A comprehensive model of vessel anchoring pressure based on machine learning to support the sustainable management of the marine environments of coastal cities

Abstract

The increased utilization of marine areas represents a significant challenge to the sustainable eco-environmental management of coastal cities. Machine learning, specifically the support-vector machine classification algorithm, was used to preprocess the massive Automatic identification System (AIS) dataset and extract anchoring vessels. Then, a comprehensive indicator evaluation model for anchoring pressure (CAPI) was constructed to evaluate the potential marine ecological pressure associated with anchoring vessels in the Bohai Sea. Spatial analysis was performed by geographic information system (GIS) to identify improper anchoring areas with high CAPI values. Finally, anchorage management in various coastal cities was assessed. The results showed that: (1) machine learning technology accurately identified anchoring vessels, (2) improper anchoring in the Bohai Sea is common, and (3) the management of anchoring activities is generally poor at boundaries between administrative regions. This study provides a rapid, feasible, and effective visualization method for marine environmental managers both theoretically and practically. The data mining method and CAPI model proposed here facilitate the management of vessel-related social issues in coastal cities, and they will help decision makers to quickly formulate targeted management measures to support the sustainable economic and environmental development of coastal cities.