文献类型：期刊文献

中文题名：南海万山群岛渔船漂移轨迹风漂系数的分段模拟

英文题名：Segmented Leeway Coefficient Simulation of Fishing Vessel Drift Trajectories in Wanshan Archipelago,South China Sea

作者：王海龙[1];仉天宇[1,2];张树钦[1];林汛[1];杨坚[1];张丽雯[1];曹倩[3];李萌[3]

机构：[1]广东海洋大学海洋与气象学院/近海海洋变化与灾害预警实验室/广东省高等学校陆架及深远海气候、资源与环境重点实验室,广东湛江524088;[2]自然资源部空间海洋遥感与应用重点实验室,北京100081;[3]山东省创新发展研究院,山东济南250101

年份：2025

卷号：45

期号：1

起止页码：134

中文期刊名：广东海洋大学学报

外文期刊名：Journal of Guangdong Ocean University

收录：北大核心2023、、北大核心

基金：国家重点研究发展计划(2021YFC3101801,2022YFC3103104);国家自然科学基金(42476219,41976200);南方海洋实验室项目(SML2022SP301);广东省教育厅创新团队项目(2023KCXTD015);广东省科技计划项目(粤西海洋野外站);广东海洋大学人才启动项目(060302032106);广东海洋大学大学生创新创业项目(S202410566018);山东省创新发展研究院智库项目。

语种：中文

中文关键词：渔船;漂流轨迹预测;AP98模型;拉格朗日粒子追踪方法;漂移动力学半解析模型;分段模拟

外文关键词：fishing vessel;drift trajectory prediction;AP98 model;Lagrangian particle tracking method;semi-analytical model for drift dynamics;segmented simulation

中文摘要：【目的】使用分段风漂系数模拟渔船的漂移轨迹,探究模拟预测的准确性,为准确预测渔船的漂流轨迹、提高海上搜救的效率和成功率提供基础。【方法】以南海万山群岛为研究区域,通过571个漂流样本建立AP98模型,并在此基础上进行分段模拟,采用风速4、6、8 m/s作为分界线建立不同风漂系数模型,同时建立漂移动力学半解析模型,通过对103个样本4组案例的拉格朗日粒子追踪模拟,对比预测模拟轨迹结果。【结果与结论】AP98模型线性回归结果表明,风速与风致漂移速度(L)和顺风向漂移速度(DWL)的决定系数介于0.200~0.800之间,与侧风向漂移速度(CWL)的决定系数介于0~0.100之间,说明风速与L和DWL之间线性关系明显,与CWL之间的线性关系不明显。不同分段处理的模拟结果表明,在案例1中,4和8 m/s的分段预测结果得到优化;在案例2中,4 m/s的分段预测结果得到优化;在案例3中,4和6 m/s的分段预测结果得到优化;在案例4中,4、6和8 m/s的分段预测结果均得以优化。分段模拟结果相比于未分段普遍有改善趋势,尤以风速4 m/s分段的模拟结果有较为明显的优化趋势,其他分段的模拟结果有好有差。合理选择分段点及确保数据分布的均衡性对提高模拟精度具有重要作用。

外文摘要：【Objective】By using segmented leeway coefficients,the study aims to accurately predict the drift trajectory of fishing vessel and improve the efficiency and success rate of maritime search and rescue.【Method】Taking Wanshan Archipelago in South China Sea as the research area,the AP98 model was established based on 571 drifting samples.Then segmented simulations were conducted on this basis,dividing the model into different leeway coefficients based on wind speeds of 4,6,and 8 m/s as the dividing lines.A semi-analytical model of drift dynamics was established at the same time,and the results of the predicted simulation trajectories were compared through Lagrangian particle tracking simulation of 103 samples in four groups.【Result and Conclusion】The linear regression results of the AP98 model indicate that the coefficient of determination(R2)between wind speed and wind-induced drift speed(L)as well as downwind drift speed(DWL)ranges between 0.200 and 0.800,whereas the coefficient of determination between wind speed and crosswind drift speed(CWL)ranges between 0 and 0.100.This suggests a significant linear relationship between wind speed and both L and DWL,but an insignificant linear relationship between wind speed and CWL.The simulation results from different segmentation treatments show that,in Case 1,the segmented predictions at 4 and 8 m/s were optimized;in Case 2,the segmented prediction at 4 m/s was optimized;in Case 3,the segmented predictions at 4 and 6 m/s were optimized;and in Case 4,the segmented predictions at 4,6,and 8 m/s were all optimized.It is evident that segmented simulations generally exhibit an improvement trend compared to non-segmented simulations,with the most noticeable optimization observed in the 4 m/s segmentation.The results of other segmentations vary,highlighting the importance of selecting appropriate segmentation points and ensuring balanced data distribution to enhance simulation accuracy.