In the later part of the nineteenth century urban growth made which of the following a necessity

4.1 Effects of Density, Grain, and Connectivity

Land use patterns can be parsimoniously characterized in four dimensions: degree of centralization or decentralization (urban form), ratio of population or jobs to area (density), diversity of functional land uses such as residential and industrial (grain), and extent of interrelation and availability of multiple modes of circulation for people and goods among local destinations (connectivity). The use of resources per capita diminishes as urban form becomes more centralized, density goes up, grain becomes finer, and connectivity shrinks. Metropolitan land use patterns in the United States after World War II show increased energy use due to increasing regional populations, decentralization, decreasing density, rougher grain, and increased connectivity.

Throughout the 19th century, most people in the United States lived in small towns and villages. With the advent of the 20th century, many people moved to industrial cities for jobs, a trend that peaked in the 1920s and was compounded by overseas immigration. Following the interruptions of the Great Depression and World War II, the pent-up demand for housing was met by a conscious process of suburbanization. Achieving the dream of home ownership became feasible for many Americans as new federal mortgage guarantee policies removed financial barriers to home ownership, as developers such as William Levitt perfected mass production of affordable housing units on greenfield sites, and as federal dollars poured into road building. Discrimination in lending and housing markets prevented most black Americans from participating in this exodus. Yet, by 1960, the suburban lifestyle was the conventional land use practice in the United States.

Suburban lifestyle has brought about sprawling, low-density suburban communities; according to the 1990 census, from 1970 to 1990, the density of urban population in the United States decreased by 23%. From 1970 to 1990, more than 30,000 square miles (19 million acres) of once-rural lands in the United States became urban areas, an area equal to one-third of Oregon's total land area. From 1969 to 1989, the population of the United States increased by 22.5%, and the number of miles traveled by that population (“vehicle miles traveled”) increased by 98.4%.

Anthony Downs defines the term “sprawl” as (1) unlimited outward extension, (2) low-density residential and commercial settlements, (3) leapfrog development, (4) fragmentation of powers over land use among many small localities, (5) dominance of transportation by private automotive vehicles, (6) no centralized planning or control of land use, (7) widespread strip commercial development, (8) great fiscal disparities among localities, (9) segregation of types of land use in different zones, and (10) reliance mainly on the trickle-down, or filtering, process to promote housing to low-income households. The impacts of urban sprawl have caused increasing traffic congestion and commute times, air pollution, inefficient energy consumption and greater reliance on foreign oil, inability to provide adequate urban infrastructures, loss of open space and habitat, inequitable distribution of economic resources, and the loss of a sense of community.

When land use patterns in the United States are compared to other countries, they show significant differences. In comparisons of metropolitan density, U.S. cities are of low density in residential and business areas whereas European cities are three to four times denser. Asian cities are 12 times denser compared to American cities (Table I).

Table I. Intensity of Land Use in Global Cities, 1990a

The average energy use for urban transportation in American cities is 64.3 gigajoules (GJ) of fuel per capita compared to 39.5 GJ in Australia, 39.2 GJ in Canada, 25.7 GJ in Europe, and 12.9 GJ in Asia. Energy use in American cities is five times more than in Asian cities. In addition, European cities consume two times more energy compared to Asian cities. This shows that energy use in transportation is closely related to land use patterns and income levels (Table II).

Table II. Transportation Energy Use per Capita in Global Regions, 1990a

Monocentricity and Polycentricity of Cities

The most classical measures are monocentricity and polycentricity of cities or urban areas. As cities expand with prevalent usage of automobiles, there has been a higher level of urban decentralization, which is referred to the movement of population and businesses away from the established city center toward the urban fringe.

To measure decentralization, population and employment density gradients can be used. One example of a reliable estimation procedure of density gradient is the two-point estimation procedure. The estimates of density gradients can be made using available data on two areas: central city and total urbanized area. More specifically, the mathematical notation of the assumption that population density in an urban area declines as a negative exponential function of distance from the city center is D(x) = be−ax, where D(x) represents population density at distance x from the city center, b is the intercept (or the central density), and a is the density gradient of the function. A lower value in the density gradient indicates that an urban population or employment is more dispersed or spread out over the urban land area. For example, a value of 1.5 for the density gradient indicates a highly concentrated urban population or employment, whereas a value of 0.01 for the density gradient indicates a dispersed urban population or employment with equal density over the urban area.

Urban decentralization depicts two waves of urban expansion in the United States: decentralization of both population and employment. In the first wave of decentralization, which started in the late nineteenth century, the population moved to the suburbs. As late as in 1960, however, most jobs (63%) were still in the central city while 51% of the population lived in suburbs. In the subsequent second wave of decentralization, jobs were also decentralized. By the year 2000, people both lived and worked in the suburbs in the United States. Across regions, the share of employment within three miles from the city center is rarely more than 19%. Consequently, employment subcenters emerged, establishing polycentricity of urban areas. Many studies have defined employment subcenters using a variety of methods, such as density maps, employment and population size, and minimum-density thresholds (Giuliano and Small, 1991). For examples, Giuliano and Small (1991) defined 32 employment subcenters in the five-county Los Angeles region by identifying a contiguous set of zones, each with 10 employees per acre, and with a combined minimum of 10 000 employees. Cho et al. (2008) identified subcenters for Mecklenburg County, part of the Charlotte metropolitan area in North Carolina. Because their study area is a much smaller metropolitan area than Los Angles (LA) in terms of population (its population is 6.58% of LA's), size (1.55% of LA's), and employment (10.35% of LA's), they lowered the minimum thresholds to four employees per acre for the employment density – that is, the mean employment density in the study area. As a result, a total of 10 subcenters, including the Central Business District (CBD), were identified in Mecklenburg County.

Similar decentralization patterns have been found in developing countries in recent decades. For example, Song (2013) investigates how road infrastructure in Chinese cities has shaped the degree of decentralization of Chinese urban regions in the last 20 years. There has been strong evidence that population density has been decreasing in the central city areas in Chinese cities. The improved mobility because of investments in road infrastructure, along with other factors such as increased urban income, has led Chinese cities to decentralize and evolve from monocentric to polycentric cities. It is evident that identification of population and employment subcenters can help understand the urban spatial structure of cities.

Applications of Models in Human Geography

Broadly, human geography models are used for (1) statistical investigation, (2) optimization, or (3) simulation and forecasting.

Statistical investigation is perhaps the most common, often used method for hypothesis testing. For example, human geographers have used specialized regression models to investigate issues such as urban decentralization, regional income inequality, historical voting patterns, and socioeconomic aspects of deforestation. What distinguishes these types of investigations from those in other social sciences is that the spatial dimensions assume a central focus.

Two broad classes of statistical modeling in geography are recognized. The first is exploratory, where the focus is on the identification of spatial pattern, testing for spatial clustering, and departures from random sampling. Models and methods for exploratory spatial data analysis have attracted a great deal of attention (Getis, 1999).

The second approach to statistical modeling is known as confirmatory spatial data analysis. Here the field of spatial econometrics has played a central role in developing new methods to deal with the treatment of spatial autocorrelation and spatial structure in regression models (Anselin, 2000). Spatial autocorrelation presents a fundamental problem to the application of traditional (i.e., a-spatial) regression methods, given that their application for inferential purposes is based on an assumption of random sampling, which is rendered invalid in the presence of spatial autocorrelation.

Optimization models are used for normative purposes to identify the optimal solution for a given problem. All optimization models have a quantity that is to be maximized (or minimized) as reflected in an objective function, for example, in the case of the well-known transportation model:

Minimize L=∑i=1n∑j=1nTi,jQi,j

subject to:

Dj≤∑inQi,j∀j

Si≥∑jnQi ,j∀i

Ti,jQi,j≥0∀i, j

where the objective is to minimize total transportation costs. The costs are a function of the unit transport cost, Ti,j and the number of units shipped between location i and j, Qi,j. The solution has to respect a number of constraints such that, the amount shipped to each region satisfies demand in that region Dj, the amount shipped from a region is no greater than the supply in the region Si, and all costs are nonnegative.

Several types of optimization models have been developed within human geography. Location-allocation models (Gosh and Rushton, 1987) typically involve a set of consumers distributed over a given area and a second set of facilities (stores, public services) to serve them. These models are used to allocate the facilities to best serve the customers. Optimization methods developed by geographers have also been used to delineate regions for specific objectives, such as school and political zone redistricting.

Simulation and forecasting models are used to study and compare the responses of a system to policy or environmental changes. Examples of these applications can be found in the areas of integrated environmental human analysis (Knight, 2000) and the analysis of trade policy impacts on regional economies (Warf and Randall, 1994).

The Automobile City

Spiralling levels of car ownership in the developed world have prompted a social ‘ungluing process’ where “use of automobile became not so much a choice but a necessity” according to two prominent researchers in the field of automobile dependency, Peter Newman and Jeff Kenworthy, who coined the term ‘Auto City’ (1999: 32) with North America’s modern, low-density cities in mind. An obvious problem inherent in ‘automobile necessity’ is that large sections of a given population will invariably be unable to drive – being too young or old, ill, poor, disqualified, or otherwise. Thus, where the authors state that “the Auto City began to lose much of its traditional community support processes” (Newman and Kenworthy, 1999: 32) it is not a sentimental lament, but an assertion of concern over the mobility of the car-less in an overwhelmingly car-dominated society. With the loss of the services and amenities associated with the ‘traditional’ neighbourhood – the corner shop, school, church, and so forth – has come the loss of the physical ‘pillars’ that helped to bind a community together, and the removal of essential local services and employment opportunities. As noted earlier, for many the freedom of personal travel brought by the private automobile has turned the decision on where to locate the home in relation to the place of work into one based on time rather than physical proximity and accessibility by public transport. The same is true for shops and services: with 10 min spent driving to the grocery store, rather than 10 min spent walking to the neighbourhood shop.

The dislocation of shops and services away from housing has been accompanied by the shift of employment away from the neighbourhood and the city centre areas to ‘edge cities’, which perpetuate the need to travel for the most basic life needs. This continuing process of urban decentralisation and segregated land use has been as evident in the historic cities of Europe as it has been in North America. In the United Kingdom, “three-quarters of those who work outside of London do so by private vehicles” (Giuliano and Gillespie, 2002: 30). Without intervention, this pattern becomes self-reinforcing “as employment continues to decentralise, the shift away from public transit will continue” (Giuliano and Gillespie, 2002: 30), leading to further social polarisation because:

The urban poor, primarily members of racial and ethnic minority groups who have relatively low levels of technical skill and own automobiles at lower rates than richer and whiter components of the population… have decreasing access to employment opportunities, which increasingly occur at low densities on the urban fringe. (Wachs, 2002: 23)

If the car has altered location and land use patterns, it has also had a profound impact on the layout of residential areas. Traffic engineering has sought to increase the efficiency and safety of residential roads, resulting in the gentle curves, road widths, parking ratios and culs-de-sac of the modern housing estate. Whilst design codes have greatly assisted the movement of automobiles through residential areas, they have also had some less desirable effects in converting significant proportions of residential space into roadways and parking areas, increasing travel distances and discouraging other travel modes. The modern cul-de-sac serves the same purpose today as it did in ancient Rome: to improve the local environment by stopping through-traffic. The resulting ‘dendritic’ structure of the modern housing estate (Figure 2) leads to circuitous routes having to be taken to travel from one point to another and a layout that severely restricts the ability of buses to deviate from main roads in order to collect passengers (Marshall, 2005).

Car-centric residential layouts can also disadvantage pedestrians by increasing walking distances on the pavements that accompany the roads in a modern housing estate. Poor natural surveillance resulting from housing that is set back, has no road frontage, or is located along paths that are badly lit at night can also deter pedestrian activity, particularly along estate access roads. These issues apply equally to segregated foot and cycle paths which aim to improve nonmotorised connectivity (Figure 3). Le Corbusier – the giant of Modernist planning – took the concept to a different scale with the entire segregation of pedestrians and vehicles into separate route networks. This theoretically ‘neat’ design solution to potential conflicts between cars and people developed in early twentieth-century North America and integrated into Clarence Stein’s new town at Radburn and later British new towns of the 1970s. In reality, however, the concept has sometimes fallen short in its application: for example in Milton Keynes, due to poor natural surveillance, and because of the gradients entailed in the bridges and underpasses. Like the road layouts themselves, such paths do not necessarily take pedestrians exactly to where they need to go or by the most direct means.

An opposite approach to segregation is to mix traffic and pedestrians together in a ‘shared spaces’ approach – a concept developed by Dutch engineer, Hans Monderman. Shared spaces shift responsibility away from highway engineering with its clutter of barriers, warning signs, and traffic calming, and invest it in the user by creating single surface streets which operate on the trust generated in eye contact between driver and pedestrian. Now a popular approach to mitigating the effects of traffic in residential areas, across the Netherlands, the concept has rapidly spread – both in its direct application and in variants, including the ‘decluttering’ of streets with the removal of signage and barriers, and in ‘home zones’ which attempt to calm traffic without use of intrusive humps and ramps. This ‘back to basics’ approach constitutes part of the wider rediscovery of historic planning principles, epitomised by ‘new’ car-free developments that form the focus of the next section.

5 Applications of Models in Human Geography

Model applications can be placed into one of three classes: (a) statistical investigation; (b) optimization; and (c) simulation and forecasting. Statistical investigation is perhaps the most common use of models in human geography. Often this takes the form of using a statistical model to test a hypothesis. For example, human geographers have used specialized regression models to investigate such issues as urban decentralization, regional income inequality, historical voting patterns, and socioeconomic aspects of deforestation. The spatial dimensions are central in these investigations.

Statistical modeling in geography can be either exploratory, where the focus is on the identification of spatial pattern, testing for spatial clustering and departures from random sampling, or confirmatory, where the focus is on spatial autocorrelation and spatial structure in regression models (Anselin 2000). Spatial autocorrelation presents a fundamental problem to the application of traditional (i.e., a-spatial) regression methods, given that their application for inferential purposes is based on an assumption of random sampling, which is rendered invalid in the presence of spatial auto-correlation.

Optimization models identify the optimal solution for a given problem. All optimization models have a quantity that is to maximized (or minimized) as reflected in an objective function, for example in the case of the well-known transportation model:

(3)MinimizeL=∑i=1n ∑j=1nTi,jQi,j

subject to:

(4)Di⩽∑jnQi,j∀i

(5)Si⩽∑jnQi,j∀ i

(6)Ti,jQi,j⩾0∀i,j

where the objective is to minimize total transportation costs. The costs are a function of the unit transport cost Ti, j and Qi, j, the number of units shipped between location i and j. The solution has to respect a number of constraints such that the amount shipped to each region satisfies demand in that region Dj, the amount shipped from a region is no greater than supply in the region Sj and all costs are non-negative (Eqn. (4)).

Several different types of optimization models have been developed within human geography. Location-allocation models (Ghosh and Rushton 1987) typically involve a set of consumers distributed over a given area and a second set of facilities (stores, public services) to serve them. These models are used to allocate the facilities to best serve customers. Optimization methods developed by geographers have also been used to delineate regions for specific objectives, such as school and political zone redistricting.

Simulation and forecasting models are used to study and compare the responses of a system to policy or environmental changes. Examples can be found in the areas of integrated environmental-human analysis (Knight 2000) and the analysis of trade policy impacts on regional economies.

Regional Planning as Resource Development: 1920–50

Regionalism emerged as a significant political force in Europe and North America in the second half of the nineteenth century, primarily in conjunction with the consolidation of national markets and the spread of nationalist ideologies aimed at sustaining rapid industrialization. In the US, regional divisions between North and South exploded in a bloody Civil War, while less violent regional uprisings multiplied throughout Europe as new nation-states arose in Germany and Italy and older states such as France, Great Britain, and Spain attempted to integrate diverse regional cultures into unified national territories. Many of the cultural regionalisms that were politically asserted during this period, including those of Catalonia, Scotland, the Basque country, and the US South, have continued to the present as important political, economic, and cultural forces.

Regional planning as a distinct form of public intervention, however, first consolidated in the US in the 1920s during a period known as the Progressive Era, when many new experiments in urban government, public administration, and planning were initiated. This first phase drew upon European traditions of utopian socialism, anarchism, regional geography and sociology, and the regional political movements that arose in response to the homogenization of national markets and the deepening urban poverty that characterized late nineteenth-century Europe. Particularly influential were such anarchist thinkers as Pierre-Joseph Proudhon and Peter Kropotkin, and two leading planning innovators, the polymath Scotsman Patrick Geddes and Ebenezer Howard, founder of the garden city movement.

This first regional planning doctrine or paradigm viewed the large industrial capitalist city, with its teeming densities and unhealthy slum housing, as the primary cause of major social and environmental problems. It emphasized urban decentralization, typically into quasi-socialist new towns and garden cities that would combine the advantages of both the city and the countryside, while hopefully ameliorating the problems of each through some form of common or public ownership. Although the European critique of urban-industrial capitalism and the particular form it was taking in the industrial capitalist city significantly influenced this first phase of regional planning, its more radical European regionalism was watered down in its Americanization. What remained, however, were a strong anti-urban bias, an idealized vision of the countryside and the environment, and a specifically spatial strategy of reform.

The appropriate planning region tended to be defined in almost organic or ecological terms, building on the social cohesion that was presumed to be coincident with the physical environment. The work of regional physical geographers was of particular importance here. The most innovative and exemplary expression of this first round of regional planning was the TVA, a semiautonomous multistate unit of government responsible for resource development and environmental preservation in the large Tennessee River drainage basin.

The TVA model of regional resource development, which would later be replicated in India, Mexico, the Soviet Union, and many other countries, linked together the two main branches of early regionalist thought in the US, one centered in New York in the Regional Planning Association of America, with Lewis Mumford its best known figure, and the other representing the southern states of the US, giving voice to problems of regional underdevelopment and, to a lesser extent, racial discrimination. Metropolitan regional planning was also emerging at the same time in the US, but mostly as an adjunct to urban planning and often in antagonism with the resource-based and larger-scale ecological approach of the northern and southern regionalists.

This first phase, especially through the TVA, would have a significant influence on US national policy in the 1920s and early 1930s, but would lose nearly all its influence as the Great Depression deepened and national priorities shifted in response to the threat and reality of World War II. Once a grand utopian experiment, the TVA became little more than a munitions factory and a generator of cheap electricity for northern industries. By the time World War II ended, regional planning had almost disappeared in the US, although many other countries continued to create specialized agencies and initiate river basin hydroelectric and resource development schemes similar to the TVA for decades afterward.

There was very little theorizing about uneven regional development during this first phase. There was some concern for urban poverty and a widespread critique of the overconcentration of both wealth and poverty in the booming industrial capitalist cities, but the regional discourse rarely addressed directly the rootedness of these problems in the capitalist development process. During this period, the discipline of geography for the most part was theoretically moribund and increasingly isolated from developments in the social sciences as well as in professional planning practice. After the swift expansion of regional planning and regionalist thinking in the first quarter of the twentieth century, peaking in the formation of the TVA, an equally rapid decline took place in the second quarter century.

4.1.4 Operational trade-offs

•

Site selection. Selecting a site is challenging because multiple factors need to be in equilibrium: the centrality of the location and the higher cost of land, as well as potential bureaucratic issues connected to land ownership (e.g. securing leases from landowners) and subsequent evacuation of an attractive area; the choice of cheaper non-urbanised areas and the cost of preparation works; the capacity of the selected site and that of the existing grid; the availability of public facilities to overtake (e.g. schools, sports areas) and the resulting lack of services; the choice of closer terrain with a difficult topography and the need of water drainage and other infrastructure [7,82,87]. Furthermore, the size and configuration of the initial settlement needs to be considered in the choice, because the creation of TH sites in small settlements can foster urban decentralisation [88–90].

Houses, site design and infrastructure. The design and construction of both hard and soft infrastructure should foster tangible and intangible aspects of community development and be timely integrated with the design of the houses and sites' layout, considering ownership issues [91]. Among other things, this can: improve accessibility and thus logistics [87], walkability, care and social gathering [54]; enable the implementation of passive design strategies [71]; sustain urban agriculture via water recycling [92]; help reducing the risk of violence and abuses on vulnerable categories of people such as women and children, e.g. by locating latrines in well-lit areas close to the housing units [93] or by adding a second door and a way out on the rear of the transitional houses [94]. Additionally, this can avoid situations where infrastructure construction threatens the stability of houses, but rather supports synergies that reduce housing exposure to natural elements [23]. It can, however, bring some challenges to pre-design and planning. For example, pre-installed solar panels coming with prefabricated structures may be orientated in a suboptimal way due to the position of the houses on site, which reduces power production capacity as well as the possibility to compensate for their share of embodied carbon [95].

Typologies, materials and building technology. When choosing housing typologies, materials and construction technologies there is a need to balance components' durability with the many uncertainties connected to the houses' lifespan, intended and real. Some researchers have proposed a ‘speed and seed’ approach where building components made of durable materials are engineered to be flexible and adaptable, allowing incremental changes [96]. Moreover, past research has shown the impact of typology and materials on the psychological wellbeing of the affected population [97]. Additional considerations concern: the use of windproof materials and the need of ventilation [98]; the use of high performance materials and their local availability [47]; the choice of a performative building technology and the capacity of local people to adopt it at a later time for safely adapting TH units to their needs [14]; the use of traditional building methods and the availability of necessary materials [23]; the use of new efficient technologies that add complexity to old threats [31]; the presence of buy-back options and the unavailability of clear criteria to establish end-of-use prices [81].