上海海洋大学学报2024，Vol.33Issue(4)：1020-1030,11.DOI:10.12024/jsou.20240404512
中西太平洋金枪鱼延绳钓钓钩深度分布及其影响因素
Hook depth distribution and influencing factors of tuna longline fishing in Western and Central Pacific Ocean
摘要
Abstract
To comprehensively examine the hook depth distribution and influencing factors of tuna longline fishing in Western and Central Pacific Ocean,data collected aboard the vessel"Songhang"from August to October 2021 were analyzed.A combination of Support Vector Regression(SVR)and Generalized Additive Model(GAM)was employed to investigat the distribution of hook depths across different operational parameters and environmental variables.The objective was to elucidate the impact of each factor on hook depth.The results showed that:(1)Among the calculated values of hook depth,Hook No.1 had the smallest value at(130.03±3.32)m,while Hook No.8 had the largest value at(363.25±36.52)m.Regarding the measured values,Hook No.1 had the smallest depth at(130.81±17.94)m,whereas Hook No.7 exhibited the largest depth at(329.91±54.37)m.(2)The SVR model exhibited varying goodness of fit,with Hook No.1 displaying the lowest(R2=0.38)and Hook No.6 demonstrating the highest(R2=0.77)overall fit.The mean squared error(MSE)for training data was smallest for Hook No.6(0.138)and largest for Hook No.8(0.309),while for testing data,it was smallest for Hook No.4(0.086)and largest for Hook No.5(0.282).(3)GAM analysis reveals that water layer average velocities at 0-64 m and 128-192 m,as well as the casting speed,exhibit the highest correlations with hook depth across all hooks.Following these,water layer average velocities at 64-128 m,192-256 m,and 256-320 m,along with vessel speed,show significant correlations.However,water layer average velocity at 320-384 m and wind speed demonstrate no significant correlations with hook depth.(4)The average differences between predicted hook depths by SVR and GAM models and the actual measured values are 1.75％and 6.38％,respectively.The predictions from the SVR model align well with the measured values,showing consistent trends with the variation in hook positions,thus demonstrating good agreement.The results of this study will contribute to understanding the distribution pattern of tuna longline fishing hooks in operation,and provide a basic reference for optimizing the performance of longline.关键词
中西太平洋/金枪鱼延绳钓/钓钩深度/支持向量回归机模型/广义加性模型Key words
Western and Central Pacific Ocean/tuna longline fishing/hook depth/support vector machine/general additive model分类
农业科技引用本文复制引用
刘志强,郭绍健,王禹程,周成,吴峰,万荣..中西太平洋金枪鱼延绳钓钓钩深度分布及其影响因素[J].上海海洋大学学报,2024,33(4):1020-1030,11.基金项目
国家重点研发计划(2023YFD2401301,2023YFD2401305) （2023YFD2401301,2023YFD2401305）
农业农村部全球渔业资源调查监测评估(公海渔业资源综合科学调查)专项(D-8025-23-1003) （公海渔业资源综合科学调查）
