Journal of Tropical Oceanography >
Retrieval of diffuse attenuation coefficient in high frequency red tide area of the East China Sea based on buoy observation
Copy editor: YIN Bo
Received date: 2019-09-09
Request revised date: 2019-12-11
Online published: 2020-01-15
Supported by
National Natural Science Foundation of China(41576030)
National Natural Science Foundation of China(41776044)
National Natural Science Foundation of China(41776045)
National Natural Science Foundation of China(41976172)
Science and Technology Planning Project of Guangdong Province of China(201607020041)
Key Special Project for Introduced Talents Team of Southern Marine Science and Engineering Guangdong Laboratory(Guangzhou)(GML2019ZD0305)
Copyright
The optical data measured by a buoy, with long time series and high temporal resolution, can be reliably used for obtaining the rapidly changing diffuse attenuation coefficient (Kd). The biomass of phytoplankton and the concentration of suspended sediment vary widely in the high-incidence red tide area of the East China Sea, resulting in complex changes in optical properties. In this article, the spectrum data collected by a marine optical buoy from September 2013 to January 2014 were used to obtain the apparent optical characteristics of the sea area, then an empirical algorithm of Kd(490) was established based on the correlation between Kd(490) and remote sensing reflectance (Rrs(λ)), and compared with seven kinds of existing algorithms. The results indicated that Kd(λ) and Rrs(λ) present significant features of class Ⅱ water body spectrum, Kd(490) varies from 0.01 m-1 to 4.31 m-1, and the turbidity also varies greatly. According to the good correlation of Kd(490) and Rrs band ratio, a dual-band ratio empirical algorithm was established, taking Rrs(650) / Rrs(510) and Rrs(555) / Rrs(510) as independent variables. New algorithm is superior to the other seven algorithms, the root mean square error, absolute percentage difference and coefficient of correlation coefficient are 0.27 m -1, 27.08 % and 0.77, respectively, between the new algorithm inversion Kd(490) and the measured values. The improvement of the accuracy of the algorithm is due to the fact that the Rrs selected by the new algorithm can fully reflect water body information and adapt to the changes of water composition in this sea area. This study provides a better choice for the inversion of the high-incidence red tide area in the East China Sea, and an example for the application of marine optical buoy in water environment monitoring.
ZHANG Yu , WANG Guifen , XU Zhantang , Yang Yuezhong , ZHOU Wen , ZHENG Wendi , ZENG Kai , DENG Lin . Retrieval of diffuse attenuation coefficient in high frequency red tide area of the East China Sea based on buoy observation[J]. Journal of Tropical Oceanography, 2020 , 39(5) : 71 -83 . DOI: 10.11978/2019084
表1 针对不同海域水体建立的Kd(490)反演算法Tab.1 Inversion algorithms of Kd(490) established for different sea area water bodies |
算法ID | 算法形式 | 算法来源 |
---|---|---|
Mueller算法 | ${{K}_{\text{d}}}\left( 490 \right)=-0.814{{[\frac{{{R}_{\text{rs}}}(490)}{{{R}_{\text{rs}}}(555)}]}^{2.242}}+1.373$ | Mueller等(1997) |
王晓梅算法 | ${{K}_{\text{d}}}\left( 490 \right)={{10}^{\{-0.581\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 555 \right)}+1.414[{{R}_{\text{rs}}}\left( 670 \right)+{{R}_{\text{rs}}}\left( 555 \right)]+0.299\}}}$ | 王晓梅等(2005) |
陈雨算法 | ${{K}_{\text{d}}}\left( 490 \right)={{10}^{[0.065\frac{{{R}_{\text{rs}}}\left( 590 \right)}{{{R}_{\text{rs}}}\left( 510 \right)}+0.968\frac{{{R}_{\text{rs}}}\left( 670 \right)}{{{R}_{\text{rs}}}\left( 510 \right)}-0.453]}}$ | 陈雨等(2014) |
Wang算法 | ${{K}_{\text{d}}}\left( 490 \right)=-\frac{0.823\times {{10}^{-5}}}{{{R}_{\text{rs}}}\left( 488 \right)}+2.13\frac{{{R}_{\text{rs}}}\left( 667 \right)}{{{R}_{\text{rs}}}\left( 488 \right)}+0.982\left[ 0.99{{R}_{\text{rs}}}\left( 667 \right)-0.19 \right][1-0.276{{\text{e}}^{\frac{-16.293}{{{R}_{\text{rs}}}\left( 488 \right)}-65.461\frac{{{R}_{\text{rs}}}\left( 667 \right)}{{{R}_{\text{rs}}}\left( 488 \right)}}}]$ | Wang等(2009) |
Kratzer算法 | ${{K}_{\text{d}}}\left( 490 \right)={{\exp }^{[-0.888\times \ln (\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 620 \right)})+0.41]}}+0.022$ | Kratzer等(2008) |
Zhang&Fell算法 | $\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 555 \right)}\ge 0.85{{K}_{\text{d}}}\left( 490 \right)={{10}^{\{76.22{{[\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 555 \right)}]}^{3}}-8.253{{[\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 555 \right)}]}^{2}}-2.935[\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 555 \right)}]-0.806\}+0.016}}$ $\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 555 \right)}0.85{{K}_{\text{d}}}\left( 490 \right)={{10}^{\{0.028{{[\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 665 \right)}]}^{3}}+0.649{{[\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 665 \right)}]}^{2}}-2.436[\frac{{{R}_{\text{rs}}}\left( 490 \right)}{{{R}_{\text{rs}}}\left( 665 \right)}]+0.853\}+0.016}}$ | Zhang等(2007) |
Tiwari 算法 | ${{K}_{\text{d}}}\left( 490 \right)=2.142\frac{{{R}_{\text{rs}}}(670)}{{{R}_{\text{rs}}}(490)}+0.189$ | Tiwari等(2014) |
本文算法 | ${{K}_{\text{d}}}\left( 490 \right)=2.351\frac{{{R}_{\text{rs}}}(650)}{{{R}_{\text{rs}}}(510)}-0.107\frac{{{R}_{\text{rs}}}(555)}{{{R}_{\text{rs}}}(510)}+0.146$ | 本文 |
图7 Kd(490)与400~700nm各波段Rrs的相关系数(a)和Kd(490)与400~700nm所有Rrs(λ1)/ Rrs(λ2)相关系数(b)Fig. 7 Correlation coefficient between Kd(490) and Rrs at 400 ~ 700 nm (a); Correlation coefficients distribution diagram between Kd(490) and Rrs(λ1) / Rrs(λ2) combined from 400 to 700nm (b) |
图9 验证数据集实测Kd(490)与估算的Kd(490)的散点图图中黑色实线为1:1线 Fig.9 Scatterplots of measured Kd(490) and estimated Kd(490) for the validation data set |
表2 各反演算法估算的Kd(490)与验证数据集中实测Kd(490)的比较统计参数结果Tab. 2 Statistical comparisons results between the estimated Kd(490) of different inversion algorithms and measured Kd(490) |
算法 | RMSE/ m-1 | MAPE/% | 斜率 | 截距 | R2 | N | 算法来源 |
---|---|---|---|---|---|---|---|
Mueller算法 | 0.49 | 55.10 | 0.18 | 0.78 | 0.18 | 581 | Mueller等(1997) |
王晓梅算法 | 0.50 | 43.95 | 0.15 | 0.69 | 0.15 | 581 | 王晓梅等(2005) |
陈雨算法 | 0.34 | 28.50 | 0.86 | 0.13 | 0.66 | 581 | 陈雨等(2014) |
Wang算法 | 0.44 | 48.37 | 0.74 | -0.14 | 0.79 | 581 | Wang等(2009)) |
Kratzer算法 | 7.42 | 334.09 | 2.62 | 0.17 | 0.71 | 581 | Kratzer等(2008) |
Zhang&Fell算法 | 1.51 | 55.66 | 1.28 | -0.34 | 0.84 | 581 | Zhang等(2007) |
Tiwari算法 | 0.31 | 28.13 | 0.70 | 0.28 | 0.68 | 581 | Tiwari等(2014) |
双比值算法 | 0.27 | 27.08 | 0.81 | 0.20 | 0.77 | 581 | 本文 |
注: RMSE为均方根误差; MAPE为平均相对误差百分比; R2为拟合决定系数; N为验证数据集的数据量 |
[1] |
艾威, 李茂田, 刘晓强, 等, 2018. 长江口南槽最大浑浊带枯季大小潮悬沙峰特征及其动力机制[J]. 海洋与湖沼, 49(4):769-778.
|
[2] |
曹沛奎, 严肃庄, 1996. 长江口悬沙锋及其对物质输移的影响[J]. 华东师范大学学报(自然科学版), (1):85-94.
|
[3] |
陈雨, 沈芳, 2014. 长江口邻近海域冬季漫衰减系数及其遥感反演[J]. 海洋湖沼通报, (4):27-34.
|
[4] |
何文珊, 陆健健, 2001. 高浓度悬沙对长江河口水域初级生产力的影响[J]. 中国生态农业学报, 9(4):24-27.
|
[5] |
胡连波, 张亭禄, 2012. 一种东中国海水体漫衰减系数Kd(490)的反演方法[J]. 海洋技术, 31(4):60-63.
|
[6] |
乐成峰, 李云梅, 查勇, 等, 2009. 太湖水体漫射衰减系数的光学特性及其遥感反演模型[J]. 应用生态学报, 20(2):337-343.
|
[7] |
李素菊, 吴倩, 王学军, 等, 2002. 巢湖浮游植物叶绿素含量与反射光谱特征的关系[J]. 湖泊科学, 14(3):228-234.
|
[8] |
丘仲锋, 崔廷伟, 何宜军, 2011. 基于水体光谱特性的赤潮分布信息MODIS遥感提取[J]. 光谱学与光谱分析, 31(8):2233-2237.
|
[9] |
沈焕庭, 贺松林, 潘定安, 等, 1992. 长江河口最大浑浊带研究[J]. 地理学报, 47(5):472-479.
|
[10] |
沈明, 段洪涛, 曹志刚, 等, 2017. 适用于多种卫星数据的太湖水体漫衰减系数估算算法[J]. 湖泊科学, 29(6):1473-1484.
|
[11] |
孙德勇, 李云梅, 王桥, 等, 2009. 基于实测高光谱的太湖水体悬浮物浓度遥感估算研究[J]. 红外与毫米波学报, 28(2):124-128.
|
[12] |
唐军武, 王晓梅, 宋庆君, 等, 2004. 黄、东海二类水体水色要素的统计反演模式[J]. 海洋科学进展, 22(S1):1-7.
|
[13] |
唐军武, 丁静, 田纪伟, 等, 2005. 黄东海二类水体三要素浓度反演的神经网络模型[J]. 高技术通讯, 15(3):83-88.
|
[14] |
王晓梅, 唐军武, 丁静, 等, 2005. 黄海、东海二类水体漫衰减系数与透明度反演模式研究[J]. 海洋学报, 27(5):38-45.
|
[15] |
杨跃忠, 曹文熙, 孙兆华, 等, 2009. 海洋高光谱辐射实时观测系统的研制[J]. 光学学报, 29(1):102-107.
|
[16] |
张清凌, 陈楚群, 施平, 2003. 南沙群岛海域水体漫衰减系数Kd(490)的特性研究[J]. 热带海洋学报, 22(1):9-16.
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
IOCCG, 2000. Remote Sensing of Ocean Colour in Coastal, and Other Optically-Complex, Waters[R]// SATHYENDRANATH S. Reports of the International Ocean-Colour Coordinating Group, No. 3. Dartmouth, Canada: IOCCG.
|
[24] |
|
[25] |
|
[26] |
|
[27] |
|
[28] |
|
[29] |
|
[30] |
|
[31] |
|
[32] |
|
[33] |
|
[34] |
|
[35] |
|
[36] |
|
[37] |
|
[38] |
|
[39] |
|
[40] |
|
[41] |
|
[42] |
|
[43] |
|
[44] |
|
[45] |
|
[46] |
|
/
〈 | 〉 |