The information in this special collection is provided by the the Watersheds of the World_CD, a product of the Water Resources eAtlas. Four organizations worked to gather and organize this data: WRI, IUCN, IWMI, and the Ramsar Convention on Wetlands. Please refer to the Water Resource eAtlas/Watersheds of the World site (http://www.iucn.org/themes/wani/eatlas/index.htm) for more information on this ongoing effort.

The information, sources and methodology to create the global maps for the Watersheds of the World_CD are incorporated in each map description under “mapping details” and “sources.” The technical notes presented here, refer specifically to the data, maps, and indicators that appear with the individual basin profiles. Data for all indicators were not available for all basins and subbasins. Non-available data are indicated in the profiles by a dash symbol (“-“). Most of the variables and indicators in the profiles were compiled through geographic information system (GIS) analysis of multiple datasets.

Watershed Boundaries
River basin boundaries are based on two datasets: a revised version of the Major Watersheds of the World dataset distributed on the GlobalARC CD-ROM by the U.S. Army Corps of Engineers Construction Engineering Research Laboratories (CERL) and the EROS Data Center HYDRO1k basin boundaries developed at the U.S. Geological Survey (http://edcdaac.usgs.gov/gtopo30/hydro/). The CERL basins were digitally derived using ETOPO5, 5-minute gridded elevation data, and known locations of rivers. The HYDRO1k is a geographic database derived from the USGS' 30 arc-second digital elevation model of the world (GTOPO30). Because of the low resolution of the elevation data used to derive the CERL base data layer, boundaries are coarse and an effort was made to refine the basin boundaries as follows. WRI revised and checked basin boundaries by overlaying ArcWorld 1:3 million rivers. In cases where rivers (except canals) crossed basin boundaries, the boundary was edited using a 1-kilometer Digital Elevation Model as a guide and redrawing the boundaries along identifiable ridges. In some cases, polygons were split to separate subbasins. After editing the boundaries, all subbasins were identified and labeled for each primary basin, using several world atlases as references. The resulting basin boundaries are from Revenga, C., S. Murray, J. Abramovitz, and A. Hammond, 1998. Watersheds of the World: Ecological Value and Vulnerability. Washington, DC: World Resources Institute.

Land Cover and Use Maps and Variables
The USGS Global Land Cover Characterization Database (GLCCD) with the International Geosphere Biosphere Programme (IGBP) classification was used to create each land cover map and to identify the extent of different land cover types within each basin (GLCCD Version 1.2. 1998. Loveland, T.R., B.C. Reed, J.F. Brown, D.O. Ohlen, Z. Zhu, L. Yang, and J. Merchant. 2000. “Development of a Global Land Cover Characteristics Data base and IGBP DISCover from 1-km AVHRR data.” International Journal of Remote Sensing 21 (6-7): 1303–1330. Data available on-line at: http://edcdaac.usgs.gov/glcc/glcc.html.) The GLCCD is derived from 1-kilometer resolution satellite imagery spanning April 1992 through March 1993. Because these data are most useful for analyzing general land cover patterns at a continental or large scale, these data are less reliable for smaller watersheds.

Each land cover basin profile map shows the different aggregated land cover types from IGBP. Please note that the extent of grasslands, savanna and shrubland are not aggregated in the basin land cover map to show the differences between strictly grassland areas and areas with increasing amounts of woody vegetation. In the land cover variables, however, grasslands, savanna and shrubland are aggregated into a single percentage value. In addition, in the basin profile land cover maps the IGBP wetland class (i.e., permanent wetland) is displayed, however, because the GLCCD does not provide much detail on wetlands coverage, the variables indicating percent wetland area are derived from the UNEP-WCMC wetlands dataset (see description of the variable below). The land cover and use variables used in the basin profiles and derived from the GLCCD include the following categories:

• Percent forest indicates the percentage of the basin classified as evergreen needleleaf forest, evergreen broadleaf forest, deciduous needleleaf forest, deciduous broadleaf forest, and mixed forest.
• Percent cropland indicates the percentage of the basin classified as cropland or as cropland/natural vegetation mosaic.
• Percent grassland indicates the percentage of the basin classified as primarily herbaceous grassland, to grassland with increasing amounts of woody vegetation including open shrubland, closed shrubland, woody savanna, and savanna.

Other variables in this basin profile section, but not derived from the GLCCD include:
Percent wetlands indicates the percentage of the basin defined as wetland according to UNEP-WCMC’s Biodiversity Map Library (UNEP-WCMC. 1998. Biodiversity Map Library. Cambridge, U.K.). Wetlands are defined as bogs, marshes, lakes, seasonal, permanent, freshwater, tidal, mangroves, and lagoons. UNEP-WCMC’s dataset was derived from Operational Navigation Charts, existing maps and expert opinion. This dataset includes wetland types that are not easily captured by remote sensing-derived data. Therefore WCMC data provide more detail than the GLCCD or other global, coarse resolution data sets that use potential vegetation, soils, and terrain to delineate wetlands. However, because of its scale, it underestimates wetlands extent, particularly seasonal wetlands, flooded forests, and wetlands in valley bottoms. In North America the area occupied by the Great Lakes, the Great Bear Lake, the Great Slave Lake, and Lake Winnipeg are not included in the calculation of wetland area. For North America wetland polygons were not differentiated by type of wetlands, instead the class field identified the proportion of the polygon--represented by a range--occupied by wetlands. To calculate wetland area, polygons were converted to a 1 square kilometer grid using the minimum of the range. Because this method assumes that wetlands are evenly distributed across each polygon, wetland extent may be overestimated in the North American basins.

Percent irrigated cropland indicates the percentage of the basin that has irrigated agriculture. The percentage was calculated by overlaying the boundaries of the major watersheds on an irrigated area map developed by the Center for Environmental Systems Research, University of Kassel (Global Map of Irrigated Areas, 1999. Kassel, Germany). The irrigated area map is a 0.5 by 0.5 degree grid depicting the percentage of the area equipped for irrigation in 1995. The map was derived by combining information from large-scale irrigation maps, and national, subnational, and drainage basin level data on irrigated area. In calculating the percentage, it was assumed that the irrigated areas are evenly distributed across each grid cell.

Percent dryland area is based on UNEP’s global aridity zone map from the World Atlas of Desertification (United Nations Environment Programme.1997. World Atlas of Desertification, 2nd edition. Edited by N. Middleton and D. Thomas. London: UNEP). This global dataset is a 30-minute resolution map that divides the world into six aridity zones based on the ratio of precipitation to potential evapotranspiration (PET). PET is the amount of moisture that, if it were available, would be removed from a given land area by evaporation and transpiration. This variable shows the percentage of the basin area that is classified as dryland according to the UNCCD definition– made up of arid, semi-arid, or dry subhumid aridity zones (excluding polar and sub-polar regions). In these zones, ratios of mean annual precipitation to mean annual PET range from 0.05 to 0.65. Additional aridity zones include hyperarid environments where ratios are less than 0.05 and humid areas where ratios are greater then 0.65.

Percent urban and industrial area is from the City Lights dataset, a 1 kilometer by 1 kilometer resolution map derived from nighttime imagery from the Defense Meteorological Satellite Program Operational Linescan System of the United States (National Oceanic and Atmospheric Administration-National Geophysical Data Center. 1998. Stable Lights and Radiance Calibrated Lights of the World CD-ROM. Nighttime Lights of the World data base available on-line at: http://julius.ngdc.noaa.gov:8080/ production/html/BIOMASS/night.html. Boulder, Colorado, U.S.A.: NOAA-NGDC). The dataset contains the locations of stable-lights, including frequently observed light sources such as gas flares and oil drilling sites. Time series analysis was used to exclude transient light sources such as fires and lightning. The extent of “lit” area may be slightly overestimated because of the sensor’s resolution and factors such as reflection from water and other surface features. The data better represent urban areas with highly developed economies indicated by extensive electricity networks, street lighting, and industrial activities, such as refineries. The data underestimate urban areas within countries with less developed economies (the mean settlement size required to produce enough light to be detected is much greater in developing countries than in industrialized countries because of differences in energy consumption). The variable is a good indicator of the spatial distribution of settlements and infrastructure, but should not be interpreted as a measure of population density. The percent urban and industrial area is calculated by dividing the area within a watershed indicated as lit, by the total area of the watershed.

Percent loss of original forest indicates the difference between the original forest extent and the current forest extent. Current forest cover refers to closed canopy forest in existence today, and was created using many global and regional biogeographic maps and expert opinion. Original forest cover refers to an estimate of the extent of closed canopy forest in existence 8,000 years ago in the absence of human impact, based on climate, topography, and other variables. Current and original forest cover maps were developed by the World Conservation Monitoring Center of the United Nations Environment Programme (UNEP-WCMC) in 1996 in collaboration with the World Wildlife Fund and the Center for International Forestry Research for inclusion in the WRI report, The Last Frontier Forests: Ecosystems and Economies on the Edge, published in 1997 (Billington, C., V. Kapos, M. Edwards, S. Blyth and S. Iremonger. 1996. Estimated Original Forest Cover Map–A First Attempt and The World Forest Map. Cambridge, UK: WCMC.)

Population Density Map and Basin Indicators
Each population density map shows people per square kilometer by basin, as well as the location of the main large cities. The maps were created from the Gridded Population of the World (GPW) database overlaid with basin boundaries (Center for International Earth Science Information Network, International Food Policy Research Institute, and World Resources Institute. 2000. Gridded Population of the World, Version 2. Palisades, New York: CIESIN and Columbia University. Available on-line at: http://sedac.ciesin.org/plue/gpw). The GPW dataset was compiled from the latest available census data for over 120,000 administrative units worldwide, at a variety of subnational district levels. The population figures were standardized to 1995. The polygon data were converted to 2.5' grid with an assumption that the population is evenly distributed within a census unit. The average population density for each basin was calculated by clipping the population grid to each basin’s extent in an equal-area projection and then using a GIS to calculate the mean of all population density grid cell values in each basin.

Basin area was estimated by summing the number of 1 square kilometer cells within each basin using GIS. Basins were modeled based on elevation data, therefore these values only reflect horizontal extent (slopes are not accounted for) and may underestimate total land surface in the drainage area. Intermittent tributaries are included in most cases; for example the northern part of the Kalahari Desert in Bostwana is included in the Okavango basin, as well as many of the intermittent tributaries within the Lake Chad basin. Water surface of rivers and lakes (i.e., Great Lakes in St. Lawrence River basin) are included in the total basin area; however the tidal portions of rivers, such as the St. Lawrence and Ob, are not included in the drainage area calculation.

Number of large cities is from the World Cities dataset (Environmental Systems Research Institute, Inc. (ESRI), 2000. World Cities. Redlands, California, USA: ESRI). Large cities are defined as cities with greater than 100,000 people in an urban area. The World Cities dataset provides a base map layer of the cities for the world, as well as corresponding population categories. Only those cities in the population category greater than 100,000 were counted and displayed on the map. The base data used to create the World Cities database comes from Geographic and Global Issues Quarterly, Volume 3, Number 4 (United States Department of State, Bureau of Intelligence and Research, 1993/1994), The World Factbook (United States Central Intelligence Agency, 2000), Missions Database (Global Mapping International, 2000), and the World Port Index (U.S. National Imagery and Mapping Agency (NIMA), 2000).

Water supply per person (1995) indicates the average annual renewable water supply per person for individual river basins as of 1995. Estimates are in cubic meters per person per year. Water experts define areas where per capita water supply drops below 1,700 m3/year as experiencing “water stress”—a situation in which disruptive water shortages can frequently occur. In areas where annual water supplies drop below 1,000 m3 per person per year, the consequences can be more severe and lead to problems with food production and economic development unless the region is wealthy enough to apply new technologies for water use, conservation, or reuse. This map is based on the analysis carried out by WRI for the Pilot Analysis of Global Ecosystems: Freshwater Systems (PAGE) (Revenga, C., J. Brunner, N. Henninger, K. Kassem, and R. Payne. 2000. Pilot Analysis of Global Ecosystems: Freshwater Systems. Washington DC: World Resources Institute). These estimates were developed by combining a global population database for 1995 and a global runoff database developed by the University of New Hampshire and the WMO/Global Runoff Data Centre (Fekete et al. 1999). The runoff database combines observed discharge data from monitoring stations with a water balance model driven by climate variables such as temperature and precipitation combined with variables on land cover, and soil information. For those regions where discharged data were available, the modeled runoff was adjusted to match the observed values; for regions with no observed data, the modeled estimates of runoff were used.

Degree of river fragmentation is an indicator of the level of modification of a river system due to dams, reservoirs, canals, interbasin transfers and irrigation consumption. Irrigation consumption refers to the water that is evaporated or used by crops through transpiration, but excludes the amount of water returned to the river after irrigation. For the regions analyzed, rivers with a historical virgin mean annual discharge equal to or above 350 m3 per second were selected. River systems are classified into three levels of fragmentation: high, medium, and low. These categories are based on the number of dams in the main river channel and tributaries, the level of flow regulation, and the length of the main channel segment without dams in relation to the entire length of the river. Generally rivers with low fragmentation do not have dams in the main channel, and if present, dams on tributaries do not change the river’s discharge by more than 2 percent. Highly fragmented river systems include those with less than one quarter of their main channel left without dams, where the largest tributary has at least one dam, as well as rivers whose annual flow patterns have changed substantially. This indicator is from analyses carried out by Dynesius, M. and C. Nilsson. 1994. “Fragmentation and Flow Regulation of River Systems in the Northern Third of the World.” Science 266: 753–762 and Revenga, C., J. Brunner, N. Henninger, K. Kassem, and R. Payne. 2000. Pilot Analysis of Global Ecosystems: Freshwater Systems. Washington DC: World Resources Institute.

Number of dams in basin and in main stem of river. A large dam is defined by the industry as one higher than 15 meters high and a major dam as higher than 150 meters. There are more than 45,000 large dams in the world; however, information on their location is not readily available for most of the world. For the basin profiles in the United States, a complete dataset of 75,187 dams from the National Inventory of Dams (Army Corps of Engineers, 1995-96) was used. A subset of this dataset categorized by height of dam (approximately 5,055 dams), was used to calculate the number of dams (>15m high) and (>150m high) by basin for the United States basins. For all other basin profiles, as well as to identify all large and major dams in the main stem of a river for all profiles (including the United States) the World Register of Dams (1998) from the International Commission on Large Dams (ICOLD), and their 2002 update for Chinese dams were used. Because ICOLD’s on-line and CD-Rom register can only be queried by river name or by dam name, list of main tributaries for each main river were compiled by basin using several world atlases. Once the tributary names were identified, the register was queried to calculate how many large and major dams were in each basin. This same process was used to query the database for dams located on the main stem of rivers. Some countries do not report their dams to ICOLD, others, like Russia, only report a fraction of their dams or do not include the river name in the database. Finally, many rivers in China, Brazil, and other countries have identical names; therefore it was hard to identify all dams in river basins. In China, for example, only the main tributaries of major rivers were queried for number of dams. Because of these data limitations, these figures should be interpreted with caution and are probably underestimate the number of dams in each basin and rivers.

Number of dams (>60m high) under construction includes dams at least 60 meters high that were under construction in 1998 and reported in the “1998 World Atlas and Industry Guide” of the International Journal of Hydropower and Dams. The approximate location of the dams was referenced based on continental-scale maps. The number of dams was later aggregated by river basin.

Biodiversity Information, Map and Indicators
The biodiversity map by basin displays the basin boundary, the extent of wetlands, the location of Ramsar sites and the location of wetland-dependent important bird areas (IBAs). Data on wetland-dependent IBAs were not available for all basins. The Ramsar sites are coded by the presence of invasive species, either animal, plant species, or both, if any invasive species has been reported for that particular site.

Number of fish species. Data on fish species were compiled from multiple sources by the World Conservation Monitoring Centre of the United Nations Environment Programme (UNEP-WCMC). Fish species counts may include diadromous and introduced species. Additional data were added by WRI from technical papers and expert opinion. Data were referenced to major rivers or basins. Because there are several potential sources of error in the species richness values, these numbers should be taken as general indicators of fish diversity and not actual measures. Sources of error include: the amount of research done in a particular basin; species extinctions; and introductions of non-native species. Some rivers, for example, have been highly sampled and most species present identified, while others, particularly in the tropics, have not been thoroughly studied and may contain many not-yet-identified species. Updates include the following. Mekong River fish species number is from Coates, D. “Biodiversity and Fisheries Management Opportunities in the Mekong River Basin.” Case study presented at the Blue Millennium: Managing Global Fisheries for Biodiversity, a thematic workshop held in Victoria, BC in June 2001 and funded by the GEF and UNEP. The case study is available on-line at: http://www.unep.org/bpsp/Fisheries/Fisheries%20Case%20Studies/COATES.pdf. The number of fish species for the Fly River is from Swales, S. 2001. “Fish and Fisheries of the Fly River, Papua New Guinea: Population Changes Associated with Natural and Anthropogenic Factors and Lessons to be Learned.” In Blue Millennium: Managing Global Fisheries for Biodiversity. Victoria, BC, Canada: World Fisheries Trust. Available on-line at: http://www.worldfish.org/. The number of fish species for Lake Tanganyika is from Ichthyology Web Resources available on-line at: http://www2.biology.ualberta.ca/jackson.hp/IWR/Regions/Africa/Lake_Tanganyika.php. Finally, number of fish species for the Murray-Darling basin is from the Murray-Darling Basin Commission website (available on-line at: http://www.mdbc.gov.au/.)

Number of fish endemics. Data on fish endemics were compiled from multiple sources by World Conservation Monitoring Centre of the United Nations Environment Programme (UNEP-WCMC) for inclusion in the WRI report Watersheds of the World: Ecological Value and Vulnerability. 1998. Washington, DC: World Resources Institute. Data were referenced to major rivers or basins. Because there are several potential sources of error in the species endemics values, these numbers should be taken as general indicators of fish diversity and not actual measures. Sources of error include: the amount of research done in a particular basin; species extinctions; and introductions of non-native species. Some rivers, for example, have been highly sampled and most species present identified, while others, particularly in the tropics, have not been thoroughly studied and may contain many not-yet-identified species. Updates for the Mekong River on number of fish endemics were provided by Theo Visser, developer of the Mekong Fish Database CD, available from the Mekong River Commission (http://www.mrcmekong.org). Data on endemic fish for Lake Tanganyika is from Ichthyology Web Resources available at: http://www2.biology.ualberta.ca/jackson.hp/IWR/Regions/Africa/ Lake_Tanganyika.php. Viewed February 17, 2003.

Number of Amphibian Species. The number of amphibian species per basin was calculated using the Distribution of Amphibian Species dataset (unpublished data provided by Center for Applied Biodiversity Science at Conservation International (CABS/CI) and the IUCN Species Survival Commission (IUCN/SSC), January 2003). These data are developed through the Global Amphibian Assessment (GAA), a project of CABS/CI, IUCN/SSC, and NatureServe. The GAA data are provided by hundreds of participating scientists around the world. The GAA has not yet been completed and hence the data provided here should be considered provisional. Only the data for Africa (except for Madagascar), Europe (except for the former Soviet Union), South and Southeast Asia (except for New Guinea), Meso-America, China, the Korean peninsula, and Australia were available at the time of preparation of this CD. To complete the analysis, the amphibian species distribution GIS layers were merged together, intersected with the river basin coverage, and counted using a GIS.

Number of Ramsar sites. The information on each listed site is included in the Ramsar Database maintained by Wetlands International under contract with the Ramsar Convention. Wetlands International compiles the database using the official information submitted by each country about its Ramsar sites. Spatial accuracy of the coordinates varies from one site to another. Wetlands International provided WRI with the point location for each Ramsar site, as well as the presence or absence of invasive species reported for each site. Sites were mapped using the coordinates provided in the database and were later aggregated by river basin using a geographic information system.

Number of Wetland-Dependant Important Bird Areas. Important Bird Areas (IBAs) are globally important sites for wild bird populations. The IBAs used in this CD-ROM are solely those sites that are inland wetlands and that are globally important for waterbirds. At the time of the preparation of this CD, only the data for Europe, Middle East, and Africa were available. These data were originally published in Evans, M.I. Ed. 1994. Important Bird Areas in the Middle East. Cambridge, UK: BirdLife International, Heath, M.F. and M.I. Evans. Eds. 2000. Important Bird Areas in Europe: priority sites for conservation. 2 vols. Cambridge, UK: BirdLife International, and Fishpool, L.F. and Evans, M.I. Eds. 2001. Important Bird Areas in Africa and associated islands. Cambridge/Newbury, UK: BirdLife International/Pisces Publications. See the 'Data Zone' at ww.birdlife.org for detailed information on individual IBAs and the waterbird species that they support.

Number of endemic bird areas. This map was created using a digital dataset provided by BirdLife International with the location of endemic bird areas (EBAs) and overlaying it with the basin boundaries. There are 218 endemic bird areas worldwide, each represented by a polygon or cluster of polygons. Some EBAs overlap. The number of endemic bird areas in each basin was calculated by counting the number of bird areas that fall either partially or totally within each watershed. Forest ecosystems in the tropics and subtropics are better represented by the EBAs than other ecosystems such as desert and grasslands. The information on EBAs by basin for each basin profiles is from Revenga, C., S. Murray, J. Abramovitz, and A. Hammond, 1998. Watersheds of the World: Ecological Value and Vulnerability. Washington, DC: World Resources Institute.

Percent protected area. Data on protected areas are from the Biodiversity Map Library (1996), produced by the World Conservation Monitoring Center, now UNEP-WCMC. Most of the protected areas are represented by polygons, but some are represented by single points. For the latter, circular buffers around the point locations were generated corresponding to the reported size of the protected area. The world protected areas database is currently being revised and updated by the World Database and Protected Areas Consortium, a group of environmental organizations that hold geospatial databases on protected areas. The information on percent protected area by basin is from Revenga, C., S. Murray, J. Abramovitz, and A. Hammond, 1998. Watersheds of the World: Ecological Value and Vulnerability. Washington, DC: World Resources Institute.