Using a surrogate safety approach to prioritize hazardous segments in a rural highway in a developing country

Abstract

Road safety is a global concern particularly in developing countries where some road sections are disproportionately more vulnerable in terms of the frequency and severity of crashes. Other than using historical crash data based reactive approaches, those sections need to be identified proactively, so that mitigation measures can be applied. Moreover, those approaches are sometimes questioned mainly due to data reliability issues in developing countries. The study reported here is aimed at highlighting the applicability of traffic conflict techniques as surrogate safety measures to identify those sections of a rural highway in a developing country, which are most likely at risk. An adapted framework is demonstrated to identify traffic conflicts using combined surrogate indicators acknowledging the limited resources and facilities in developing countries. A new model is put forwarded using a count data modelling approach. Both fixed and random parameters model derivatives have been explored as an alternative methodological approach to relate the factors affecting the number and probability of conflicts. The partial effects of individual independent variables were estimated to gain a better insight of their impact. The results show that the model can predict high risk segments in terms of probability of conflicts as well as safety risk, as well as prioritize road sections according to the likelihood of their safety level. The model provides a less expensive alternative to the collection of historical crash data in order to identify hazardous road locations or black spots on two-lane highways in developing countries.11 A part of the paper was presented in the Transport Research Board (TRB) 97th Annual Meeting and an abridged version from that part was published in TRB online forum1. Mahmud, S., et al., Prioritizing Hazardous Road Sections Using Surrogate Safety Measures: A Count Data Modelling Application in a Heterogeneous Traffic Environment, in Transport Research Board (TRB) 97th Annual Meeting 2018, Transport Research Board (TRB) Washington DC. p. 18–02193. Link: http://amonline.trb.org/2017trb-1.3983622/t025-1.3995192/382-1.3995580/18-02193-1.3993134/18-02193-1.3995603?qr=1.

Keywords

• Crash prediction models
• Poisson model
• Random parameter
• Zero-inflated
• Developing countries
• Heterogeneous or mixed traffic