A near-real-time global ship emission inventory model with daily-updated technical database

Near-real-time emission inventories are essential for accurate modeling of air quality and climate impacts. While advances in transportation big data have enabled high-resolution emission inventories in various sectors, existing maritime models still face two key limitations. On one hand, ship technical parameters are updated much slower than activity data, leading to uncertainty in emission calculation. On the other hand, maintaining a dynamically updated inventory platform remains challenging. Here, to address these gaps, this study develops a near-real-time global ship emission inventory model with a daily-updated ship technical database​ and a scrubber-use simulation module. Our model identifies newly added ships from daily Automatic Identification System (AIS) data and predicts their deadweight tonnage and engine power using an XGBoost-based regression model, enabling more accurate emission factor matching. For sulfur-dioxide and particular matter emission calculation, both low-sulfur fuel use and scrubber use are considered as the potential choices of ships. Compared with models with daily-updated technical database, failure to supplement newly-identified ships results in an underestimation of approximately 12.9% to 16.1% in daily ship emissions. Results from our improved model show that from 2022 to 2024, global annual ship emissions experienced steady growth from 835 million tons to 872 million tons, consistent with the increase in maritime trade. The model also captured the abrupt decline by 40% in daily emissions in the Red Sea following the Red Sea crisis in December 2023, alongside the corresponding rise in the Indian Ocean and South Atlantic Ocean due to rerouted container ships.