Grant Ian Thrall,
Geo Info Systems,
Mark McClanahan,
and Susan Elshaw-Thrall
" Ninety Years of Urban Growth
as Described With GIS: An Historic Geography"
volume 5, Number 4, April 1995; pages 20-27, 45.
Grant Ian Thrall,
Geo Info Systems,
Mark McClanahan,
and Susan Elshaw-Thrall
" Ninety Years of Urban Growth
as Described With GIS: An Historic Geography"
volume 5, Number 4, April 1995; pages 20-27, 45.
Susan Elshaw Thrall is a cofounder of Spatial Decisions and Analysis Inc., and serves as its Vice President. She is a Professor of Computer Science, College of Business, Lake City Community College. She has a Ph.D. in Geographic Information Systems, Computer Sciences and Computer Education from University of Florida.
A vague understanding of how cities grow results in muddled and misguided solutions to urban problems and vague awareness of opportunities that arise from that growth. In this column, we discuss the use of geographic information systems (GIS) to describe how the land use of a country has changed during the past 90 years.
Thrall (1995) writes that description is the first step in reasoning with GIS. In this column, we provide a framework for describing urban growth more precisely and more productively than would be possible without GIS technology. The framework outlined here can be duplicated to describe urban growth elsewhere. This example is from St. Lucie County, Florida, USA (see Thrall and Elshaw-Thrall, 1991; and McClanahan, 1993). Every professional whose interests are directly or indirectly linked to the urban built environment benefits from knowledge about how a city grows. Planners need this information to anticipate where infrastructure should be installed. Housing developments require spatial information to take advantage of the local market: where and how much housing can be constructed and sold (in the real-estate industry this is referred to as housing absorption). Financial institutions use this information to evaluate the risk of their loan portfolio. Environmentalists can use information about city growth to be proactive in acquiring land and working to get zoning ordinances in place long before a construction team arrives on a site.
Spatial Patterns of Urban Growth
The history of understanding the geographic regularity for land-use patterns begins with the contributions of Johann von Thunen (1826). Early in the nineteenth century, von Thunen introduced a pioneering approach to problem solving that today would be comparable to a "spreadsheet." Von Thunen used his mathematical constructs to perform "what-if' type scenarios for changes in land-use patterns: What if the agricultural market price increases? what if the cost of production changes? what if productivity increases? what if the cost of transportation increases? and so on. What will happen to market land values by specific location? How and where will the spatial patterns of land use change?
Alfred Marshall (I890) addressed how market forces affect change in urban land use patterns. He was perhaps the greatest economist of the late nineteenth century. In building upon von Thunen's economic geography work, Marshall raised fundamental and still-provocative issues about spatial patterns of urban growth, which Thrall (1991) has grouped into five axioms:
* Land is a factor of production and "is the foundation of much that is most interesting
and most difficult in economic science" (Marshall, 1890).
* Land has area and density, and is a complement to capital and labor. "[Tjhere is a
certain application of capital and labour to the acre which gives the highest return" (Marshall,
1890, page 447).
* The "situation value" and "site value" of land are measures of the importance of
relative spatial location (Marshall, 1890, page 441).
* Land is the recipient of external economies from natural and built environments.
Localization of manufacturing industry arises from agglomeration economies that result from
density and scope of labor market, the spatial distribution of physical resources, and threshold
demand. Concentrations of service industry by saving customers transportation cost and time.
Because of increased volume of customers and increased visibility, higher-order facilities benefit
by agglomerating with other higher-order facilities (Marshall, 1890, pages 267-273).
* Land value and intensity of land use are the result of spatial equilibrium processes. The
system comes to be in spatial equilibrium by way of the movement of people and industry, which
in turn changes population density and land values (Marshall, 1890, pages 429 and 449).
Marshall (as interpreted by Thrall, 1991) believed that the process of suburban land-use
transition from agricultural or other nonurban use followed a repetitive and predictable
three-sequence pattern:
* Manufacturing firms locate at the urban periphery where land values are
comparatively low.
* Housing is subsequently constructed nearby for the employees of the manufacturing
firms.
* Retail sales and the service industry follow the construction of housing.
This sequence of land development increases neighborhood land values, making
further manufacturing expansion unlikely. Instead, new manufacturing expansion takes place at
the present urban periphery where the land values are less. The cycle at the urban periphery
repeats what happened earlier and closer to the core of the city.
Urban growth, according to Marshall, begins with the addition of new jobs at the urban periphery. The arrival of retailers signals the climax and end of the development process.
It is worthwhile to mention that more than a century ago manufacturing was considered to be a suburban phenomenon - not an economic activity that is attracted to the urban core. (Contemporary supporters of urban-core enterprise zones should take particular note.)
Manufacturing is not attracted to the urban core; instead, manufacturing becomes the core around which the city grows.
The study of how a city grows spatially may be referred to as urban morphology. Many descriptions and theories of urban morphology have been proposed since the seminal work of von Thunen and Marshall, including the contributions of Burgess (1925), Hoyt (1939), Harris and Ullman (1945), Alonso (1964), Thrall (1987), and many others.
Those discussions have traced how a variety of phenomena have affected city growth patterns, including changing income levels, crime incidence, racial conflict, transportation, planning controls, and amenities. Land economists generally consider transportation, or change in access, to be the most important phenomenon.
The urban, built environment has been shaped by the modes of transportation in use during the era that the development occurred. Narrow lot faces characterize the era when people relied primarily on their feet for transportation. The introduction of the trolley car allowed for increased access and more dispersal. The automobile era of the 1930s and 1940s led to new urban perimeter, middle- and upper-income neighborhoods. But these neighborhoods were still close to places of employment and traditional shopping.
Urban development of the 1950s follows the path of the freeway corridors, radiating spokewise outward from the urban cores. High-speed transportation allowed middle-income households to choose living accommodations from an expanded area. House lots became wider, commutes to work became further, and, just as Marshall predicted in 1890, the retailers followed. But it is unlikely that Marshall imagined today's shopping malls exceeding one million square feet.
In some places, multiple freeway routes transacted the city and, in the 1970s, began to be connected by the "outerbelt." Today, more than half of American households commute from their homes in the suburbs to their workplaces, which are also in the suburbs. The outerbelt has for many become more important than the freeway spokes the outerbelts are designed to link. The outerbelt urban environment has been named by Washington Post journalist Joel Garreau (1988) as the "edge city."
The development of the edge city, and the way that land is used there, is no surprise to those land economists who theorized the development of our urban landscapes. Arguably, what eluded the theorists of land economies is the enormous scale of the edge city. Edge cities dwarf in square footage and acreage the earlier urban commercial centers.
Recall that Marshall's sequence of development begins with the introduction of an urban-perimeter factory. In edge cities, manufacturing may be absent. The economic base may be retail sales and services. And for urban morphologists, it appears that the sequence of development today is different from what Marshall described in the nineteenth century. Today,
* investment in transportation infrastructure and improvement of transportation
technology increase access;
* changing access is accompanied by demographic change;
* employers (service sector in suburban office buildings) follow, locating near to
where potential employees want to live; and
* retailers still follow the demographics; now, however, those retailers and service-
sector employers are themselves significant engines of employment. In an edge city, employment
in the manufacturing sector may be totally absent.
What evidence do we have that the above narrative is true? Our intuition is supported
by our personal experiences with the contemporary urban landscape. But our intuition can be
wrong; our personal experiences may not be representative of the larger, true state of affairs.
And, though interesting, the narrative is vague.
Again, a vague understanding of the issues of how cities grow results in vague solutions to problems and vague awareness of opportunities that arise from that growth. We use GIS technology to add precision and remove the vagueness from our descriptive narrative. With the high level of precision that GIS technology allows, we can summarize the outcomes of the large spatial forces that have changed our urban, built environment from the era when households relied upon their own feet for transportation, to the contemporary freeway-era edge city.
A GIS FOR HISTORICAL GEOGRAPHY
An array for cataloging geographic phenomena (Berry, 1964) was discussed in the previous "Shop Talk" column (Thrall, 1995). In Berry's array, geographic phenomena are placed in a three-dimensional cube according to activity, time of occurrence, and location. Following this guideline, the St. Lucie County urban growth patterns will be described as they occurred from 1900 to the present. The activity is urban growth as measured by the number of single-family housing units, the square footage of eating and drinking establishments, the square footage of office space, the square footage of the service sector, and the square footage of the retail sector. The time spans the period from the beginning of development of St. Lucie County in 1900 through 1992; time is grouped by decade. The spatial component is measured by location, relative to the first urban development in the city of Fort Pierce, Florida; location is grouped by one-half-mile bands north or south of the central city.
St. Lucie County is situated north of Palm Beach and Martin counties in the southeastern coastal area of the Florida peninsula. The legal property-tax records of the county contain 132,000 parcels. All parcels were geocoded and included in this historical geography. The data base includes 41,232 houses and 365 retail properties, totaling about 1.5 million square feet. Only three attribute fields were used: Department of Revenue (DOR) Land Use Code, year built, and square footage. A detailed discussion about the construction of the geocoded data base is in Thrall and Elshaw-Thrall (1991).
We used three software programs for this analysis: Atlas GIS (Strategic Mapping Inc., Santa Clara, California, USA), dbase IV, and Quattro Pro (Borland, Scotts Valley, California).
When time trends are summarized, the phenomenon under study is aggregated into some commonly accepted time periods: hours, days, weeks, months, years, and so on. For example, the closing of a house sale occurs at a specific hour, but instead of tracking the average price of house sales by the hour, we are usually interested in only a summary of the prices houses sold for by the month or quarter (three months). The analyst chooses how to aggregate the information so that the summary is condensed but also useful. Tracking by the hour presents too much information; tracking by the century presents too little information. The unit measure of time must be at a level of aggregation to be useful input for our judgment (Thrall, 1995).
Likewise, the level of spatial aggregation must be useful. Analysts, by convention, rely on the federal government's definition of spatial units: census blocks, block groups, tracts, city boundaries, county boundaries, and so on. But these spatial units may not be useful for spatial description. GIS technology allows us to define our own level of spatial aggregation.
For this analysis, we want to set the area of spatial aggregation so that it reveals the spatial-time pattern of development. The selected spatial units should be able to reveal the effects of changing land-use patterns as anticipated from the previous discussion. Therefore, we chose to partition the urban landscape into polygon rings one-half-mile wide, circling the original urban core of St. Lucie County, the city of Fort Pierce. To capture spatially nonuniform growth patterns that result from the city growing more in one direction than another, the concentric-ring polygons were split into two parts - half lying to the north of Fort Pierce and half lying to the south of the city.
When defining spatial aggregation as concentric zones around a central node, bias can enter the analysis, including:
Directionality. Retail growth is expected by central-place theory (Christaller, 1966) to be clustered, but this procedure for aggregation will assign total retail square footage to the entire polygon. Therefore, a shopping center lying to the southeast will add to the value of the polygon that overlays the southwest as well.
Likelihood for capture. The size of the land area inside the concentric zone increases with distance from the central point, therefore increasing the likelihood that the polygon includes a given phenomenon; larger aggregate values are expected in distant concentric zones because more of the targeted phenomena can locate there.
Splitting. The concentric-zone polygon may split a commercial area, thereby dividing the activity between two adjacent polygons; but, had a different spatial increment been used - for example, two miles instead of one mile - a different aggregate summary measure may result. Also, if the commercial activity is defined as a single legal unit, the parcel may not be split by this procedure. Instead, if the activity is legally defined as multiple parcels with multiple coordinates, then the activity may be split merely by change of legal definition, not by change in geography.
These three problems of spatial bias should be kept in mind as the results are interpreted; however, the geography of the example site, St. Lucie County, somewhat lessens the problems. The county is bounded on the east by the Atlantic Ocean. Most development follows the north-south U.S. Highway I corridor. To the west of this near-linear development are orange groves and marshland.
Fifty-four half-circle polygons were constructed using Atlas GIS in the following manner.
* First, a central point was specified at the comer of Orange Avenue and 13th Street,
part of the historical central core of Fort Pierce.
* Second, the central point was buffered by a one-half-mile radius. The operation was
repeated increasing the radius by another one-half mile. This procedure was repeated until the
entire study area was covered.
* Third, these buffers were converted into polygons.
* Fourth, a polygon-over-polygon operation was executed to eliminate the portions of
the concentric zones that lay outside the county boundary.
* Fifth, the concentric zones were split in half along an east-west transect passing
through the central point. Each concentric ring polygon was given an attribute name according to
its distance from the central point and its direction - north or south of the central city.
Next, a point-in-polygon operation was performed to identify in which polygon the
property observation (a point) was situated. The point was assigned the attribute name of the
polygon it fell within. The property records were then exported from Atlas as a data file. This
last operation completes the required GIS software procedures by adding to the property record
the concentric ring in which the property is located.
We used dbase to count the number of properties built during each time period for each property type. For the description that follows, a decade-long time span was chosen. For commercial property, the macro also summed the total square feet by concentric ring for each time period. The dbase file was then exported to Quattro Pro, where three-dimensional graphs were created with the axes being activity, time, and location.
RESULTS
In the five figures, the first horizontal band (orange) is cumulative activity by location from 1900 through 1939. Successive horizontal bands measure cumulative activity from previous time bands through the particular decade. The last horizontal (red) band is 1980 through 1992.
From the beginning of St. Lucie County's urban development through 1939 (the orange band), most housing development was concentrated near Fort Pierce (Figure 1). Retail was almost entirely located in one distance band to the north of our "ground-zero" point in the central city (Figure 2). All service-sector square footage was tightly concentrated in the central city (Figure 3). Office space fell on either side of the central-city core, but with the larger aggregate square footage situated on the north side (Figure 4). Eating and drinking square footage was almost exactly balanced in distance bands to the north and south of the central-city core in the central city (Figure 5).
From 1940 through 1949 (the blue band), the net addition of housing amounted to 1,174 units, almost a threefold increase compared to the net addition the previous decade (see Figure 1). Housing development occurred in distance bands tightly concentrated around the central-city core. In the 1930s, some "leapfrog" type housing development began at about the four-mile distance band to the south of the central-city core. The 1950s (the magenta band) shows that infill occurred bridging the earlier leapfrog developments with the central city.
But, as the gap was filled, new leapfrog developments began in the 1950s around the eight-mile-distance band. The city of Port St. Lucie is situated in that part of the county and became the central core around which comparatively colossal housing development occurred in the 1970s, becoming the "red wall" by the early 1990s. By contrast, the northern part of the county was rather dormant. But evidence reveals that housing development to the north in the 1980s (the red band) began to parallel the development that had occurred to the south in the 1960s (the yellow band).
In the 1960s, Port St. Lucie's south side emerged as the focal point for new housing development. Yet retail development remained concentrated around the central city core of Fort Pierce (Figure 2). Significant suburban-retail development did not occur until long after the demographics were in place in the southern distance bands during the 1980s.
Eating and drinking establishments are a special case of pioneering businesses; they are the first to respond to the demographic transition at the edges of an urban area (Figure 5). Significant eating- and drinking-establishment development began in the 1970s (the green band) in the southern distance bands following the nearby housing development of the 1960s.
Retail growth in the central area did not follow the flight of housing until after 20 years of consistent housing decentralization; it was not until the 1980s that retail square footage stopped increasing in the Fort Pierce city center (Figure 2). Retail growth followed housing growth.
During 1980 to 1992, retail growth transplanted to the outer areas. Housing increased by 9,818 units during the decade of the 1970s, and by 22,692 in the following time period. Most of this increase occurred south of Fort Pierce in the Port St. Lucie area. At the same time, retail growth increased from 462,985 total square footage through 1979, to 1,881,096 through 1992.
Three miles north or south of the central-city core, there is very limited net addition of new housing; hence, there would be no expectation that retail square footage would increase in those areas. Moreover, Fort Pierce is not central to the new demographics (the red wall) lying to the south. Fort Pierce's retail future hinges on the prospect that during the next 20 years the lagging northern part of the county will catch up and balance the growth of the southern county. If that happens, Fort Pierce will once again be central to the demographics of the county.
Development - including housing, retail, and office space - was initially compactly clustered around a so-called ground zero" central-city core of the 1930s. By the 1990s, a transition had occurred where the greater preponderance of housing (people) was situated at the earlier urban edge. Eating and drinking commercial activity is the first to pioneer the new demographics, followed by service sector and office space. Unlike in Marshall's day, people come first. Also, it is worthwhile to note that the evidence for urban growth presented here is that urban development need not be symmetric. Indeed, urban growth can be highly skewed, leaving the original central city core at the perimeter of the new demographics.
CONCLUSION
This description of urban development relies on GIS technology and provides the professional with a perspective of urban growth that would otherwise require years of personal experience to understand. A significant benefit of GIS technology is its ability to help users visualize spatial data. In the descriptive phase of GIS analysis, data visualization ". . . assist[s] our ability to interpret information visually and improve our intuitive understanding of the distribution and interrelationships among phenomena" (Thrall et al., 1993). Many other figures of similar construction for other study sites would be required, however, before we could claim that what these trends reveal is representative of cities elsewhere.
Our analysis used the entire legal-property record of a county from its beginnings in 1900 through 1992. This presentation is one of the few, perhaps the first, to comprehensively document the spatial patterns of urban growth using GIS technology. Earlier comprehensive documentation was done by Homer Hoyt (1933) and a large staff during a several-year period. Our findings that cities are not symmetric in their development are consistent with Hoyt's findings that cities are sectorial. We have also found that retail development lags behind housing construction. As has been argued elsewhere in this "ShopTalk" series, many GIS benefits can be obtained by using inexpensive, point-based data rather than spatial data bases that offer legally accurate property boundaries, but that incur high costs of construction and long-term development horizons (Thrall and Elshaw-Thrall, 1991). The cost of GIS can be justified early in its development with useful analyses.
This work demonstrates that historical geography of the city can be the foundation for all urban constituencies to make better decisions.
REFERENCES
Alonso, W. 1964. Location and Land Use. Cambridge, Mass.: Harvard University Press.
Berry, B.J.L. 1964. "Approaches to Regional Analysis: A Synthesis." Annals of the Association of American Geographers (54):2-11.
Burgess, E.W. 1925. "The Growth of the City' " In The City, Eds. Park, R.E., Burgess, E.W.; and McKenzie, R.D. Chicago: University of Chicago Press.
Christaller, W. 1966. Central Places in Southern Germany. Translated by C.W. Baskin. Englewood Cliffs, New Jersey: Prentice-Hall.
Harris, C., and Ullman, E.L. 1945. "The Nature of Cities," Annals of the American Association of Political & Social Sciences 242:7-17.
Hoyt, H. 1933. One Hundred Years of Land Values in Chicago. Chicago: University of Chicago Press.
Hoyt, H. 1939. The Structure and Growth of Residential Neighborhoods in American Cities. Washington, D.C.: Federal Housing Administration.
Hoyt, H. 1966. According to Hoyt.- 50 Years of Homer Hoyt (1916-1966). Washington D.C.: The Homer Hoyt Institute.
Marshall, A. 1890. Principles of Economics, 8th edition, 1946. London: Macmillan.
McClanahan, M. 1993. "Ninety-Two Years of Housing and Retail Growth in St. Lucie County, Florida." Unpublished master's thesis. Gainesville: University of Florida.
Thrall, G.I. 1987. Land Use and Urban Form. London: Methuen.
Thrall, G.I. 199 1. "Production Theory of Land Rent." Environment & Planning, Series A. (23): 955-967.
Thrall, G.I. 1995. "Stages of GIS Reasoning." Geo Info Systems 5(2):46-5 1.
Thrall, G.I., and S. Elshaw-Thrall. 1991. "Reducing Investor Risk: A GIS Design for Real Estate Analysis." Geo Info Systems, 1(10):40-42.
Thrall, G.I., and S. Elshaw-Thrall, 1993. "Commercial Data for the Business GIS." Geo Info Systems, 3(7):63-68.
Thrall, G.I., S. Elshaw-Thrall, M. Ruiz, and C. Sidman. 1993. "Using GIS Tools to Analyze and Visualize Spatial Phenomena." Geo Info Systems 3(5):59-65.
von Thunen, J.H. 1826. Der Isolierte Staat in Beziehung auf Landwirtschalft und Nationalokonomie" (Hamburg); 2nd Edition reprinted 1966. (Gustav Fischer, Stuttgart); English translation: von Thunen's Isolated State. 1966, P. Hall, Ed. Translated by C.M.Wartenberg. London: Pergamon Press.
Geo Info Systems April 1995 p 20-27, 45.
