Given the 59 largest Combined Statistical Areas (CSAs) and Metropolitan Statistical Areas (MSAs) and data on the numbers of housing units by census tract within those areas from 1950 to 2010, my next task was delineating the extent of the urban areas within each MSA or CSA for each year.
Following the general approach taken by the Census in defining Urbanized Areas, these urban areas would be identified as the sets of contiguous census tracts that meet a minimum density threshold. The Census uses a minimum density of 500 persons per square mile for adding blocks or larger units to an Urbanized Area. For 2010, the Census reported an average of 2.34 persons per housing unit for the nation as a whole. Using this value, an equivalent housing unit density would be almost exactly 1 unit per three acres or 213.33 housing units per square mile. So for each year and metropolitan area, the urban area was identified that consisted of contiguous census tracts with densities exceeding this minimum value.
The urban areas could not extend beyond the boundaries of their CSA or MSA. This was only an issue when large CSAs or MSAs were adjacent to one another and the urban areas each extended out to the boundary. The CSA or MSA boundary then became the dividing line between the urban areas. Some additional details had to be addressed ranging from how to deal with areas of water to the status of tracts not meeting the density threshold surrounded by urban tracts. In general, the rules for delineating Urbanized Areas were followed, except where their use of blocks rather than census tracts as the basic elements for creating the areas made a difference.
As the interest of the research is in the patterns of the large urban areas, only the large area at the heart of the CSA or MSA was considered to be the urban area. Smaller areas of urban density tracts that later became a part of the urban area as it expanded were not identified as urban until they became a contiguous part of the larger area. However, exceptions were required for those areas that started out as two or three large, separate urban areas which later grew together to form a single contiguous urban area. An example would be Dallas and Fort Worth. In those cases, the separate urban areas were identified as components of the urban area from the beginning. This raised the question of when a second or third urban area would be considered sufficiently large in relation to the largest area to warrant such inclusion. I based this on the ratios of the Urbanized Area populations for the smaller urban areas to the largest. I used either 2010 census data if the Urbanized Areas were separate or data from the last census at which the Urbanized Areas were recognized as separate (for example, 1970 for Dallas and Fort Worth). I ordered the list and had to make a judgment call for the cutoff, for which I chose a ratio of 0.28. The last 3 areas included, with the smallest ratios, were Akron (with Cleveland), Tacoma (with Seattle), and Providence (with Boston). There was a significant gap between the areas with these ratios and the next lower values. Next were Kissimmee (with Orlando, ratio 0.21) and Concord (with Charlotte, ratio 0.17). So this choice seemed reasonable.
The urban areas are intended to be a unique set of areas defined for this research. Nevertheless, since the procedure used followed the Census Urbanized Area definition in many respects, it seemed interesting to compare the areas. A note Urbanized Area Comparison is on the Research page, as is the note Urban Patterns Dataset Description which provides more detail on the urban area definition.