A novel dynamic segmentation model for identification and prioritization of black spots based on the pattern of potential for safety improvement

Road segmentation is one of the most important steps in identification of high accident-proneness segments of a road. Based on the ratio of the Potential to Safety Improvement (PSI) along the road, the objective of the paper is to propose a novel dynamic road segmentation model. According to the fundamental model assumption, the determined segments must have the same pattern of PSI. Experimental results obtained from implementation of the proposed method took four Performance Measures (PMs) into consideration; namely, Crash Frequency, Crash Rate, Equivalent Property Damage Only, and Expected Average Crash Frequency with Empirical Bayes adjustment into the accident data obtained from Highway 37 located between two cities in Iran. Results indicated the low sensitivity of the method to PMs. In comparison with the real high accident-proneness segments, identified High Crash Road Segments (HCRS) obtained from the model, demonstrated the potential of the method to recognize the position and length of high accident-proneness segments accurately. Based on the road repair and maintenance costs limitation index for safety improvement, in an attempt to compare the proposed method of road segmentation with conventional ones, results demonstrated the efficient performance of the proposed method. So as to identify 20 percent HCRS located on a read, the proposed method showed an improvement of 38 and 57 percent in comparison with the best and worst outcomes derived from conventional road segmentation methods.