Modeling Annual Average Daily Traffic with Integrated Spatial Data from Multiple Network Buffer Bandwidths

This paper presents a geospatial method to develop models to estimate annual average daily traffic (AADT) along a link with integrated spatial data from multiple network buffer bandwidths. The method incorporated spatial data that captured principles and statistical techniques to estimate directly AADT, and thereby bypassed the sequential four-step method. Spatial data (off-network characteristics, such as demographic, socioeconomic, and land use) captured from multiple network buffer bandwidths around a link were integrated with spatial weights that decreased as the distance from the link increased. Off-network and on-network characteristics were used as independent variables. The AADT calculated from field observations was used as the dependent variable. The applicability of Poisson and negative binomial distributions to the development of the models in a generalized framework for estimating equations was examined. A chi-square statistic test was conducted to validate the developed models and assess their applicability. The results indicated that the models based on the negative binomial distribution were more suitable than the models based on the Poisson distribution to integrate spatially varying data and estimate AADT. The maximum distance necessary to capture and to integrate spatial data varied by road functional class.