Port pollution prediction and management via multi-view intelligent computing: A case study of Tianjin Port

doi:10.1016/j.rsma.2026.104805

CSpace

	Port pollution prediction and management via multi-view intelligent computing: A case study of Tianjin Port
	Xue, Tong 1; Li, Yong 1; Mei, Qiang 2,3; Bai, Yundi 1; Yang, Yang 4; Cui, Lei 2; Wang, Peng 5; Zhang, Beibei 6; Wang, Shaohua 7
	2026-03-01
发表期刊	REGIONAL STUDIES IN MARINE SCIENCE
ISSN	2352-4855
卷号	95 页码:27
摘要	Ship emissions exacerbate air pollution in ports, and their assessment and control have attracted widespread attention from both government and academia. To enable scientific tracing and management of port-related ship pollutants, this study proposes an multi-view intelligent spatiotemporal computing framework that integrates emission inventory construction and emission forecasting. The emission inventory is constructed based on AIS data, combined with vessel activity parameters and emission factor models, covering channels, anchorages, and berths to achieve high spatiotemporal resolution. In the prediction module, Transformer-, MLP-, TCN-, and RNN-based models are employed, and CEEMDAN is introduced for multi-scale decomposition to address the non-stationary nature of emission sequences. Experimental results for Tianjin Port from 2015 to 2018 show that CO2, SOX, and NOX emissions increased from 8.66 x 105 t, 1.28 x 104 t, and 6.6 x 103 t in 2015 to 1.35 x 106 t, 1.74 x 104 t, and 9.6 x 103 t in 2018, respectively. Spatially, the Xingang main channel and Beijiang Port area are emission hotspots. Among vessel types, oil tankers, bulk carriers, and container ships collectively contribute over 90% of total emissions. Source analysis indicates that main engine emissions in channels exceeded 70%, while berths and anchorages together contributed over 90%. Comparative prediction results demonstrate that CEEMDAN decomposition enhances the fine-grained representation of emission forecasts across all model types. The hybrid model SCINet_D, which integrates the strengths of SCINet and DLinear, exhibits relatively favorable predictive performance. The proposed spatiotemporal computing framework enables a coordinated analysis of emission inventory construction and trend prediction, providing scientific support for refined port air quality management, emission reduction strategies, and evaluation of domestic Emission Control Area (DECA) policies.
关键词	Pollutants Greenhouse gases Emission inventory Tianjin port AIS data Time series forecasting
DOI	10.1016/j.rsma.2026.104805
收录类别	SCI
语种	英语
WOS研究方向	Environmental Sciences & Ecology ; Marine & Freshwater Biology
WOS类目	Ecology ; Marine & Freshwater Biology
WOS记录号	WOS:001686436700001
出版者	ELSEVIER
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/42824
专题	中国科学院计算技术研究所
通讯作者	Mei, Qiang; Wang, Shaohua
作者单位	1.Beijing Univ Technol, Coll Comp Sci, Beijing 100124, Peoples R China 2.Jimei Univ, Nav Coll, Xiamen 361021, Peoples R China 3.Shanghai Maritime Univ, Merchant Marine Coll, Shanghai 201306, Peoples R China 4.East China Normal Univ, Sch Geog Sci, Shanghai 200241, Peoples R China 5.Chinese Acad Sci, Inst Comp Technol, Beijing 100086, Peoples R China 6.Zhengzhou Univ, Sch Cyber Sci & Engn, Zhengzhou 450001, Peoples R China 7.Aerosp Informat Res Inst, Chinese Acad Sci, State Key Lab Remote Sensing & Digital Earth, Beijing 100094, Peoples R China
推荐引用方式 GB/T 7714	Xue, Tong,Li, Yong,Mei, Qiang,et al. Port pollution prediction and management via multi-view intelligent computing: A case study of Tianjin Port[J]. REGIONAL STUDIES IN MARINE SCIENCE,2026,95:27.
APA	Xue, Tong.,Li, Yong.,Mei, Qiang.,Bai, Yundi.,Yang, Yang.,...&Wang, Shaohua.(2026).Port pollution prediction and management via multi-view intelligent computing: A case study of Tianjin Port.REGIONAL STUDIES IN MARINE SCIENCE,95,27.
MLA	Xue, Tong,et al."Port pollution prediction and management via multi-view intelligent computing: A case study of Tianjin Port".REGIONAL STUDIES IN MARINE SCIENCE 95(2026):27.