Introduction

Authors

Zack Taylor and Marcy Burchfield

Growing Cities

Cities are the most intricate of human creations. Because of their varying social, physical, economic, and political contexts, and complex internal dynamics, no two cities grow and develop in exactly the same way.

This study offers an "apples-to-apples" comparison of three metropolitan regions and an approach that allows for further comparisons with other cities. It represents a collaboration between the fields of remote sensing and geographical information systems (GIS) on the one hand, and policy analysis on the other.

Using new techniques for the analysis of satellite imagery in conjunction with census data, we measured changes in urban land use, density, housing stock, and population growth for three Canadian metropolitan areas: Toronto, Vancouver, and Calgary. The approach is replicable in any urban region where high-quality satellite data and fine-grained census data on dwelling counts are available.

We also systematically studied changes in planning policies and governing institutions over the long term. This allowed us to assess the degree to which the observed growth patterns correlate with planning policies, and to draw inferences as to how institutional structures may have helped or hindered their application.

The result is a rare comparative investigation not only of how different metropolitan regions have grown, but also of why they have grown as they did. In her review of evaluation in planning, Talen (1996) laments the lack of attention by practitioners and scholars to assessing the effectiveness or outcomes of plans. It is this question that this study seeks to address.

Conceptual debates

"Good" city form, metropolitan governance, and regional planning

This study is informed by several ongoing debates, some normative, others methodological. Comparison of different metropolitan areas' urbanization patterns and policies is especially relevant to long-running debates over what constitutes "good" city form and the related question of how metropolitan areas can be most efficiently and equitably governed.

On one side of this debate are those who advocate for comprehensive regulation of land use to promote a more compact and dense built form. The proposed social, economic, and environmental benefits of growth controls include the protection of agricultural land and significant natural features (Ewing, 1997), reducing automobile dependence and energy use (Newman & Kenworthy, 1999), lower public costs of infrastructure and servicing (Carruthers & Ulfarsson, 2003), and greater social cohesion (Talen, 1999). Those calling for more comprehensive regulation also tend to favour greater centralization of planning authority into metropolitan or higher-level governments to overcome jurisdictional fragmentation (Lightbody, 2006; Orfield, 1997; Rusk, 1993).

On the other side are those who argue that the evidence for these benefits is weak (Gordon & Richardson, 1997). Those who endorse the "public choice" school of political economy argue that lower levels of regulation enable market processes to allocate urban development efficiently. Seen this way, fragmentation of local decision-making authority is a virtue, as the consumer can choose to reside in a municipality that offers his or her preferred level of taxation and mix of land use and other policies (Bish, 1971; Peterson, 1981). Arguments that dispersed rather than compact urban form has been the historical norm, particularly in North America, are congruent with this claim (Bruegmann, 2005; Kotkin, 2005).

This debate cannot be resolved here. However, this study illustrates that metropolitan regions can be governed by quite different governing arrangements and that authorities in different places can make distinctive policy choices that result in different land use outcomes.

Defining metropolitan area boundaries

A second debate is methodological, concerned with how best to delineate urban regions for the purpose of comparative analysis. Political jurisdictions seldom correspond to "functional" metropolitan areas defined by land use patterns, commuter travel behaviour, housing and labour markets, the movement of goods, ecological systems, or other measures (Buckwalter & Rugg, 1986; Sancton, 1994). National census authorities typically define metropolitan areas by applying thresholds of human activity -- population density and commuter travel behaviour -- to local political units. For example, Statistics Canada (2006) defines a Census Metropolitan Area (CMA) as an "area consisting of one or more neighbouring municipalities situated around a major urban core. A census metropolitan area must have a total population of at least 100,000 of which 50,000 or more live in the urban core." As Sancton (2008) demonstrates, these technically derived boundaries rarely correspond to a single political or administrative unit. Even if they do, they cannot contain a growing region for long. Moreover, Sancton points out that a metropolitan area defined by governments for one purpose may not be applicable to the study of other phenomena.

As we will explain in greater detail later, we started this study by using CMA boundaries, but modified those boundaries to take into account additional factors, such as population growth rates, geographies, and measurements of urban form.

Interpreting the landscape

Another debate has to do with how best to identify urban and non-urban land uses within the physical landscape. This question has both normative and methodological dimensions. There is a long tradition in landscape ecology studies, geography, and planning thought of conceptualizing the landscape as a "field," "gradient," or "continuum" that incorporates a range of physical types and uses ranging from the urban to the rural (Friedmann & Miller, 1965; Geddes, 1925; Pickett et al., 1997; Weber, 2001; Zeng, Sui, & Li, 2005). Over the years, this approach has found expression in the writings and works of Lewis Mumford and the Regional Plan Association of America (see Fishman, 2000), the planning methods of landscape architect Ian McHarg (1969), and the influential "transect" planning model of neotraditional architect-planners Andres Duany and Elizabeth Plater-Zyberk (2009).

Another tradition divides the landscape into two categories: urban and non-urban. While cities have had defined boundaries since ancient times, the modern origin of planning policies based on this binary division is Ebenezer Howard's (1965 [1902]) vision of halting the outward extension of the core city with a "greenbelt" while directing growth to smaller satellite "garden cities." This notion of a clear distinction between urban and rural led directly to policies of urban containment through the comprehensive protection of rural land surrounding London and other cities in the second half of the 20^th century and, indirectly, to the establishment of urban growth boundaries and urban service areas in various North American cities (Nelson, Dawkins, & Sanchez, 2007:8-13; Pendall, Martin, & Fulton, 2002).

We must recognize that concepts such as "urban," "suburban," "exurban," or "rural" are value-laden, culturally embedded, and relative. As cultural geographers have shown, what one observer may comprehend as "urban" may appear different to someone else or in another context (Comber, Fisher, & Wadsworth, 2005; Duncan & Duncan, 1988; Meinig, 1979). The multiple meanings of landscapes are qualitative and cannot be reduced to quantitative datasets.

Nevertheless, these concepts have value in the North American context, and are frequently used in planning practice. Plans and planning policies often treat development within the existing urban area and new development on surrounding rural land separately. Indeed, the "inside game" and the "outside game" are understood to be played by different economic and political rules (Rusk, 1999). For this reason, we have chosen to treat these urban and non-urban realms separately.

Defining the extent of what can be considered urban depends on the source data used (Wolman et al., 2005). Political boundaries and census geographic units are not a reliable indicator of the extent of urbanized land because they often contain non-urban, non-developable areas such as greenspaces or water bodies. Census geographic units also differ from country to country, making international comparisons difficult (Davis & Schaub, 2005). To circumvent this problem, geographers and planners are increasingly making use of remote sensing -- analysis of satellite imagery of the physical landscape -- to differentiate between different categories of land use, particularly urban and non-urban uses (Mesev, 2003).

We have employed this approach here. While the reduction of the continuum to two either-or categories produces a less nuanced understanding of the total landscape, it makes it possible to map and measure urban growth patterns in a consistent way over time, within existing urban areas, or in comparisons of urban and non-urban or rural land, or across different metropolitan areas.

In this project, we refer to urban development within the existing urban area as intensification. This may take the form of infill on vacant lots, the redevelopment of previously built land parcels, or the renovation of existing buildings, with the effect of increasing density. The alternative to intensification is greenfield development -- the conversion of previously non-urban, countryside land to urban uses.

Three cases: Toronto, Calgary, and Vancouver

This study focuses on the Toronto, Calgary, and Vancouver metropolitan regions. These regions are of interest for three reasons. First, the population of each has grown rapidly over the past half-century and at similar rates in the decade in question. Second, each is governed by a different configuration of provincial, regional, and municipal institutions for land use planning and the provision of infrastructure. At the same time, all three are subject to the same national constitutional and legal frameworks governing property rights and, arguably, generally the same cultural norms (Garber & Imbroscio, 1996; Goldberg & Mercer, 1986). Finally, each urban area is surrounded by different types and levels of constraints on urbanization. This includes physical constraints, such as mountains and water bodies, as well as policy constraints, such as the comprehensive protection of lands for farming, for resource extraction, or for their ecological significance.

Physical and policy context

Many factors influence development patterns in metropolitan regions; therefore, it is important to understand the differences among the three regions in terms of size, physical geography, and the existence of policy areas that may constrain urban expansion. Figure 1.1 shows the built-up urbanized area as it existed in 1990 and 2001 for each of the three regions. Each region is mapped at the same scale to illustrate differences in the relative size and extent of each region. The maps also show significant "green" policy areas, including provincial plan areas and large municipal parks where urban development is restricted, as well as large-scale zones of specialized urban development, such as airports. First Nations reserves may also act as constraints on urban development. Only the policy areas that existed as of 2001 are shown. Appendix B contains more detailed maps that show topographic relief (shown using hill-shading), additional policy areas, First Nations Reserves, municipal boundaries, major roads, and higher-order transit routes.

The Toronto region's urban area is largely surrounded by agricultural land, much of it actively farmed (Wright, 2000). Other than Lake Ontario to the south, there are no physically impassible barriers to urban development. The Niagara Escarpment, first protected by the provincial government in the 1970s, is close to the urbanized parts of our study area only at the western edge of Lake Ontario, near the Town of Milton and the Cities of Burlington and Hamilton. Although urban uses are restricted within most of the Escarpment plan area, there is plenty of room for expansion of existing settlements on either side of it. The same is true of the Oak Ridges Moraine, which is protected by a plan enacted in 2001, at the end of the study period. The Niagara Escarpment and Oak Ridges Moraine plan areas are shown in green on the maps.

In the Vancouver region, there are a variety of physical constraints on urban growth -- the Pacific Ocean, mountains, rivers, and the border with the United States. Several large policy areas abut much of Vancouver's urbanized areas and restrict urban development -- the provincial Agricultural Land Reserve (ALR), created in 1973, and Metro Vancouver's Green Zone, established in 1996. It has been estimated that "nearly half of the region's developable lowlands" and nearly all of the metropolitan region's farmland are included in the Green Zone, which includes the ALR in Vancouver (GVRD, 1997). The ALR and Green Zone are shown in green on the maps.

Figure 1.1 Policy constraints on urban expansion

Calgary is located at the confluence of the Bow and Elbow Rivers. Its expansion is largely unconstrained on all sides. The principal immediate constraint on urban expansion is to the southwest, where the City abuts the Tsuu T'ina First Nations reserve. Two large parks -- Fish Creek Provincial Park in the south and Nose Hill Municipal Park in the north -- also limit urban development, as does the international airport in the northeast. The Kananaskis Country lands, which are protected and managed by the provincial government, lie about 40 kilometres to the west. With the exception of the Tsuu T'ina reserve and the sour gas lands to the northeast, Calgary has expanded outwards around most of its perimeter.

Defining the metropolitan region

The public discourse in any given place will contain multiple ideas of what constitutes the "region." In Toronto, for example, media often refer to the Greater Toronto Area, or GTA, although it has never corresponded to a governing body. More recently, people have begun including the City of Hamilton, about 70 km to the southwest, in their conception of Greater Toronto. In Vancouver, people and policymakers often speak of the Lower Mainland -- an area that extends east from Metro Vancouver to include the largely agricultural lowlands governed by the Fraser Valley Regional District. Enclosed by mountains, the valley ends at Hope, a 160-km drive east of downtown Vancouver. Calgary's sense of region in public discourse is more amorphous, extending into the prairie to the north, east, and south. To the west, the Foothills and the Rocky Mountains represent a physical and visual limit.

Figure 1.2 shows the study area boundaries for each region. The background satellite imagery is the data used to delineate built-up urban land, as described in the next chapter. Once again, each region is mapped at the same scale, to indicate differences in size and extent.

The study area for Toronto includes the Regional Municipalities of Halton, Peel, York, and Durham, as well as the Cities of Toronto and Hamilton, and comprises the majority of three contiguous CMAs (Toronto, Hamilton, and Oshawa). The Province of Ontario, on the other hand, considers these areas to be the "Inner Ring" of a larger urban region known as the Greater Golden Horseshoe for the purposes of growth management and transportation legislation and plans. The so-called "Outer Ring" contains freestanding towns and cities that are historically independent of Toronto and Hamilton.

The Vancouver and Calgary metropolitan regions are more easily defined, because they are geographically isolated from other major settlement areas. The extent of contiguous urban activity in Vancouver is largely equivalent to its CMA, which generally corresponds to the administrative area of a regional authority, Metro Vancouver (formerly the Greater Vancouver Regional District, or GVRD). Although the City of Calgary takes in one entire contiguous urban area, Calgary's CMA also includes several small, non-contiguous towns. In recent years these towns have expanded rapidly, but their spatial extent in 1990 was not large enough to include them in this study.

Figure 1.2 study areas for three regions, Toronto, Vancouver, and Calgary, mapped at the same scale

Table 1.1 Comparison of various metropolitan region definitions

	Population
	Area, 2001 (km²)	1991	2001 % Change
Toronto
(M) Inner Ring	8,309	4,687,421	5,572,094	+ 19%
Outer Ring	23,924	1,731,756	1,979,993	+ 14%
Greater Golden Horseshoe (GGH)	32,233	6,419,177	7,552,087	+ 18%
Inner Ring as % of GGH	26%	73%	74%	+ 1%
Vancouver
(M) Metro Vancouver	2,879	1,590,287	1,980,422	+ 25%
Fraser Valley Regional District (FVRD)	13,347	186,163	237,550	+ 28%
Lower Mainland	16,226	1,776,450	2,217,972	+ 25%
Metro as % of Lower Mainland	18%	90%	89%	- 1%
Calgary
(M) City of Calgary	697	710,677	878,866	+ 24%
Calgary CMA	5,086	754,033	951,494	+ 26%
Calgary CMA (excluding City)	4,389	43,238	72,628	+ 68%
Census Division 6	12,425	804,802	1,021,060	+ 27%
Census Division 6 (excluding City)	11,723	94,007	142,194	+ 51%
City as % of CMA	14%	94%	92%	- 2%
City as % of Census Division 6	6%	88%	86%	- 2%

(M) indicates the regional definition used in this study. All values from Statistics Canada, E-STAT.

Table 1.1 compares growth in population within these regional study areas between 1991 and 2001 to growth in surrounding areas. In each case, the areas chosen for this study accounted for the vast majority of the population of the larger region. This suggests that applying our analysis to a larger territory would add little to the analysis. In Toronto in 2001, the Inner Ring accounted for 74% of the population of the Greater Golden Horseshoe. Metro Vancouver accounted for 89% of the combined population of the Lower Mainland, which also contains the adjacent Fraser Valley Regional District (FVRD). The City of Calgary comprised 92% of its CMA and 86% of the population of the much larger Census Division 6 in 2001.

Between 1991 and 2001, the Inner Ring of the Greater Golden Horseshoe grew faster than the Outer Ring, while in Calgary and Vancouver, adjacent areas grew faster than the metropolitan core. Despite these trends, each area's proportions at the decade's beginning and end were similar. This may not be true for subsequent time periods. Expanding the boundaries of these study areas may be required for a more current comparison.

Summary of chapters

The remainder of this report is divided into four chapters. Chapter 2 describes the methods used to define the urbanized area for each region and to determine how much growth occurred and where. A more technical explanation of these methods is located in Appendix A. Technical monographs are available on the website of the Cartography Department at the University of Toronto (http://www.geog.utoronto.ca/research/publications/gcut/gcut_home).

Chapter 3 relates the findings of this analysis. First, we compare how the population, number of dwellings, amount of urban land, gross density, and mix of housing stock changed in each region between 1991 and 2001. Then we show where and how greenfield development took place in each region. Finally, we map and discuss the location of intensification, as well as the proportion of each region's total growth that occurred through intensification as opposed to greenfield development -- what we call the residential intensification rate.

What accounts for these distinct patterns? The second part of the report is our attempt to answer this question. The growth of any city is the outcome of the complex interaction of demographic, cultural, institutional, economic, and physical factors. The patterns described in Chapter 3 represent the result of countless decisions, many of them made one parcel at a time, by planners, politicians, landowners, developers, builders, banks, insurers, and home buyers. Nevertheless, it is possible to assess whether land development patterns correspond to objectives expressed in policies and plans, while acknowledging how the many other factors involved may either reinforce or undermine particular planning policies and governing institutional frameworks.

Chapter 4 provides an overview of the development of regional governance and planning institutions and policies in each region, especially policies relating to the distribution of residential growth. For each metropolitan region, the report summarizes the evolution over time of the structure of regional planning and municipal government institutions; planning policies and principles for greenfield development; planning policies and principles that promote intensification, either generally or in specific areas; and the role of comprehensive rural land protection in promoting urban containment.

Chapter 5 compares the observed patterns and the documented plans, and comments on the extent to which the regional land use patterns observed correspond to the long-term planning ideas and principles active in each region. In other words, do the patterns fit the plans? The answer is a qualified yes.

This is an exploratory study and the Neptis Foundation hopes it will stimulate discussion both on methods for measuring and mapping growth and on the extent to which planning policies account for the observed patterns. The report concludes with a brief review of directions for future research, which include the application of similar methods to other urban regions in North America to further test both the methods and the results of the analysis.

Notes

1. This and other specialized terms marked in bold sans-serif type are defined in the glossary.

2. Bourne (1996) offers a more encompassing definition of "reurbanization" that is focused on processes of urban development, whether or not they result in an increase in density. He distinguishes between several processes, each with distinctive dynamics and outcomes: repopulation, intensification, replacement, conversion and adaptive reuse, economic renewal and restructuring, and revitalization.

3. The Greater Toronto Area, or "GTA," is a term used since the 1980s to refer to the City of Toronto (formerly the Municipality of Metropolitan Toronto) and the surrounding regional municipalities: Halton, Peel, York, and Durham. The GTA is generally understood to exclude Hamilton.

4. The fact that Statistics Canada still identifies three CMAs in this urban area is an artifact of the historic criteria used in Canada to define a CMA (Puderer, 2008). Under the criteria used in the United States, the Hamilton, Toronto, and Oshawa CMAs would be considered a single metropolitan area.