| 基于船载走航观测高光谱颗粒物吸收系数反演浮游植物粒级结构* |
| |
| 引用本文： | 王桂芬,徐文龙,周雯,许占堂,曹文熙.基于船载走航观测高光谱颗粒物吸收系数反演浮游植物粒级结构*[J].热带海洋学报,2018,37(5):50-61. |
| |
| 作者姓名： | 王桂芬 徐文龙 周雯 许占堂 曹文熙 |
| |
| 作者单位： | 1. 河海大学海洋学院, 江苏 南京 2100982. 热带海洋环境国家重点实验室(中国科学院南海海洋研究所), 广东 广州 5103013. 中国科学院大学, 北京100049; |
| |
| 基金项目： | 国家自然科学基金(41776045, 41576030, 41776044);中央高校基本科研业务费专项资金(2017B06714);热带海洋环境国家重点实验室自主研究项目(LTOZZ1602);广州市科技计划项目(201607020041) |
| |
| 摘 要： | 利用高光谱监测数据反演浮游植物种群组成是当前海洋光学和水色遥感的研究热点。文章采用大西洋经向断面航次中走航式观测系统测量的海水总颗粒物吸收光谱数据, 尝试建立了两种模型对浮游植物粒级结构(Phytoplankton size class, 简称PSC)进行反演和比较讨论。一类模型是基于总颗粒物吸收光谱高斯分解获得的典型波段高斯带强度与色素浓度之间的关系, 建立了偏最小二乘回归模型(Partial Least Squares regression model, 简称PLS回归模型); 另一类模型是采用长波波段吸收基线高度推算海水总叶绿素a浓度, 进而根据Brewin等(2010)生物量算法推算PSC的三组分模型(简称三组分模型)。模型比较验证结果显示, 两类模型对海水总叶绿素浓度的反演都有较高的精度, 相对偏差ME在15%左右; 对于三个粒级浮游植物对应的叶绿素浓度(Pico级Cp, Nano级Cn, Micro级Cm)的反演效果也相当, PLS回归模型反演的ME分别为28.4%、31.9%和41%, 三组分模型反演的ME分别为31%、35.9%、37.7%。研究结果初步表明了采用高光谱吸收系数反演浮游植物种群结构的潜在优势, 可为不同海域走航式高光谱观测系统的推广应用提供思路。
|
| 关 键 词： | 浮游植物粒级结构 高光谱 颗粒物吸收系数 船载走航观测 |
| 收稿时间： | 2018-05-03 |
| 修稿时间： | 2018-06-14 |
Retrieving phytoplankton size class from hyperspectral particulate absorption data measured by the ship-based underway flow-through system |
| |
| Guifen WANG,Wenlong XU,Wen ZHOU,Zhantang XU,Wenxi CAO.Retrieving phytoplankton size class from hyperspectral particulate absorption data measured by the ship-based underway flow-through system[J].Journal of Tropical Oceanography,2018,37(5):50-61. |
| |
| Authors: | Guifen WANG Wenlong XU Wen ZHOU Zhantang XU Wenxi CAO |
| |
| Institution: | 1. College of Oceanography, Hohai University, Nanjing 210098, China2. State Key Laboratory of Tropical Oceanography (South China Sea Institute of Oceanology, Chinese Academy of Sciences), Guangzhou 510301, China3. University of Chinese Academy of Sciences, Beijing 100049, China; |
| |
| Abstract: | Retrieving of phytoplankton groups from hyperspectral optical data has been a hot topic in recent years for studies of marine optics and ocean color remote sensing. In this paper, based on the hyperspectral particulate absorption data (ap(λ)) measured by the ship’s flow-through system during the Atlantic Meridional Transect (AMT) cruise, two models were proposed and compared for estimating phytoplankton size class (PSC) from the ap(λ) spectra. An algorithm that defines phytoplankton pigment absorption as a sum of Gaussian functions was used for decomposing the total particulate absorption, and a partial least squares model (PLS_model) was developed for estimating PSC using the Gaussian magnitudes at specific wavelengths. Another model (3component_model) is a modified abundance-based model proposed by Brewin et al (2010), in which the total chlorophyll-a concentration (C, mg·m-3) was estimated from the absorption line height at 676 nm. Results showed that both models give good results for retrieving PSC from the particulate absorption spectra, with the mean relative errors (ME) for estimating total C in the range of 15%. ME values for estimating the size-specific chlorophyll a concentration (Cp for pico-, Cn for nano- and Cm for micro-plankton) are 28.4%, 31.9% and 41% for the PLS_model, and are 31%, 35.9% and 37.7% for the 3component_model. Our results confirmed the potential advantages of hyperspectral optical measurements for retrieving phytoplankton groups, which also provided much insights for the application of ship-based underway flow-through system in different sea areas. |
| |
| Keywords: | phytoplankton size class hyperspectral data particulate absorption coefficient ship-based underway measurements |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《热带海洋学报》浏览原始摘要信息 |
| 点击此处可从《热带海洋学报》下载免费的PDF全文 |
|
