Identification of accident black spots to improve the safety on highways.

Abstract

Road accidents remain a critical global concern, necessitating efficient transportation system management. Identifying high-risk areas, known as "Black Spots (BS)," is crucial for road safety. The research objectives involve conducting a comprehensive analysis of past accident records, comparing the outcomes produced by various statistical methods, and evaluating how road geometry contributes to improve safety at identified BS locations. Four methods, Accident Point Weightage, Accident Rate Screening, Empirical Bayesian, and Spatial Autocorrelation (Moran’s I) Method, coupled with the Getis-Ord Gi function were applied to five years of traffic accident data (2018-2022) from Padeniya - Anuradhapura Road (0 km to 54.4 km) in Sri Lanka. The analysis reveals a 19% fatality rate, with rear-end collisions (26%), angle collisions (18%), and pedestrian accidents (17%) being prevalent. Forty-nine road segments were identified as BS locations by at least one method, showing consistency among APW, ARS, and EB methods. Spatial Autocorrelation method results differed but still identified high-risk areas. This suggests that these methods can be favorably applied to roads with similar characteristics as those selected for this study. Considerably, each BS method yielded both concordant and disparate BS locations, with enhanced accuracy observed for all methods.