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published in the 2002 State of the Nation’s Ecosystems Report as well as the 2003 and 2005 Web-Only Updates have been superseded by the 2008 Report and thus should be used with caution.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.

The Indicator

This indicator reports the total number of acres that are classified as “urban and suburban” and the amount of the various “undeveloped” land cover types within these areas. More detail is provided below, but “urban and suburban” is defined here as land that is substantially covered by one of the following land cover types: low-intensity residential, high-intensity residential, commercial–industrial–transportation, or urban and recreational grass. These categories are based on remote-sensing classification of land cover and are defined at http://landcover.usgs.gov/classes.asp.

It was our intent that urban and suburban areas should include all major metropolises and their outlying suburbs as well as smaller settlements across the country that have a similar character even though they may not be adjacent to a metropolis. Our goal was to define those areas across the United States that should be classified as “urban and suburban”; The Heinz Center examined several possibilities before choosing the approach used here.

The use of the Census Bureau’s metropolitan statistical areas (MSAs) was the coarsest approach considered. MSAs include entire counties (or cities and townships in New England) rather than only the large urban centers and those outlying areas that are connected to them in some fashion. In the West especially, vast counties are included in MSAs even though only a fraction of the county area is actually urban or suburban. MSAs account for about 20% of the land area of the lower 48 states; The Heinz Center believes this is a significant overestimate of the area covered by cities and suburbs.

Urbanized areas (UAs), also defined by the Census Bureau, offer a more refined but still incomplete solution. Metropolises and their outlying areas are included in UAs, but smaller settlements, which share many of their characteristics with suburbs, are not included. A drawback to using UAs is that they are determined in part by political/jurisdictional boundaries, in addition to the degree of development. A potentially larger confounding issue is that the rules for delineating UAs have changed significantly since their first use in 1950. The Census Bureau is well aware of The State of the Nation’s Ecosystems: Technical Notes 264 this shortcoming and will be releasing newly constructed UA boundaries in early 2002. The U.S. Department of Agriculture (USDA) Economic Research Service (ERS) has estimated urban land area since 1950. ERS’s estimate has incorporated the area of UAs as well as the amount of area in Census-defined “places” that have a population of at least 2,500 people. We have used ERS’s estimate to gauge the change in urban land area over time in the national extent indicator; however, due to the limitations of UAs we chose not to rely on these estimates exclusively to define urban/suburban areas.

A third and still more refined option considered would have relied totally on Block Groups (BGs), which are small regions based on political boundaries within which the Census Bureau counts the population. It would be possible to choose a density threshold—1000 people per square mile is generally accepted as “urbanized” by the Census Bureau—and define those BGs that meet or exceed this density as urban and suburban. A shortcoming of this approach is that BGs dominated by warehouses or railroad yards, for example, which are certainly urban in character, would be excluded because of their low population densities. As is discussed below, the approach chosen for this indicator does, indeed, include most BGs with densities at or above 1000 people per square mile.

A fourth option was to adopt the estimates for developed lands made by the USDA Natural Resources Conservation Service’s National Resources Inventory (NRI). The definitions used by the NRI agree, in principle, with those for urban/suburban lands. However, NRI reports on any and all developed areas—including those down to about one-quarter acre. In contrast, this project focuses on those areas with sufficient density and size to qualify as “suburban” in character, as well as areas that are undeniably “urban.” As noted below, this project’s definition requires an area to be at least 270 acres in size before it is included within the “urban/suburban” definitions. In addition, the NRI data are derived from a statistical sampling rather than a cataloging of all developed lands. Thus, it would not have been possible to delineate individual urban/suburban areas on a map (as is done along with the national extent indicator), which would be necessary to implement several of the other indicators included in this report.

The approach adopted here (see “Data Manipulation” below) uses satellite data to classify land cover. The advantage of this method is that it includes virtually all the BGs with at least 1000 people per square mile, as well as other developed but lightly populated land, such as warehouse districts. In addition, by overlaying BGs on the urban/suburban areas, it was possible to estimate that about 75% of the 1990 population lived in these areas (note that the data used to generate urban/suburban areas came from 1992). As described in more detail below, a series of steps have been used to define the outlines of urban and suburban areas based on four different satellite land cover classifications.

A potential shortcoming of using a satellite-based approach rather than a delineation based in Census data is that it will be more challenging to correlate environmental quality trends like air and water quality with human demographic and health data. However, a geographic information system (GIS) can be generated to associate Census BGs, for example, with urban/suburban areas. This would permit such correlations to be done for studies of demographics and human health.

It may be useful in future editions of this report to consider presenting the data on urban/suburban lands based on the number of people associated with them. This would require shifting priorities for the indicator and a GIS analysis as described above. Also, as data become available, it would be good to add the proportion of native and non-native species to the graph showing the composition of the undeveloped portion of urban and suburban lands.

The Data

Data Source: Satellite data are derived from the National Land Cover Dataset (NLCD), a product of the Multi-Resolution Land Characterization (MRLC) Consortium, which is a partnership between the U.S. Geographical Survey (USGS), the U.S. Forest Service, the National Oceanographic and Atmospheric Administration (NOAA), and the Environmental Protection Agency (EPA) (see http://www.epa.gov/mrlc/ or http://landcover.usgs.gov/nationallandcover.html).

Data Collection Methodology: Please refer to the national extent indicator technical note for a discussion of the NLCD.

Data Manipulation: The NLCD divides the lower 48 states of the United States into several billion square pixels that are about 100 feet on a side. The data presented for this indicator are based on analysis of larger pixels (1000 ft on a side), each of which contains 100 of the smaller pixels. The first step was to classify any 1000-ft pixel as urban and suburban if a majority of the 100-ft pixels within it fell into one of the four “developed” land cover types available in the NLCD: low-intensity residential, high-intensity residential, commercial–industrial–transportation, or urban and recreational grasses. Very large aggregates of the 1000-ft pixels, which were found for metropolises such as New York City, were “smoothed” to some degree; that is, small clusters of “undeveloped land” pixels that were wholly included within a metropolis were subsumed in the urban and suburban areas. Other clusters of undeveloped-land pixels within an urban and suburban area, although connected to the perimeter by one or more pixels on a diagonal, were also included in the urban and suburban area. For clusters of developed-land pixels to be counted as urban/suburban in outlying areas, at least 13 of the 1000-ft pixels had to touch at their sides or corners for a minimum size of 270 acres. The final step for this indicator was to evaluate the proportion of different land cover types within the 1000-foot pixels. This process yielded estimates of the amount of both developed land and undeveloped land (in several categories) by region.

Data Quality/Caveats: It is important to note that the methods used to establish the NLCD relied on two different satellite images of a given area, plus ancillary data. An image taken during the “leaf-off ” period in the late fall to early spring was often more important to the classification process than the fully vegetated image. This was especially true in urban settings with a good deal of tree-lined streets; the foliage of deciduous trees should not have obscured the constructed surfaces during the leaf-off period and, therefore, should not have led to an underestimate of developed lands in these regions.

Given that the method used here to establish urban/suburban areas is based on square pixels that are roughly 100 feet on a side, some detail would have been missed in a typical urban setting. Specifically, the trees on a tree-lined streets would most likely not be distinguished from the street and sidewalk. However, a large expanse of trees, such as a heavily wooded median strip or a small park, may well have been classified as forest.

Data Access: All these analyses were conducted at the Land Cover Applications Center at USGS’s Earth Resources Observations Systems Data Center. The raw data from which this indicator was developed are available at no cost from the MRLC Consortium (http://edcwww.cr.usgs.gov/programs/lccp/mrlcreg.html), but vast computing power was necessary for this analysis. Note: The data available at the Web site listed here are the “raw” data from which estimates of urban/suburban area, and the size of natural areas within, were prepared. The actual data presented in this report were prepared specially for The Heinz Center for this report.