IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group III: Mitigation of Climate Change

9.3 Regional and global trends in terrestrial greenhouse gas emissions and removals

Mitigation measures will occur against the background of ongoing change in greenhouse gas emissions and removals. Understanding current trends is critical for evaluation of additional effects from mitigation measures. Moreover, the potential for mitigation depends on the legacy of past and present patterns of change in land-use and associated emissions and removals. The contribution of the forest sector to greenhouse gas emissions and removals from the atmosphere remained the subject of active research, which produced an extensive body of literature (Table 9.2 and IPCC, 2007a: Chapter 7 and 10).

Globally during the 1990s, deforestation in the tropics and forest regrowth in the temperate zone and parts of the boreal zone were the major factors responsible for emissions and removals, respectively (Table 9.2; Figure 9.2). However, the extent to which carbon loss due to tropical deforestation is offset by expanding forest areas and accumulating woody biomass in the boreal and temperate zones is the area of disagreement between land observations and estimates by top-down models. The top-down methods based on inversion of atmospheric transport models estimate the net terrestrial carbon sink for the 1990s, which is the balance of sinks in northern latitudes and source in tropics (Gurney et al., 2002). The latest estimates are consistent with the increase found in the terrestrial carbon sink in the 1990s over the 1980s.

Figure 9.2

Figure 9.2: Historical forest carbon balance (MtCO2) per region, 1855-2000.

Notes: green = sink. EECCA=Countries of Eastern Europe, the Caucasus and Central Asia. Data averaged per 5-year period, year marks starting year of period.

Source: Houghton, 2003b.

Denman et al. (2007) reports the latest estimates for gross residual terrestrial sink for the 1990s at 9,500 MtCO2/yr, while their estimate for emissions from deforestation amounts to 5,800 MtCO2/yr. The residual sink estimate is significantly higher than any land-based global sink estimate and in the upper range of estimates produced by inversion of atmospheric transport models (Table 9.2). It includes the sum of biases in estimates of other global fluxes (fossil fuel burning, cement production, ocean uptake, and land-use change) and the flux in terrestrial ecosystems that are not undergoing change in land use.

Improved spatial resolution allowed separate estimates of the land-atmosphere carbon flux for some continents (Table 9.2). These estimates generally suggest greater sink or smaller source than the bottom-up estimates based on analysis of forest inventories and remote sensing of change in land-cover (Houghton, 2005). While the estimates of forest expansion and regrowth in the temperate and boreal zones appear relatively well constrained by available data and consistent across published results, the rates of tropical deforestation are uncertain and hotly debated (Table 9.2; Fearnside and Laurance, 2004). Studies based on remote sensing of forest cover report lower rates than UN-ECE/FAO (2000) and lower carbon emissions carbon (Achard et al., 2004).

Recent analyses highlight the important role of other carbon flows. These flows were largely overlooked by earlier research and include carbon export through river systems (Raymond and Cole, 2003), volcanic activity and other geological processes (Richey et al., 2002), transfers of material in and out of products pool (Pacala et al., 2001), and uptake in freshwater ecosystems (Janssens et al., 2003).

Attribution of estimated carbon sink in forests to the short- and long-term effects of the historic land-use change and shifting natural disturbance patterns on one hand, and to the effects of N and CO2 fertilization and climate change on the other, remains problematic (Houghton, 2003b). For the USA, for example, the fraction of carbon sink attributable to changes in land-use and land management might be as high as 98% (Caspersen et al., 2000), or as low as 40% (Schimel et al., 2001). Forest expansion and regrowth and associated carbon sinks were reported in many regions (Table 9.2; Figure 9.2). The expanding tree cover in South Western USA is attributed to the long-term effects of fire control but the gain in carbon storage was smaller than previously thought. The lack of consensus on factors that control the carbon balance is an obstacle to development of effective mitigations strategies.

Large year-to-year and decade scale variation of regional carbon sinks (Rodenbeck et al., 2003) make it difficult to define distinct trends. The variation reflects the effects of climatic variability, both as a direct impact on vegetation and through the effects of wild fires and other natural disturbances. There are indications that higher temperatures in boreal regions will increase fire frequency; possible drying of the Amazon basin would increase fire frequency there as well (Cox et al., 2004). Global emissions from fires in the 1997/98 El Nino year are estimated at 7,700 MtCO2/yr, 90% from tropics (Werf et al., 2004).

The picture emerging from Table 9.2 is complex because available estimates differ in the land-use types included and in the use of gross fluxes versus net carbon balance, among other variables. This makes it impossible to set a widely accepted baseline for the forestry sector globally. Thus, we had to rely on the baselines used in each regional study separately (Section, or used in each global study (Section However, this approach creates large uncertainty in assessing the overall mitigation potential in the forest sector. Baseline CO2 emissions from land-use change and forestry in 2030 are the same as or slightly lower than in 2000 (see Chapter 3, Figure 3.10).

Table 9.2: Selected estimates of carbon exchange of forests and other terrestrial vegetation with the atmosphere (in MtCO2/yr)

Regions Annual carbon flux based on international statistics Annual carbon flux during 1990s 
UN-ECE, 2000 Based on inversion of atmospheric transport models Based on land observations 
OECD North America   1,833 ± 2,2009 0 ÷ 1,1005  
Separately: Canada 340      
USA 610     
OECD Pacific 224    0±7331 
Europe 316 495 ± 7526  0 ± 7331 
Countries in Transition 1,726 3,777 ± 3,4472 1,100 ± 2,9339 
     1,181 ÷ -1,5887 
Separately: Russia  1,572 4,767 ± 2,9339 1,907± 4698 
Northern Africa   623 ± 3,5932   
Sub-Saharan Africa     -576 ±2353  
     -440 ± 1104  
     -1,283 ± 7331 
Caribbean, Central and South America   -2,310 -1,617 ± 9723 
     -1,577 ± 7334 
     -2,750 ± 1,1001 
Separately: Brazil     ± 73312 
Developing countries of South and East Asia and Middle East    -2,493 ± 2,7132  -3,997 ± 1,8331 
     -1,734 ± 5503 
     -1,283 ± 5504 
Separately: China   2,273 ± 2,4202  - 110 ± 7331 
      128 ± 9513 
Global total    4,767 ± 5,5009  -7,993 ± 2,9331 
    2,567 ± 2,93310 -3,300 ÷ 7,7005 
   4,9132 -4,00015  
    951617 -5,800 16  
Annex I (excluding Russia)     130019 

Notes: Positive values represent carbon sink, negative values represent source. Sign ÷ indicates a range of values; sign ± indicates error term.

Because of differences in methods and scope of studies (see footnotes), values from different publications are not directly comparable. They represent a sample of reported results.

1 Houghton 2003a (flux from changes in land use and land management based on land inventories); 2 Gurney et al., 2002 (inversion of atmospheric transport models, estimate for Countries in Transition applies to Europe and boreal Asia; estimate for China applies to temperate Asia); 3 Achard et al., 2004 (estimates based on remote sensing for tropical regions only); 4 DeFries, 2002 (estimates based on remote sensing for tropical regions only); 5 Potter et al., 2003 (NEP estimates based on remote sensing for 1982-1998 and ecosystem modelling, the range reflects inter-annual variability); 6 Janssens et al., 2003 (combined use of inversion and land observations; includes forest, agricultural lands and peatlands between Atlantic Ocean and Ural Mountains, excludes Turkey and Mediterranean isles); 7 Shvidenko and Nilson, 2003 (forests only, range represents difference in calculation methods); 8 Nilsson et al., 2003 (includes all vegetation); 9 Ciais et al., 2000 (inversion of atmospheric transport models, estimate for Russia applies to Siberia only); 10 Plattner et al., 2002 (revised estimate for 1980’s is 400±700); 11Nabuurs et al., 2003 (forests only); 12 Houghton et al., 2000 (Brazilian Amazon only, losses from deforestation are offset by regrowth and carbon sink in undisturbed forests); 13 Fang et al., 2005; 14 Pan et al., 2004, 15 FAO, 2006a (global net biomass loss resulting from deforestation and regrowth); 16 Denman et al.,2007 (estimate of biomass loss from deforestation), 17 Denman et al.,2007 (Residual terrestrial carbon sink), 18 EDGAR database for agriculture and forestry (see Chapter 1, Figure 1.3a/b (Olivier et al., 2005)). These include emissions from bog fires and delayed emissions from soils after land- use change, 19 (Olivier et al., 2005).