Forests are found predominantly in moist climates that enjoy at least one moderately warm season. Within this biome – the total assemblage of plant and animal life – ecologists define anywhere between five and twelve major forest types.
Land cover maps define forest areas according to minimum thresholds of area, tree height, and percentage of land area covered by the tree canopy (percent canopy cover). These thresholds are a necessary, but essentially arbitrary, means of distinguishing forests from neighboring ecosystems, such as woodlands or savanna.
Sometimes forest boundaries are naturally distinct, as in the tree line that marks the upper limit of tree growth on mountain slopes. Human modification often creates clear and abrupt transitions in vegetation cover, for example, where agricultural land abuts closed canopy forest. In many other areas, forests shade gradually into other ecosystems, in a complex mosaic of vegetation types. Under such circumstances, land cover maps pose artificial, discrete boundaries where none exist on the ground.
Experts have attempted repeatedly to map the extent and distribution of the world’s forests and woodlands. Betwen 1923 and 1985, at least 26 calculations of closed forest land were made; they ranged from 24 million square kilometers to 65 million square kilometers, with no discernible trend over time.
The difficulty stems from aggregating national inventory data, which date from different years and use different definitions of forest, particularly in the distinction between closed and open forest. Definitions vary among countries and international organizations, and even among agencies within countries. Over the past decade, efforts to understand the global climate, and the regulating role played by the world’s ecosystems, has encouraged “top down” analysis, in which satellite data are used to generate data at the regional, continental, and global levels. This approach provides a more consistent picture of forest cover than attempts to sum and harmonize national data, but requires verification in the field.
The following summarizes key findings of the PAGE study regarding the extent of, and changes in, the world’s forested areas, as defined by a number of leading land cover classification schemes.
Conditions and trends
- Forests cover about one quarter of the world’s land surface, excluding Greenland and Antarctica. Just over half are found in developing countries.
- Global forest cover has been reduced by at least 20 percent since preagricultural times, possibly by 50 percent. Forest area has increased slightly since 1980 in industrial countries, but has declined by at least 10 percent in developing countries.
- Tropical deforestation rates are uncertain, but probably exceed 130,000 km2 per year. About 40 percent of forests are relatively undisturbed by human activity, though nearly half of these are likely to be developed soon.
- Nearly all forests in Europe and the United States are under some degree of management.
- Mixed forest/agriculture zones are spreading rapidly at the edges of formerly intact forest, but this form of land use change is often not recorded as forest conversion.
- Roads are a useful proxy indicator of habitat fragmentation and degradation. The world’s expanding road network is opening up remote forests to logging, mining, and pioneer settlement. Roads also increase hunting and poaching.
- The area burned by natural forest fires is now insignificant in comparison with human-initiated fires. Tropical forest fires have increased in area and intensity in recent years, because of drought, clearance for agriculture, and land tenure disputes.
Information status and needs
- National level forest maps are often outdated and forest inventories unreliable in developing countries. Global estimates of forest area are complicated by different definitions of forest land and deforestation.
- Remote sensing data expected to become available in the next few years should improve the information base. Priority information needs include more frequent satellite surveys and higher sampling rates to catch nonrandomly distributed deforestation. Ground truthing will remain important to verify maps generated by remote sensing data.
- Knowledge of forest biological condition lags behind that for forest extent. Classification schemes for forest condition are simplistic but still difficult to implement. There is a need for agreement on what constitutes good condition in different forest types, managed for different purposes, and for indicators to monitor change, applicable at the national and subnational levels.
- Data on mixed forest/cropland land cover are poor. Vegetation classification based on thresholds and discrete boundaries work against fine scale interpretation of land use data. There is a need for higher resolution remote sensing data and information on biomass quantities.
- The global roads dataset is out of date. Information is poor in developing countries where the road network is expanding fastest. Updated digitized information on existing and planned roads would be useful.
- There is an urgent need is to improve national and international ability to estimate forest fire potential and to detect and monitor wildfires while they are still small enough to control. A number of satellite systems have been evaluated for fire detection, including AVHRR, the Defense Meteorological Satellite Program (DMSP) Operational Linescan System sensor, and the NOAA Geostationary Operational Environmental Satellite (GOES) sensor. At present all three systems, each with unique characteristics, are required to provide the best results.
Quality and availability of data
PAGE measures and indicators
Data sources and comments
Global forest cover
Food and Agriculture Organization of the United Nations (FAO). 1997a. State of the World’s Forests, 1997. Rome: FAO.
Area estimates based on national inventories, maps, some remote sensing data.
International Geosphere Biosphere Programme (IGBP). 1998. IGBP Data and Information System. IGBP-DIS Global 1-km Land Cover Set DISCover.
DeFries, R.S., M.C. Hansen, J.R.G. Townshend, A.C. Janetos, and T.R. Loveland. 2000. “A New Global 1-km Data Set of Percentage Tree Cover Derived from Remote Sensing.” Global Change Biology 6: 247-254.
Both based on AVHRR 1-km resolution remote sensing data.
Historic forest loss Matthews, E. 1983. “Global Vegetation and Land Use: New High-Resolution Data Bases for Climatic Studies.” Journal of Climate and Applied Meteorology 22: 474-487.
Estimates of historic forest loss supplemented with FAO data for post-1980 period. WCMC. Estimates of original forest cover for WRI’s Frontiers Forests study (Bryant, D., D. Nielsen, and L. Tangley. 1997. The Last Frontier Forests: Ecosystems and economies on the edge. Washington, D.C.: World Resources Institute.).
Recent deforestation
- Food and Agriculture Organization of the United Nations (FAO). 1997a. State of the World’s Forests, 1997. Rome: FAO.
- Analysis of remote sensing data for Indonesia.
- Holmes, D. 2000. Draft of 25 February. Deforestation in Indonesia: A Review of the Situation in Sumatra, Kalimantan, and Sulawesi. Draft report in preparation for the World Bank, based on mapping carried out by the Indonesian Ministry of Forestry and Estate Crops. Data are subject to final revision but are not expected to change significantly.
- Analysis of remote sensing data for Indonesia.
Degree of naturalness
- Bryant, D., D. Nielsen, and L. Tangley. 1997. The Last Frontier Forests: Ecosystems and economies on the edge. Washington, D.C.: World Resources Institute.
- Forest intactness determined by presence of roads, other development and expert opinion.
- FAO, FRA 2000 (unpublished).
- Forest naturalness determined by intensity of human intervention.
Forest/cropland transition zones Global Land Cover Characteristics Database (GLCCD), Version 1.2. 1998. Data available on-line at: http://edcdaac.usgs.gov/glcc/glcc.html. See also Loveland, T.R., B.C. Reed, J.F. Borwn, D.O. Ohlen, Z. Zhu, L. Yang, and J. Merchant. 2000. “Development of a Global Land Cover Characteristics Database and IGBP DISCover from 1-km AVHRR Data.” International Journal of Remote Sensing 21 (6-7): 1303-1330.
Land cover classification scheme modified by WRI. Methodology may overstate degree of forest modification.
Fragmentation by roads Central Africa Regional Program on the Environment (CARPE). 1998. Data CD-ROM. Published by World Resources Institute, World Wildlife Fund, World Conservation Monitoring Centre. Online at: http://carpe.gecp.virginia.edu/.
Roads database, updated by WRI for 6 Central African countries.
Forest fires Various remote sensing sources. Fires can be detected and monitored through thermal and mid-infrared imaging during the day, and by the light they emit at night.
