Emily Matthews is Senior
Associate in the Information
Program at the World Resources
Institute, Washington, DC,
is a Senior
Lecturer in the School of
Geography, University of
Leeds, United Kingdom.
A survey of organizations and individuals with a professional interest in global forest data looks at the methodology and findings of FAO's Global Forest Resources Assessment, and suggests future approaches.
Users of forest data include not only national governments seeking to support the development of forest policies and programmes, but also the international scientific community, conservation planners, forestry professionals, environmental non-governmental organizations (NGOs), bilateral and multilateral lending agencies, forest products industries and intergovernmental policy bodies. Many of these users have expressed frustration over the perceived continuing inaccuracy of the global data available. The release of the results of FAO's Global Forest Resources Assessment 2000 (FRA 2000) - the only source of globally consistent data and information on forest area, condition and management - thus stirred an unusual level of interest and controversy. This article summarizes the viewpoints of a number of organizations and individuals with a strong professional interest in global forest data, based on a survey carried out by the World Resources Institute at FAO's request. Users were invited to share their thoughts on the methodology and findings of FRA 2000 and its predecessors, and on the objectives and design of future global forest assessments.
The survey was limited, involving telephone interviews and e-mail exchanges with contacts at 13 institutions (see Box below) and seven individuals speaking on their own behalf. While this is not a statistically significant sample of the user community, an effort was made to sample organizations with particular interest, expertise and/or influence in global forest data collection, analysis and policy development.
The article focuses primarily on comments related to forest area and forest area change over time, because these issues proved to be of greatest concern to the users surveyed. Although users welcomed FAO's increasing emphasis on qualitative forest information, they also expressed concern that this emphasis may spread too thinly the resources available for FAO's programme in forest resources assessment. It was implied that expansion of FRA could make new or broadened institutional arrangements necessary, probably involving more use of partnerships. Users were sensitive to the political and cultural issues surrounding any global monitoring exercise involving the collection and analysis of data on national resources. However, many respondents doubted that the current approach, based primarily on national data sources, can adequately serve the needs of global scientific research and policy-making. The article closes with a set of suggested options for achieving the "new vision and approach" called for by FAO for the next Global Forest Resources Assessment.
Users appreciated FRA 2000 for its coverage of forest inventory, environmental and management issues. This broad scope provides strong endorsement for the view - once controversial - that a comprehensive approach, going beyond traditional parameters relevant to timber production, should be the goal of forest monitoring and assessment.
The report provides in one volume considerable detail on forest cover and characteristics at the national and regional levels. The data harmonization methods used by FAO aim to synthesize this information into a global "big picture" that is unavailable from any other organization. The extensive data tables enable comparisons among countries, and the ten-year time span of the report provides trend data for changes in forest area.
FAO made available much of the data and supporting documentation on its Web site well in advance of publication of the final report. Comments were invited and were received, both positive and negative. Such openness was welcomed by users.
FRA 2000 was compiled in a more collaborative manner than earlier assessments. FAO developed the methodology through a process of consultation with experts outside the Organization. Significant elements of work were undertaken by other institutions. For example, the United Nations Environment Programme's World Conservation Monitoring Centre (UNEP-WCMC) provided analysis based on its protected area database; Rutgers University in the United States conducted a literature review of scientific publications on tropical deforestation; and the new global maps were produced with support from the United States Geological Survey's Eros Data Center (EDC), UNEP-WCMC, the International Institute for Applied Systems Analysis (IIASA) and other institutes and government agencies around the world. Users commented that more collaboration of this kind will be essential to the success of future forest resources assessments, as the range and depth of issues to be covered continue to increase.
Users of the report welcomed FAO's reversion to the use of linear projections and expert opinion, rather than modelling, in adjusting forest area estimates to a common reference year of 2000 and estimating deforestation rates for countries with inadequate data sources. This approach is believed to reflect more accurately the uncertainty surrounding inventory data on forest cover and change rates in much of the tropics.
Users welcomed the greater transparency regarding the collection and analysis of underlying data in FRA 2000. For the first time, users have ready access to the original data on forest cover provided by countries, allowing some insight into the process by which FAO reclassified these national data to conform to a global classification system. Detailed technical papers available on the FAO Web site provide supplemental explanations of the FRA 2000 process and methodologies.
FRA 2000 is the first FAO assessment to present global maps of forest cover, ecological zones and protected areas. The maps are not adequate to develop national-level estimates of forest cover (nor are they intended to do so); however, one expert noted that the map of ecological zones is an improvement on the ecofloristic zone map used in FRA 1990, which some sources had criticized, especially in relation to the estimates for the main types of tropical forest, particularly in Africa (Grainger, 1996). The new map, based on the Köppen climatic zone system used by Bailey (1989) to delineate the boundaries of major biomes, represents tropical biomes better than earlier maps. For example, estimates of the boundary between tropical rain forest and tropical moist deciduous forest were considered more realistic, although the boundary between tropical moist deciduous forest and tropical dry forest in Africa is still set in such a way as to include much open savannah woodland within the tropical moist deciduous category.
The map of ecological zones used in FRA 2000 is an improvement over FRA 1990, although the boundary between tropical moist deciduous forest and tropical dry forest in Africa is still set in such a way as to include much open savannah woodland (shown in Senegal) within the tropical moist deciduous category
- FAO FORESTRY DEPARTMENT/FO-0414/C. PALMBERG-LERCHE
The main areas of concern to users were accuracy, comparability with earlier reports and definitions.
Despite a slight improvement over the past decade in the availability of recent inventory and survey data, the accuracy of estimates for many developing countries was still considered to be relatively poor.
Tropical countries. Respondents reported frustration over the unreliability of estimates of the extent of tropical forests and the rate at which they are declining in area (deforestation) and quality (degradation). FAO acknowledges that the quality of primary data available on tropical forest resources remains very poor. Users interviewed noted that the accuracy of national estimates supplied to FAO was affected by two major sources of error.
First, forests are not monitored comprehensively or frequently enough in most tropical countries to map their extent accurately or to track their rate of change. In the absence of inventory data for specific dates (e.g. 1990 and 2000), FAO's latest estimates of forest area and change over time were often based on projections and expert opinion and thus, in the opinion of most of those interviewed, remain educated guesses. FAO acknowledges satellite imagery as a useful supplement to inventory information, but because of limited resources it has relied on a 10 percent sample of tropical forests to track deforestation since 1980. The statistical validity of this level of sampling has been strongly challenged (e.g. Tucker and Townshend, 2000). Clearly, even if the sample results can legitimately be extrapolated to the tropical region as a whole, they cannot be used as the basis for country estimates. In some countries with very poor inventory data, just one or two satellite scenes appear to have been the prime source of new information.
Second, as FAO has acknowledged (FAO, 2001), estimates of open woodland areas are far less accurate than those of closed forest because it is difficult to monitor woodlands by remote-sensing techniques and government forestry agencies tend not to survey them as part of normal forest inventories. Differences in definitions used among countries further complicate this issue (see below). The authors note that if open woodlands account for about 40 percent of all tropical forests as estimated in FRA 1980, the error attached to their area makes a significant contribution to the error for the area of tropical forests as a whole. This is likely to be a problem for years to come.
Both sources of error are outside FAO's control if global estimates are based on data provided by national governments and if capacity building in poor countries remains underfunded.
Estimates of open woodland areas (shown here, Chile) are less accurate than those of closed forest because it is difficult to monitor woodlands by remote sensing techniques and government forestry agencies tend not to survey them as part of normal forest inventories
- GLOBAL FOREST WATCH CHILE
Countries in the temperate and boreal regions. The quality of data from developed countries is generally better than that from developing countries, but some respondents pointed out that differences in national forestry definitions and systems of measurement, and the use of different reference periods, were still causing problems. In particular, the methodologies used by the Russian Federation and Canada to define and measure natural forest area and change were seen to be inconsistent with those of other countries. Because these two countries account for some 65 percent of all forests in the developed country survey, the results are likely to be skewed for the entire temperate and boreal forest region.
The Russian Federation (pictured) and Canada, which account for some 65 percent of all forests in the developed country survey, used different methods than the other countries to define and measure natural forest area and change; this may influence results for the entire temperate and boreal forest region
- GLOBAL FOREST WATCH RUSSIA/GREENPEACE
Forest carbon stores. The limitations of national forest area data have an adverse effect on investigations related to climate change and biodiversity. FAO has stated that the FRA data, as the world's most comprehensive data set on the world's forests, may become an important basis for carbon accounting under the Kyoto Protocol (FAO, 2000a). However, the FRA 2000 report is frank in acknowledging that volume and biomass data were not widely available for the developing countries and that existing data were based on widely differing standards, terms and definitions (FAO, 2001).
In FRA 2000, volume and biomass data were derived by multiplying estimated volume and biomass per hectare for various forest types with the estimated area of each forest type. Perhaps no more could have been done given the data available, but many respondents noted the need for a systematic effort to measure global carbon stores more accurately in the future. The tropical countries with the greatest carbon stores (which stand to benefit most from the Clean Development Mechanism) are frequently those without adequate capacity to measure them. Respondents feared that carbon credits and trading systems could be developed based on sometimes erroneous data, which could discredit climate change mitigation efforts based on carbon sinks. The authors note that trading in a variety of ecosystem goods and services is likely to become more widespread in future, presenting a major opportunity to press the case for improved carbon stock monitoring.
Biodiversity. The number and size of protected areas in forests are important proxy indicators of biodiversity. FRA 1990 simply listed the number and area of sites protected under the World Conservation Union (IUCN) categories I to II and categories III to V (FAO, 1995). In FRA 2000 a more sophisticated methodology was introduced, but some users objected that the results could be misinterpreted. For developed countries, FRA 2000 aggregated areas in IUCN categories I through VI. Categories V and VI represent very low levels of protection and may include areas under forest management for productive purposes (as FAO acknowledges), so it is unfortunate that these two areas were not listed separately.
The developing countries were not asked to provide information on protected areas. FAO overlaid the UNEP-WCMC protected areas database on the new FRA 2000 global forest map to determine protected forest areas and subsequently disaggregated them by forest type. With this approach the area of protected forests could be estimated at the regional level, but disaggregated data could not be provided at the national level. Good mapped data on protected areas to use for the overlay were limited; boundary data were either unavailable or of uncertain accuracy for many of the world's protected areas. FAO cited the technical difficulties involved in generating area data from point reference data, which were all that were available for some tropical countries (FAO, 2001).
FRA 2000 does not contain information on the protection status of protected areas or on the status of biodiversity in forests outside protected areas.
Reporting on forest biodiversity status is a critical but immense task which some respondents believed should not be undertaken by FAO. They argued that the attempt to cover all the environmental and management issues relevant to global forest resources could spread the resources available to FAO's Forest Resources Assessment Programme too thinly, and that largely environmental issues such as biodiversity are increasingly well covered by other organizations. Some respondents suggested that FAO should concentrate on collecting basic forest statistics (monitoring) while a number of collaborative partners could take responsibility for specific areas of analysis (assessment).
FAO's policy is to publish the best available estimates on forests and forest cover, even if they show some discontinuity with results of earlier assessments. Many respondents, however, complained that successive FRA reports differ so much in methodologies and presentation that clear comparisons among the data are impossible. Consistent data time series do not exist beyond the decade spanned by each report. Lack of information on long-term trends in forest area was one of the main concerns of users surveyed.
Changing baselines. In FRA 2000 a globally consistent definition of forests was adopted: a 10 percent canopy cover threshold was used for all countries. A 20 percent threshold was formerly used for the developed countries. In addition, new methodologies were introduced to estimate forest area in the developing countries. These changes and adjustments were projected retrospectively to develop a revised estimate of global forest area in 1990, the baseline year from which changes in forest area were calculated. The new baseline is 15 percent higher than that reported in FRA 1990. The new baseline may or may not be more accurate than the original estimate, but the methodology-induced 15 percent increase in global forest area is actually greater than the decrease in global forest area estimated to have occurred during the 1990s. In the tropics, there were no major changes in definitions, but the differences between forest area estimates in successive FRA reports is still striking. As an example, various estimates of natural forest area and forest loss in tropical Africa since 1980 are shown in the Figure. Even specialists find such shifts frustrating, and they are likely to be impenetrable to lay readers. Different interpretations of the FRA data could lead to further inconsistencies in publications based on these data.
Successive estimates of natural forest cover in tropical Africa
Aggregating natural and planted forests. Most respondents were concerned that natural forest area and plantation forest area were not reported separately in FRA 2000, as they were in FRA 1990. To compare a country's natural forest area or planted forest area as reported in FRA 2000 with previous estimates readers must first disaggregate natural forest area and plantation area using the global tables appended to the report. Respondents noted that this type of comparison will be confounded by two problems. First, because of the revised baseline developed for FRA 2000, readers cannot use national estimates of natural forest area from FRA 1990, but must calculate natural forest area 1990 from FRA 2000. One way to do this is to make a subtraction from total forest area, using the plantation area 2000 and calculating back to 1990 using the average annual planting rate. However, because national annual planting rates are often available for only one year, and are used by FAO to project year 2000 plantation areas in the first place, comparisons of plantation area - and hence natural forest area - made in this way are circular. Second, because FRA 2000 does not provide revised estimates of plantation area in 1990 at the country level, readers might use 1990 plantation area data that were published in FRA 1990, and subtract these from total forest area 1990 provided in FRA 2000. However, because FAO's methodology for estimating plantation area has changed over the 1990s, the two sets of plantation area data should not be compared in this way.
Many FRA users surveyed disagreed in principle with the aggregation of natural forest and plantation area. Although it might be interesting in some respects to monitor the change over time in total forest, the two types of forest often differ in composition, biodiversity levels, wood production and methods and intensity of harvesting and exploitation. Some respondents also worried that the aggregation of natural forest and plantations could encourage the perception that an increase in plantation area can offset the various impacts of an equivalent fall in natural forest area.
Aggregating forest area losses and gains. Deforestation rates and forest expansion rates are aggregated in FRA 2000 to give a net forest area change rate. Natural forest losses (through conversion to other land uses) and gains (through reforestation and afforestation) are disaggregated at the global and tropical/non-tropical levels, but not at the regional or country levels. Therefore no meaningful country-level comparisons can be made with natural forest loss rates provided in earlier FRA reports. The problem, noted by all respondents, is that deforestation (largely caused by clearance for agriculture) and reforestation or afforestation (usually responding to demand for wood fibre or tree crops) are fundamentally different processes; they felt that to understand what is happening in the world's forests both trends should be monitored and reported independently.
Respondents were also concerned that since deforestation is spatially concentrated in the tropics, publication of a global mean net change rate can be misleading. They indicated that they would have preferred a greater emphasis on the disaggregated regional trends, rather than on the global rate of change.
Changing ecological zones. Other comparability problems were reported as possible impediments to more specialized uses of the data from FRA 2000, such as charting of trends in particular forest types, for example, tropical moist forest or open savannah woodlands. In FRA 2000 only global figures were published for the areas of each forest type. The percentages of total forest in each ecological zone are listed for each country, but not the areas (the ecological zone map being too coarse in scale to allow such differentiation). Although the new ecological zone map used in FRA 2000 is an improvement over the ecofloristic zone map used in FRA 1990, its estimates for different types of both tropical and temperate forests are, by definition, not comparable with those in FRA 1990.
Use of expert opinion. It was noted that serious comparability problems result from the use of expert judgement to estimate tropical forest area and deforestation rates when adequate inventory data are not available. Expert opinion is subjective, and consistency in subsequent FRA reports cannot be guaranteed. The margins of error cannot be quantified with any degree of reliability.
Most respondents were concerned that natural forest area and forest plantation area were not shown separately in FRA 2000; pictured, a plantation in Zimbabwe
- FAO FORESTRY DEPARTMENT/FO-0402/C. PALMBERG-LERCHE
Common, agreed definitions of such terms as "forest" and "deforestation" are critical to consistent, replicable global forest assessments. Yet many respondents pointed out that definitions that are satisfactory to all potential users of the FRA reports remain elusive.
The definition of forest used in FRA 2000, as agreed by international experts at the FAO Expert Consultation on Global Forest Resources Assessment 2000, known as Kotka III (Finnish Forest Research Institute, 1996), is land with tree crown cover of more than 10 percent and area of more than 0.5 ha, which is not used primarily for agricultural purposes. The definition of deforestation used in FRA 2000 (after some modifications) is "the conversion of forest to another land use or the long-term reduction of the tree canopy cover below the minimum 10 percent threshold" (FAO, 2001). A note explains that the definition specifically excludes areas where trees have been removed as a result of harvesting or logging and where the forest is expected to regenerate naturally or following silvicultural intervention. Definitions for both terms thus encompass both land cover and land use elements.
A land use definition of forest that includes land currently bare of trees can be useful for policy-making related to forest management for timber production. But it was noted that forest land that has been cleared of trees or thinned in the course of selective felling has ecological characteristics very different from those of closed forest. Ecologists, climate change scientists, conservationists and others focused on aspects of forest quality (for example, degree of fragmentation) have argued strongly in favour of a land cover definition, and preferably one that distinguishes several different canopy cover thresholds. Data on harvested area are provided in FRA 2000, but were available for only a few developing countries.
Some of the respondents pointed out that a practical problem with the land use definition concerns the difficulty of anticipating whether forest "expected to regenerate naturally" will actually do so, or whether planned replanting will actually occur. Distinguishing between temporary and permanent removal of tree cover is often impossible in the tropics, where land use tends to be unpredictable because of volatile economics and insecure land tenure. Harvesting, clearing for cultivation and abandonment to possible regrowth or replanting tend to cause rapid changes which, so far as possible, need to be monitored as they occur.
Equally serious difficulties arise in attempts to reconcile differing national definitions of various types of forest - montane forest, swamp forest, wooded grassland, wooded savannah and so on. The harmonization process involves reclassifying numerous categories of forest land into the broad categories of closed forest, open forest, plantations, shrubs, forest fallow and other land. Respondents commented that the harmonization process appears to be particularly inconsistent in central Africa, where wooded savannah is classified as closed forest in some cases and as open forest in others. For example, FAO forest area data for Gabon, Cameroon and the Central African Republic were considered inconsistent with other estimates based on maps or remote-sensing images of vegetation cover that record "where the trees are" but do not attempt to reconcile national definitions of forest (e.g. Mayaux, Achard and Malingreau, 1998).
Distinguishing between temporary and permanent removal of tree cover is often impossible in the tropics (shown here Sumatra, Indonesia), where land use tends to be unpredictable and changes rapid
- FOREST WATCH INDONESIA
In the light of these responses to FRA 2000, what is the best way to move forward to provide the next comprehensive global forest data set and analysis of that data set? FAO has already called for a "new vision and approach" for the next Global Forest Resources Assessment (FAO, 2000b). The authors believe that it is first imperative to identify the major stakeholders in the FRA process - which include governments, subnational groups, researchers, advocacy groups and the the private sector - and determine their needs.
The following options, formulated by the authors based on discussions with the respondents, may help define alternative visions and approaches.
Enhance the present organizational design. FAO has continued to refine the design of FRA since 1980. In the absence of any major policy shifts, it is likely that incremental improvement would continue. In this scenario, the next FRA would differ in some respects from FRA 2000, but national forest statistics would continue to be the main source of data. This approach has the advantage of continuity and it maintains FAO's key partnership with national governments. On the other hand, if this approach is kept it is likely that the same constraints that affected FRA 2000 will also restrict the accuracy and usefulness of FRA 2010.
Seek funding for dedicated research centres in member countries. A second option, which is consistent with FAO's partnership with national governments and its policy of strengthening the capabilities of developing countries, would be to stimulate funding for the establishment and staffing of dedicated research centres in key countries that have a large proportion of the world's tropical forests and are in proximity to other heavily forested countries. FAO has already launched a pilot programme in a limited number of countries to improve forest monitoring through a genuine partnership approach; this may have potential for expansion. A drawback to this approach would be probable delays in raising of funds, training and construction of facilities.
Adopt a consortium design for data collection. Another strategy would be for FAO to establish and coordinate a consortium of a small number of remote-sensing centres, each responsible for data collection for one or more regions or subregions. This approach could produce results in time for the next FRA. However, the remote-sensing data would not be a substitute for national on-the-ground inventories. The approach might also be criticized as too centralized, but in time each member could form its own regional consortium to arrive at a more decentralized mode of operation.
The processes already developed under the FRA Programme would enable the linkage of global data and information sets produced by the consortium with national-level data gathering and field validation of results. A regular monitoring process that combines advanced technologies with the current bottom-up approach could exceed the current performance of either approach alone. If FAO's burden of data acquisition were reduced, FAO would be free to focus on supporting the regular collection and analysis of data for a limited set of core variables, providing the foundation for integration of additional information, modelling and analysis by independent research institutes, NGOs and other competent bodies.
Adopt a consortium design for data processing. As indicated above, a number of respondents felt that the resources allocated to FAO's FRA Programme have been spread too thinly in the attempt to provide substantive information on trends in forest quality in addition to basic data on forest area and rates of change. Another consortium option (which could be pursued in parallel with or in place of the preceding option) would be for FAO to "contract out" to other national and international bodies the analysis of the implications of changes in forest area. Relevant issues for analysis would include biomass and carbon stocks, biodiversity, sustainability of forest management and timber reserves and production. Areas such as forest use and management, production and trade of forest products and the socio-economic role of forests as sources of employment could be given greater priority than in previous assessments.
Many universities, research institutes and other groups around the world are already collecting and analysing forest-related information at the national, regional and global levels. Their contributions could be greatly strengthened if their efforts were harmonized around a commonly agreed global forest information agenda. Such an agenda, with the mandate of FAO behind it, would identify priority monitoring and assessment needs for the coming decade and beyond, based on an analysis of user needs and the degree to which those needs are being filled by existing efforts. Some logical partners include the United Nations Commission for Europe (ECE) in Geneva, the International Tropical Timber Organization (ITTO) and the Center for International Forestry Research (CIFOR).
Propose the establishment of a global forest monitoring organization. The alternative to relying on outside bodies to collect and process satellite and other forest data would be to work towards the formation of a dedicated global forest monitoring organization. FAO has already proposed a World Forest Survey which bears some similarities to this idea, but the proposal has not so far attracted the necessary financial support. Nor has the international community yet agreed to establish a forest convention, which might give FAO (if specified as the lead agency) the mandate to establish a comprehensive global forest monitoring programme.
In the absence of a forest convention, another strategy would be to propose the establishment of a United Nations Global Environmental Monitoring Organization with which FAO and other UN agencies would collaborate, possibly together with public and private-sector organizations outside the UN system. Such a body would supply much of the basic data on forest areas and rates of change necessary for the FRA reports. This would allow FAO to continue to collate estimates of forest resources submitted by its member countries and to work with a consortium to add value to its basic data as described in the preceding option.
It is essential to recognize the demands of forest information users and the possibility that other bodies will strive to supply these demands if there is a perceived "gap in the market". The increasing availability and quality of satellite imagery is already leading to the emergence of other forest surveys from outside the UN system. In the 1990s, bodies such as the TREES Programme of the European Commission Joint Research Centre and WCMC produced estimates of world forest cover. More recently, UNEP, working with the United States Geological Survey, produced an estimate of the world's remaining closed forest area and highlighted differences with FAO findings (UNEP, 2001). None of these surveys matches the FRA reports in terms of global coverage, and their differing methodologies mean that regional results cannot be combined to produce a global whole. Nor do they have the mandate to report national information to an international audience. However, these deficiencies could in principle be overcome if there were the international will to develop a consistent global data set. This raises a challenge for FAO as it faces the next Global Forest Resources Assessment. Will the Organization seize the initiative in trying to shape the global forest monitoring community of the future, or will it stand by while alternative systems of global forest monitoring become established that could create confusion among users seeking authoritative forest data?
Users surveyed concurred that the state of knowledge regarding forests today falls short of what is needed and of what could be achieved with greater resources, more collaboration and more use of new technologies. In the opinion of most of the survey respondents, current monitoring efforts are not adequately meeting the increasingly pressing needs of policy-makers or scientists for information on the global extent and quality of forest resources. The global forest estate is diminishing in quantity and - in some respects - quality, even as understanding of the importance of forest products and environmental services grows. New international environmental agreements will stand or fall on the trustworthiness of data underpinning national goals and commitments to meet them.
FAO, its member countries and other interested organizations together can develop new information-gathering and communication tools that will capitalize on the existing strengths of FAO and its networks and on the tremendous capacity of remote sensing and other new technologies. Some changes in approach will be needed to resolve the problems summarized in this article. With increasing political and public interest in forests, and with new technologies now making better forest data available at affordable cost, the next FRA represents an unprecedented opportunity for improvements.
Bailey, R.G. 1989. Explanatory supplement to ecoregions map of the continents. Environmental Conservation, 16(4): 307-309.
FAO. 1995. Forest Resources Assessment 1990: global synthesis. FAO Forestry Paper No. 124. Rome.
FAO. 2000a. Climate change and the Kyoto Protocol: key forestry-related issues. Secretariat Note. Fifteenth session of the Committee on Forestry, Rome, 12-16 March 2001. Rome.
FAO. 2000b. Global forest survey: concept paper. Forest Resources Assessment Programme Working Paper No. 28. Rome.
FAO. 2001. Global Forest Resources Assessment 2000 - main report. FAO Forestry Paper No. 140. Rome.
Finnish Forest Research Institute. 1996. Proceedings of FAO Expert Consultation on Global Forest Resources Assessment 2000 in cooperation with ECE and UNEP with the support of the Government of Finland (Kotka III), Kotka, Finland, 10-14 June 1996, ed. A. Nyyssönen & A. Ahti. Research Papers No. 620. Helsinki, Finland.
Grainger, A. 1996. An evaluation of FAO's Tropical Forest Resource Assessment 1990. Geographical Journal, 162: 73-79.
Mayaux, P., Achard, F. & Malingreau, J.-P. 1998. Global tropical forest area measurements derived from coarse resolution satellite imagery: a comparison with other approaches. Environmental Conservation, 25.
Tucker, C.J. & Townshend, J.R.G. 2000. Strategies for monitoring tropical deforestation using satellite data. International Journal of Remote Sensing, 21(6): 1461-1472.
United Nations Environment Programme (UNEP). 2001. An assessment of the status of the world's remaining closed forests. UNEP/DEWA/TR 01-2. Nairobi, Kenya.