Modeling the complexity of sustainable cities: The interdependence between infrastructure systems and the socioeconomic environment
MetadataShow full item record
As a critical component of the city, urban infrastructures emerge through the interactions with the socioeconomic environment. Managing the complexity behind the interactions can make the city more sustainable. By this, we mean if we provide more sustainable amenities that people desire, a greater adoption of more sustainable infrastructures will likely occur. Two categories of infrastructure have emerged in recent years as exemplars of more sustainable development: green infrastructure and transit-oriented development. At the same time, new digital tools have emerged to better predict market acceptance of these infrastructures. This dissertation employs agent-based modeling, a latent-class analysis of survey results, and an online survey to model the potential of adoption of these infrastructures and the public benefits. The principal research content of the dissertation consists of two parts. First, understanding social preference and adoption of green infrastructure (e.g., low-impact development (LID) to control storm water), and transit-oriented development (TOD) to reduce car dependence and incentivize denser land use; Second, by developing an urban model that accounts for the complexity of the urban system, the purpose is to predict the emergent property of the city (e.g., land use, water consumption, tax revenues and carbon emissions). These two aspects constitute the research content of this dissertation. The principal findings of the dissertation are: 1) the use of digital feedback tools to inform the modeling of complex urban systems; 2) the future development of the metro Atlanta area can be more compact and sustainable with implementations of LID, TOD, and the proper policy. This dissertation consists of four sections. In the first section, I have developed an agent-based model (ABM) to predict the land use pattern. The ABM is an approach suited to simulating and understanding the dynamics of the complex system. To reduce the complexity and uncertainty of the ABM, the model simulates the decisions and interaction of agents (i.e., home buyer, the developer and the local government) at the neighborhood scale. The output of the ABM serves as the baseline scenario of land use pattern for evaluating the effect of tax investment and fees on the adoption of green infrastructure designs and more compact land use patterns. Second, with the help of the ABM, I evaluated and compared the policies (i.e., impact fees, subsidy) on the adoption of green infrastructure designs and more compact land use pattern. I developed a more sustainable development (MSD) scenario that introduces an impact fee that developers must pay if they choose not to use LID (i.e., rainwater harvesting, porous pavement) to build houses or apartment homes. Model simulations show homeowners selecting apartment homes 60% of the time after 30 years of development in MSD. In contrast, only 35% homeowners selected apartment homes after 30 years of development in a business as usual (BAU) scenario where there is no impact fee for LID. The increased adoption of apartment homes results from the lower cost of using LID (i.e., rain garden, native vegetation and porous pavements) in public spaces and improved quality of life for apartment homes relative to single-family homes. The MSD scenario generates more tax revenues and water savings than does BAU. Third, as an initial effort to calibrate the home buyer’s preference for community design in the ABM, I developed an analytic model based on an existing community preference survey. The data available for this effort is from National Association of Realtors’ 2011 community preference survey. I applied a latent class choice model to this data, and discovered four classes of individuals that reveal distinctive behaviors when choosing smart growth neighborhoods, based on the interplay between aspects of community design, socioeconomic characteristics and personal attitudes. Linking the results of the latent class choice to an agent-based market diffusion model enables planners to evaluate the effectiveness of a proposed smart growth neighborhood design in inducing less sprawling development. In the fourth section, I developed a survey that focuses on preferences of metropolitan Atlanta residents for LID and TOD. With the responses collected using Mechanical Turk, I developed a latent-class residential community choice model of four distinctive classes that reveal heterogeneous preferences for community designs. Spatial distribution of the four classes was mapped out to visualize the locations of the demand for different community designs in metropolitan Atlanta. The analysis of the impact of increase in housing price on the adoption of LID and TOD shows a low risk of investing in LID and TOD in metro area. Residents are willing to adopt the community with LID and TOD as compared to the corresponding one without LID and TOD. It turns out that LID and TOD have a great potential for adoption in metro Atlanta. Further, I integrated the individual residential community choice simulation into an agent-based market diffusion model to predict the emergent land use pattern and explore polices that can drive the adoption of more compact development. Results show that the current policy requiring single-family houses to implement LID based on individual sites should be switched to one that requires community-based LID for single-family houses. Such a policy switch will lead to a higher adoption of apartment homes with LID and TOD. Lastly, I estimated a 28% carbon emission reduction from more compact development driven by LID and TOD. This thesis is the very beginning of using digital feedback tools to anticipate market responses to more sustainable development alternatives. On the basis of the progress made in this dissertation, future work is recommended in terms of the development of an integrated platform that supports the integration of individual modules (e.g., land use, traffic simulation, air quality, and water resource management) for modeling the complexity, big data analytic techniques (e.g., Twitter, GPS data, sensors) for uncovering the interdependencies between infrastructures and socioeconomic development, and the exploration of sustainability metrics for public communication to build citizen capacity for sustainable cities.