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.
