|
A brief description of the Western Futures development maps
David M. Theobald
Natural Resource Ecology Laboratory
Colorado State University
A central goal of the Western Futures project was to produce maps of historical, current, and likely future development patterns of the eleven western states, to better envision what landscape changes have occurred, and what future changes are likely to occur. To support this goal, I produced spatially-detailed maps of development using housing data from the 1990 US Census block-groups and blocks, which are subdivisions of the familiar census tract. A typical block-group contains between 250 and 550 housing units, and there are over one-quarter million block-groups in the US.
Historical data was gleaned from data compiled from Question H17 from the Census: "Approximately what year was your house built?" These responses were aggregated to each decennial year so that the number of housing units between 1940 and 1990 can be mapped. However, nationwide, the number of housing units was underestimated by 8.3% in 1980 with 1.8% of counties less, by 14.4% in 1970, and by 27.0% in 1960. I corrected these estimates by computing a correction factor for each county, which is the ratio of the number of units in the historical census, divided by the total housing units summed from the H17 block-group estimates.
To understand likely future landscape changes, I developed a forecast model also based on block-group data. The primary goals of the forecast model were to create a straightforward, easy-to-interpret model, make use of national demographic data sets rather than specialized land use data, and be easily extendable if local knowledge and finer-scale data are available. The modeling approach is a simplified version of a supply/ demand/allocation model, which is not driven by a particular economic theory, but is rooted in practical assumptions and limitations on development. That is, for each block group, I computed: the number of units available to be developed (supply); the number of units needed to meet the needs of the projected population (demand); and the locations where new units will be placed first, and where they will spread to if demand exceeds supply (allocation).
Probably the most challenging, but important, part of the model is the allocation component. The critical question here is: At what density does a block-group fill up with housing units so that development "spills-over" into adjacent areas? Although zoning regulations that restrict development types and densities are the main factor that determine build-out density, it is practically impossible to compile all available zoning for the West - indeed many counties do not even have zoning regulations in place. In lieu of finer-scale information, I assumed that future development will occur in a pattern similar to the recent past. So, the operating rule here is that for a given decade, a block-group’s density will not exceed the average density of its neighboring block-groups. This allows development to organically grow up and spread over time, and because average density is calculated locally, it is specific to each county - indeed sub-sections of counties can have markedly different growth patterns.
The rate of growth is driven by the demand component, which is generated by county-level population projections. We chose to develop a conservative set of county projections using state-level population projections to the year 2025 from the U.S. Census Bureau. We projected growth for each county out to 2025 using county-specific 1990-99 growth rates, but constrained them so that the sum of the county population did not exceed the state-level projection from the Census data. Population estimates for 2050 were created by simply assuming that the same finite increase in population during 2000-2025 would occur again from 2025 to 2050---this implies a slowing rate of growth that we think makes it a conservative projection.
I classified housing density into four general classes: urban, suburban, exurban, and rural. Urban densities are typically defined as areas with greater than 1,000 people per square mile (1.6 people per acre). Assuming an average of 2.5 people per housing unit, this translates to roughly 0.7 units per acre (~1 unit per 1.6 acres). I define urban density as greater than 2.0 units per acre, which is slightly more relaxed than the Census definition of urban. Suburban density ranges from 0.1 to 0.5 units per acre (1 unit per 2 to 10 acres). This identifies areas that are lower-density subdivisions. Exurban density ranges from 0.025 to 0.1 units per acre (1 unit per 10 to 40 acres). This class identifies very low-density development, including "ranchette" development in Colorado, which occurs at 1 per 35 - 40 acres. Rural density is below 0.025 units per acre (1 per 40 acres and more). Typically this includes working farmsteads and ranches, but also includes some remote vacation and second houses on the public lands interface.
For additional information, refer to: Theobald, D.M. 2001. Technical description of mapping historical, current, and future housing densities in the US using Census block-groups. Natural Resource Ecology Lab, Colorado State University. 28 March 2001.
http://www.ndis.nrel.colostate.edu/davet/dev_patterns.htm
|
