Abstract

One way of planning for a region's future transportation infrastructure and capacity needs is to forecast travel demand. Integrated land use-transportation modeling tools do this in a way that is sensitive to how the performance of a transportation system affects peoples' decisions of where to live and firms' decisions of where to locate. As such, they are helpful tools for analyzing different transportation and land use policy scenarios. The transportation system is a factor in how new urban areas develop. Policies attempt to regulate dispersed urban development, known as urban sprawl, however integrated modeling frameworks can only evaluate those policies that affect the extent of non-urban land legislated as developable, or those within urban areas.
Incorporating other policies relating to sprawl into integrated models is limited by their ability to represent geometric changes to the landscape; those associated with the transition of non-urban land to residential use. Building on current methods for representing geometric landscape changes, this thesis is about models and algorithms for representing the specific forms these changes can take. There are a number of algorithms, taking distinct approaches to subdividing blocks into parcels and generating roads, suggesting different algorithms are better for generating different forms. There is little guidance on when to use which algorithm, potentially resulting in sub-optimal geometric representation of future urban areas.
This thesis outlines a process for representing the spatial distribution of urban form in future urban areas within integrated models. To this end, it has estimated a model for predicting the spatial distribution of road network patterns in future residential neighborhoods and identified the block subdivision algorithm most suited to subdividing each of the possible road network patterns. Results can serve as guidelines for deciding which algorithms to use on which road network types. They also present a possible way of estimating the type of future road network in a local area as a function of slope of terrain, period of development, proximity to a river and adjacency to a road network of the same type, among others. This knowledge could help improve the accuracy of population predictions and potentially be implemented within the modeling process of integrated models once these are better able to represent geometric changes to the landscape.