Megaregions: Literature Review of the Implications for U.S. Infrastructure Investment and Transportation Planning

Section II. Foundations and Methods of Delineation

C. Defining a Region

The act of defining and delineating the boundaries of a region can take many forms, from the accumulation of community visions and perceptions, to historical precedence, to data-driven methods based on economic and urban theories. This section provides a background on the historical methods of regional delineation in the United States and reflects on preliminary delineation methods for megaregions in the United States and abroad.

1. Historic Delineation Methods

Since U.S. megaregions are commonly defined as connected networks of metropolitan areas (Delgado et al., 2006a; University of Pennsylvania, 2006), we revisit the definition of metropolitan areas as a starting point of reviewing delineation methods.

According to the 1990 definition, a Metropolitan Statistical Area (MSA) consists of a city with a population of 50,000 or Census Bureau-defined "urbanized area"¹³ with a population of at least 50,000 if the component county's or counties' population exceeds 100,000 (US Census Bureau). For the 2000 census, the MSA is more simply defined by including a population criterion (at least 50,000) of a central county or counties with a Census Bureau-defined "urbanized area" and commuting criteria for adjacent counties. In addition, the Micropolitan Statistical Area is newly defined in 2000 in a similar way that it includes a central county or counties of Census Bureau-defined "urban clusters"¹⁴ with population size between 10,000 and 49,999, and adjacent counties that have high interaction with the central county. Both metropolitan statistical areas and micropolitan statistical areas are included in the Core Based Statistical Area (CBSA).

Similarly, the European Union uses a threshold of population to delineate the statistical areas (NUTS) of Member States. The concept inherent in this method has been used in delineating boundaries of urban centers at a metropolitan level. Giuliano & Small (1991) and McMillen & McDonald (1998) use a minimum employment and employment density procedure to delineate the boundaries of urban centers in Los Angeles and Chicago metropolitan areas. They use different "cut-off" points for different metropolitan areas. For example, Giuliano and Small (1991) use 10 employees per acre as a minimum density and a total of at least 10,000 employees in a set of contiguous areas for the Los Angeles metropolitan area. However, McMillen and McDonald (1998) use 20 employees per acre and a total of 20,000 employees as minimum "cut-off" points for the Chicago metropolitan area. This implies that it would be difficult to apply a unified minimum-value procedure for delineating boundaries in different geographic areas, particularly, in a large country like the United States.

For these reasons, researchers have developed mathematical or econometric methodologies such as graph analysis (Dietvorst & Wever, 1977; Nystuen & Dacey, 1961; Campbell, 1972; Kipnis, 1985; Puebla, 1987) and Markov chain (Brown & Holmes, 1971; Dietvorst & Wever, 1977) for identifying linkages between regions using commuting flows, telephone flows, and migration flows, and multivariate analysis methods (Kim & Woo, 2000; Clayton, 1974) and density functions (McMillen, 2001; McMillen & McDonald, 1997; Muniz et al., 2003; Lee, 2007; Woo, 2007). Different kinds of multivariate analysis, including cluster analysis, factor analysis, and principal component analysis, have been used to delineate boundaries of places with a high degree of centrality and their spheres of influences using diverse criteria, such as population, employment, commuting flows, or telephone call flows. Multivariate analysis is useful in that several criteria are simultaneously considered in the model (Kim & Woo, 2000). While it is difficult to consider several criteria in one density function, this method has been broadly used in identifying areas with high concentrations of population or economic functions. Recently, Woo (2007) and Lee (2007) used mixed procedures employing those methods above, showing two-stage procedures in delineating the boundaries of centers.

Figure 19. Gottmann's Megalopolis (Gottmann 1957)

While these methods are useful to identify central regions and their relationships and to delineate the boundaries of their exurban areas, they have not been applied to megaregion analysis in the United States. These delineation methods have been used specifically in a context of polycentric spatial patterns at the metropolitan level. As European spatial planning has emphasized developing polycentricity at the interregional scales (Davoudi, 2003; Green, 2005) and each megaregion consists of hierarchical centers, these methods would provide useful information in developing delineation methods for megaregions in the United States.

2. Preliminary Delineations of Megaregions

In 1950, the Bureau of the Census presented a map, which showed the economic continuation of metropolitan areas from Hillsborough County, NH (the north of Boston) to Fairfax County, VA (the south of Washington, D.C.), on a pamphlet of statistics on ‘States Economic Areas' (Figure 19). Jean Gottmann named this continuous metropolitan economy ‘Megalopolis', which he described as an economic continuation of metropolitan areas. It was reflection on Gottmann's analysis of megalopolis beginning in 2004 by University of Pennsylvania and the Lincoln Institute of Land Policy that began a national debate on a new geography for aspects of United States planning.

In recognition of the markedly varied and largely nonurban character of the United States, researchers have agreed upon new terminology, adopting "megaregion" as the descriptor of regions with a multiplicity of relationships composed of numerous political bodies, from local to state, and even national scales. A research team from the University of Pennsylvania defines megaregions as "large, connected networks of metropolitan regions that are driving an increasing share of global production and trade." The Great Lakes Megaregion plan similarly defines megaregions as "a set of interconnected metropolitan areas". While U.S. megaregions have been delineated with bounded areas, the megaregional concept in Asia and Europe has been used to address regional or transnational networks, often using high-speed rail and separated goods movement systems (Metcalf & Terplan, 2007). All of these efforts commonly argue that a large scale of analysis and planning is necessary to address economic and environmental problems that frequently take place beyond traditional political boundaries.

a) Nationwide Delineation Approaches in the United States
The definition and delineation of megaregions in the United States have been explored by several researchers using diverse criteria (Metcalf & Terplan, 2007; Ross et al., 2007; Zhang et al., 2007; Delgado et al., 2006a; Regional Plan Association, 2006; University of Pennsylvania, 2006; Contant et al., 2005; Lang & Dhavale, 2005; Seltzer et al., 2005). Among the attempts to delineate large scale regions, two have received the greatest attention - Virginia Tech's Metropolitan Institute's Megapolitans and the Regional Plan Association (RPA)'s Megaregions. Although the titles are different, they are both intended to recognize large-scale economic centers.

Figure 20 (left). The RPA's megaregions (RPA, 2006)

Figure 21 (right). Metropolitan Institute's Megapolitan Areas (Lang and Dhavale, 2005)

The RPA (2006) has identified five criteria that define 10 emerging or existing megaregions (Figure 20). The criteria include "environmental systems and topography, infrastructure systems, economic linkages, settlement patterns and land use, and shared culture and history" (Regional Plan Association, 2006). The RPA assumes that an area that shares many of these criteria will be a cohesive megaregion.

Similarly, Virginia Tech's Metropolitan Institute identifies 10 "megapolitans" using a different set of criteria and procedures (Figure 21). They define megapolitans as clusters of counties that combine at least two metropolitan areas and have a total of more than 10 million residents by 2040. Based on Lang and Dhavale (2005), the following process and criteria are used to identify megapolitans:

• Map the counties in U.S. Census defined micropolitan and metropolitan statistical areas. Any string of contiguous metropolitan and micropolitan areas is a candidate for a megapolitan.
• Non-metropolitan counties that are completely encircled by metropolitan or micropolitan counties are included in the megapolitan.
• The type of connectedness is examined for all candidate counties. For example, the "CSA-to-CSA" relationship has the highest priority, followed by "metro-to-CSA", "metro-to-metro", "metro-to-micro", and "micro-to-micro" link.
• The candidate megapolitans are tested by population maps to check if they are consistent with urban development patterns.
• Boundaries are smoothed by other criteria, such as transportation networks, development trends, topographic elements, etc.
• Finally, those clusters of counties with at least two metropolitan areas and a total of 10 million of projected residents by 2040 are defined as megapolitans.

Table 7 documents the differences between the RPA and Metropolitan Institute approaches to define large-scale regions in the United States.

Table 7. Comparison of delineating criteria of megaregions in the United States

	Metropolitan Institute	Regional Plan Association
Analysis Unit	County	County
Requirements of megaregions	More than 2 metropolitan areas & 10 million population by 2040	N/A
Analysis Criteria	- Population size - Contiguity - Cultural and historical geography - Physical environment - Links of large centers - Growth projections - Goods and service flows	- Environmental systems and topography - Infrastructure system - Economic linkage - Settlement patterns and land use - Shared culture and history

As the RPA megaregion and Metropolitan Institute megapolitans maps show, these different approaches produce similar, yet not identical, definitions of the regions. Both the Metropolitan Institute and RPA identify 10 megaregions across the country. The Metropolitan Institute includes Oklahoma City and Tulsa (OK) as megapolitans, focusing on the Interstate 35 corridor and including Houston in the Gulf Coast Megapolitan. However, RPA excludes those Oklahoma metropolitan areas from megaregions, identifying a triangular form of metropolitan areas within Texas. The Metropolitan Institute identifies a broader area as the Northeast Megapolitan, including Richmond, while RPA's region stretches south by Washington D.C. metropolitan area. However, the Metropolitan Institute delineates a smaller area for the Midwest Megapolitan while RPA identifies a broader area including Minneapolis (MN) and Buffalo (NY). In addition, RPA excludes Knoxville (TN) from the Piedmont Atlantic Megaregion and the Metropolitan Institute excludes Jacksonville (FL) from the Florida Megapolitan.

Table 8 compares the results of the megapolitan versus megaregion approaches by describing which metropolitan areas are contained in each.

Table 8. Comparison of identified megaregions/megapolitans across the country

Metropolitan Institute (10 Megapolitan Areas)	Regional Plan Association (10 Megaregions)
1.Northeast (Including Richmond (VA))	1.Northeast (Excluding Richmond and Virginia Beach (VA) of Chesapeake)
2.Midwest (Including Chicago (IL), Detroit (MI), Indianapolis (IN), Cincinnati (OH), Columbus (OH), Pittsburgh (PA), Cleveland (OH))	2.Great Lakes (Including Minneapolis (MN), Chicago (IL), St. Louis (MO), Indianapolis (IN), Louisville (KY), Cincinnati (OH), Columbus (OH), Cleveland (OH), Detroit (MI), Pittsburgh (PA), Buffalo (NY))
3.Piedmont (Including Knoxville (TN))	3.Piedmont Atlantic (Excluding Knoxville (TN))
4.Peninsula (Excluding Jacksonville, FL)	4.Florida (Including Jacksonville, FL)
5.Gulf Coast (Including coast areas of LA, MS, AL, TX, and FL)	5.Gulf Coast (Including coast areas of LA, MS, AL, TX, and FL)
6. I-35 Corridor (Including Tulsa (OK), Oklahoma City (OK), Dallas-Fort Worth (TX),  San Antonio (TX), Austin (TX))	6.Texas Triangle (Including Dallas-Fort Worth, Houston, San Antonio, Austin)
7. Valley of the Sun	7.Arizona Sun Corridor
8.Cascadia	8.Cascadia
9.NorCal	9.Northern California
10.Southland	10.Southern California

b) Regional Delineation Approaches in the United States
While somewhat different criteria are used in delineating megaregions based on their own regional characteristics, most institutions use counties and MSAs as the analysis units, and the data of those criteria are incorporated by a mapping method with Geographic Information System (GIS) and a descriptive analysis. Table 9 presents the different types of analysis used in identifying megaregions in different parts of the United States.

In the Piedmont Atlantic Megaregion (PAM), researchers identify urban cores and their surrounding regions, considering watersheds, geography, ecologically sensitive areas, commuting patterns, and freight movement. The urban cores are verified by the analysis of settlement patterns and transportation flows. The boundaries are smoothed by some factors including commodity flows (western boundaries), cultural factors (northeastern and southern boundaries), and natural features (eastern and northwestern boundaries) along the county lines (Contant et al., 2005).

The researchers examining the Northeast Megaregion use two components to define the region: urban core and support zones (Dewar & Epstein, 2007). The urban core is defined as contiguous MSAs and adjacent counties that have more than 15 percent of Employment Interchange Measure (EIM). The support zone, an outer boundary to the urban core, includes adjoining counties that have large preserved open spaces, water and recreational resources, environmentally sensitive areas, and river basins that are related to the developments from the urban core (University of Pennsylvania, 2005). In addition, Vicino et al. (2007) have recently delineated the Northeast Megaregion using only two criteria: population density and contiguity of metropolitan areas.

The Northern California study identifies a core area and its sphere of influence by overlapping four feature maps, including travel times, population growth and land consumption, environmental features, and "government-defined" regional groupings, such as multi-county regional councils. The urban core of the Northern California region is defined by commuting sheds of greater Sacramento (Metcalf & Terplan, 2007).

The Great Lakes Megaregion selects the largest metropolitan areas, which have common industrial histories and large population size within states bordering Great Lakes, as megaregions. Although some researchers include Minneapolis and St. Paul (MN) in the Midwest megaregion, they were excluded in the Great Lakes Megaregion because of the dissimilar industrial histories and challenges. Adjacent counties that have greater than 8 percent of population growth rate are added to the region to consider issues of land consumption and environment degradation in the future. Also, other counties adjacent to one of the Great Lakes and those surrounded on more than two sides by megaregion counties are included (Delgado et al., 2006a).

Zhang et al. (2007) have verified the Texas Triangle Megaregion of the Regional Plan Association mostly based on maps of transportation networks, industrial flows, and eco-zones of four metropolitan areas, including Dallas-Fort Worth, Houston, San Antonio, and Austin.

However, while there have been efforts to understand the challenges and opportunities for other megaregions, including Southern California, Southern Florida, the Gulf Coast, Cascadia, and the Arizona Sun Corridor, specific methodologies to identify their boundaries have not been detailed at the regional level.

Table 9. Comparison of delineating methods of selected megaregions in the United States

	Piedmont Atlantic	Northeast	Northern California	Great Lakes	Texas Triangle
Institution	Georgia Institute of Technology	University of Pennsylvania	San Francisco Planning & Urban Research	University of Michigan	University of Texas, Austin
Source	Contant, Ross et al. (2005)	University of Pennsylvania (2005)	Metcalf & Terplan (2007)	Delgado et al. (2006a)	Zhang et al. (2007)
Analysis Unit	County, MSA	County, MSA	County	County, MSA	County, MSA
Structure	Urban core & surrounding regions	Urban core & support zone	Core area & sphere of influence	Major MSAs & adjoining areas	Major MSAs & counties within the triangle area
Analysis Criteria	- Population growth - Settlement patterns - Watersheds - Geography - Ecologically sensitive areas - Linkage of highways - Commuting patterns - Freight movement*Urban cores are verified by the analysis of settlement patterns and transportation flow.	Urban core: - Contiguous MSAs - EIM (Employment Interchange Measure)Support zone: - preserved open space - water & recreational resources, - environmentally sensitive area - river basin	- Travel times - Population growth - Land consumption - Environmental features - Regional groupings (e.g. multi county councils)* Core area is defined by commute sheds of greater Sacramento	Major MSAs: - Industrial history - Population sizeSupport areas: - Population growth (8%) - Adjacency to the Great Lakes	- Transportation Networks - Industrial flows - eco-zones
Smoothing Boundaries	Consider commodity flows, cultural factors, and natural features			Include counties surrounded on two or more sides by Megaregion counties
Incorporating Data	Descriptive analysis & Mapping

Several studies (including those of the Piedmont Atlantic, Northeast, Northern California, and Great Lakes megaregions) divide the megaregion into two parts: core areas (or major MSAs) and areas of influence (areas that play a role in the economic activity of the core and are affected, both positively and negatively, by the core). There exist outer areas of metropolitan areas on the maps of the Texas Triangle studies. However, authors did not explicitly characterize these areas. On the other hand, Lang and Dhavale (2005) assert that "Megapolitan Areas should have discrete boundaries, as do metropolitan and micropolitan areas." However, the identification of areas of influence is important as Sassen (2007) mentions that "rural enterprise zones" require megaregional investment to provide low-wage jobs as one strategy to play a more complete role in the global economy.

The criteria used in the previous research can be categorized by "essentialist" criteria and "relational" criteria, as Healey (2004) presents. As seen in Table 10, essentialist geography is a traditional approach where the relationship between places is hierarchically determined by their activities, and proximity and contiguity are important factors that characterize their relationship. On the other hand, the relational approach emphasizes that spatial effects cannot be determined by physical proximities because the effects may occur from a distance as well as nearby. In addition, a place is not necessarily bounded by specific territories, and the nodes and borders are continually changing according to their relationship with others.

Specifically, in the megaregion studies, several criteria, including proximity, contiguity, population growth, settlement patterns, land consumption, and political boundaries, expressed as essentialist criteria, were overlaid to identify candidate megaregions. In addition, relational criteria, such as commuting patterns, industrial flows (goods and service flows), historical characteristics, and international passenger traffic and labor migration, have been analyzed to examine interactions between regions. In a spatial context, the relational approach is important to understanding broad economic issues in a national and global economy. However, it should be noted that the traditional (essentialist) approach is useful in policy implementation. For example, specific boundaries of regions are needed for effective physical infrastructure planning and implementation. In addition, such boundaries and distance criteria are useful to designate and protect environmentally sensitive areas. Thus, a strategic approach that may incorporate the advantages of both essentialist and relational approaches should be explored.

Table 10. Essentialist and relational geography in spatial planning (reorganized from Healey, 2004)

	Essentialist Geography	Relational Geography
Spatial Effect	Physical proximity is an important factor. Connections are governed by transport routes with simple distance-decay characteristics.	Spatial effects cannot be analyzed only by physical proximities. Instead, they may occur from a distance as well as nearby.
Place	The boundaries of distinct regions or settlements are identified.	Nodes and borders are continually emergent, not derived by specific model of "socio-spatial organization".
Relationship between places	Places are divided into several zones based on their activities (e.g. core & periphery).	Place is seen as a social construct, where "meanings" are given to particular places, areas, or nodes. Places are located in particular relational networks, and the distance between places and nodal points in relevant networks is important.
Development	Development is treated in a linear way from "less-developed" to "more-developed" areas.	Different places have different potentials because of the interactions between their histories and relationships with others.

c) Delineation Approaches from Abroad
The methods of delineating megaregions in foreign countries are not directly applicable to the proposed American megaregions because their political systems and administrative units differ from those in the United States. However, useful information can be drawn from the decision criteria used in delineating or identifying megaregions in those countries (see Table 11 for a comparison for methodologies). Since European and East Asian countries (except China) are relatively small, some megaregions appear to be transnational. As a result, these megaregions are more focused on the connectivity between major cities and countries via infrastructure, such as high-speed railway system and under-sea canal, rather than the delineation of boundaries of hinterlands.

Table 11. The criteria of delineating megaregions in foreign countries

	Randstad (Netherlands)	NUTS (European Union)	Zhujiang Delta (China)	North-East Asian Megaregions
Analysis Unit	City	City	-	City
Analysis Criteria	Core: - Employment density - Population sizeHinterlands: - Commuting flows	NUTS: - Population sizeNon-administrative units: - Continuity - Geography - Socio-economics - History - Culture - Environment	Physical Geography: - Flood zones - Coastal lines - River basins	International interaction: - Air passenger traffic - Flow of information - International labor migration

The "Functional Urban Regions" (FURs) of the Randstad, in the Netherlands, include a core and a hinterland. A core area consists of municipalities that have a total of at least 20,000 jobs and a job density of 7 jobs per hectare (17.29 jobs per acre). Commuting data are used to identify hinterland areas. Similar to the concept of Employment Interchange Measure (EIM) of the U.S. Census, municipalities that send more than 10 percent of commuters to the core areas are assigned to the hinterlands. As seen from Figure 22, 25 functional urban regions are identified in the Randstad area (Randstad, 2005).

The European Union (EU) divides the Member States into three regions. The Nomenclature of Territorial Units for Statistics (NUTS) is used to designate administrative divisions of the Member States for statistical purposes. The NUTS is a hierarchical classification of areas because it subdivides the Member States into NUTS 1, NUTS 2, and NUTS 3. The NUTS 2 and NUTS 3 regions are subdivisions of the NUTS 1 and the NUTS 2, respectively (Figure 23). The EU defines territorial units using normative criteria: population size and geographical areas (existing administrative units). Table 12 presents the thresholds of the average population size to divide the regions into NUTS 1, NUTS 2, and NUTS 3.

However, if there are no administrative units in these population scales in a Member State, a new geographical units ("non-administrative units"), which constitutes existing small administrative units, can be aggregated considering such relevant characteristics as contiguity, geography, socio-economy, history, culture, or environment (The European Parliament and the Council of the European Union, 2003).

Table 12. The minimum and maximum thresholds for the average population size of the NUTS regions.

	Minimum	Maximum
NUTS 1	3,000,000	7,000,000
NUTS 2	800,000	3,000,000
NUTS 3	150,000	800,000

Source: Article 2 of the Regulation (EC) No 1059/2003 of the European Parliament and of the Council of 26 May 2003 on the establishment of a common classification of territorial units for statistics (NUTS)

In the study of South-East Asia, Macleod & McGee (1996) divide urban regions into three regions: urban core, metropolitan areas, and extended metropolitan regions (EMRs), which are later called mega-urban regions (MURs). The EMRs are similar to hinterlands of the U.S. megaregions. While authors do not specify the boundaries of the region, they observe that urban cores are filled with globally linked finance and producer service functions (Douglass, 2000). Chu (1996) describes one of Chinese megaregion, the Hong Kong-Zhujiang Delta, in a framework of the world city system. The region is defined by physical characteristics, such as flood zones, coastal lines, and river basins, and includes 7 municipalities and 21 xians with a total area of 42,600 km2 (16,448 mi2).

After the Cold War era, policy makers in China, Japan, and Korea have increasingly emphasized the three countries' mutual political and economic relationships. Choe (1996) presents two emerging North-East Asian megaregions: the Yellow Sea Rim and the East Sea Rim regions (Figure 24). Without the issue of specific boundaries of the megaregions, he identifies transnational interactions between major cities using several criteria, such as air passenger traffic, the flow of information (the volume of international telephone calls), and international labor migration.

---

13 The U.S. Census Bureau defines an urbanized area as "densely settled territory that contains 50,000 or more people" (U.S. Census Bureau 2000).

14 An urban cluster consists of densely settled territory with population between 2,500 and 50,000 (U.S. Census Bureau 2000).

To provide Feedback, Suggestions or Comments for this page contact Supin Yoder (supin.yoder@dot.gov).