空间插值法在北部湾渔业资源密度评估中的应用

Application of a spatial interpolation method for the assessment of fishery resources in the Beibu Gulf

为弥补采样过程中可能存在的缺陷,进而获得更加合理的渔业资源密度分布情况,本研究将空间插值方法与渔业声学技术相结合,利用Echoview渔业声学数据处理系统(Version 6.1)对2014年7月至8月间的北部湾渔业声学调查数据进行预分析,再对实际探测数据进行反距离权重插值(IDW)、普通克里格插值(OK)、简单克里格插值(SK)。通过计算对比几种插值模型的决定系数(R^2)、平均误差(ME)、均方根误差(RMSE)等,筛选最优插值模型。结果表明,鱼类密度总体分布较均匀,相对集中的区域较少,分别在中东部沿岸区域、中越共同渔区中西部区域和北部远岸海域,而在中南部远岸海域分布稀少;比较3种插值参数发现,R^2和RMSE显示的优劣顺序相似,OK>SK>IDW,ME显示的优劣顺序为OK>IDW>SK,说明在该区域进行的渔业声学资源密度评估中,普通克里格插值(OK)能够更好的预测北部湾海域的资源密度分布情况。

As one of the most important inshore fishery waters, the Beibu Gulf, is attracting increasing attention with regard to the density of fishery resources and resource changes. The fishery acoustic technology is increas- ingly being applied for fishery resource assessment, because of its advantages such as rapid, continuous, greater sampling rate, and rich spatio-temporal data. Nonetheless, this technology is influenced by some factors such as weather, sea conditions, and survey route; hence, the echo data is unreliable. In order to overcome the possible defects that exist in the sampling process, and to obtain a more reasonable density distribution of fishery resources, we combined the spatial interpolation methods with the fishery acoustic technology. We conducted a preliminary analysis of the Beibu Gulf fishery acoustic survey data obtained during July to August of 2014 by using Echoview （a fishery acoustic survey data processing system, Version 6.1）. Next, Inverse Distance Weighted （IDW）, Ordinary Kriging （OK）, and Simple Kriging （SK） were applied to analyze the actual data after logarithmic transformation of the practical detection data, in order to obtain normally distributed data. By calculating and comparing the deter- mination coefficients （R2）, mean error （ME）, and root mean square error （RMSE） of the three interpolation models （IDW, OK, and SK）, we selected the optimal interpolation model fitting the fishery acoustic survey on the Beibu Gulf. The results showed that the distribution of fish density was relatively uniform, which indicated relatively less concentrated distribution in the survey area. Further, fish were mainly concentrated in the mideast coastal area and the midwestern region of the common fishing zone. In contrast, the distribution of fish in the south-central offshore waters was rare. By comparing the three interpolation parameters, we found that R2 and RMSE showed a good precedence similarity in the following order： OK〉SK〉IDW. In contrast, ME showed the order of OK〉IDW〉SK. Our findings suggest that OK could be a better predictor of resource density distribution of the Beibu Gulf waters for the evaluation of fishery acoustic resource density of the area and is useful for potential application in the combination analysis of fishery acoustic technology and spatial interpolation.