Note that the data published in the 2002 State of the Nation’s Ecosystems Report as well as the 2003 and 2005 Web-Only Updates have been superseded by the 2008 Report and thus should be used with caution. For the most recent data, purchase the 2008 Report from Island Press.

Note: Several other indicators refer to this technical note for the discussion of remote-sensing data (National Land Cover Dataset) included at the end of this note.

The Indicator

“Coasts and oceans” are indicated by the area of brackish water off U.S. coasts. Brackish water is defined as all waters that have a salinity greater than about 1 part per thousand (ppt) and less than about 30 ppt (measurements are actually made in units called practical salinity units, which are quite close to parts per thousand).

Brackish water systems, including estuaries, are among the most productive ecosystems in the world––before 1985, estuarinedependent fish species accounted for more than 50% of U.S. fish landings. Brackish water is a mixture of fresh water and seawater, and its distribution is a fundamental parameter of the distribution, abundance, and productivity of estuarine-dependent organisms and of essential fish habitats such as tidal wetlands (mangrove swamps, salt and brackish marshes, and intertidal flats), submerged attached vegetation (macroalgae and vascular plants), and oyster reefs.

Most variability in the salinity of coastal ecosystems is related to freshwater runoff and groundwater discharge. Thus, the areal extent of brackish water is an index of the amount of freshwater that flows from the continent to coastal waters and can be used as a surrogate for nutrient enrichment, sediment loading, and contaminant inputs.

“Croplands” includes the following U.S. Department of Agriculture (USDA) Economic Research Service (ERS) categories: cropland harvested, crop failure, cultivated summer fallow, cropland used only for pasture, and idle cropland; Conservation Reserve Program lands are included. Note: In the Farmlands chapter of this report (see Total Cropland), we present multiple estimates of the area of croplands; the ERS was selected for use in this national indicator as illustrative of long-term trends.

“Forests” is defined by the USDA Forest Service as areas of at least one acre with a certain density of trees (at least 10% cover). See also the forest area indicator and its associated technical note.

“Fresh waters” includes lakes and streams, as well as wetlands; however, only wetland acreage is reported in this indicator. Wetlands are defined according to the U.S. Fish and Wildlife Service’s wetland classification system, which is the national standard, and include the following types of freshwater wetlands: principally palustrine forested wetlands, palustrine scrub-shrub wetlands, and palustrine emergent wetlands.

“Grassland/shrubland areas” for the lower 48 states are defined according to the National Land Cover Dataset (NLCD; see below) and include several land cover categories. Definitions of land cover in Alaska are from a separate study (see below).

“Urban/suburban areas” is generally defined here as land that is substantially covered by one of the following land cover types: low-intensity residential, high-intensity residential, commercial or industrial or transportation lands, and urban and recreational grasses. These categories are based on remote-sensing classification of land cover (see NLCD description below). A series of steps were taken to limit these areas to those thought to be most representative of urban and suburban issues (see the urban/suburban extent technical note for a thorough description). There are several other ways that urban areas have been defined by various programs (again, see urban/suburban technical note). The approach adopted by the ERS that relies on U.S. Census Bureau data for urban areas is a consistent dataset, however, it is based on different assumptions than the definition of urban/suburban areas in this report. The ERS time series is shown to give a sense of the relative change in urban/suburban areas over the past 50 years.

The land cover and ocean depth (bathymetry) map displays the geographic location of the various ecosystems. Data for forests, grass/shrublands, croplands, and urban/suburban were derived from the definitions in the NLCD. Only those rivers with flow rates exceeding 1000 cubic feet per second (cfs) are shown. Bathymetry data in coastal waters show the depth to the ocean floor in several ranges.

The Data

Coasts and Oceans: Data are not adequate for national reporting. Some data for the salinity of open waters of the U.S. Exclusive Economic Zone are available from the National Oceanographic Data Center (see http://www.nodc.noaa.gov/General/salinity.html). Local and regional data for semi-enclosed bodies of water are collected by a variety of federal and state agencies, but these data have not been compiled into a single source.

Cropland: Data on cropland extent come from the USDA ERS, and are available at http://www.ers.usda.gov/Emphases/Harmony/issues/arei2000/. ERS relies on data provided by the National Agricultural Statistics Service, as well as a variety of other sources. The ERS estimate for croplands is a reasonable estimate; however, it is not the only credible estimate. Specifically, the croplands extent measure provides estimates of the extent of cropland from other agencies and programs; these estimates of mid-1990s cropland extent range from a low of 431 million acres (USDA Census of Agriculture) to a high of 510 million acres (USDA National Resources Inventory and NLCD). Data from ERS (455 million acres) are used here, but without additional research into which data source is more accurate, it would be equally fair to use any of the other estimates.

Forests: Data on forest extent are from the USDA Forest Service Forest Inventory and Analysis (FIA) program (see http://fia.fs.fed.us). FIA is a survey-based program that has operated since the late 1940s, collecting information on a variety of forest characteristics. See the technical note for the forest area indicator for additional details.

Fresh Waters: Data on freshwater wetlands are from the U.S. Fish and Wildlife Service’s National Wetlands Inventory (NWI). See Dahl (2000); data also available at ftp://wetlands.fws.gov/status-trends/SandT2000Report_lowres.pdf. The NWI produces periodic reports on the extent of wetlands in the United States. See also the technical note on freshwater extent. River data are from the U.S. EPA River Reach File (see http://www.epa.gov/region02/gis/atlas/rf3.htm), which was constrained so that only those rivers with flow rates of at least 1000 cfs were used. Procter & Gamble’s Miami Valley Laboratory conducted this analysis for The Heinz Center.

Grasslands and Shrublands: Data on the extent of grasslands and shrublands (lower 48 states) are from the NLCD—see the detailed description below.

Grassland/shrubland data for Alaska are from a vegetation map of Alaska, based on Advanced Very High Resolution Radiometer (AVHRR) remote-sensing images with an approximate resolution of 1 kilometer on a side (see complete description below). The following groupings of classes were used (see http://agdc.usgs.gov/data/projects/fhm/#G, Statewide Vegetation/ Land Cover; other classifications listed below): alpine tundra & barrens (#3); dwarf shrub tundra (#4); tussock sedge/dwarf shrub tundra (#5); moist herbaceous/shrub tundra (#6); wet sedge tundra (#7); low shrub/lichen tundra (#8); low & dwarf shrub (#9); tall shrub (#10); and tall & low shrub (#23). See the Area of Grasslands and Shrublands technical note for information on the pre-settlement estimates of these lands from Klopatek et al. (1979).

Urban and Suburban Areas: Grassland and shrubland data are a relatively straightforward presentation of NLCD vegetation classes, but urban and suburban area data required additional processing. Basically, this involved identification of areas with urban/suburban land cover (using the NLCD classes; see below), then making adjustments to account for the intermixed land use at the edges of urban areas. For example, undeveloped parcels or large parks located within developed areas were included as “urban/suburban” even though they might have been classified as forest or grass/shrub according to the satellite data. See the technical notes on urban/suburban extent for additional information.

The Land Cover and Ocean Bathymetry Map: The map shown in this indicator was constructed from several datasets by USGS’s Earth Resources Observations Systems Data Center. These datasets are described below.

Data Quality/Caveats

Because these data are from multiple sources, some caution is appropriate. Different programs use different definitions and may be conducted in different years. Every effort has been made here to identify consistent land cover categories and time periods.

Given the diversity of programs, definitions, techniques, and time periods, there are inevitable conflicts between these various estimates. For example, satellite data (described below) indicate that there are about 690 million acres of forest in the United States (all 50 states), while the USDA Forest Service’s FIA program estimates that there are about 747 million acres of forest. Satellite remote sensing, which can provide data on the entire U.S. land surface, may serve as a common reference point, against which other programs that count only forest, for example, or only private lands, or only cropland, may be compared.

The National Land Cover Dataset (NLCD): In the 1990s, a federal interagency consortium was created to coordinate access to and use of land cover data from the Landsat 5 Thematic Mapper. Using Landsat data and a variety of ancillary data, the consortium processed data from a series of 1992 Landsat images, to create the NLCD on a square grid covering the lower 48 states. Each square in the grid, or “pixel,” is approximately 100 ft on a side.

Each pixel was assigned one of 21 land cover classes, which are described at http://landcover.usgs.gov/classes.asp. The steps of this classification process, which can be found in detail elsewhere (see Vogelmann et al. 2001 and Vogelmann et al. 1998), are summarized here. First, an automated process is used to create clusters of pixels for a given regional area. Second, these clusters were interpreted and labeled with the help of aerial photographs. Third, in cases where clusters of pixels included multiple land cover types (i.e., “confused clusters”), models that utilize ancillary data, such as elevation or population density, were used to help assign land cover classes. Finally, lands that are bare—especially clear cuts and quarries—and many grass areas, such as parks, golf courses, and large lawn, are not easily distinguished from other land cover classes during the automated process, so a process of on-screen verifications was used as clarification. These four steps were the general process, and additional steps were taken in certain regions in order to further improve the accuracy of classifications (see http://landcover.usgs.gov/accuracy/ for a discussion of NLCD error analysis).

Note that classification of pixels was based in part on the character of surrounding squares in the grid; thus, a pixel of grass-like land cover surrounded by residential pixels would probably be classified as “urban and recreational grasses” rather than as “pastureland.” Where appropriate, the agencies also made use of data from both the Census Bureau and the U.S. Fish and Wildlife Service’s National Wetlands Inventory data to help make such distinctions. Satellite data offer an unprecedented opportunity to classify land cover on a consistent basis over very large areas (i.e., the entire country). However, the accuracy of any classification is not perfect. The accuracy of satellite-derived classifications is related to many factors: amount of data available (i.e., many dates of imagery rather than just one), the detail of the required land cover information (i.e., forest vs. deciduous forest vs. sugar maple/beech/yellow birch), classification methods, computing power, and, of course, time and money. Assessments of the NLCD for the eastern United States indicate an accuracy of approximately 80% or higher for general land cover categories (e.g., forest, agriculture, developed). Accuracy assessments for the western United States are currently under way. Improving technology and techniques offer the potential to increase accuracy of the next NLCD (2000) currently being assembled by the Multi-Resolution Land Characterization Consortium. The land cover classes associated with the 30-m (100-foot) square pixels were grouped for the different ecosystems as follows (the number in parenthesis is the NLDC land cover class reference):

• Forests: deciduous (#41); evergreen (#42); mixed forest (#43)
• Croplands: pasture/hay (#81); rowcrops (#82); small grains (#83); fallow (#84); orchards/vineyards/other (#61)
• Grass/Shrub: shrubland (#51); grasslands/herbaceous (#71); bare rock/sand/clay (#31)
• Water: open water (#11); wetlands (#91 & #92)
• Developed: low-intensity residential (#21); high-intensity residential (#22); commercial/industrial/transportation (#23); urban/recreational grasses (#85)
• Other: quarries/strip mines/gravel pits (#32); transitional (#33); perennial ice/snow (#12)

Data Quality/Caveats: The power of satellite-derived classifications is that satellite data can easily cover the entire country and the classification process can be automated (though not completely). This makes it possible to compile a nationally consistent land cover dataset; however, any land cover classification is subject to error. The NLCD for the eastern United States has an accuracy of approximately 80% or higher for the general land cover categories used for our study (see http://landcover.usgs.gov/accuracy/). Some of the known misclassifications that occur in the dataset include suburban areas or tree farms classified as forest; grasslands classified as agriculture, or vice versa; and fallow agricultural fields classified as barren lands.

Data Access: NLCD data are available at http://landcover.usgs.gov/mrlcreg.html. Further detail is also available from Vogelmann et al. (2001). Other data can be obtained from the sources cited in this note.

Coastal Bathymetry Data: These data come from the National Geophysical Data Center, and are known as ETOPO5 data. They were generated from a digital database of land and sea-floor elevations on a 5-minute latitude/longitude grid. The resolution of the gridded data varies from true 5-minute for the ocean floors, the United States, Europe, Japan, and Australia to 1 degree in data-deficient parts of Asia, South America, northern Canada, and Africa. Data sources are as follows:

Ocean Areas: U.S. Naval Oceanographic Office; United States, W. Europe, Japan/Korea: U.S. Defense Mapping Agency; Australia: Bureau of Mineral Resources, Australia; New Zealand: Department of Industrial and Scientific Research, New Zealand; balance of world land masses: U.S. Navy Fleet Numerical Oceanographic Center. These various databases were originally assembled in 1988 into the worldwide 5-minute grid by Margo Edwards, then at Washington University, St. Louis, Missouri. Data have been described in NOAA (1988). The version of the data making up ETOPO5 is from May 1988, with the exception of a small area in Canada (120-130° W, 65-70° N), which was regridded in 1990; the data are available at: http://www.ngdc.noaa.gov/mgg/global/seltopo.html.

Alaskan Land Cover Data: Data for Alaska are from a vegetation map of Alaska by Flemming (1996), based on AVHRR remote-sensing images with an approximate resolution of 1 kilometer on a side. The following groupings of classes were used (see http://agdc.usgs.gov/data/projects/fhm/#G [Statewide Vegetation/Land Cover]):

• Freshwater: water (#1); glaciers and snow (#2)
• Grass/Shrub: alpine tundra & barrens (#3); dwarf shrub tundra (#4); tussock sedge/dwarf shrub tundra (#5); moist herbaceous/shrub tundra (#6); wet sedge tundra (#7); low shrub/lichen tundra (#8); low & dwarf shrub (#9); tall shrub (#10); tall & low shrub (#23)
• Forest: closed broadleaf & closed mixed forest (#11); closed mixed forest (#12); closed spruce forest (#13); spruce woodland/shrub (#14); open spruce forest/shrub/bog mosaic (#15); spruce & broadleaf forest (#16); open & closed spruce forest (#17); open spruce & closed mixed forest mosaic (#18); closed spruce & hemlock forest (#19)
• Other:1991 fires (#21); 1990 fires & gravel bars (#22)

Hawaiian Land Cover Data: These data came from the National Oceanographic and Atmospheric Administration’s (NOAA) Coastal Change and Analysis Program (C-CAP), which is a national effort to develop and distribute regional land cover and change analysis data for the coastal zone by using remote-sensing technology. The data used in this program are created from a combination of satellites and fieldwork. C-CAP classifies land cover types into 22 standardized classes that include forested areas, urban areas, and wetlands. C-CAP land cover data are derived from Landsat Thematic Mapper satellite imagery and are available at http://www.csc.noaa.gov/crs/lca/m_eight.html.

2003 Web Site Update: Data for 2002 for forests were added to this update. Data were acquired from the Forest Service and are available on the Web at http://fia.fs.fed.us.

References

Dahl, T.E. 2000. Status and trends of wetlands in the conterminous United States 1986 to 1997. Washington, DC: U.S. Department of the Interior, Fish and Wildlife Service.

Flemming, M.D. 1996. A statewide vegetation map of Alaska using a phenological classification of AVHRR data. 1996 Alaska Surveying and Mapping Conference, Anchorage, Alaska.

Klopatek, J.M., R.J. Olson, C.J. Emerson, and J.L. Joness. 1979. Land-use conflicts with natural vegetation in the United States. Environmental Conservation 6:191–199.

NOAA. 1988. Data announcement 88-MGG-02. Digital relief of the surface of the earth. National Geophysical Data Center, Boulder, Colorado.

Vogelmann, J.E., S.M. Howard, L. Yang, C.R. Larson, B.K. Wylie, and N. van Driel. 2001. Completion of the 1990s national land cover data set for the conterminous United States from Landsat Thematic Mapper data and ancillary data sources. Photogrammetric Engineering & Remote Sensing 67:650–662.

Vogelmann, J.E., T.L. Sohl, P.V. Campbell, and D.M. Shaw. 1998. Regional land cover characterization using LANDSAT Thematic Mapper data and ancillary data sources. Environmental Monitoring and Assessments 51: 415–428.