Port ship congestion and Port-oriented cities air pollution: the role of machine learning models in transportation environmental governance

Abstract

Port-oriented cities worldwide are facing significant challenges due to port congestion and environmental concerns. However, research quantifying the relationship between port congestion and air pollution in port towns is limited. This study used deep learning predictive models to examine the influence of port congestion on particulate matter (PM) levels within port-oriented cities. The study centered on Shanghai, a major Chinese port city, analyzing 30,590 records over six years (January 1, 2017, to December 30, 2022) on PM concentrations, meteorological conditions, and port congestion. This study evaluated three deep learning models (LSTM, BILSTM, and CNN-LSTM) for long-term time series forecasting using two datasets: one with air pollutants and meteorological data, and another adding port congestion data. Performance was assessed using MAE, MSE, and RMSE metrics. The results show that the CNN-LSTM models exhibit the best prediction performance and all models improve when port congestion data is included. This indicates that air pollution in port-oriented cities is influenced by port congestion dynamics. Specifically, This study elucidates the intricate relationship between port congestion and air pollution in port-oriented cities through machine learning modeling. These findings offer significant decision-making assistance for shipping businesses and policymakers regarding port-oriented cities strategic planning and environmental risk management.