Connecting Land Use and Transportation Toward Sustainable Development: A Case Study of Houston-Galveston Metropolitan Area

How do land use characteristics affect individual and household travel behavior in a regional context? Can the investigation justify the land use policies to reduce automobile dependence and achieve the goals of sustainable development in the metropolitan areas? Previous research enhanced our understanding of the connections between land use and travel behavior. It also provided implications for managing automobile-dependent travel behavior. However, there are questions still left unanswered about the causal connections between them, and the effectiveness of the land use policies to manage travel demand. To address the issues, attention is focused on the effects of land use measures on travel behavior outcomes from different modeling perspectives. The travel demand modeling explores the associations between land use and travel behavior. In addition, the causal modeling helps clarify the causal connections between them. It includes the structural equation models (SEMs) and the directed acyclic graphs (DAGs). The study focuses on six counties of the Houston-Galveston Area Council (HGAC) area. Travel behavior outcomes contain individual mode choice, household automobile trip generation and household total vehicle miles traveled (VMT). Three dimensions (i.e., density, diversity and design) of six land use measures are considered, which are computed using quarter-mile buffers for both trip origins and destinations. Different travel outcomes and modeling strategies are examined for different travel purposes. The significance of land use measures in affecting travel behavior is found to be evident, while varying to a certain degree according to trip purposes, travel outcomes and methodologies. For individual model choice, multinomial logit (MNL) models, the SEMs and the DAGs for different trip purposes support the hypothesis that land use measures directly affect individual mode choice behavior when other factors are kept constant. There is also evidence from causal models that land use factors indirectly influence it through travel time. For household automobile trip generation, there is no evidence to assert that land use measures at origin significantly affect household automobile trip rates when travel cost and socioeconomic variables are controlled. However, it is confirmed that land use measures have indirect causal connections with automobile trips through travel costs for all trip purposes. For household total VMT, it is found that land use patterns around residential locations are not only significantly associated, but also causally connected with household VMT. To summarize, compact development with high density and improved network design generally contribute to the reduction in automobile dependent travel patterns in the HGAC region