Mechanism analysis of highway parking facility layout impacts on traffic flow based on ETC spatio-temporal trajectory data

Highway parking facilities improve service accessibility but may disturb mainline traffic through deceleration, lane changing, merging, and diverging associated with parking access. This study develops an electronic toll collection (ETC) trajectory-based framework to quantify how parking facility layout affects highway traffic operation. Thirty days of ETC records from a 168.73 km corridor containing 26 gantries and 11 parking facilities were used to reconstruct vehicle trajectories and identify parking events. Facility density, spacing, capacity, and parking demand-supply ratio were related to average speed, speed fluctuation, and congestion index. Spatial influence was examined by distance-based comparison and exponential/power-law decay models, while XGBoost with SHAP interpretation was used to assess nonlinear effects. Results show that traffic disturbance is strongest within 1 km of parking facilities and becomes weak beyond approximately 3-5 km. Higher facility density and demand-supply pressure are associated with lower operating speed and greater fluctuation and congestion, whereas larger spacing generally mitigates disturbance. The exponential decay model provides a better representation of spatial attenuation than the power-law model, with an average R² of about 0.91. XGBoost achieves R² values of 0.93, 0.86, and 0.89 for speed, speed fluctuation, and congestion, respectively. Facility density, spacing, and demand-supply balance are consistently more influential than capacity alone. The findings support data-driven evaluation and coordinated planning of highway parking facilities.