First International Conference on Spatial Data Acquisition in GIS
Title
Measuring Urban Bulge: The Case of Settlement Density in Sacramento County

Author
Lia Sullivan
American River College, Geography 350: Data Acquisition in GIS; Spring 2010

Abstract
The presence of low density development in Sacramento County is examined using Development Footprint, a digital raster authored by the California Department of Forestry and Fire Protection.

Introduction

“The density of cities is their most basic advantage over any other kind of settlement. Without density of settlement most of what we learn, produce, construct, organize, consume, and provide as a service in the world would simply be too expensive. Density increases the sheer efficiency by which we can pursue an economic activity.” - Jeb Brugmann, Welcome to the Urban Revolution
Definitions of sprawl are often as wide-ranging as the phenomenon the word itself seeks to describe. This is due in large measure to the indictment of urban sprawl for a host of crimes, including, but not limited to usurping open space and rural farmland, threatening species and watershed, and increasing air pollution and traffic congestion. Surrounding these assertions a debate has arisen about the actual versus perceived harm caused by sprawl. In his book, The Limitless City, William Gillham asserts, “Even the supporters of the status quo no longer deny the existence of sprawl, but any consensus stops right there, because not everyone agrees about exactly what sprawl is.” Without a precise definition for what constitutes sprawl it becomes impossible to identify what separates sprawl from the natural trajectory of urban growth by way of economic and demographic factors. How then for the geographer to overcome this problem? Establish a definitive nomenclature for sprawl and identify a data set which joins defined classes to cartographic display.

The aim of this project was to apply density measures to detect the presence of sprawl within the boundaries of Sacramento County. Have the newly incorporated cities of Elk Grove, Rancho Cordova, and Citrus Heights achieved levels of urban density? For this purpose Development Footprint, a data set of high spatial and thematic resolution, authored by the Fire and Resource Assessment Program (FRAP) of the California Department of Forestry and Fire Protection, was employed for analysis.

Background
Professor Reid Ewing of Florida International University identifies a set of characteristics unique to sprawl that have been recognized collectively by urban scholars dating back to 1957 (Gillham, 2002). The four most commonly cited attributes of sprawl are: leapfrog or scattered development, commercial strip development, low density, and large expanses of single-use development (ibid). In the scope of this project, only leapfrog and low density development are considered, since these two characteristics are most easily determined from the FRAP’s Development Footprint . They are also casually linked so that the presence of one makes the other easily detectable.

Gillham defines leapfrog development as subdivisions that have ‘leapfrogged’ over interceding land. The result is a patchwork of empty tracts of land between subdivisions that appear almost like outposts. The affect of leapfrog settlements on the landscape is best captured from the air.

this is my leapfrog photo



The single most prevailing ground characteristic of sprawl is low density. There are several ways to quantify density: the number of people per acre of square mile, the number of dwelling units per acre, or floor area ratio (FAR) (Gillham, 2002). FAR is the ratio of the number of square feet of built area to land area; generally included in FAR calculations are commercial buildings (ibid). While the methods of quanitative calculation may differ, the resulting measurements remain constant: sprawl settlements are neither as high as the urban core nor as low as the rural countryside; they are somewhere in between (ibid). And this generalization holds true despite the fact that the central density of a city varies regionally as a result of geography, history, and economics. Thus a city’s core density becomes the standard for comparison to its suburban counterpart.

Methods
To determine density levels for the county, I downloaded Development Footprint (DEVELOP05_1) from Cal Fire's website and brought it into ArcMap for analysis. Development Footprint, a raster digital data set, is an attempt by the Fire and Resource Assessment Program (FRAP) to “spatially define the footprint of development”, both commercial and residential, throughout the state of California from 1992 through 2000. The strength of this data set lies in its use of multiple sources to create specific classes of development density in the region. Three separate fields contained within the dataset were used for analysis: NLCD, DenClass10, Develop10. The NLCD is a National Land Cover Dataset from USGS, captured at 30 meter resolution and resampled by FRAP to 100 meters. FRAP modified the data set so that low intensity residential areas were limited only to census defined urbanized areas; all high intensity residential areas were left intact. The DenClass10 field was derived from 2000 census block data on housing; classifications 1 through 10 are based on housing units per square mile. FRAP created the Develop10 dataset using NLCD classes to modify DenClass10. The result is a comprehensive scale of development levels encompassing both commercial and residential uses.

To begin, I took the FRAP raster for the state of California and clipped it to the boundaries of Sacramento County. I acquired a communities shapefile from the Sacramento County GIS database and added it to my layer. All layers were reprojected to Teale Albers to match FRAP’s DEVELOP05_1 raster. For display and analysis, I then created a series of unique values maps for each of the three fields described.

Results
When creating symbology for the NLCD maps only commercial and residential fields were chosen for display. Low intensity residential, which is a mixture of vegetation and constructed materials, is shown in bright pink. High intensity residential, in which constructed materials account for 80 to 100 percent, is purple. Commercial areas, highly developed areas not classified as high intensity residential, are yellow. Not surprisingly, low intensity residential areas comprise nearly all of the settlement within our county. Detached single homes under the cover of our urban arboretum are a defining characteristic of our county’s urbanized areas, so the ability to discriminate density differences between neighborhoods and communities by NLCD classes was infeasible. The only indicators of sprawl are the patchy appearances of pink, low-residential areas, most likely leapfrog settlements. I did, however, find the layer useful for comparison to an historical USGS land cover and land use data set from the 1970 and 80’s. Though based on a different classification system (Anderson Level II Classification) than the 1990’s NLCD from FRAP, I employed identical colors for all residential and commercial areas to permit a visual comparison.

this is my nlcd map


What was lacking in the NLCD map, a series of gradations based on quantitative classes, I found in the DenClass10 map. The class breaks for DenClass10, which are scaled according to the number of housing units per square mile, are found in the table below. The colors were chosen to display rural areas in shades of red and orange; low-density urban settlements in shades of yellow, green, and light blue; and high density in dark blue and purple.

DenClass Units per Square Mile
1 None
2 <4
3 4 - 16
4 16 - 32
5 32 - 64
6 64 - 128
7 128 - 640
8 640 - 1280
9 1280 - 3200
10 > = 3200

this is my denclass10 map

Since it is expected that density levels for sprawl are somewhere between the lowest densities of rural areas and the highest densities of compact urban areas, density classes four through seven would be most indicative of sprawl settlement. The choice for such a break is supported both by FRAP, which defines “urban” as Density Class 8 or greater (see FRAP metadata) and by Gillham’s definition of sprawl. However, since DenClass10 shows only levels of residential development it is not a complete “picture”. Areas that are commercially developed are falsely identified as class 1, alongside undeveloped or agricultural land. That explains the rose-colored shadings for commercial areas of downtown.

By integrating the NLCD and DenClass10, Develop10 is an attempt by FRAP to resolve the ambiguities and deficiencies of the former datasets and establish a comprehensive scale of human development for California. Where NLCD data was absent Density Classes are identical to Develop10 classes. Additionally where density classifications were 8 or above (DenClass10>=8) Develop10 classes are identical (DenClass 8,9, & 10 = Develop 8, 9, & 10): no modifications were made on the basis of NLCD data. Density classes that were less than class 8 were modified on the basis of NLCD classifications. If DenClass10 values were less than 8 and NLCD equaled low-intensity residential it became class 8 under Develop10 (DenClass10 < 8 and NLCD = 21, Develop10 = 8). Low intensity residential classification within an urbanized area affectively trumped housing density classes under FRAP’s schema, but not enough to receive the highest classification of Develop10. However, if DenClass10 values were less than 8 and NLCD equaled commercial or high-intensity residential it was effectively upgraded to a Develop10 Class 10, indicating the highest intensity use (DenClass10 < 8 and NLCD = 22 or 23, Develop10 = 10). Thus Develop10 affectively maintains classes 9 and 10 for the highest levels of urbanization, a sufficient criterion for the detection of sprawl settlement.

this is my comparison map


Finally, the communities of Elk Grove, Citrus Heights, Carmichael, and Rancho Cordova were selected for further analysis and comparison using the Develop10 dataset. With the exception of Carmichael (chosen as a “control” for the group), each of these communities has been recently incorporated, marking a maturing process in their growth trajectory from suburb to city. Do these newly formed cities demonstrate levels of development density characteristic of a city? Or are they merely low density suburbs with city status? On the basis of the comparative map below a set of conclusions were drawn.

this is my community comparison map



Analysis
Rancho Cordova, incorporated in 2003, shows predominantly classes of 8, 9, and 10 development density concentrated in the northwestern area. The large swath of undeveloped land, to the east (colored light rose) offers Rancho Cordova, projected by the Sacramento Area Council of Governments to be the fastest growing community in both housing, employment, and population growth within the next few decades, room to expand; and the established pattern of dense development absent leapfrog settlements suggests that the community is equipped to both effectively absorb, sustain, and benefit from the commercial and residential growth of the future. Citrus Heights, which incorporated as early as 1997, is covered mostly by dark blue shades and appears as the densest settlement in the sample group along with Carmichael, a census designated place that till now has not shed its suburban status for cityhood.

Interestingly, it is only Elk Grove, a city since 2000, which shows significant areas of low-density development. Most of this class 4 through 7 development lies on its eastern side, but even areas along its western corridor where the majority of development is clustered display only sparse patches of class 9 and 10. Admittedly, in the years since 2000, when the dataset was created, Elk Grove underwent explosive housing growth, and if a policy of structured growth and infill had been adhered to the result would be higher levels of development density than displayed by Develop10; so to test the reliability of this hypothesis, NAIP 2009 summer imagery was examined.

this is my elk NAIP image


this is my elk NAIP image with Develop10 transparency

Develop10 transparency layer over NAIP 2009 aerial imagery.

Despite the growth of the last decade, FRAP’s development assessment remains surprisingly accurate for Elk Grove: an overall case of low density despite new buildling. The significant gaps puncturing the landscape suggest a consistent pattern of leapfrogging. It is fair to conclude that the recent growth was sporadic and unplanned since no attempt was made to locate new settlements in absolute proximity to one another. The picture of poor development density for Elk Grove remains unchallenged.

Conclusion
In his book Sprawl: A Compact History, Peter Brugman, uses historical London as a paradigm case of urban growth. Brugman cites the flight of nineteenth century residents away from the squalor, contagion, and overcrowded center of industrial London to lower density settlements around the periphery. Overtime, as the process was repeated, the central density of London declined while the suburbs’ increased. The fact that such a process has been repeated elsewhere and at other times makes the “problem of sprawl” appear as a short term phenomenon - or rather phantom - with no teeth.

After Census 2000 a similar demographic shift was noted in the United States: suburbs located in the West reached levels of urban density and many had developed into cities. Examining FRAP’s Development Footprint, it becomes evident that the communities of Rancho Cordova, Carmichael, and Citrus Heights followed this suburban-urbanizing trend. So why worry about a sprawling Elk Grove? One should expect that overtime it too will follow course. Ironically, during the same 2000 Census, it was also noted that the suburbs of the South had become more sprawling. This suggests that suburban densification is not inevitable; nor is a low density pattern necessarily reversible.

Settlement patterns become “facts on the ground” that are difficult to change when future circumstances call for alteration, so that a certain pattern of low-density settlement accompanies a set of unknown costs purchased at a future date. Will existing patterns compromise a community's ability to accommodate natural population growth and the influx of new settlers? And if not what sort of tensions might arise between communities and between communities and the land as a result of such strains? Presently, there are other hidden costs to low density development. For example, in lower density areas residents generally pay the same price for water, sewer, and trash collections, though it costs the servicing agency much more to provide these services in terms of infrastructure and labor (Brugmann, 2009). Increased density contributes fundamentally to the economic and social efficacy of an urban community (ibid).

At this time, it hardly appears that Elk Grove is contemplating a change of course. For the past two years Elk Grove has been seeking to annex roughly 14,000 acres of pasture land and open space to the south and east of its present boundary. The move has come under some criticism from citizens and county officials alike who want the city to pursue growth within its current boundaries first. Unfortunately, data sets are not proscriptions: they simply mirror facts on the ground. Yet, if the data could talk it would undoubtedly say, “Elk Grove does not need more land, it needs to use the land it has more effectively.”

References

Bruegman, Peter. Sprawl: A Compact History. Chicago: University of Chicago, 2005. Print.

Brugmann, Jeb. Welcome to the Urban Revolution: How Cities Are Changing The World. New York: Bloomsbury Press, 2009. Print.

California Department of Fire and Forestry Protection. Development Footprint. (DEVELOP05_1).
CDF Fire and Resource Assessment Program, 2 Dec. 2005.

Firestone, David. “90's Suburbs of West and South: Denser in One, Sprawling in Other.”
The New York Times. Sprawl City, 17 Apr. 2001. Web.

Fulton, Bill. “Elk Grove: 7-Year-Old City Seeks Land for Expansion.” California Planning and Development Report. CP&DR, 30 Oct. 2007.

Gillham, Oliver. The Limitless City: A Primer on the Urban Sprawl Debate. Washington, DC:
Island Press, 2002. Print.

Kalb, Loretta. “District 5 Candidates Debate Elk Grove Expansion.” The Sacramento Bee.
Sac Bee Online, 3 May 2010.

Sacramento Area Council of Goverments. SACOG Information Center: Demographics. SACOG. n.d.

Appendices
The metadata accompanying the Development Footprint was considerably entensive, but by no means complete. The metadata in its entirety has been catalogued online by Conservation Biology Institute. See the full (metadata).