Category Archives: Urban sprawl

Measuring urban sprawl: some results

I measure sprawl by looking at the pattern of development in the suburban portions of 59 large urban areas in the United States (see my earlier post).  I have delineated the urban areas for the census years from 1950 to 2010 using census tract data for housing units. Urban areas are defined as the contiguous tracts meeting a minimum density threshold in much the same way that the Census defines Urbanized Areas. The suburban areas are then defined as consisting of those tracts added to the urban areas after 1950.

Multiple measures of density in the suburban areas are used for the creation of the sprawl index. These include density, housing-unit-weighted density, and the percentages of housing units in census tracts with densities less than the first density quartile and the median density for all of the suburban tracts. Standardized scores are used to combine these measures into a single sprawl index for 2010.

First looking at which areas sprawl the most: Greenville-Spartanburg, South Carolina tops the list. But sprawling areas are not limited to the South, as Albany-Schenectady-Troy comes in second. And they are also not limited to the smaller urban areas in the dataset, as Boston-Providence is the eighth most sprawling of the 59 urban areas, which may be a surprise to many. But remember we are not talking about the very urban environments of Boston or Cambridge, we are talking about the levels of sprawl in the suburban areas developed after 1950.

The least spawling areas will be equally surprising. Las Vegas has the lowest level of sprawl, followed by Miami-Fort Lauderdale-West Palm Beach. And Los Angeles and Phoenix, areas seen by some as exemplars of sprawl, are the areas with the fourth and fifth lowest levels of sprawl.

Looking at the means of the sprawl index for the areas in the 4 census regions, average levels of sprawl are by far the highest for the areas in the Northeast. And they are equally low for the areas in the West. Areas in the Midwest and the South have mean levels of sprawl closer to the overall mean for all areas.

One factor that may be contributing to lower levels of sprawl in some areas is the presence of physical barriers to the expansion of the urban area. Urban areas up against mountains and wetlands had much lower levels of sprawl than other urban areas.

For those questioning this approach to measuring sprawl because it places Boston as high sprawl and Los Angeles low, consider this: The suburban areas around Boston had a density of 584 housing units per square mile. For Los Angeles it was 1,335. And 46 percent of all suburban housing units in Boston were located in tracts with densities less than the median for all of the urban areas, 611 units per square miles compared with only 9 percent for Los Angeles.

More on the measurement of sprawl and these results can be found in my paper, “An Alternative Approach to the Measurement of Urban Sprawl” which can be downloaded here.


Measuring urban sprawl: what is sprawl?

The previous post described my intentions to measure urban sprawl by considering the patterns of development in the suburban portions of urban areas. This necessarily requires specifying what aspect of that pattern is associated with degrees of urban sprawl. In other words, I need to define sprawl–a topic on which there has been little agreement.

Some have simply said of sprawl that, like pornography, “I know it when I see it.” Obviously this doesn’t help for measurement. Nor does citing areas as exemplars of sprawl or describing sprawl using aesthetic standards, e.g., ugly development.

Some have defined sprawl as unplanned development. But nearly all jurisdictions in the United States have planning and zoning, yet sprawl certainly is common. So is sprawl development that results from “bad” planning? How would you define this? It better not be planning that leads to sprawl.

Another approach has been to define and measure sprawl based on its causes or consequences. For example, automobile dependency might be seen as an indicator of sprawl. But then how would one account for other consequences of sprawl, say the loss of agricultural land or negative health effects? By saying that automobile dependency is associated with certain patterns of development? But then the patterns of development would seem to take precedence and be the actual indication of sprawl.

I come to the conclusion that sprawl must be defined by some aspects of the pattern of development. And low density and scattered or leapfrog development are most often cited as characteristics of sprawl. Virtually all studies that use a single measure for sprawl have used measures of either density or fragmentation.

And I can take one further step in simplifying things. Scattered development is separated by vacant land. Considering the pattern of development over somewhat larger areas, this vacant land results in low densities. So it is reasonable to take some measures of density for the measurement of urban sprawl.

A full description of my use of multiple measures of housing-unit density for the measurement of sprawl can be found in my paper, “An Alternative Approach to the Measurement of Urban Sprawl” which can be downloaded here.

Measuring urban sprawl: where is sprawl?

As I began my urban patterns research, I had intended to stay away from the characterization and measurement of urban sprawl. This is, after all, such a value-laden topic with much of the writing shedding more heat than light. But as I worked with my data, some ideas about the measurement of sprawl emerged that I thought might be a unique and useful contribution. So I undertook the research, wrote a paper, and am posting about it on this blog. I am doing  a series of 3 posts. This first one addresses what is the most distinctive contribution of my approach, asking what areas should be considered when measuring sprawl. The next discusses the definition of sprawl and its measurement in those areas. The third gives some results for sprawl for large urban areas in 2010.

The term “urban sprawl” is generally used to refer to patterns of suburban developemnt. Sprawl is seen as a characteristic of suburban areas built since about the middle of the last century. Indeed, numbers of people have used the term “suburban sprawl” to mean exactly the same thing.

So I have a modest proposal: If sprawl is a characteristic of suburban development, then the measurement of the extent of urban sprawl should consider the patterns of development in suburban areas. I think this seems perfectly logical and reasonable. However it contradicts virtually all efforts to measure and compare sprawl across urban areas, which have considered the patterns of development in entire urban areas.

If two urban areas have identical suburbs with the same amount of sprawl, why should it matter if the older, denser, nonsprawling portion of one area is denser than the other? Why should that urban area be considered to have a lower level of sprawl? But that is exactly what results from looking at the pattern of the entire urban area.

Measuring sprawl by looking at the pattern of the entire urban area can actually create a paradoxical situation. An urban area with a very large, very dense older core surrounded by a very sprawling suburban area could actually be seen as less sprawling than another area with a smaller and less dense core surrounded by suburbs with low levels of sprawl.

Further arguments for focusing on suburban areas when measuring sprawl and the use of this approach for such measurement can be found in my paper, “An Alternative Approach to the Measurement of Urban Sprawl” which can be downloaded here.

Images of Infinite Suburbia

Several years ago, I did a series of posts discussing how articles and books critical of urban sprawl were inevitably illustrated with overhead views of fairly dense single-family development (first post), while work promoting new urbanism and smart growth more frequently showed those environments from the ground-level perspective of their residents (second post). And indeed, some of those new urbanist developments looked surprisingly similar to the sprawling areas when vieweed from above (third post).

The new book Infinite Suburbia (edited by Alan Berger and Joel Kotkin) contains scores of articles looking at suburban development from a wide range of perspectives. (I will post some additional comments on this book when I have finished reading it. (The book is huge–over 700 pages, about 6 pounds!) Many different points of view are expressed in the various pieces. The book is certainly not an unambiguous critique of urban sprawl.

But despite taking a balanced view of suburban development, the book gives in to the idea that such development should be illustrated primarily with views from above. The book includes many 2-page color spreads (nearly 50) of aerial photos of suburban areas from around the world, frequently showing areas discussed in the accompanying articles. The pictures are gorgeous, but they reinforce the notion that suburban patterns should be perceived and understood from above, not from the points of view of their inhabitants.

2-page aerial view from Infinite Suburbia

2-page aerial view from Infinite Suburbia

To be sure, many of the articles include additional images of the suburban environments being discussed. But these are much smaller than the 2-page aerial-view spreads. (Some of the largest of the pictures within the articles were one-sixth the size (in terms of area) of the overhead images.) And many of these were additional overhead images, not views from the perspective of residents of the areas. The message, unfortunately, is unambigous: Suburban development is best viewed from above.

Loss of agricultural land in an uncertain world

One of the negative consequences attributed to urban sprawl is the loss of farmland that is converted to urban use.

In their paper questioning the desirability of compact cities and the problems associated with urban sprawl, Peter Gordon and Harry Richardson begin by presenting arguments that sufficient agricultural land is available in both the United States and the rest of the world such that any losses should not be a reason for concern. (See “Are Compact Cities a Desirable Planning Goal?” Journal of the American Planning Association, 1997). In his response, Reid Ewing accepts some of the evidence Gordon and Richardson presented but raises numbers of countervailing arguments. (See “Is Los Angeles-Style Sprawl Desirable” in the same issue.)

Obviously different conclusions can be drawn from these conflicting presentations. But one issue was not raised in discussing the significance of losses of agricultural land–climate change and global warming. (Ewing did mention greenhouse gas emissions and climate change when discussing energy use but not in relation to the loss of agricultural land.) Higher temperatures and related climate change could have significant effects on agricultural productivity. Both the climatic effects and their impacts on food production are highly uncertain. And this uncertainty should heighten concern over the potential consequences of the loss of agricultural land.

Density in Houston without zoning

Houston is well-known for being one of the only cities of any size not to have zoning. So an obvious question is how the urban pattern of Houston might differ from that of other large urban areas. Let’s consider the population density.

The density of the Houston Urbanized Area in 2010 was 2,978 persons per square mile. Dallas-Fort Worth seems to be a reasonable area for comparison. It is another large urban area in Texas, so aside from zoning, one might suspect other factors affecting density might be similar. The density of the Dallas-Fort Worth Urbanized Area in 2010: 2,879 persons per square mile, very close to Houston.

What about other large urban areas? Seattle is about as far away from Houston both in terms of distance and many other characteritics as you can get. The density of the Seattle Urbanized Area in 2010: 3,028 persons per square mile, just a tiny amount larger than Houston. Or how about the Philadelphia and Detroit Urbanized Areas, with densities of 2,746 and 2,793 persons per square mile, slightly below Houston.

If Houston doesn’t differ in terms of density, what about the other measures of the urban pattern that I have been using (see this post? Here are the values for 2010 for the Houston and Dallas-Fort Worth areas as defined for my research (note density is now in housing units per square mile):

Urban area Density Dissimilarity (Variation) Centralization Ratio Moran’s I (Clustering)
Houston 1,066 0.31 0.12 0.41
Dallas-Fort Worth 1,066 0.31 0.17 0.51

Density and variation are identical (and their densities were nearly identical in 1950 as well). Houston is somewhat lower than Dallas-Fort Worth with respect to centralization and clustering. Not sure what to make, if anything, of these small differences. And remember that Dallas-Fort Worth is an area resulting from two separate urban areas growing together, so there’s that difference as well.

My major conclusion is that, despite the absence of zoning, the urban pattern in Houston does not look that different from other large urban areas.

Always compromises–never black and white

I live in Upland, California, a suburb about 35 miles east of downtown Los Angeles. A short block from our house is Euclid Avenue, a broad, busy boulevard running from south of the city north to the foothills of the San Gabriel Mountains.

Euclid Avenue is the signature feature of Upland, created when the city was originally laid out in the latter 1800s. The street is primarily lined by residences, becoming larger and more impressive as one approaches the mountains. Its 65-foot wide median is lined with large old pepper trees and has a spacious path in the middle constantly used by walkers and runners, the occasional horse, and I once even saw a llama. (The median was originally used for a streetcar line, first powered by a mule which pulled the car up the slope to the top of Euclid Avenue and then got to ride down on a trailer at the back as the car descended powered by gravity.)

Euclid Avenue, Upland, California

Euclid Avenue, Upland, California

Philip Langdon, in his book A Better Place to Live, excoriates contemporary suburbs for their boring and placeless design and their obstacles to mobility given their street patterns. Yet he devotes 8 pages of the book to describing and praising Euclid Avenue. (This is worth reading; his account is accurate and comprehensive.) He praises the frequent cross streets, 28 in 7 miles, as providing pedestrian access and not causing problems with the heavy traffic on Euclid Avenue.

Here’s my problem: Langdon gives the count of the number of cross streets, but he does not make the obvious point that these come each quarter mile. The blocks, in the north-south direction on Euclid, are a quarter-mile long. This separation of the cross streets is not unreasonable for a heavily traveled arterial. But these are awfully long blocks from the perspective of Langdon’s desire for a grid of connected streets and maximum pedestrian opportunities. A compromise reasonably required by the presence of the busy boulevard.

And Langdon fails to address the rest of the story. Upland is not this jewel in the midst of the automobile-oriented faceless suburbia that he rails against. Not only are the blocks a rather long quarter mile in the north-south direction, many extend a full half mile in an east-west direction. Some of these huge blocks are cut through by more-or-less straight streets creating smaller blocks. But many of these large blocks are pierced only by cul du sacs and meandering streets that are the antithesis of the grid that Landon espouses.

Things are seldom black-and-white, good or bad. Before moving to Upland, I lived in Zionsville, Indiana, a suburb of Indianapolis. At its core was the original town settled in the mid-nineteen century with a street grid with small blocks, old houses on small lots, and a small, quaint downtown on a brick street. This part of Zionsville was affectionately referred to as the “village” and was the major element of the town’s attraction. And the village was surrounded by newer suburban development of exactly the sort Langdon decries.

Next door to Upland is the city of Claremont, home, of course, to the Claremont Colleges. It too has an older, somewhat denser core and an extensive, lively downtown area (attributable at least in part to the presence of the colleges). Claremont likewise refers to this portion of the city as the “village.” And most of the remainder of the city consists of typical suburban development.