The amount of carbon held in the vegetation and soils of terrestrial ecosystems varies spatially and temporally as a result of natural processes and human activities. Sources and sinks of non-CO2 GHGs (methane and nitrous oxide) are also affected by changes in land use. The issue addressed here is whether there are methods of measuring stocks, losses, and accumulations of carbon, as well as changes in the flux of CO2 and non-CO2 GHGs, in ways that are compatible with the requirements of the Kyoto Protocol-in particular, whether the precision and costs of such methods will enable changes in carbon stocks and changes in the flux of GHGs to be determined satisfactorily over the commitment period 2008-2012.
Measuring carbon sequestration or release during the commitment period as a result of human-induced activities depends on two factors: first, the type of activities to be included-namely, activities considered in Articles 3.3 (ARD), 3.4 (other LULUCF activities), and 3.7 of the Protocol-or full carbon accounting; and second, the scale of interest (project or national level). Table 2-7 presents an analysis of the various methods in terms of scale of applicability, suitability for project-level versus national-level carbon monitoring, costs, accuracy, and so forth.
Methods exist for measuring the amount of carbon in all components of terrestrial ecosystems, as well as for measuring changes in this amount. The methods vary in complexity, precision, accuracy, and cost. Different methods are appropriate for different pools and components of terrestrial carbon and for different temporal and spatial scales (see Table 2-7 and Figure 2-7). Methods used to measure carbon, or a change in carbon, are different from those used to attribute cause to an observed change in carbon (e.g., direct human activity versus natural causes). This distinction is important because the Protocol is concerned with human-induced, rather than total, changes in carbon. Even the most direct measurements on small plots do not distinguish mechanisms or yield attribution. Attribution must be inferred from controlled experiments or from ecosystem process models that are based on the mechanism thought to be held responsible for the change (e.g., land-use change versus CO2 fertilization).Table 2-7: Characteristics of methods for determining changes in carbon storage. | |||||||||||
|
|||||||||||
Scale of Applicability | Time Span | Parameter Assessed | Suitable to Monitor ARD Activities | Applicable when Alternative Definitions for ARD are Chosen | Suitable to Monitor Soils and Additional Activities1 | Suitable for Full Carbon Accounting |
Sampling Density |
Costs2 | Accuracy3 | Verifiability | |
|
|||||||||||
Vegetation Inventory | 0.01-109 ha | 1-100 yr | Aboveground stemwood volume and increment, harvesting and mortality; derived from whole-tree biomass | Yes, but design of sampling must be adapted to cover Kyoto lands | Yes, provided forest definitions used in inventories are adapted | Mainly for specific additional measures that impact forest C stock, such as thinning, fertilization, etc. | No-usually excludes soils | Project basis: 400 plots on 5,000 ha; in national-scale inventories: 1 plot represents 1,000 ha | US$ 0.05-0.6 ha-1 in national scale inventories; US$11-18 at project levels (10,000 ha) | Area: s.e. = 0.4% Growing Stock: s.e. = 0.7% Increment: s.e. = 1.1% (Tomppo, 1996) | Relatively easy |
|
|||||||||||
Soil Inventory | 0.1-103 ha | 10-1,000 yr | SOC stock and changes over time | Yes | Yes | Yes | Assesses one compartment only | Depending on soil heterogeneity ~300 sample points per 10,000 ha; one sample for every 10-cm depth | US$ 3-20 per sample | 2-3% error for analytical precision; total error much higher due to spatial heterogeneity and sampling error | Relatively easy |
|
|||||||||||
Eddy Flux | -20 ha | Day - 10 yr | Net Ecosystem Production | For verification only | For verification only | For verification only | No-excludes harvesting and decay of wood products | Required sampling density to obtain a large area representative flux must still be determined | US$ 100,000 per site initial costs; US$20,000 yr-1 running costs | 10-20% | Relatively easy through forest inventory and soil analyses |
|
|||||||||||
Flask Measurements | -109 ha | Decades | Atmospheric CO2 concentration | No | No | No | No-excludes wood products | Required sampling density to obtain a large area representative flux must still be determined | Unknown | Sample analysis is very accurate | Verification of analysis is relatively easy |
|
|||||||||||
Satellite Remote Sensing | 0.05-109 ha | Day - decades | Area (sometimes derived estimates of biomass and NPP) | Yes | Yes, provided spatial resolution is adequate | Suitable for monitoring, e.g., fire management; in general, all area-related parameters of non-ARD | No, mainly to assess areas | ~80 sites in Northern Hemisphere Integral coverage through pixel size | US$ 0.0002 ha-1 for the picture and same amount for labor to process it; aircraft-derived pictures more expensive | Precise for area measurements (15%); for biomass, less precise | Relatively easy with ground truth data |
|
|||||||||||
Ecosystem Modeling | 0.1-1 ha | Day - 100s of years | NPP, NEP per compartment | When validated with on-site data | Yes, provided model can be parameterized with new forest data | Yes, when management activities can be modeled | Yes, if all components of C cycle are included in model | Usually integral coverage | Cheap once model is developed | Uncertain; subject to many assumptions | Difficult in long term |
|
|||||||||||
Biome Models | Grid - 109 ha | Day - 100s of years | NPP, NEP per grid | No | No | Often soils are included in these | Yes, if all components of C cycle are included in model | Usually integral coverage | Cheap once model is developed | Uncertain; subject to many assumptions | Difficult |
1 Additional activities could be low tillage, drainage of peatlands, reduced-impact logging, thinning, wood products recycling. 2 Costs cannot be scaled up to larger regions using these per hectare estimates. Average costs per hectare will vary more with heterogeneity than with absolute area. 3 In the absence of a reference, estimates of accuracy are based on expert judgment. |
Other reports in this collection |