Modeling the effects of horizontal and vertical road geometries on traffic crashes in rural roads

Abstract

This study investigates the relationship between road geometry and traffic crashes on rural state roads in Zonguldak, Turkey, between 2016 and 2022. The objective is to quantify how specific combinations of geometric elements influence average crash frequency per kilometer. A total of 941 crashes were analyzed across approximately 156 km of roadway, using officially authorized crash reports and project-based geometric data. Crash locations were spatially matched with corresponding alignment segments. Four geometric combinations were evaluated: horizontal curve–vertical curve, horizontal curve–longitudinal grade, horizontal tangent–vertical curve, and horizontal tangent–longitudinal grade. Multiple linear regression models were developed to assess their joint effects. The results indicate that horizontal curves result in higher crash frequencies than horizontal tangents, while vertical curves exhibit greater risk than longitudinal grades. The highest crash rates were observed where horizontal and vertical curves coincided, suggesting a potential interaction effect between geometric elements. The final models demonstrated statistically significant relationships (p < 0.05) and satisfactory explanatory power, identifying critical geometric configurations associated with increased crash occurrence. These findings provide quantitative evidence for prioritizing safety improvements on existing rural roads and offer guidance for safer geometric design in new roadway projects.