A novel framework for identifying fishing grounds from AIS data containing vessels of unknown types

Changhai Huang ^1* Jialong Kong ¹ Jian Zheng ² Yuli Chen ¹ JINGEN ZHOU ^3*

• ¹ Merchant Marine College, Shanghai Maritime University, pudong, China
• ² College of Transport and Communications, Shanghai Maritime University, pudong, Shanghai, China
• ³ Kedge Business School, Talence, Aquitaine, France

To address the issue of precisely identifying fishing grounds in vast sea areas, this study proposes a framework that includes a fishing behavior detection model and a fishing ground identification model, considering vessels of unknown types. The absence of information regarding unknown vessels can result in incomplete identification of fishing grounds, which in turn leads to regulatory oversight, and these unidentified fishing areas might be hotspots for illegal fishing activities. Identifying these missing fishing grounds is crucial for enhancing regulatory efforts and for vessels go through these areas to plan their routes more effectively in advance. This helps in finding illegal fishing and optimizes the operational efficiency of fishing vessels. Firstly, the Speed-Direction-Based Stops and Moves of Trajectories (SDB-SMOT) algorithm is proposed. Based on this algorithm, a fishing behavior detection model is developed to identify fishing activity trajectories from AIS data that encompasses vessels of unknown types. Subsequently, an algorithm that integrates the Data Field and OPTICS (DF-OPTICS) algorithm is proposed, and a model for identifying fishing grounds is constructed based on the DF-OPTICS algorithm. The efficiency and effectiveness of this framework are validated by identifying fishing grounds from AIS data that contains both fishing vessels and vessels of unknown types in the South China Sea. The Davies-Bouldin Index of DF-OPTICS algorithm reached 0.267, 0.224, 0.203, the Silhouette Coefficient Index reached 0.560, 0.598, 0.633 and the Calinski-Harabasz Index reached 2213939, 3296101, 4320688 under three sets of hyperparameters. This framework not only bridges the gap in identifying fishing grounds from AIS data containing vessels of unknown types but also improves the efficiency of the fishing ground identification process.