisation” effect occurs directly through enhanced photo-synthesis, and indirectly through effects such as increased water use efficiency). At high CO2 concentration (800 to 1,000 ppm) any further direct CO2 fertilisation effect is likely to be small. The effectiveness of terrestrial uptake as a carbon sink depends on the transfer of carbon to forms with long residence times (wood or modified soil organic matter). Management practices can enhance the carbon sink because of the inertia of these “slow” carbon pools.

Feedbacks in the carbon cycle due to climate change

  • Warming reduces the solubility of CO2 and therefore reduces uptake of CO2 by the ocean.
  • Increased vertical stratification in the ocean is likely to accompany increasing global temperature. The likely consequences include reduced outgassing of upwelled CO2, reduced transport of excess carbon to the deep ocean, and changes in biological productivity.
  • On short time-scales, warming increases the rate of heterotrophic respiration on land, but the extent to which this effect can alter land-atmosphere fluxes over longer time-scales is not yet clear. Warming, and regional changes in precipitation patterns and cloudiness, are also likely to bring about changes in terrestrial ecosystem structure, geographic distribution and primary production. The net effect of climate on NBP depends on regional patterns of climate change.

Other impacts on the carbon cycle

  • Changes in management practices are very likely to have significant effects on the terrestrial carbon cycle. In addition to deforestation and afforestation/reforestation, more subtle management effects can be important. For example, fire suppression (e.g., in savannas) reduces CO2 emissions from burning, and encourages woody plant biomass to increase. On agricultural lands, some of the soil carbon lost when land was cleared and tilled can be regained through adoption of low-tillage agriculture.
  • Anthropogenic N deposition is increasing terrestrial NPP in some regions; excess tropospheric ozone (O3) is likely to be reducing NPP.
  • Anthropogenic inputs of nutrients to the oceans by rivers and atmospheric dust may influence marine biological productivity, although such effects are poorly quantified.

Modelling and projection of CO2 concentration

Process-based models of oceanic and terrestrial carbon cycling have been developed, compared and tested against in situ measurements and atmospheric measurements. The following are consistent results based on several models.

  • Modelled ocean-atmosphere flux during 1980 to 1989 was in the range -1.5 to -2.2 PgC/yr for the 1980s, consistent with earlier model estimates and consistent with the atmospheric budget.
  • Modelled land-atmosphere flux during 1980 to 1989 was in the range -0.3 to -1.5 PgC/yr, consistent with or slightly more negative than the land-atmosphere flux as indicated by the atmospheric budget. CO2 fertilisation and anthropogenic N deposition effects contributed significantly: their combined effect was estimated as -1.5 to -3.1 PgC/yr. Effects of climate change during the 1980s were small, and of uncertain sign.
  • In future projections with ocean models, driven by CO2 concentrations derived from the IS92a scenario (for illustration and comparison with earlier work), ocean uptake becomes progressively larger towards the end of the century, but represents a smaller fraction of emissions than today. When climate change feedbacks are included, ocean uptake becomes less in all models, when compared with the situation without climate feedbacks.
  • In analogous projections with terrestrial models, the rate of uptake by the land due to CO2 fertilisation increases until mid-century, but the models project smaller increases, or no increase, after that time. When climate change feedbacks are included, land uptake becomes less in all models, when compared with the situation without climate feedbacks. Some models have shown a rapid decline in carbon uptake after the mid-century.

Two simplified, fast models (ISAM and Bern-CC) were used to project future CO2 concentrations under IS92a and six SRES scenarios, and to project future emissions under five CO2 stabilisation scenarios. Both models represent ocean and terrestrial climate feedbacks, in a way consistent with process-based models, and allow for uncertainties in climate sensitivity and in ocean and terrestrial responses to CO2 and climate.

  • The reference case projections (which include climate feedbacks) of both models under IS92a are, by coincidence, close to those made in the SAR (which neglected feedbacks).
  • The SRES scenarios lead to divergent CO2 concentration trajectories. Among the six emissions scenarios considered, the projected range of CO2 concentrations at the end of the century is 550 to 970 ppm (ISAM model) or 540 to 960 ppm (Bern-CC model).
  • Variations in climate sensitivity and ocean and terrestrial model responses add at least -10 to +30% uncertainty to these values, and to the emissions implied by the stabilisation scenarios.
  • The net effect of land and ocean climate feedbacks is always to increase projected atmospheric CO2 concentrations. This is equivalent to reducing the allowable emissions for stabilisation at any one CO2 concentration.
  • New studies with general circulation models including interactive land and ocean carbon cycle components also indicate that climate feedbacks have the potential to increase atmospheric CO2 but with large uncertainty about the magnitude of the terrestrial biosphere feedback.


CO2 emissions from fossil fuel burning are virtually certain to be the dominant factor determining CO2 concentrations during the 21st century. There is scope for land-use changes to increase or decrease CO2 concentrations on this time-scale. If all of the carbon so far released by land-use changes could
be restored to the terrestrial biosphere, CO2 at the end of the century would be 40 to 70 ppm less than it would be if no such intervention had occurred. By comparison, global deforestation would add two to four times more CO2 to the atmosphere than reforestation of all cleared areas would subtract.

There is sufficient uptake capacity in the ocean to incorporate 70 to 80% of foreseeable anthropogenic CO2 emissions to the atmosphere, this process takes centuries due to the rate of ocean mixing. As a result, even several centuries after emissions occurred, about a quarter of the increase in concentration caused by these emissions is still present in the atmosphere. CO2 stabilisation at 450, 650 or 1,000 ppm would require global anthropogenic CO2 emissions to drop below 1990 levels, within a few decades, about a century, or about two centuries respectively, and continue to steadily decrease thereafter. Stabilisation requires that net anthropogenic CO2 emissions ultimately decline to the level of persistent natural land and ocean
sinks, which are expected to be small (<0.2 PgC/yr).

Other reports in this collection

IPCC Homepage