Inventory studies measure changes in carbon stocks over large areas, and can thus provide spatially aggregated estimates of large-scale fluxes of CO₂ over multi-annual time-scales (Box 3.1). Mid- and high latitude forests are covered by extensive national inventories based on repeated measurements of many thousands of plots. Inventories in the tropics are by comparison generally inadequate, particularly in view of the high rates of land-use change and extremely heterogeneous carbon density in many tropical ecosystems. There are still therefore large uncertainties in attempting to balance the terrestrial carbon budget on a global scale using inventory data.

The FAO Temperate and Boreal Forest Resource Assessment (TBFRA-2000) is a recent synthesis of inventories of forests and other wooded lands in Annex I (developed) countries for the early 1990s (UN-ECE/FAO, 2000). Many countries reported substantial increases in forest areas in recent years, as well as increasing carbon density in existing forests. According to TBFRA-2000, the land-atmosphere flux was -0.9 PgC/yr for all Annex I countries combined (the net annual increment of trees accounted for -1.5 PgC/yr, while losses due to fellings were 0.6 PgC/yr). Of this flux, -0.8 PgC/yr was due to uptake in “northern” forests (Europe, CIS, Japan and North America). An earlier review of individual regional and national studies by Dixon et al. (1994), highlighted in the IPCC WGII Second Assessment Report (IPCC, 1996b; Brown et al., 1996), gave a range of -0.6 to -0.9 PgC/yr for the land atmosphere flux in northern forests. While TBFRA-2000 estimated biomass of woody vegetation only, the analyses reviewed in Dixon et al. (1994) included other vegetation, soils, litter and wood products. Under the United Nations Framework Convention for Climate Change (UNFCCC) signatory countries are required to report greenhouse gas emissions, including those from land-use change and forestry. Compilation of these data implies a land-atmosphere flux of -0.6 PgC/yr for all Annex I countries, and -0.6 PgC/yr for Annex I countries in the northern latitudes only (UNFCCC, 2000). While the TBFRA synthesised country statistics and adjusted data to fit FAO definitions and methodologies for calculating carbon stocks, the UNFCCC report summarises emissions data reported by each country according to IPCC guidelines; interpretation of guidelines is variable, and not all countries had reported data on land use. The implications of definitions and methodologies in calculating carbon fluxes, particularly in relation to implementation of the Kyoto Protocol, is discussed in detail in the SRLULUCF (IPCC, 2000a).

A recent compilation of data from 478 permanent plots in mature tropical moist forests throughout the tropics over at least two decades found these were taking up carbon due to increasing rates of tree growth. Extrapolation from these plots led to an estimated land-atmosphere flux of (0.6 ± 0.3 PgC/yr in Latin America; growth trends in African and Asian forests were not significantly different from zero (Phillips et al., 1998). This net uptake is offset by emissions due to deforestation. Dixon et al. (1994) estimated tropical forests overall to be a net source of carbon with a land-atmosphere flux 1.7 ± 0.4, based mostly on FAO (1993b) inventory data and simple models of the effect of land-use change (Houghton, 1995). It will not be possible to assess trends and fluxes for the 1990s in the tropics from inventory data until a full data set is available from the FAO Global Forest Resources Assessment 2000. Among those countries that have reported land-use emissions data to the UNFCCC, there are significant discrepancies between the primary data used in emissions inventories and the data available in international surveys; for example, rates of deforestation differ from rates reported by FAO (1993b) by as much as a factor of six (Houghton and Ramakrishna, 1999).

The results of globally aggregated forest inventories show a greater uptake of carbon in forest growth than model-based calculations of the marginal effects of land-use change (e.g., Houghton, 2000). Thus, inventory studies provide independent evidence for the existence of a residual terrestrial sink; and they show that a substantial part of this sink, at least, is located in northern extratropical and tropical forests. Additional evidence from individual inventory studies in mature forests that have not undergone land-use changes shows that carbon stocks in such forests are increasing (e.g., Lugo and Brown, 1993; Phillips et al., 1998; Schulze et al., 1999). The difference between the northern extra-tropical land-atmosphere flux of around -0.8 PgC/yr calculated by inventories (TBFRA-2000) and that of -0.1 PgC/yr from land-use statistics (Houghton, 2000), both for the early 1990s, implies a residual terrestrial sink on the order of -0.7 PgC/yr in northern mid- and high latitudes. Combining this with the estimated sink of -0.6 PgC/yr in mature tropical moist forests (Phillips et al., 1998) makes it plausible that at least a significant fraction of the current global terrestrial sink (Table 3.1) could be explained by an increase of carbon stocks in extant forests. The inventory-based estimate of land-atmosphere flux in northern forests (-0.8 PgC/yr) is at the positive end of the range calculated by inverse modelling studies for the >30°N latitude band from 1990 to 1996 (-1.8 to -0.7 PgC/yr, Section 3.5.3), either because of biases in inverse modelling that might tend to increase apparent uptake in the north (Section 3.5.3), or because possible sinks in other ecosystems (e.g., temperate grassland soils) have not been considered in the inventories. In the tropics, the difference between the uptake of carbon estimated by inventory studies in mature forests of Latin America (-0.6 PgC/yr) (Phillips et al., 1998) and the estimated emissions due to deforestation in the tropics of 1.7 PgC/yr (Houghton, 2000) yields an estimated land-atmosphere flux of 1.1 PgC/yr, which is at the positive end of the range calculated by inverse modelling studies for 30°S to 30°N (-1.3 to +1.1 PgC/yr, Section 3.5.3). Again, it should be noted that possible additional sinks (e.g., in savannas) are neglected by the land-use and inventory-based calculations.

IPCC Homepage