THE WORKSHOP HAS RAISED SEVERAL SCIENTIFICALLY IMPORTANT ISSUES BUT COULD NOT GIVE THEM SUFFICIENT ATTENTION.
Conceptually, the estimation of terrestrial carbon sources and sinks through the bottom-up approach requires:
(a) flux tower site measurements that serve in the development and validation of models mimicking ecosystem interactions with the atmosphere;
(b) measurements at additional sites to ensure that the models represent fluxes and processes in a larger region;
(c) gridded data sets that permit application of the models across the terrestrial landmass, from landscape to globe; these are obtained from satellite measurements where possible.
The detailed observation requirements are given in Tables 2 and 4.
Landscape heterogeneity, caused by natural (topography, disturbances) and human (land use) agents, introduces considerable complexity into the sampling design that is necessary to ensure the adequacy of the coverage for items (a) and (b) above.
For (a), global terrestrial monitoring/validation sites should encompass the complete range of climate and biome type combinations (Running et al., 1999). When the current flux towers are mapped over the annual temperature/precipitation climate space of current global vegetation, some important biomes are underrepresented (Churkina and Running, 1998; Terrestrial Observation Panel for Climate, 1998). Also, the correspondence between flux tower locations and permanent ecological field sites is low, illustrating the need for more combined flux and ecological measurements in order to more rigorously test the models that estimate fluxes across the terrestrial ecosystems. Thus, the design of a comprehensive flux tower network requires further analysis, taking into consideration various axes of the 'ecological space' such as climate, vegetation functional type, disturbance, succession, land use, and possibly others. While a flux tower network for TCO must take advantage of existing sites to the maximum extent possible, future sites should be established to fill gaps in coverage that are identified through the above process.
For (b), the greatest potential is offered by ecological measurements of variables such as biomass and NPP. In addition, the potential of roving flux towers should be explored. Each such tower is associated with a nearby fixed tower and is placed at a site for a limited time period (days to weeks). One tower thus permits a more detailed sampling of the 'ecological space', providing a richer data set for model validation and scaling than would be possible with fixed towers. Measurement and characterization methodologies have been developed in the context of satellite product validation (Campbell et al., 1999; Chen et al., 1999). They rely on in situ observations in conjunction higher resolution satellite data, as an intermediate scale at which an entire pixel can be characterized on the ground.
A similar site selection issue concerns the top-down approach. The existing trace gas sampling sites are mostly at remote marine locations, and thus are insufficient for a detailed atmospheric inversion. The selection of additional sites should be guided by the requirement to maximally reduce uncertainties in the estimated distribution of sources and sinks. While the locations are not obvious, objective analysis and modelling methods can be used to identify these, as further discussed on page 38.
A systematic examination and application of the above concepts in the context of TCO remains. This should be done for both top-down and bottom-up approaches, in a coordinated manner. It should result in the establishment of objective criteria for the selection of sites for the different observation methods. From these criteria and given the present distributions, the conceptual and practical feasibility of enhancing current networks to fill the gaps can be established.
The application of the dual constraint method requires the knowledge of certain variables at all points where carbon fluxes are to be estimated (Table 1). Some of these may be obtained from satellite measurements, presently or in the near future. Others cannot be obtained now, but the evolution of satellite technologies offers a realistic prospect that their measurement will be possible in the foreseeable future (e.g. canopy chemistry). Yet others cannot be expected from satellite observations, for different reasons. They include climate, hydrological data, and others. For terrestrial carbon observations, especially critical are soil physical and chemical properties, and land use (including disturbances) history. Although data sets for soils and land use have been compiled (e.g. Global Soils Data Task, 1999; Imam et al., 1999), they have been limited by the accessibility of data to the international scientific community. More detailed data sets exist, but often at sub-national levels and their compilation would thus entail significant effort. Before undertaking additional activities in this respect, it would be important to know which areas/aspects are in greatest need for improvement from the viewpoint of estimating terrestrial carbon fluxes.
Data on carbon emissions caused by human activities are essential for the use of the dual constraint approach to estimating carbon sources and sinks. Such data are available at the national level due to the UNFCCC reporting requirements (IPCC, 1996) but their timely availability at the regional to local levels is virtually non-existent. In a longer term, satellite sensors may provide three-dimensional, temporal distribution of CO2 in the atmosphere. The issue of near-term observation requirements for such emissions and potential solutions needs to be discussed. Further attention also needs to be given to observations that would be required to complement satellite CO2 measurements and the accuracy/resolution of the latter.
The transfer of carbon between pools becomes important in characterizing the global carbon budget when it takes place between pools that differ widely in their present or future turnover rates. At the workshop, discussion has focused mostly on carbon fluxes between the ecosystem and the atmosphere. The most important transfers involve terrestrial dissolved organic (DOC) and inorganic (DIC) carbon. Carbon moves from soils undergoing relatively rapid carbon turnover (tens of years) through groundwater to fresh water (lakes, rivers, reservoirs), or directly from soils to fresh water. There, the dissolved carbon is captured in slow-turnover lakes (thousands of years) or carried to the sea for eventual incorporation into very-slow turnover sediments (millions of years). Schlesinger's (1997) analysis of extant data concluded that globally, over 0.8 Gt C yr-1 enters rivers and groundwater, with DIC slightly exceeding DOC. Both quantities are likely to increase when plants are immersed in higher atmospheric CO2 concentrations, as carbon transfer between above- and below-ground terrestrial pools increases.
While these transfers do not affect the fluxes between the ecosystems and the atmosphere, they are important for understanding the global carbon cycle and thus its predictability. It should also be noted that it is not the magnitude of the different fluxes but changes in these that are of key importance. Observations that can provide reliable estimates of the DOC/DIC include separate measures of carbon in soils, groundwater, streams and lakes. The data are generally not amenable to remote sensing but rather must be gathered in situ. Measurements in soils and groundwater will be most useful if they include recurring measurements of the age of the carbon in each of these pools. This issue also required further attention, and its implications for TCO need to be examined.
Discussions at the workshop focused on the CO2 as the most important greenhouse gas. However, other carbon gases are also involved in the fluxes between the biosphere and the atmosphere. While NMVOC gases play a significant role under some circumstances, CH4 is arguably the most important, especially in wetlands where it is produced under anaerobic conditions. Globally, the increase in atmospheric CH4 concentration has stabilized in recent years, but the understanding of the CH4 status and trend is important for the knowledge of the global carbon cycle (and thus its predictability). Satellite sensing technology will also play an important role at the regional to global scales in the near future, with the successful launch of MOPITT on Terra in December 1999. The need for satellite and in situ observations of these gases and the implications for TCO require further discussion.