Advisor: Karen C. Seto
Problem Statement: Cities are responsible for most of the global final energy use and carbon emissions. Energy use and emissions are coupled to urban form (the physical configuration of a city) through land use conversion, transportation, and consumer preferences. Urban form in India is undergoing extensive transformation due to changing lifestyle preferences and decentralized expansion into arable land (Tiwari, 2002). Urban growth in one city affects growth of other cities on a regional scale. Exchanges of labor, material, information, and energy between clustered cities results in tremendous growth in some cities and marginal growth in others.
Urban layout and built form produce significant effects on regional and global climate through localized heat transfer effects and emissions (Seto, 2011). To understand how Indian cities will grow, it is necessary to understand the processes that lead to the growth of individual cities as well as the mechanisms by which cities interact on a regional scale. My proposed project will investigate these local and regional urban processes. This information about the dynamics of urban growth in India will be incorporated into a numerical model to assess the relative strengths of these mechanisms and to forecast future urban growth at a regional scale.
Objectives: What determines the type of growth experienced by a city? How does the growth of a city affect the growth of cities to which it is economically clustered? What are the mechanisms by which regional urban clusters communicate? How will future investment in infrastructure and resource sectors affect regional growth and the strength of interactions between cities? What are the anticipated effects of these external perturbations on regional growth with respect to land use conversion, loss of agricultural land, and pollution?
Background: Potential for altering urban patterns, including density and resource distribution, depends critically on understanding the strong nonlinear interactions between the built and social environments. Additionally, cities can be viewed as open systems through which materials, people, information, and energy flow. Previous work has identified drivers of urbanization, which include foreign investment and infrastructure development (Seto, 2011). Existing research has also identified significant feedbacks that counteract or enhance these drivers, such as disease resulting from high population density and lack of sanitation services, or housing tenure loss due to infrastructure development and subsequent eviction. This study will investigate how drivers on local and regional scales affect the growth of individual cities and the clusters of interacting cities.
Field Site Justification: Field work over the summer will employ interviews with corporations and city agencies to investigate why corporations move to one city over another, how planning decisions are made to accommodate and attract growth, and what sorts of flows exist and are driven by urban growth in an individual city. The following cities will be used for comparison: Delhi, Chandigarh, Agra, Jaipur, Ahmedabad, Mumbai, Pune, and Hyderabad. These cities have diverse patterns of urbanization, population change, and industrial activity, making them an ideal sample for understanding how urbanizing forces vary across states. India is undergoing some of the most rapid urbanization worldwide making it an important case study.
Methodology: In each city, I will meet with representatives of city planning agencies, real estate companies, and land development and multinational corporations to understand how each city has grown and how these actors make decisions about which cities to invest in. My advisor, Karen Seto, and her collaborators at TERI University in Delhi will assist me in obtaining contacts to access these representatives from these entities.
Using information from the field, I will develop an agent-based model (ABM) that will represent behaviors and decisions made by socioeconomic agents, and feedbacks resulting from interactions between the agents, following methods from Werner (2007) and McNamara (2008). Preliminary review of the literature suggests that the following feedbacks are integral to the city
system and I will explicitly represent these in the model (pending adjustments informed by field interviews). Agents that represent households will form social networks, and these networks will be the basis for propagation of cultural values, lifestyle preferences, and employment and housing opportunities (Banerjee, 1981). Household level agents will make decisions in the labor market (formal and informal) about which jobs to hold based on information made available through their social networks (Mitra, 2004). City agency agents will make decisions for building infrastructure and transit networks, and slum clearance ordinances, which are affected by foreign direct investment, federal investment and predictions of urban growth patterns (Amis, 2000). Industrial and technology firm agents will make decisions on production, investment in expansion, and location to maximize labor market access and minimize costs (Shaw, 1999). Investment decisions will incorporate predictions of future economic conditions and availability of skilled labor being trained in local universities. Foreign investment decisions will affect slum clearance agendas and the demolition of informal settlements will simultaneously increase the desirability of localized investment opportunities (Dupont, 2008). Land developer agents will make decisions on a spatially explicit grid to develop tracts for housing and commercial property in response to housing and land markets as well as proximity to transit and sanitation infrastructure. Transit and sanitation networks, in turn, will respond to development by increasing network extent. Agents will employ a suite of adaptive behavioral and utility optimization strategies in conjunction with heterogeneous projections of future market conditions to make decisions following Feigenbaum (2003).
Agents will make decisions on a spatially explicit grid derived from high-spatial- resolution satellite imagery. Gridded simulation output will be compared to historic land cover maps derived from remotely sensed data. The simulation will be compared to historical land cover in terms of the relative area of land cover types, total area of land conversion over time, and spatial pattern characteristics such as shape and edge metrics. Parameters and initial conditions that have been estimated from data will be incrementally adjusted until metrics closely approximate historical urbanization, following Brown (2005).I will calibrate the model with the Night Time Lights dataset, which gives historical indices of urban intensity on a coarse scale. A comparison of simulated and remotely sensed data over time will be used to assess how well the model predicts regional urban expansion. The calibrated model will be fed simulated shocks to foreign investment and energy prices, for example, to estimate how sensitive the system is to perturbations, and how the interactions between cities are affected.