A Kernel Density Method for Aggregating Boundary Collision Data into Areal Units

Boundary collisions are motor accidents which occur on the boundaries of areal units. In some areas, boundary collisions may account for a large proportion of the total collisions. Generally, it is a critical step to aggregate boundary collisions into areal units before collision analysis. Exaggerated or underestimated aggregation results may hamper traffic safety analysis and management. In this paper, we propose a boundary collision aggregation approach based on the collision density ratio. The proposed method is compared with two other boundary collision aggregation methods. Two regions (downtown region and south region) in City of Edmonton are selected as the study area. The assessment result shows that the proposed method outperforms the other two methods.