基于船位数据的西南大西洋公海鱿钓渔船作业状态特征提取和分析OA

To investigate the operational state characteristics and spatiotemporal distribution patterns of squid jigging vessels targeting Argentine shortfin squid(Illex argentinus)in the Southwest Atlantic,vessel monitoring system(VMS)data from 10 representative squid jigging vessels from December 2023 to June 2024 were selected.Using a Gaussian mixture model and a multi-layer filtering method,the operational state of the vessels was classified into four types:fishing,drifting,sailing,and storm avoidance.The VMS-derived operational states were then calibrated using fishing logbook data.The results showed that the fishing duration and daily catch extracted from three data sources(VMS,VMS corrected by logbooks,and fishing logbooks)all exhibited high linear fitting degrees(r values of 0.27,0.30,and 0.25,respectively).Moreover,on over 92%of the operational days,the difference between the fishing positions derived from VMS and those recorded in logbooks was less than 15 nmile,validating the reliability of extracting operational state from VMS data.The peak fishing season for Illex argentinus in the Southwest Atlantic in 2024 was from February to May,with the fishing grounds located in the high seas outside Argentina's exclusive economic zone(60°W-61°W,45°S-47°S and 57°W-59°W,41°S-43°S).The prediction of the four operational states of squid jigging vessels in this area using an attention-based deep neural network(ADNN)model achieved an average accuracy of 95.2%.Notably,the accuracy for storm avoidance reached as high as 99%,while the accuracies for sailing,fishing,and drifting all exceeded 93%.This approach can provide technical support for dynamic monitoring of vessel behavior and fisheries resource management.