Accident Prediction Model for Horizontal Curves on State Highways Using Spatial Variation

Abstract

Accidents have become one of the primary cause of fatalities on highways. Road accidents are one of the significant issue around the globe, but in context of India, the severity is more due to immense growth in road networks and traffic capacity. Curve are at higher range of potential risks of accidents because of inadequate sight distance and speed measures. This study aims to develop accident prediction model using regression analysis. Location selected for study was State Highway-1 in Udupi district, Karnataka. Ten curves are selected on the road and comparative study of model prepared is checked to verify the model reliability. Datasets used for model calibration and development is Highway Geometric data, past accidents records, and spot speed of vehicles. Geometric data for the road sections are obtained from satellite imageries, and GIS data and speed data are collected using speed camera. Model generation was done using statistical computing by using multi-linear regression. The model showed that curve details and speed reduction between successive features were strongly related to accident frequency. Sharper curves are more tend to accidents, and speed reduction is higher at curves with smaller radius. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.