Multiscale Analysis of Urban Walkability and Pedestrian's Destination Choice

Much attention has been paid to the relationship between the built environment and walking behavior. However, the findings are mixed due to different quantitative measures based on different research purposes or at different geographical scales. Various geographical scales to define neighborhood increase spatial uncertainty, which could make research outcomes biased. Drawing upon a household travel survey in Salt Lake County, Utah, this study implements a multiscale analysis to explore the people's walking behavior. A spatial filtering method is employed to examine the determinants of the walking density of the trip ends at the block level. We also construct a destination choice model to explore the determinants of people's mobility by walking at the trip level. The results suggest that a compact design and high population density contribute to high walking density but decrease walking mobility. Furthermore, mixed-land use does not help generate walking trips but creates destinations for walking and increases walking mobility. The geographically weighted regression (GWR) model suggests that the impact of mix-land use on people's walking behavior is sensitive to the local context. We also find that air pollution and traffic volume negatively influence the walking density and mobility.