Quantifying the contribution of ship emissions to black carbon pollution along the coast of the East China sea using machine learning approach

Abstract

Ship emissions are a significant source of atmospheric black carbon (BC) aerosols, impacting both human health and climate, particularly in nearshore marine environments. However, the influence of ship activity on BC pollution and its extent remains poorly constrained. In this study, we quantified the contribution of ship emissions to BC pollution along the East China Sea coast near the Ningbo-Zhoushan harbor during the fishing season. This was achieved using an Extreme Gradient Boosting (XGBoost) model combined with ship activity data obtained from Automatic Identification System (AIS) signals. Our findings reveal that ship speed and ship count are critical factors influencing BC mass variability, alongside seasonal changes in emission intensity (reflected by Unix time) and meteorological factors. Increased ship speeds, particularly for fishing vessels (over 6 knots), and higher ship counts were generally associated with elevated BC mass concentrations. By training the XGBoost model on data from non-ship periods, we effectively separated BC contributions from ship emissions during the entire observation period. Ship emissions were found to contribute 10 %–60 % of the total BC mass, exhibiting a characteristic diurnal pattern with higher contributions in the morning and evening rush hour, corresponding to ship departures and returns at the port. These results underline the potential of machine learning approaches for evaluating ship emissions and highlight the significant role of ship activity in BC pollution in the nearshore environments of the East China Sea.