Prediction equation for vehicle-pedestrian crash and safety analysis at signalled intersections in Ghana

Abstract

The occurrences of vehicle-pedestrian crashes at signalized intersections were investigated using a 3 year (2004-2006) crash records of 82 signalized intersections in Accra, Kumasi, Tema, Sekondi-Takoradi and Tamale. The data were analyzed using Micro-computer Accident Analysis Package. Traffic flow characteristics, intersection features and socio- economic activities at the intersections were used as inputs to develop mathematical model for predicting pedestrian crashes. On the assumption that negative binomial errors control over dispersion characteristic of the crash data, a Generalized Linear Model was applied to estimate parameters of the predicting equations. Accra accounted for more than 57.6 % of the pedestrian crashes with highest casualty rate of 34 injuries per annum (including deaths). Crossing was the dominant pedestrian action leading to about 70 % of the injuries. Vehicular traffic flow, crossing pedestrian flow and mean vehicle approach speed were the key predictor variables that had positive correlations with the crash rates, while approach number of lanes and presence of median on the road minimized occurrence of pedestrian crashes. Road designers and planners could apply the pedestrian crash model to predict the effects of traffic growth, vehicular speeds, intersection geometric characte-ristics and dynamics of socio- economic activities on crash frequencies at signalized intersections in Ghana.