How local street networks contribute to the traffic accident occurrence: Mersin city case

Abstract

In this study, the issue of the accident frequency/occurrence in local residential streets where vehicle counts are not obtained is discussed. The accident reports received from the Ministry of Interior include fatal and injury accidents. But since the number of deadly accidents is minor, a binomial regression was not employed in the statistical analysis. This study investigates the role of morphological characteristics of the street network where main arterials are excluded in the safety of residential neighborhoods of Mersin, Turkey. Results showed that commercial and industrial uses generate more traffic than residential neighborhoods, therefore, responsible for higher rates of accidents. The number of vehicles, employment, and population increase the accident rate. Empirical model results showed that population density, street length per capita (meters), and number of intersections significantly affect accident occurrence. Number of intersections increases accident frequencies while street length per capita reduces. In contrast, the number of links, street density, link to node ratio, and average link length have no significant effect on accident occurrence. Traffic safety cannot be achieved if the street length is not proportional to the number of vehicles. In addition, street design is also a matter of efficient use of urban space.