A Data-Driven Method to Define and Identify Bicycle Crash Corridors
Dr. Aditya Medury
Dr. Offer Grembek
California Department of Transportation (Caltrans)
While spot safety approaches are effective and necessary, they address the safety problem in a reactive manner and on a very small scale. To complement spot identification methods, transportation safety practitioners have shown an increased interest in developing approaches that can also lead to the implementation of improvements in a proactive manner such as corridors. This research proposes a high-level specification for safety corridors, and studies the performance of three approaches to identify bicycle crash corridors:
- A method based on calculating the variance of inter-crash spacing;
- A modified Continuous Risk Profile approach; and
- A density-based spatial clustering of applications with noise (DBSCAN) (see Figure 1).
The methods were compared in their ability to detect corridors in terms of detection output, detection procedure, and robustness. The variance-based approach provided a slightly better understanding of the corridor, while DBSCAN required fewer input parameters and is more robust.
Figure 1. An illustration of DBSCAN-based corridor identification method
Figure 2. Heatmap of corridors (in blue lines) identified along the state highway system
To further facilitate the identification of appropriate countermeasures for identified corridors, the research is also developing a typology for bicycle corridors based on dominant attributes of crashes within the corridor (location and types of crashes, types of movements, etc.).