Urban Dynamics

Temporal Urban Mapping - Compilation Procedures

The temporal urban database was compiled using a variety of image processing techniques including manual photo-interpretation and automated spectral classification. The type of source materials available determine the actual procedures that are used. A geographic information system (GIS) is used to assemble and integrate the data.

Compilation of Urban or Built-up Land

Historical Sources
When using historical maps, urban extent was compiled by map interpretation and delineation on mylar overlays. The interpretations were based on map features, such as tint, street density, and building locations. After the mylar overlays were annotated, they were digitized into a geographic information system (GIS), edited, and projected to a Universal Transverse Mercator (UTM) grid. A GIS file was generated for each map and when verified all files were appended into a single GIS coverage which represented the specific time period. Any cities that were not otherwise identified were assigned a seed point if they were listed as historically significant cities by the U.S. Bureau of the Census in their tabular list of dates and cities with at least 500 people.



Flow chart depicting the steps involved in delineating urban extent from topographic maps.

Contemporary Sources

When using contemporary sources urban development is determined by various image processing procedures that combine the spectral information from Landsat data with the dense systematic street pattern derived from the DLG road network, and the GIRAS land use and land cover interpretations.

Landsat TM data was resampled to a 30-meter grid using the North American Datum of 1927. A multi-spectral classification was carried out using unsupervised training statistics and a maximum likelihood classifier. Interpretation of the unsupervised classes identified urban categories by spectral signatures recognized as concrete, asphalt, buildings, roads, residential neighborhoods, and commercial buildings. In addition to the Landsat classification a rasterized version of the DLG road network was processed to derive a built-up delineation based on the proximity of roads. The GIRAS land use/land cover vector data was also rasterized to a 30-meter raster UTM grid. Misclassification or spectral confusion in the TM data interpretation was minimized by combining the Landsat interpretations with GIRAS LULC and/or the DLG-derived interpretations.

Compilation of Principal Transportation

Principal routes for each time period were derived from map interpretations. Implicit and explicit compilation criteria were used to designate these routes. The map interpretations were then used as a reference to manually select the appropriate arc segments using the 1:100,000-scale transportation digital line graphs (DLG) as a template. Using a GIS the selected arc segments were assigned the appropriate attribute from the classification system being used. Missing arcs were manually digitized if they were not found in the DLG data.

Map interpretation differed for each of the different time periods due to the variety sources throughout the years. In the early years, around 1900, cartographic features on the maps required complete reliance on implicit criteria for interpretation. When mapping efforts across the United States picked up after World War II, cartographic representation of transportation routes became more explicit and interpretation was less implicit and more derived from map symbology. After the 1960's when major highway systems were being installed across the U.S., map symbology indicating "primary highway, hard surface", "heavy duty, greater than or equal to four lanes", or "interstates" became the main tool for interpretation of principal transportation routes.

In certain time periods, DLG road feature designations such as "Primary Route", "Secondary Route", "Interstate", "U.S. Route", or "State Route" was used as explicit criteria to indicate principal roads. These road segments were selected directly from the DLG attribute tables. This process provided digital templates that were reviewed and edited to remove minor omission or commission errors resulting from the automated selection of arcs by DLG attribute rather than by manual selection of each feature. Principal transportation features newer than the DLG data were manually digitized using commercial road maps and the Landsat Thematic Mapper data as a reference.