Truck trip and long-haul truck tour identification based on national-level truck GPS data

Truck Global Positioning System (GPS) data provide continuous location observations for a large sample of trucks and thus are valuable inputs for truck travel demand estimation. However, truck GPS data only record where and when each truck was observed instead of the actual loading and unloading stays, maintenance stops, and rest breaks. In particular, long-haul truck depots and tours lack sufficient investigation for a coherent estimation of long-term large-scale truck travel demand. To facilitate the GPS data-driven truck travel demand estimation, this study proposes a comprehensive framework of both heuristic rules and data-driven methods for imputing trips using spatial-temporal features of the GPS pings, constructing trip chains using truck stay features and point-of-interest data, and revealing tours for four types of long-haul trucks based on the existence of fixed depots. The framework is proven efficient when it uses a one-month nationwide truck GPS dataset in the United States.