Linear Mixed-Effects Models for Paint Pavement-Marking Retroreflectivity Data

Abstract

The objective of this paper is to present an improved paint pavement-marking retroreflectivity degradation model. Previous degradation-modeling methods were reviewed and studied. The paint-marking retroreflectivity data were collected on 25 two-lane rural highways in North Carolina from November 2007 to May 2010. A linear mixed-effects model (LMEM) was selected as most appropriate for the paint-marking retroreflectivity data. The LMEM took into account the correlation among repeated measurements of the data set and produced more accurate predictions than other methods. LMEMs were established for white edge and yellow center paint pavement markings. The LMEM results showed that the estimated white edge pavement marking life cycle is 37.5 months and yellow center marking life cycle is 38.9 months on average when using 100 and 65 mcd/m2/lx as the minimum acceptable retroreflectivity values for white and yellow paint markings. A model comparison indicated that the LMEM provided a more accurate prediction than linear-regression models when multiple retroreflectivity measurements are collected on a specific road.