A range of methods exists for measuring and estimating the flux of GHGs from the land's surface over different spatial scales. Some of these methods are discussed in the following sections. The development of methods is an active area of research.
Flux measurements are the only practical way of measuring non-CO2 GHGs. Unlike carbon, methane and nitrous oxide do not exist as stocks on land, and changes in inventories cannot be used to infer fluxes to the atmosphere. These fluxes must be determined through direct measurement of changes in the concentration of the gas in chambers that enclose a piece of an ecosystem, through emissions ratios that relate the non-CO2 fluxes to CO2 flux, or with models that calculate net exchange from knowledge of soil properties. Estimates of flux for these gases are generally more uncertain than fluxes of CO2 because of their greater variability over time and space.
Chambers may be used to enclose a small area, or an individual component, of an ecosystem (e.g., soils, stems, leaves). Changes in the concentration of CO2 (or other gases) within the chamber or differences between the concentrations in incoming and outgoing air are used to calculate flux.
Several techniques exist for measuring the flux of CO2 for an entire ecosystem (less than 1 km2) without enclosures. The technique in most common use today is the eddy correlation/ covariance technique (described below), which measures flux directly. Other techniques that are applicable at the same spatial scale infer a flux rather than measuring it directly. These techniques include the boundary layer gradient method (Griffith and Galle, 1999), tracer techniques (Leuning et al., 1999), upwind-downwind measurements (Denmead et al., 1998), and long-path infrared spectroscopy.
The eddy covariance method is a well-developed method for measuring the exchange of CO2 between terrestrial ecosystems and the atmosphere-that is, Net Ecosystem Production (NEP) (e.g., Baldocchi et al., 1988, 1996; Moncrieff et al., 1996; Valentini et al., 2000; Aubinet et al., 2000). Measurements are continuous and semi-automatic (often with an hourly time step), and the net flux of CO2 entering or leaving the ecosystem integrates an area typically on the order of 20 ha. The precision integrated annually has a confidence interval of �30 g C m-2 yr-1 [0.3 t C ha-1 yr-1]. A slight modification of the method, a relaxed eddy correlation technique, is applicable to non-CO2 GHGs (Rinne et al., 1999).
In some sites that are affected by complex topography, ecosystem nighttime respiration can be underestimated because of air advection (e.g., drainage flows), leading to an overestimation of the annual carbon balance (greater sink or smaller source of carbon). Methods have been developed to recognize when these cases occur, and corrections have been proposed (Aubinet et al., 2000).
Automated eddy covariance measurements of CO2 fluxes have now been made at more than 65 sites worldwide (FLUXNET16), with standard measurement protocols and storage systems. A typical cost for a complete eddy covariance system is on the order of US$50,000. The cost of site infrastructure is additional and will vary according to the remoteness of the site, the height of the vegetation (whether a tall tower must be built), and the existence of other facilities. The cost of a small mast over a pasture and an insulated container to house computers may be less than US$500; the typical cost of a 30-m walk-up scaffold tower for flux measurements is US$20,000.
The eddy covariance method determines overall net carbon exchanges at stand level. It is repeatable in time and is nondestructive. It can assist in the determination of carbon budgets of forests and other terrestrial vegetation on a project level, in the validation of forest growth models, and in verifying studies of carbon balances determined by stock change methods.
The rapidly expanding network of flux towers greatly assists in the understanding of land-atmosphere exchanges but does not yet provide reliable evidence of the magnitude and location of carbon sinks because a tower site does not operate over a large enough and sufficiently unbiased sample area to represent a land cover throughout its disturbance cycle. Furthermore, the method is limited to generally flat terrain with uniform vegetation. Nevertheless, the data confirm evidence for carbon sinks quantified by stock change measurements on a statistically representative sample.
Other reports in this collection