Disaggregation analyses of spatial interaction residential location models
Spatial Interaction Land Use Models are the most applied Urban Simulation Models in the United States for Integrated Transportation and Land Use Planning. However, they have three major limitations. First of all, there is incompatibility between land use models and transportation models in terms of geographic detail as land use models typically use larger zones than transportation models. This is in part caused by the difficulty to obtain employment data at fine geography and the fact that employment and household models typically use the same zonal system. Secondly, there is incompatibility between land use models and transportation models in terms of households' socioeconomic detail. A transportation model usually requires more detailed household information while a typical land use model classifies households only by income. Finally, housing variables are not explicitly considered in land use models. Consequently, these models do not fully capture households' behavior and are incapable of analyzing housing policies.
This dissertation addresses the spatial incompatibility problem by developing a nested two-zonal modeling system for household allocation, i.e., a larger zonal system for employment and a nested smaller zonal system comparable to Traffic Analysis Zones for households (Spatial Disaggregation). The calibration results of Atlanta region data show that nested two-zonal modeling system is flexible, stable, consistent, and not sensitive to employment aggregation. This dissertation also experiments on certain housing variables as additional gauge of locational attractiveness (Attractiveness Disaggregation). The calibration results show that Percent of Rental Housing Units and Rent improve model's Goodness-of-Fit and help to explain households' locational choices for certain household types. This dissertation finally experiments on additional classifications of households (Locator Disaggregation). The calibration results show that different household groups (e.g. classified by marriage status, presence of children, or the number of vehicles owned) behave quite differently. This new approach not only captures the diversity of households, but also provides more informative input to the transportation models.
Area planning & development;
0999: Area planning & development