全文下载： 
202302016.pdf
文章编号： 1672-6987（2023）02-0116-11； DOI： 10.16351/j.1672-6987.2023.02.016
汤贵艳， 朱善良*， 周伟峰， 杨树国(青岛科技大学 数理学院；数学与交叉科学研究院;青岛市人工智能海洋技术创新中心，山东 青岛 266061)
摘要： 海洋次表层的热力结构对于海洋环流和全球气候变化具有重要的意义。提出一种新的融合高斯混合模型(gaussion mixture model, GMM)和轻量级梯度提升机(light gradient boosting machine, LightGBM)算法的海洋次表层温度(ocean subsurface temperature, OST)反演模型，利用海表温度(sea surface temperature, SST)、海表盐度(sea surface salinity, SSS)、海表高度(sea surface height, SSH)、海表风场(sea surface wind, SSW)的水平分量(USSW)和垂直分量(VSSW)等多源海表参数对印度洋海域的次表层热力结构进行反演，并采用均方根误差和决定系数对模型进行验证。结果表明:所提出的模型可以准确反演印度洋海域的OST分布特征和季节变化规律。在此基础上，设计了不同海表参数输入组合的3种对比实验来定量分析不同海表参数对LightGBM模型的影响。结果表明:所有海表参数对模型都有积极作用，但5个输入参数(SST、SSS、SSH、USSW和VSSW)的LightGBM模型反演效果最好，3个输入参数(SST、SSS和SSH)和2个输入参数(SST和SSH)的LightGBM模型次之。另外，与已有的极限梯度增强(extreme gradient boosting, XGBoost)反演模型相比，5个输入参数的LightGBM模型具有更好的模拟能力。
关键词： 高斯混合模型； 轻量级梯度提升机； 机器学习； 海洋次表层温度
中图分类号： P 732文献标志码： A
引用格式： 汤贵艳，朱善良，周伟峰，等.基于高斯混合聚类和LightGBM算法的印度洋次表层温度反演研究［J］. 青岛科技大学学报(自然科学版), 2023, 44(2): 116-126.
TANG Guiyan, ZHU Shanliang, ZHOU Weifeng, et al. Estimation of Indian Ocean subsurface thermal structure based on Gaussion mixture clustering and LightGBM algorithm［J］. Journal of Qingdao University of Science and Technology（Natural Science Edition), 2023， 44（2）： 116-126.
Estimation of Indian Ocean Subsurface Thermal Structure Based
on Gaussion Mixture Clustering and LightGBM Algorithm
TANG Guiyan, ZHU Shanliang, ZHOU Weifeng, YANG Shuguo
(College of Mathematics and Physics; Research Institute for Mathematics and Interdisciplinary Sciences； Qingdao Innovation Center
of Artifical Intelligence Ocean Technology, Qingdao University of Science and Technology, Qingdao 266061, China)
Abstract: The thermal structure of the ocean subsurface is of great significance to ocean circulation and global climate change. In this paper, a new ocean subsurface temperature (OST) estimation model combining the Gaussian mixture model (GMM) and light gradient boosting machine (LightGBM) algorithm. The model uses multisource sea surface parameters including sea surface temperature (SST), sea surface salinity (SSS), sea surface height (SSH), northward and eastward components of sea surface wind (USSW and VSSW) to retrieve the OST of the Indian Ocean. Moreover, the root mean square error and coefficients of determination are employed to assess the performance of the model. The results show that the model proposed in this paper can accurately reflect the distribution characteristics and seasonal variation of the OST in the Indian Ocean. On this basis, three comparative experiments with different input combinations of sea surface parameters were designed to quantitatively analyze the influence of different input variables on the LightGBM model. The experimental results show that all sea surface parameters have a positive effect on the model. Moreover, the LightGBM model with five input parameters (SST, SSS, SSH, USSW, VSSW) has the best estimation effect, followed by the LightGBM model with three input parameters (SST, SSS, SSH) and two input parameters (SST, SSH). In addition, compared with the existing eXtreme gradient boosting (XGBoost) estimation model, the LightGBM model with five input parameters has better simulation capabilities.
Key words: Gaussian mixture model; light gradient boosting machine; machine learning; ocean subsurface thermal
收稿日期： 2022-05-11
基金项目： 中国科学院海洋环流与波动重点实验室开放基金项目(KLOCW2003).
作者简介： 汤贵艳(1995—), 女, 硕士研究生. *通信联系人.
