Citation

Keay K, Simmonds I. Accid. Anal. Prev. 2005; 37(1): 109-124.

Affiliation

School of Earth Sciences, University of Melbourne, Victoria 3010, Australia.

Copyright

(Copyright © 2005, Elsevier Publishing)

DOI

10.1016/j.aap.2004.07.005

PMID

15607282

Abstract

An investigation into the effect of weather variables on traffic flow at a site in Melbourne, Australia, for the period 1989-1996 was performed. Rainfall was the strongest correlated weather parameter and it had the greatest impact in winter and spring, when traffic volume is reduced on wet days. There are statistically significant decreases of 1.35 and 2.11% in traffic volume on wet days in winter and spring. The reduction increases to 2-3% over the 2-10mm range, the largest being 3.43% for the 2-5mm class in spring. For the first time, our study considers separately daytime and nighttime periods. We found a reduction of 1.86% in winter and 2.16% in spring during daytime rainfall. The reduction at nighttime is significant over all seasons, ranging from 0.87% in winter to 2.91% in spring. We have explored an application where the traffic volume was used to normalise the road accident count and found the rain effect to increase by 2.4, 1.9 and 5.2% relative to the daily, daytime and nighttime dry mean accident count. Generally, the normalised count is greater than the raw count, with a larger increase for the higher rainfall classes.