| 摘要: |
| 为了解东海北部海域三疣梭子蟹（Portunus trituberculatus）资源时空分布规律，探索更适合三疣梭子蟹资源量预测的模型方法，根据2006-2007年共四个季度在东海北部海域的底拖网调查数据，运用梯度提升回归树（gradient boosting regression tree，GBRT）和支持向量机（support vector machine，SVM）这两种机器学习方法，分析了三疣梭子蟹时空分布与环境因子之间的关系，同时使用方差解释率（VE）、相对均方根误差（RMSE）以及决定系数R2等指标对不同模型的拟合效果、预测性能以及稳定性等进行了比较，选择其中最佳模型对东海北部海域三疣梭子蟹资源分布进行预测。结果显示，GBRT模型的拟合效果相对优于SVM模型，两种模型的拟合结果均显示底层海水盐度（SBS）为影响三疣梭子蟹资源分布最为显著的环境因子。GBRT模型的预测性能较高且模型较为稳定，其预测结果显示夏季的资源量高于其他三个季节，且各季节所研究海域的东南部均存在一个资源分布的低值区。研究结果预期可为三疣梭子蟹资源分布及资源量预测新方法的探索和分析提供技术指导。 |
| 关键词: 三疣梭子蟹 梯度提升回归树(GBRT) 支持向量机(SVM) 资源量 |
| DOI：10.11693/hyhz20210200050 |
| 分类号:S931.1;S932.5 |
| 基金项目:国家重点研究发展计划，2017YFA0604902号；浙江省基础公益计划项目，LGN21C190009号。 |
|
| SPATIOTEMPORAL DISTRIBUTION OF PORTUNUS TRITUBERCULATUS IN THE NORTHERN EAST CHINA SEA BASED ON TWO MACHINE LEARNING METHODS |
|
LI Xiao-Dong, WANG Jing, YANG Chun-Hui, WANG Ying-Bin
|
|
School of Fishery, Zhejiang Ocean University, Zhoushan 316022, China
|
| Abstract: |
| To understand the temporal and spatial distribution of Portunus trituberculatus resources in the northern East China Sea, and to explore a more suitable model for the prediction of P. trituberculatus resources, two machine learning methods:gradient boosting regression tree (GBRT) and support vector machine (SVM), were used to analyze the relationship between spatiotemporal distribution of P. trituberculatus and environmental factors based on the survey data in the northern East China Sea from 2006 to 2007. The fitting effect, predictive performance, and stability of the two models were compared in the variance explained (VE), the root mean square error (RMSE), and the coefficient of determination (R2). The optimal model was selected to predict the distribution of P. trituberculatus in the northern East China Sea. Results show that the fitting effect of GBRT model was better than that of SVM model, and the bottom seawater salinity (SBS) was the most significant environmental factor affecting the distribution of P. trituberculatus. The predictive performance and stability of GBRT model were better than those of SVM. The predictive results show that the abundance of P. trituberculatus in summer was higher than those in the other three seasons, and there was a small abundance area in the southeast of the studied sea area in each season. This study provided a technical tool for exploring new methods of prediction for the resource distribution and abundance of P. trituberculatus. |
| Key words: Portunus trituberculatus gradient boosting regression tree (GBRT) support vector machine (SVM) resources |
