Journal of Tropical Oceanography >
The spectral characteristics of phytoplankton absorption coefficient and assessment of MODIS-Aqua products in typical sea areas of the South China Sea
Received date: 2017-06-08
Request revised date: 2017-08-15
Online published: 2018-05-03
Supported by
National Natural Science Foundation of China (41506202)
Natural Science Foundation of Guangdong Province (2014A030310287)
State Key Laboratory of Tropical Oceanography Under Independent Project (LTO1509)
Project of Basic Scientific Research Expenses Supported by the Central Public Welfare Research Institute (South China Institute of Environmental Sciences, Ministry of Environmental Protection, PM-zx703-201601-014)
Ministry of Environmental Protection department budget Project (The policy research of coastal areas pollution prevention and control technology)
Science and Technology Planning Project of Guangxi (AB16380339)
Natural Science Foundation of Hubei Province (BZY15028)
Copyright
Using remote sensing to accurately estimate phytoplankton absorption coefficient aph(l) can provide basic data and useful method to distinguish different functions of phytoplankton species for long time and large spatial scale. In this paper, the characteristics of aph(l) spectral are compared and analyzed in four typical areas of the South China Sea (SCS), east area of Qiongdong (QD), Guangdong Coastal area (GD), and the Pearl River Estuary (PE) by using field data collected during2003-2012.Then, the phytoplankton population structure differences are preliminarily identified. Furthermore, the performances of MODIS-Aqua aph(l) products derived from the semi-analytical algorithm QAA and empirical algorithm PL by using MODIS-Aqua remote sensing reflectance Rrs(l) products are compared in the SCS and QD waters based on the relaxed match-ups between MODIS-Aqua products and field data. The results show the differences of aph(l) spectral features are obvious among the clear water represented by the SCS and QD and turbid waters represented by GD and PE. In the clear waters, the aph(l) value is small but in a dominant position of particle absorption, while in the GD and PE areas, the aph(l) value is relatively large but not in a dominant position. The aph(l) coefficient have obvious spatial differences, and the possible causes are pigment packaging effect and the variation of pigment composition and concentration. MODIS-Aqua aph(l) products derived from the empirical algorithm PL perform better than those from the semi-analytical algorithm QAA. The algorithm QAA-derived aph(l) products underestimate the results compared to the field data, while the algorithm PL overestimate the results, with the average relative error (APD)less than 22% for both algorithms. There is a great improvement in the accuracy of the PL algorithm by using the Chl-a products derived from the optimized algorithm of OCI (named algorithm NOCI), with the APD less than 14%. In summary, there are strong application prospects to discuss different functions of ocean phytoplankton species by using remote sensing products.
ZHAO Wenjing , CAO Wenxi , HU shuibo , WANG Guifen , LIU Zhenyu , XU Min . The spectral characteristics of phytoplankton absorption coefficient and assessment of MODIS-Aqua products in typical sea areas of the South China Sea[J]. Journal of Tropical Oceanography, 2018 , 37(3) : 35 -44 . DOI: 10.11978/2017067
Fig. 1 Typical area of the South China Sea (SCS), and spatial distribution of in situ aph(l) data (open circles).The dots represent the spatial distribution of relaxed (cross) match-ups for MODIS-Aqua Rrs(l) and aph(l). (a) The open sea of the SCS; (b) the coastal area of Qingdong (QD); (c) the coastal area of Guangdong (GD); and (d) the Pearl River Estuary (PE)图1 2004—2012年南海各典型海区位置及aph(l)现场观测站位(空心圆圈表示)空间分布示意 |
Fig. 2 In situ data spectra of aph(λ) in the SCS, QD, GD, and PE waters图2 南海(SCS)、琼东(QD)、广东沿岸(GD)和珠江口附近海域(PE)水体浮游植物吸收aph(λ)光谱 |
Fig. 3 In situ data spectra of ap(λ) in the SCS, QD, GD, and PE waters图3 南海(SCS)、琼东(QD)、广东沿岸(GD)和珠江口附近海域(PE)水体颗粒物吸收ap(λ)光谱 |
Fig. 4 In situ data spectra of aph*(λ) in the SCS, QD, GD, and PE waters图4 南海(SCS)、琼东(QD)、广东沿岸(GD)和珠江口附近海域(PE)水体浮游植物单位吸收系数aph*(λ)光谱 |
Fig. 5 Variations of aph*(443)(a) and aph*(667) (b) with CChl图5 aph*(443)(a)和aph*(667)(b)分别随CChl的变化 |
Tab. 1 The ranges of aph*(443), aph*(667) and aph*(443)/ aph*(667) in each typical area of the SCS表1 南海各典型海区单位吸收系数aph*(443), aph*(667)和aph*(443)/ aph*(667)量值范围 |
海区 | aph*(443) | aph*(667) | aph*(443)/aph*(667) |
---|---|---|---|
南海 | 0.02~0.90 (0.16) | 0.006~0.32 (0.05) | 1.33~7.13 (3.85) |
琼东 | 0.08~0.35 (0.17) | 0.02~0.07 (0.04) | 3.13~7.61 (4.81) |
广东 | 0.01~0.22 (0.08) | 0.008~0.11 (0.04) | 1.15~4.52 (2.26) |
珠江口 | 0.006~0.1 (0.05) | 0.002~0.05 (0.02) | 1.79~3.82 (2.58) |
注: 表格中的数值格式为“最小值~最大值(平均值)”。 |
Fig.6 Scatter plots of match-ups between in situ and satellite-derived aph(λ) (λ=412, 443, 490) using the PL (a, b, c) and QAA (d, e, f) algorithms in the SCS and QD sea waters for MODIS-Aqua of relaxed match-ups (N=23). The solid line is 1︰1 line图6 在宽松匹配法则下采用PL和QAA算法获得的南海开阔海域MODIS-Aqua aph(λ) (λ=412, 443, 490)产品值与相应实测值(N=23)之间的散点图 |
Tab. 2 Statistics comparing in situ aph(λ) (λ=412, 443, 490) with MODIS-Aqua products basing on QAA and PL algorithms in the SCS and QD sea waters when relaxed match-ups are included表2 在宽松的匹配法则下, 基于经验算法PL和半分析算法QAA反演获得的南海清洁海域(南海海区和琼东海区)MODIS aph(λ) (λ=412, 443, 490)产品值与实测值(N=23)的统计评价结果 |
QAA aph(412) | QAA aph(443) | QAA aph(490) | PL aph(412) | PL aph(443) | PL aph(490) | |
---|---|---|---|---|---|---|
APD/% | 22.49 | 29.68 | 46.89 | 20.15 | 17.94 | 21.65 |
RPD/% | -18.11 | -29.33 | -42.84 | 10.42 | 10.82 | 14.14 |
RMS | 0.0023 | 0.0040 | 0.0039 | 0.0022 | 0.0024 | 0.0018 |
Ratio | 0.7800 | 0.7099 | 0.5778 | 1.0543 | 1.0744 | 1.1448 |
SIQR | 0.1139 | 0.0981 | 0.1706 | 0.1804 | 0.1507 | 0.1311 |
R2 | 0.6938 | 0.7231 | 0.3642 | 0.6178 | 0.7160 | 0.6682 |
Slope | 0.8622 | 0.7434 | 0.6959 | 0.9210 | 0.8504 | 0.8312 |
Interception | -0.0004 | -0.0004 | -0.0008 | 0.0015 | 0.0028 | 0.0020 |
注: APD为平均绝对误差, RPD为平均相对误差, RMS为均方根误差, Ratio为遥感产品值与实测值之比的中值, SIQR为遥感产品值与实测值之比的四分位, R2为决定系数, slope为斜率, Interception为截距。 |
Fig. 7 Scatter plots of match-ups between in situ and satellite-derived aph(λ) (λ=412, 443, 490) using the PL algorithm based on regional algorithm NOCI-derived Chl-a in the SCS and QD sea waters for MODIS-Aqua of relaxed match-ups (N=23). The solid line is the 1︰1 line图7 基于算法NOCI反演得到的Chl-a产品作为输入, 采用经验算法PL反演获得的MODIS-Aqua aph(λ)(λ=412, 443, 490)产品值与相应实测值(N=23)之间的散点图(图中实线为1:1直线) |
Tab. 3 Statistics comparing in situ aph(412), aph(443) and aph(490) with MODIS-Aqua products using the PL algorithm, which is based on Chl-a products derived from the regional algorithm NOCI in the SCS and QD sea waters表3 基于NOCI算法反演获得的Chl-a产品作为输入, 采用经验算法PL反演获得的南海海域和琼东海域MODIS aph(412), aph(443), aph(490)产品值与实测值(N=23)的统计评价结果 |
NOCI aph(412) | NOCI aph(443) | NOCI aph(490) | |
---|---|---|---|
APD/% | 12.14 | 11.87 | 13.89 |
RPD/% | -4.02 | -2.07 | 0.31 |
RMS | 0.0014 | 0.0019 | 0.0014 |
Ratio | 0.9331 | 0.9551 | 0.9842 |
SIQR | 0.0859 | 0.0869 | 0.1024 |
R2 | 0.8012 | 0.8297 | 0.7538 |
Slope | 0.7129 | 0.6250 | 0.6007 |
Interception | 0.0020 | 0.0038 | 0.0026 |
注: APD为平均绝对误差, RPD为平均相对误差, RMS为均方根误差, Ratio为遥感产品值与实测值之比的中值, SIQR为遥感产品值与实测值之比的四分位, R2为决定系数, slope为斜率, Interception为截距。 |
The authors have declared that no competing interests exist.
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
|
[26] |
|
[27] |
|
[28] |
|
[29] |
|
[30] |
|
[31] |
|
[32] |
|
/
〈 |
|
〉 |