USGS - science for a changing world

The USGS Land Cover Institute (LCI)

|| Urban Dynamics |  Temporal Mapping |  Objectives |  Benefits |  Regional Assesments |  Papers/Abstracts |  Data/Images ||


Cindy Bell
Johnson Controls World Services
NASA Ames Research Center, MS 242-4
Moffett Field, CA 94035

William Acevedo
US Geological Survey
NASA Ames Research Center, MS 242-4
Moffett Field, CA 94035

Janis Taylor Buchanan
Johnson Controls World Services
NASA Ames Research Center, MS 242-4
Moffett Field, CA 94035

Abstract: A methodology has been developed to document the tremendous growth large metropolitan regions have experienced over time. A geographic information system (GIS) was used to compile a database of urbanization for the San Francisco/ Sacramento urban region spanning 140 years. Historical records, USGS topographic maps, aerial photographs and Landsat imagery were used to identify the urban spatial extent. Digital transportation data and tabular census data were also incorporated into the database to provide a more complete picture of changes occurring over time. A time-series animation of urban growth for the urbanized region depicts the alarming growth patterns the area experienced between the mid-1800s and the 1990s. The same process is being used to document growth in other urban regions, such as the Baltimore-Washington area. This innovative use of temporal spatial data and animation focuses attention on the dramatic increases in urban development and the spatial patterns that have developed over time.


The United States Geological Survey (USGS) has a long tradition of mapping land use and land cover, both current and potential. Initially charged by Congress with the "classification of the public lands", the USGS began topographic and geologic mapping in 1879. Early mapping activities took place in the largely uninhabited Western United States. In the 1970s, using high-altitude aircraft data, the USGS developed a land use and land cover classification system for the nation (Anderson,, 1976). A program was implemented where a series of land use and land cover maps were produced by aerial photo interpretation. Total coverage of the coterminous United States was completed by 1986.

The USGS also developed techniques for mapping land cover using satellite data. A project is now underway to produce a data base for North America consisting of three Landsat Multispectral Scanner scenes representing land cover in the 1970s, 1980s and 1990s. USGS also produces bi-weekly composites of Advanced Very High Resolution Radiometer (AVHRR) data for the conterminous US from which a prototype land cover characteristics data set was derived (Loveland et. al., 1991). Current plans are to produce a Multi-Resolution Land Characteristics Monitoring System in collaboration with several federal agencies.

In 1993, the USGS began a Human-Induced Land Transformations (HILT) project as a contribution to the US Global Change Research Program. The objective of this project is to understand urban growth and its associated patterns from a historical and a multi-scale perspective. Temporal mapping of major urban regions will enable others to assess ecological, environmental and climatic impacts of urban change and to model and predict future urbanization patterns and impacts. To understand the urbanization process, we first focused our efforts on a rapidly growing region in Northern California. Visualizing the results of this project on a regional scale depicts the alarming rate at which urban areas are growing together. We have also expanded our efforts to map the growth of other large urban areas, including the Washington-Baltimore area.

Study Area

The urban area selected for this study includes the San Francisco Bay Area and a portion of the Central Valley centered on Sacramento. (Figure 1) This region encompasses a diverse environment from the Pacific Ocean to the Sierra Nevada foothills and includes two large metropolitan areas (San Francisco and Sacramento), and several rapidly growing cities (Stockton, Modesto, Tracy, etc.) These urban areas were chosen because they exemplify the fast pace of growth occurring both within and between urban areas that were once thought to be quite separate.

Figure 1


Table 1


	Year			Data Source				Scale

1850 Coast Survey Topographic Maps Unknown

Historical accounts

1900 USGS Topographic Maps 1:62,500

1940 USGS Topographic Maps 1:62,500

Army Map Service Maps 1:50,000

1954 ABAG* Landuse map

USGS Topographic Maps 1:24,000

1962 ABAG* Land use map

USGS Topographic Maps 1:24,000

1974 Landsat MSS 60 m/pixel

1990 Landsat TM 30 m/pixel

* Association of Bay Area Governments



Urban boundary information was derived from a variety of sources ranging in scale and resolution.(Table 1) Data sources included historical accounts, topographic maps, aerial photography and Landsat Multispectral Scanner and Thematic Mapper data. Based on the available materials, seven data layers between 1850 and 1990 were compiled for the San Francisco-Sacramento urban regions.

Urban Definitions

Delineating an urban boundary first requires selecting a definition of urban land. Urban land can be broadly divided into functional and physical definitions. "Urban" in functional terms relates to activities such as industrial, residential, agricultural, etc. However, there are often problems determining which activities should be adopted as urban. Similarly, "urban" can be defined in physical terms, relating either to population density or to land cover, where any developed land is considered urban regardless of its function. There are also variations in the intensity of land uses that influence the definition of urban area, such as high or low housing density. For example, the US Bureau of the Census primarily bases its definition on population size and density:

An urbanized area comprises a place and the adjacent densely settled surrounding territory that together have a minimum population of 50,000 people. The "densely settled surrounding territory" adjacent to the place consists of the following:

1. Territory made up of one or more contiguous blocks having a population density of at least 1000 people per square mile...(US Census Bureau, 1990)

Depending on the specific purpose and the sources used for a study, there can be more than one definition of urban land. We utilized a variety of sources in this study, therefore several urban definitions were required. These sources included satellite imagery, topographic maps and regional land use maps.

Landsat TM and MSS. Using satellite imagery, the spectral reflectance value of an urban surface was used to identify urban land cover (Colwell, 1983). In color infrared images, urban surfaces, such as concrete and asphalt, are characteristically bluish to white in color. Vegetation appears light to dark red, and soil appears yellow to light brown. Individual buildings can be identified if they are large and contrast with the background reflectance color. The generalized urban boundary may contain undeveloped land that is completely surrounded by developed areas. Parks, golf courses and other area whose natural vegetation has been significantly altered is included in the urban category.

Topographic Maps. Modern USGS Topographic maps typically use a pink tint to indicate urban areas and a purple tint to indicate updated photo-revised urban areas. These areas are defined by building and road network density. However, earlier maps constructed around the 1900s did not contain tinting , just the existing road network. Early cities were typically constructed in a gridded road network pattern. In these cases, the urban boundary was defined by these gridded areas, or the density of the road network. (Figure 2) The pink and purple tinted areas were used as urban boundaries for their respective dates. We often found that the tinted areas did not include some commercial/industrial buildings (indicated as black squares on the maps) and dense road networks that existed on the urban fringe. In those cases we included those areas as part of the urban development.

Figure 2. Portion of 1900 USGS Topographic Map.

ABAG Land Use Maps. The Land Use maps compiled by the Association of Bay Area Governments were also used to derive urban delineations. These maps, originally derived from aerial photo interpretation, consisted of five residential classes, one commercial class, one institutional class, three industrial classes, six open area classes, one military class, one transportation class and one water class. Our interest was only to distinguish urban from non-urban areas, so we defined urban as all residential classes, except the very low density class, industrial, commercial, institutional, military, transportation, and cemeteries (open) classes. We excluded the very low density (<= 1 families per net residential acre) residential class because those areas were not discernible as urban on the TM satellite imagery.

Database Development

1990 and 1974. Two adjacent Landsat Thematic Mapper (TM) scenes, dated June 6, 1990, and two Multispectral Scanner scenes dated July 16, 1974 were used in this study. The spatial resolution of TM imagery is 28.5 meters and of MSS imagery is 57 x 79 meters. The TM imagery was used to define which cities would be included in the database and to delineate urban areas for 1990. A 150-square meter area (5x5 TM pixels or 3x3 MSS pixels) was considered the minimum mapping unit for delineation. Any area that was visible on the TM scene was to be included in the data base and delineated on the historic maps. If an area was not visible, i.e. it was too small or was located in a dense tree cover area, to maintain consistency, those areas were excluded from the study. The imagery was first registered to a UTM coordinate system, then used as a backdrop in the GIS to digitize polygons representing urban areas on-screen. Aerial photography, local maps and local knowledge of the region were also used to help interpret questionable areas on the imagery. Urban areas were more difficult to identify on the MSS imagery because of its coarser resolution. The 1990 data layer was used to help visually interpret the 1974 MSS imagery.

1954 and 1962. The ABAG maps were used to identify urban areas in the nine Bay Area counties for 1954 and 1962. Several cities, including Sacramento, were not included in the ABAG maps so supplemental data were obtained from USGS 1:24,000-scale topographic maps. The 1990 data layer, based on the TM imagery, was used as a base to identify ABAG urban polygons that would be included and those that would be excluded. Urban areas that were on the 1990 data layer (i.e. that could be identified in the satellite imagery) were included in the 1954 and 1962 data layers, if they existed. Those urban polygons that were in the 1954 and 1962 data layers but did not exist in the 1990 coverage were deleted. For example, areas classified as institutional and military were found to include large undeveloped areas located outside the closest urban boundary. These classes were either reinterpreted and edited to include only the developed portions of those categories, or deleted.

1900 and 1940. Historical topographic maps were used to identify urban areas for 1900 and 1940. The 1900 urban extent was based on topographic maps at a scale of 1:62,500 ranging in dates from 1897 to 1906. The Sacramento area was mapped from a 1:125,000-scale map published in 1887 and from a 1:250,000-scale map representative of the Sacramento Valley from 1903 to 1910. The 1900 maps did not contain the characteristic urban tint, so street patterns and street density were used to define the urban boundary. USGS and Army Service maps derived from aerial photos taken 1937 to 1940 were used to create the 1940 data layer.

The topographic maps typically contained only one or two urban areas per map, so a large quantity of maps were required to cover the entire study area. We were unable to delineate urban areas directly on to the topographic maps, so mylar was overlayed on the maps and the urban areas with associated tic coordinates were transferred. The mylar overlays were then digitized and input into a GIS system. After each map was digitized, the polygons were labeled with their city names. Finally, all polygons for the year being assembled were integrated to create a complete data layer.

In order to ensure that the data for each year represented all known urban areas, numerous iterations of gathering maps, transferring information to mylar, digitizing and displaying resultant polygons was required. We used GIS capabilities to overlay the polygons for each year on each other to check for inconsistencies. When the data layer for a particular year was completed, it was converted into raster form with a 30 meter grid cell size.


In addition to the urban extent, historical transportation data layers were also assembled in the GIS. The transportation layers were created for the same years as those for the urban extent: 1850, 1900, 1940, 1954, 1962, 1974 and 1990. The 1990 data layer was derived from USGS 1:100,000-scale Digital Line Graphs (DLGs) of the region. We were only interested in the primary road network , so only Class 1 and 2 roads were included. We defined the primary road network as those roads that have been numbered by the Federal transportation department (US DOT) or the State Department of Transportation (Caltrans). As a result, many Class 2 roads were also deleted. DLGs were derived from data collected between 1976 and 1982. To update the 1990 data layer, the DLG was overlaid on to the 1990 Landsat TM scene and any changes (primarily additions) were made by digitizing on-screen.

Historic transportation maps were obtained from the California Department of Transportation (Caltrans). We identified those roads with numbers as primary roads and used the 1990 data layer to create the historical data layers. Several roads that were primary in earlier years, i.e. 1954, were not considered primary roads in 1990. The original DLG was used to identify those roads and add them to the data layer. The result was one data set that included all primary roads for all six years. The attributes assigned to each road included whether it was primary for a certain year (Y or N) and what the number of the road was for each year. The GIS enabled us to select only those roads that were primary for each respective year.


The temporal GIS built for the study area ideally lends itself to time-series animation and data visualization. The seven data layers of urban extent were converted into raster images and used to produce and single frame animation or movie depicting the growth patterns experienced over time.

The San Francisco-Sacramento animation (140 years) consisted of the seven original source data layers and an additional 133 images created by linear interpolation. The additional images were necessary to produce a smooth animation. Since video resolution is rather coarse, the original vector data was gridded to a 300-meter spatial resolution. The image interpolation uses a set of primary images (i.e. 1900 and 1940) as input using the first as a starting date and the second as the ending date. The program written for the study computes a linear distance to the ending boundary for each pixel in the area between the starting and ending boundaries and outputs images based on the pixels distance.

The yearly frames were then constructed by compositing the desired data layers. For San Francisco, we first overlayed built-up areas on top of a shaded relief image base. We then overlayed water bodies and a shoreline on top of the built-up and shaded relief composite. Other overlays included text indicating the year and title.

A video was created using specialized video recording equipment. The equipment allowed us to record single images on a computer screen to the frames on a video tape. For San Francisco, we recorded each of our images to six frames on the video tape, recording a total of 840 frames. An additional 30 frames of the 1850 and 1990 images were recorded to start and end the movie, for a total of 900 frames on the video tape. Since video tape is played back at 30 frames per second, the San Francisco movie runs 30 seconds.


Figure 3 illustrates the data layers created in the GIS and utilized in the animation which corresponds with the areal data in Table 2 and the population data in Table 3. These data layers exemplify the explosive growth that has occurred in the San Francisco-Sacramento urbanized region over time. The pattern of development in the San Francisco area has been limited by physical constraints of the bay and the surrounding steep hillsides. Development in this area first centered around the ports of San Francisco and Oakland with a smaller

Table 2. Growth of San Francisco-Sacramento urbanized region over time.

Table 3. Population growth of San Francisco-Sacramento urbanized region.(US Census Bureau)

commercial area centering around San Jose in the south bay. Sacramento's development began at the junction of the Sacramento and American Rivers and eventually extended into their floodplains and up into the foothills of the Sierra Nevada mountains.

As roads and other forms of transportation became more developed, the migration of middle-income residents from the central city to the suburbs steadily increased between 1900 and 1940. As developable land became increasingly scarce, the bay was infilled in some areas to allow for additional development. The largest wave of suburban expansion occurred between 1940 and 1954 as federal and state housing and highway construction programs proliferated to meet the needs of service personnel returning from World War II (Gerkens, 1988). This growth continued well in to the 70s. Although growth in the San Francisco region slowed considerably between 1974 and 1990, the Sacramento Valley region continues to grow at an alarming rate. The interstate highway that connects the San Francisco and Sacramento regions has now, itself become an urbanized corridor, transforming once two separate urban areas into one large urbanized region. Although the travel distance between the cities of San Francisco and Sacramento is approximately two hours, the gap between their adjacent suburban areas has disappeared.


The USGS has traditionally utilized various methods to map land use/land cover, but this project signifies the first attempt to focus on urban regions. The methodology developed here integrates historical maps, topographic maps, digital imagery and animation techniques to portray a unique perspective of regional urban development over time. We intend to broaden the scope of this mapping project by incorporating population density, wetlands, and protected lands. Historical topographic maps and satellite imagery is available for every large metropolitan area in this country. Producing dynamic maps of these areas will enable scientists, planners, policy makers and teachers to better understand the impacts of growth on the regional and global environments and to model future growth.


Anderson, J.R., E.E. Hardy, J.T. Roach, and R.E. Witmer, 1976. "A Land Use and Land Cover Classification System for Use with Remote Sensor Data", US Geological Survey, Professional Paper 964.

Gerkens, L.C., 1988., "Historical Development of American City Planning" in Frank So et. al. (eds.), The Practice of Local Government Planning, Washington DC: International City Management Association, pp. 32.

Jensen, John R., 1983. "Urban/Suburban Land Use Analysis" in Manual of Remote Sensing, Vol. II, Edited by John Estes, Falls Church, Virginia, American Society of Photogrammetry pp. 1571-1666.

Loveland, T.R., J. Merchant, D.O. Ohlen and J. Brown, (1991). "Development of a Land Cover Characteristics Data Base for the Coterminous U.S." in Photogrammetric Engineering and Remote Sensing, v. 57, pp. 1453-1463.

U.S. Census Bureau, 1990. Federal Register, Vol. 1, No. 204.

Accessibility FOIA Privacy Policies and Notices logo U.S. Department of the Interior | U.S. Geological Survey
Page Contact Information: LCI Administrator
Page Last Modified: November 2016