Document Type : Research Article
Faculty of Engineering, Imam Khomeini International University, Qazvin, Iran
Department of Civil Engineering, Imam Khomeini International University
Motor vehicle crashes are currently among the ten major mortality factors all around the world. Iran follows a similar pattern and has a high annual rate of fatal crashes. The majority of these fatal crashes occur on rural routes, particularly on two-lane roads. In this study, the effective factors on the occurrence of accidents on Birjand-Qayen route (South Khorasan Province, Iran) is studied. In order to investigate the effects of being in two-lane segments, horizontal curves with more than 2% slope, and segments with insufficient lighting, on the fatal crash count and crash severity, standard Poisson regression model (fatal crash count per km as the dependent variable) and standard ordered logit model (severity of crashes as the dependent variable) were calibrated using the 2013-2016 data. Zero-inflated negative binomial and zero-inflated Poisson models were also applied to investigate the effect of true zero in fatal crash count modeling. Furthermore, a panel analysis of crash severity ordered logit model was carried out to consider the spatio-temporal effects. The results of the fatal crash count modeling showed that three or four-lane segments (compared to two-lane segments) reduce the number of fatal crashes, whereas curved segments with more than 2% slope (compared to straight segments) increase this number. Moreover, according to the crash severity model results, segments with sufficient lighting as well as three-lane or four-lane segments (compared to two-lane segments) reduce the severity of crashes; whereas crash severity in curved segments with more than 2% slope (compared to straight segments) is increased.