Previous Page Table of Contents Next Page


GRIDDED DATA PRODUCTS


Forest inventory data
Non-forest inventories
Lateral carbon transfers
Input gridded data sets
Priority needs and recommendations

STUDIES OF THE TERRESTRIAL CARBON CYCLE POSE A SERIOUS METHODOLOGICAL PROBLEM OF SCALE. TERRESTRIAL ECOSYSTEM PROCESSES EXHIBIT HIGH VARIABILITY IN SPACE AND TIME THAT CAUSE LOCAL TO REGIONAL IMPACTS WHICH ARE OF POLICY AND ECONOMIC INTEREST. The aggregated impact of these ecosystems on the atmosphere is also important from the perspective of the global carbon cycle. Spatial variability is high enough that point measurements alone cannot provide adequate estimates of fluxes (or changes in stocks) over large regions, implying that models must be used.

Without regional "wall-to-wall" observations, such models cannot be convincingly evaluated because of an incomplete sampling associated with in situ measurements. Spatial/gridded in situ data sets are therefore needed for several reasons: as input to models; as constraints for model dynamics and parameterisations, and for verification of model results. Issues related to data fusion and data scaling methods need to be dealt with in model calculations so that a transparent methodology is available for review and assessment of uncertainties or error analysis.

Several types of measurements provide regional data not derived from point measurements: remote sensing, atmospheric composition measurements, and resource inventories. Remote sensing techniques have advanced substantially over the past few years, but by themselves provide only an incomplete constraint on the carbon budget of ecosystems.

Atmospheric CO2 measurements can be analysed to produce estimates of regional to continental net CO2 exchange, but these presently have wide confidence intervals and a very coarse spatial resolution. Thus regional, gridded data with a reasonable spatial resolution (£~100 km2) are in principle an important component of a convincing validation of modelled fluxes.

In situ data on key variables are needed for global model calculations of carbon fluxes and changes in stocks. These include (Table 1) soils, meteorology, tropospheric chemistry (particularly nitrogen deposition and possibly ozone), atmospheric concentration of CO2, disturbance regimes, and land use history. In addition, global or regional observations are needed for constraining and where feasible, verifying estimates of carbon fluxes and pools. For this purpose, useful parameters are forest cover, composition, biomass, age, management, products, coarse woody debris, fine litter; crop/grassland yield/biomass, management, and soil carbon and nitrogen content.

Table 1 provides a framework for assessing capabilities, gaps and priorities in relation to the use of in situ data for TCO product validation at different scales (see also page 16). Appendix 3 contains detailed listings of data sets for specific countries provided by the meeting participants (Canada, China, Europe, New Zealand, Russia, United Kingdom and United States of America). It is evident that the data sets available in individual countries differ in both type and characteristics, including measurement methods, and thus the preparation of harmonized data sets over large areas is complex.

In the following sections, conceptual issues are discussed in more detail for: inventory data (forest, page 21 and non-forest, page 33), data sets characterizing lateral carbon transfer (page 33), and data products required as input into spatial ecosystem models (page 34).

Forest inventory data

Among a country's national terrestrial carbon- related inventories, forest inventory data are usually the most extensive. Such data sets could, in principle, be used to assess modelled carbon flux estimates. Forest inventories are the basis for almost all country reports of CO2 emissions or removals due to land-use change and forestry (LUCF) under UNFCCC. According to the accepted reporting framework (IPCC, 1996), the net emissions/removals of atmospheric carbon by land-use change and forestry are assumed to be equal to the changes of carbon stocks in biomass and soil per country per year.

Since carbon flux is not a parameter included in traditional inventories, transformation of the inventory data into biomass is required. Carbon flux may then be estimated from change in biomass, using the relatively stable relationship between biomass and its carbon content. Although the 'change in stock' approach is not sufficiently sensitive to validate model-derived fluxes at annual or shorter time steps (page 13), the spatial data sets are important for increasing confidence in the model outputs. If the data are of high quality (accurate, temporally consistent) and taken over long time scales they can also provide an alternative estimate of carbon sequestration by the inventoried ecosystems.

Since forest inventories have generally been carried out for management purposes they commonly provide estimates of standing volume by tree species, but rarely include above-ground biomass and typically ignore below-ground biomass. As a remedy, an estimate of above-ground biomass may be produced from attributes such as (in the order of decreasing quality): species-specific relationships to volume; cover and crown closure relationships to volume; cover type and a 'biome-typical' regional biomass default value. The latter conversions are likely to produce results of very limited value for a rigorous evaluation of modelled fluxes and in any case, such use of inventory data must be undertaken with great care.

In this section, some of the questions in using inventories for carbon flux estimation are addressed. This is based primarily on a comparison of methods employed by member states of the European Union (EU) for estimating CO2 emissions and sinks from Land Use Change and Forestry European experience (Löewe et al., 2000; www.bib.fsagx.ac.be/coste21/report/2000-09-28.html/).

A lack of transparency, consistency, and completeness were identified by the EU analysis as the major limitations for the reporting on Chapter 5 of the National GHG inventories of the 15 EU Member States for UNFCCC and the EU Monitoring Mechanism. The state of the art of forest inventorying and carbon-sinks/emission reporting was found to be highly variable in different European countries. Some countries have a long-term forestry tradition and assessed sampling errors for the total timber volume as low as 0.6%. Others have never performed a forest inventory but sent questionnaires to forest owners. Yet others conducted an inventory some decades ago or over a period of some decades. Even within a country, highly different procedures may have been applied in different regions. Aside from the quality of the underlying forest inventory, there remains the overriding problem that forest inventories are performed for assessing marketable stem-wood and not for carbon sinks/emission reporting.

Data required

Carbon flux data is what is needed to be extracted from forest inventories. Since fluxes are not directly determined in inventory programmes, other measures may be obtained as a partial substitute. These are (in order of decreasing accuracy/value for carbon flux assessment):

To estimate carbon fluxes from periodic timber volume inventories, forest growth rates must be estimated first, with the intermediate result being the annual volume increment of roundwood overbark (m3 ha-1 y-1). This annual increment must then be expanded to the whole-tree volume increment (including leaves, fruits, branches, coarse roots, and fine roots). The next step is the conversion of the individual biomass compartments to dry matter-biomass. The carbon fraction of dry matter (summed up across all compartments) provides the carbon increment per unit area. A similar procedure is applied to calculate annual carbon release through harvest. Finally, the net annual CO2 emission or removal is calculated the difference between growth increment and harvest to be multiplied by the forest area, the latter being applicable to the individual tree species or species groups in the country.

Data available

Forest management inventory data should be available for all countries (TBFRA, 2000a and b; FAO, 2001a and b). They vary greatly among countries and regions, from nation-wide digitally stored annual inventories to data unsuitable for systematic forest monitoring because of the absence of appropriate inventory programmes. The national forest inventory data generally differ in:

From national inventories, biomass estimates needed for carbon flux estimates may in principle be obtained in various ways (in decreasing order of accuracy and usefulness for carbon flux estimation): The usefulness of the conversion of forest inventory into biomass conversion for carbon flux assessment is limited by, among others: Most countries with a significant forest industry have information from which biomass can be estimated, within constraints as those mentioned above. In practice, more accurate and reliable biomass estimates can be obtained if the inventory is based on a sample of plots which are re-measured at regular intervals. This is more accurate than re-mapping (usually from aerial photographs) with limited field sampling, especially if the sampled sites vary between the inventories. Thus, with various degrees of success it is possible to infer biomass from inventory data at a national level.

Attempts are underway to reconcile differences among national inventory-derived information in a regionally or globally consistent framework (FAO, 2001a,b and Goodale et al., 2001). For example the Contribution of Forests and Forestry to Mitigate Greenhouse Effects (COST E21, www.bib.fsagx.ac.be/coste21/report/2000-09-28.html) is a joint project of national inventory experts from European countries of the UN-ECE region, designed to harmonize and improve methods for reporting carbon pools, pool changes, and effects of management practices on carbon sequestration.

A key requirement in obtaining more reliable estimates of carbon sequestration is the combining of traditional forest inventories with data from networks of long-term forest ecosystem research sites that represent major forest types and growth areas. The most relevant network in Europe is the International Cooperative Programme Forest, which was launched in 1985 by thirty-eight countries of UN-ECE in cooperation with the European Union (www.icp-forests.org/).

Limitations of current inventories cause large uncertainties in the estimation of biomass, and especially of biomass change, and does not provide accurate flux estimates to compare with model-derived fluxes. Nevertheless, inventories are performed in all countries with a substantive forestry industry. A large number of plots have been surveyed in many countries (e.g. ~100000 plots in North America and Europe) and currently represent the main source of information for reporting of carbon sources and sinks from LUCF.

The recent "Bonn Agreement" (6th Conference of the Parties, UNFCCC) and its definitions and rules in regard to Land Use, Land-Use Change and Forestry (LULUCF, www.unfccc.int/resource/docs/cop6secpart/l07.pdf/), underline the need to improve forestry inventory systems and strengthen the scientific foundation for transforming timber volume growth increments and harvest into carbon emissions and sinks.

In parallel, there is a clear need to define and develop tools for independent source/sinks verification at national to regional scales. Improved inventories could also serve to provide independent estimates of fluxes to assess TCO models and derived products, especially over longer time intervals (e.g. 5 years). It should be noted that an Intergovernmental Panel on Climate Change (IPCC) Expert Group has been established recently to prepare 'Good Practice Guidance in Land Use Change and Forestry for the Revised 1996 Guidelines for National Greenhouse Gas Inventories'.

Other practical issues

Once available data has been identified, will it be possible to harmonize inventory data and derived biomass information? To facilitate such a process, the characteristics of the source inventories, data conversion, scale of source data, and other relevant characteristics must be determined. These characteristics will permit an assessment of the accuracy of the derived information.

The success of this process is limited from the outset by the initially collected data. This is why FAO is in the process of developing a new approach to future global forest resources assessments. The approach, known as Global Forest Survey (FAO, 2000), will allow monitoring of the forestry and related resources in each country on the basis of a standardized sampling design using biophysical and socio-economic variables. This will help build national capacities and establish a cooperative framework among parties concerned. It will also allow improved and better management of forest knowledge, thus meeting the needs of a broader range of users.

Important practical issues also include: data availability and access (e.g. data freely available vs. data with distribution limitations); costs (i.e. cost recovery programmes or industrially generated data); adequate supporting information (metadata); and data form and format as well as the related cost of conversion/transformation for easier use.

Non-forest inventories

Non-forest areas must be captured to ensure that the complete range of global carbon flux conditions are calculated. The inventory and analysis of all major ecosystem types need to be developed so as to evaluate the current stores of carbon in ecosystem compartments. Plant biomass measured from different plant pools and soil measurements of carbon pools are needed from deserts, grasslands, woodlands, wetlands, croplands, etc. Lateral fluxes from these ecosystems due to erosion and leaching need to be quantified. In addition, a method is needed to account for carbon fluxes caused by the harvest and export of these products. Current carbon accounting systems do not adequately estimate these fluxes in carbon stocks.

Lateral carbon transfers

Horizontal carbon transfers among regions are an important component of a comprehensive analysis of the terrestrial carbon budget. This is because such carbon may be part of a sink in one region and subsequently a source in another. On the other hand, atmospheric inversion studies only infer the net atmosphere-surface CO2 flux in a particular region. The horizontal flux pattern thus constitutes critical additional information for the interpretation of the inversion results. The most important horizontal transfers are:

Plant material and soils can be transported by wind and water from one location to another. Matter can be sequested in sediments of rivers, reservoirs, lakes or coastal areas. The amount of carbon transferred in this manner is difficult to quantify, but constitutes a significant contribution in regional carbon balance estimates (Pacala et al., 2001).

Quantification of the rates of loss and the rates of turnover of the deposited material is an important issue in closing the carbon accounting of terrestrial systems. The flux of dissolved organic carbon (DOC) in rivers is likely to be important, but it is also not well quantified. The human removal of harvested material in wood, fibre, or carbohydrates also needs to be accounted for in the total carbon flux from managed ecosystems.

The net export of carbon in the goods and products from one region to another is quite substantial, and the quantity and fate of the translocated carbon needs to be accounted for in atmospheric inversion calculations. For the understanding of the global carbon cycle it is also important to distinguish between anthropogenically induced (e.g. carbon trade fluxes) and natural, "background" lateral carbon fluxes (e.g. due to the river export of DOC).

Input gridded data sets

Ecosystem models are the principal method to link terrestrial and atmosphere components of TCO. However, such models are strongly dependent on spatially comprehensive and temporally detailed data inputs in gridded form or on parameters derived from detailed but spatially confined process studies.

The vast majority of data are however obtained through in situ measurements such as from flux towers or inventories that are extrapolated to provide spatial coverage. Thus, the models are constrained by accurate representation of landscape heterogeneity and its variation over time determined through landscape syntheses, whether from direct point sampling or indirect spatially comprehensive observations.

The 1995 NPP intercomparison of terrestrial biosphere models by Cramer et al. (1999) illustrates the main characteristics and data requirements of such models. Although the models themselves have improved since 1995, the basic data inputs have not changed. It is therefore possible to identify broad categories where global gridded data sets are required by some if not all models (refer to Table 1), these are: climate (including radiation, precipitation, temperature); topography; hydrology; land use/management and land cover; vegetation and soil characteristics; and disturbance. A review of the main existing data sets for the above categories is provided in Appendix 5, page 104.

It should be noted that more up-to-date or improved gridded data sets continue to be prepared through various programmes. The sponsors of these activities are national space agencies, international organizations (FAO, ICSU, WMO, etc.) national meteorological agencies, and national research programmes (e.g. land use history and disturbance data sets).

There are also projects focused on improving data access. For example, the International Satellite Land Surface Climatology Project (ISLSCP) has undertaken the assembly and publication of numerous gridded data sets, made consistent in spatial and temporal dimensions. In 1994, ISLSCP published a five-volume "Initiative I" CD-ROM collection of global data sets to support energy, water and biogeochemical cycling studies that has been extensively used by the scientific community.

In 1999, the production and publication of an expanded collection of data as ISLSCP Initiative II was started. This data set will consist of a global 10-year collection (1986 to 1995), with higher spatial (one-quarter to one degree) and temporal resolution, and improved data generation algorithms. It will also include carbon data sets designed to support global carbon cycling studies.

As with Initiative I, data collection will consist of model driver data sets, surface boundary conditions, and selected model validation data. ISLSCP II will expand the data collection in each of these areas by adding carbon data sets, topography, soils, runoff, and basin boundaries. A final Initiative II data collection is to be available in the summer of 2002. The final version will be distributed on appropriate permanent media such as CD-ROM. As data sets become available from the data providers, they will be placed on line at the Goddard Space Flight Center ISLSCP II web site (http://islscp2.gsfc.nasa.gov/). Appendix 5, page 72, provides a list of data sets included in this ISLSCP initiative.

Priority needs and recommendations

A number of issues have been raised that are important in using gridded inventory data for carbon budget studies. In the following section some of these issues are translated into recommendations.

Forest and other inventories:

Lateral transfers:


Previous Page Top of Page Next Page