基于体散射函数及吸收系数的南海水体漫射衰减系数研究  

Studying on diffuse attenuation coefficient in the South China Sea based on volume scattering function and absorption coefficient

作　　者：张现清 李彩[1,3] 周雯 刘聪[1] 许占堂 曹文熙 杨跃忠[1,3] ZHANG Xianqing;LI Cai;Zhou Wen;LIU Cong;XU Zhantang;CAO Wenxi;YANG Yuezhong(State Key Laboratory of Tropical Oceanography(South China Sea Institute of Oceanology,Chinese Academy of Sciences),Guangzhou 510301,China;University of Chinese Academy of Sciences,Beijing 100049,China;Southern Marine Science and Engineering Guangdong Laboratory,Guangzhou 511458,China)

机构地区：[1]热带海洋环境重点实验室(中国科学院南海海洋研究所),广东广州510300 [2]中国科学院大学,北京100049 [3]南方海洋科学与工程广东省实验室,广东广州511458

出　　处：《热带海洋学报》2023年第3期86-95,共10页Journal of Tropical Oceanography

基　　金：国家自然科学基金(41976181、41976172、41976170);广州市科技计划重点项目(201707020023);南方海洋科学与工程广东省实验室(广州)人才团队引进重大专项项目(GML2019ZD0305);热带海洋环境国家重点实验室自主研究项目(LTOZZ1602)。

摘　　要：漫射衰减系数K_(d)(z,λ)是估算水下光场及水色要素剖面分布、研究浮游植物光合作用及赤潮灾害预警方法的重要参数,它是一个准固有光学特性参数,是波长λ和剖面深度z的函数,除与水体吸收、散射或后向散射有关外,对归一化的水体体散射函数即散射相函数的角度分布极为敏感。本文基于广角体散射函数测量仪(volume scattering and attenuation meter,VSAM)、吸收衰减系数测量仪ac-9和ac-s以及海洋光学剖面仪ProfilerⅡOCI/R-200I和Hyper ProⅡ在南海海域实测数据,利用Light GBM、随机森林(random forest,RF)、Cat Boost三种高效机器学习方法,首次构建了基于体散射函数β(ψ)、吸收系数a及对应剖面深度z的漫射衰减系数K_(d)(650)剖面分布估算模型,并综合R^(2)、RMSE、MAPE以及估算与实测数据的对比进行模型评价,结果表明,三种机器学习模型中,Cat Boost模型的R2和RMSE分别为0.8534和0.0472m^(-1),均优于RF和Light GBM;Cat Boost模型的MAPE为11.0585%,低于RF模型但略高于Light GBM模型;通过对比估算和实测结果发现,Cat Boost模型估算结果与实测结果最为相近,是K_(d)(650)最优估算模型。利用Cat Boost模型,结合实测体散射函数β(ψ)、吸收系数a及其相应剖面深度z,对南海北部多个站点15m以浅K_(d)(650)的剖面分布估算表明,上述站点K_(d)(650)在5、10、15m三个水层变化范围为0.275~0.7m^(-1),5m水层的K_(d)(650)较为平稳,10与15m水层K_(d)(650)跨度较大。本研究方法考虑了多角度体散射函数分布对漫射衰减系数的贡献,为基于固有光学特性参数估算K_(d)(z,λ)提供了新方法思路。Diffuse attenuation coefficient of downwelling irradiance K_(d)(z,λ)is an important parameter for estimating the profile distribution of underwater light filed and water constituents,and studying the photosynthesis of the phytoplankton and warning method of harmful algae bloom.K_(d)(z,λ)is a“quasi-inherent”optical property as a function of wavelength𝜆and depth𝑧.Not only is it sensitive to absorption and scattering/backscattering coefficient,but also sensitive to the angular distribution of the normalized volume scattering function(i.e.,scattering phase function).In this study,based on the volume scattering function[VSF,β(ψ,z)]in seven directions determined with a custom in situ device called VSAM(volume scattering and attenuation meter),the absorption coefficient a(z)determined with the WET Labs ac9 and ac-s,and the downwelling irradiance Ed(z)determined with the Satlantic ProfilerⅡOCI/R-200 and HyperProⅡin the north South China Sea(SCS)with a broad range,using LightGBM,Random Forest(RF)and CatBoost,three machine learning models for estimating the profile distribution of K_(d)(z,650)were developed at first,and they were then evaluated by the key indicators including R^(2)、RMSE、MAPE,as well as the comparison between in situ measured K_(d)(650)and estimated K_(d)(650).The evaluation indicated that the CatBoost model performed the best with R2 of 0.8534,RMSE of 0.0472 m^(-1),MAPE of 11.0585%,and the estimated K_(d)(650)was also closest to the measured K_(d)(650).Using the established CatBoost model,input inherent optical properties(IOPs)were the absorption coefficient,the volume scattering function(VSF),and their profile depth,the K_(d)(650)profile distribution among 15 m in the north SCS was estimated.The result shows that K_(d)(650)varies from 0.275 to 0.7 m^(-1) at 5,10 and 15 m underwater.At 5 m,K_(d)(650)is relatively stable while it varies greatly at 10 and 15 m.The contribution of volume scattering function distribution to K_(d)(z,λ)is considered in this study,which provides a new idea

关 键 词：漫射衰减系数K_(d)(650) 体散射函数 吸收系数 机器学习 

分 类 号：P731.14[天文地球—海洋科学]