Spatial Heterogeneity Model of Impact of Community Built Environment on Vehicle Miles Traveled

To promote a green travel environment in the community life circle planning, this paper analyses the spatial heterogeneity of the impact of the community built environment on vehicle miles traveled (VMT). Based on the 5D-dimension built environment, six indicators were used to describe the built environment of the community, such as population density, land use diversity, and bus stop density. Based on the technical guide for community life circle planning, the study defined the scale differentiation of community life circle combined with walking speed, non-linear coefficient and other indicators. The built environment was measured by the point of interest (POI) data, road network and other geospatial data. Taking the travel behavior survey data of Baoding residents as the empirical research data source, the study developed a multi-scale geographically weighted regression model (MGWR) considering the scale variation of independent variables. The results indicate that: (1) compared to the ordinary least squares regression (OLS) model and the traditional geographically weighted regression (GWR) model, the MGWR model with variable scale heterogeneity reduces the autocorrelation of the residual, and the adjusted R square value is the highest, which is respectively 1.8 times and 6.0 times higher than the GWR Model and OLS model. (2) From the standardized coefficient, land use mixing degree and bus service level have the greatest impact on VMT. (3) The road density and intersection density are close to the global scale, and the spatial heterogeneity is weak. Other built environment variables have strong spatial heterogeneity, so the differentiated spatial design is needed. (4) The spatial distribution pattern of the local regression coefficient shows the trend of "center-periphery", which is strongly coupled with the urban form. © 2022 Science Press. All rights reserved.