Pavement markings provide useful visual and navigational guidance information to motorists. Current warrants for the application of pavement markings in the United States are based on traffic volume, traveled way width, and the number of travel lanes. To be effective, pavement markings must be visible to drivers, particularly at night. Little research exists to quantify the relationship between pavement marking retroreflectivity and traffic crash frequency. The purpose of this paper is to perform an exploratory analysis to determine if there exists a statistical association betweenpavement marking retroreflectivity and traffic crash frequency. First, models of pavement marking retroreflectivity degradation were developed from selected highways in North Carolina using artificial neural networks. Subsequently, monthly estimates of pavement marking retroreflectivity levelswere combined with roadway inventory and crash frequency data. Generalized estimating equations were used to estimate the monthly target crash frequency. The results indicate that the regression parameter estimates for yellow and white edgeline pavement markings were negative, but neither was statistically significant for the two-lane highway nighttime target crash frequency model. For multi-lane highways, all of the pavement marking retroreflectivity parameter estimates were statistically significant. The white pavement marking retroreflectivity parameter estimates were negative, as expected. The yellow pavement marking retroreflectivity parameter estimate was positive. This finding was unexpected and may suggest that the pavement marking system is providing different navigational cues for drivers on multi-lane highways.
Abstract