Relationship between driving behaviors and capacity drop using empirical trajectory data

Capacity drop, referring to the reduction in bottleneck discharge rate, is a primary contributor to freeway delays. This paper aims to provide empirical evidence of how driving behaviors trigger capacity drop through the analysis of vehicle trajectory data collected before its onset. By quantifying the spatiotemporal impacts of driving behaviors on the discharge rate, we find that the impact of lane-changing (LC) on the discharge rate increased significantly before the onset of capacity drop. Gaps produced by LCs in the anticipation phase contribute most significantly to capacity drop, followed by gaps in the relaxation phase. In contrast, gaps left in the original lane are filled by other vehicles, thereby increasing the discharge rate. The contribution of car-following (CF) behavior to capacity drop is relatively limited. However, after capacity drop occurs, CFs gradually reduce the discharge rate within the bottleneck segment. These empirical results are further validated using several trajectory datasets.