Long-Term Trend of Temporal Distribution of Roadway Traffic

Abstract

This paper investigates the long-term trend of temporal distribution of roadway traffic using hourly traffic-count data collected in Seoul, South Korea, from 1995 to 2010. Interpolations using quadratic local polynomial regressions and a Gaussian mixture model were applied in order to identify changes in temporal distributions at a scale of less than one hour. The identified changes in distribution characteristics suggested that more attention should be given to outbound traffic, which appeared to be worsening because of the increased population and jobs outside of Seoul. Although this study did not attempt to thoroughly investigate the causal relationship between the changes of temporal distributions and the associated factors, the framework and methodology suggested are expected to play a foundational role in identifying the long-term trend of traffic flow using archived data.