西北太平洋渔业资源调查采样设计比较

为兼顾调查成本和调查效果并实现多目标远洋渔业资源调查，本研究根据2021—2023年西北太平洋渔业资源调查数据，以澳洲鲐、日本鲭和远东拟沙丁鱼的相对资源量以及总相对资源量为调查目标，使用Kriging插值法模拟了调查目标的空间分布，利用计算机模拟方法比较了简单随机抽样（Design1）、系统抽样（Design2）、基于层面积分配方法的分层随机抽样（Design3）和基于层资源量分配方法的分层随机抽样（Design4）4种采样设计对调查目标的评估效果，以相对误差（Relative estimation error, REE）和相对偏差（Relative bias, RB）衡量评估效果的精度，采用准确度变化率（Accuracy change rate, ACR）对调查站点数进行选择。结果显示，4种采样设计对调查目标的评估精度不同，表现为Design1 < Design2 < Design3 < Design4。Design1、Design3和Design4的REE值随站点数量增加而降低，Design2的REE值随站点数量增加波动较大。Design4对日本鲭、远东拟沙丁鱼和总相对资源量的评估精度更高，但高估了日本鲭相对资源量。当采用Design4对进行模拟采样时，综合考虑所有调查目标，28个站位为最适调查站位数量。结果表明，分层随机抽样（Design4）可优先作为西北太平洋渔业资源调查采样设计方案。

A Comparative Study of Different Sampling Designs for Fishery Resources survey in the Northwest Pacific Ocean

This study aimed to develop a multi-objective distant-water fisheries survey scheme that balances survey cost and effectiveness. Based on the survey data from 2021 to 2023 in the Northwest Pacific, the relative abundance distribution of Scomber australasicus, Scomber japonicus, Sardinops sagax was simulated using Kriging interpolation, and assume relative resource as their ‘true’ values. The effectiveness of the Simple random sampling (Design1), Systematic sampling (Design2), Stratified random sampling based on stratum area size allocation method (Design3), and Stratified random sampling based on fishery resource biomass allocation method (Design4) was assessed through computer simulation. Two indicators, Relative estimation error (REE) and Relative bias (RB) were used to evaluate the accuracy of the assessment effects, and the number of sampling stations was selected using the Accuracy change rate (ACR). The results showed that the estimation accuracy of the four sampling designs were different. In general, Design4 had the highest accuracy, followed by Design3, Design2 and Design1. The REE values of target species abundance index estimation of Design2 fluctuated with the increasing number of sampling stations. Design4 exhibited higher accuracy in estimating the abundance index of S. japonicus, S. sagax and the total fishery resources. However, Design4 demonstrated an overestimation in estimating the abundance index of Chub mackerel. Using Design4 for simulated sampling, and considering all objectives, the optimal number of sampling stations could be set at 28. In conclusion, the Stratified sampling method (Design4) can be prioritized as the survey station design approach for the Northwest Pacific fishery resource survey.