Jianwei Wei and Zhongping Lee, "Retrieval of phytoplankton and colored detrital matter absorption coefficients with remote sensing reflectance in an ultraviolet band," Appl. Opt. 54, 636-649 (2015)
The light absorption of phytoplankton and colored detrital matter (CDM), which includes contribution of gelbstoff and detrital matters, has distinctive yet overlapping features in the ultraviolet (UV) and visible domain. The CDM absorption () increases exponentially with decreasing wavelength while the absorption coefficient of phytoplankton () generally decreases toward the shorter bands for the range of 350–450 nm. It has long been envisioned that including ocean color measurements in the UV range may help the separation of these two components from the remotely sensed ocean color spectrum. An attempt is made in this study to provide an analytical assessment of this expectation. We started with the development of an absorption decomposition model [quasi-analytical algorithm (QAA)-UV], analogous to the QAA, that partitions the total absorption coefficient using information at bands 380 and 440 nm. Compared to the retrieval results relying on the absorption information at 410 and 440 nm of the original QAA, our analyses indicate that QAA-UV can improve the retrieval of and , although the improvement in accuracy is not significant for values at 440 nm. The performance of the UV-based algorithm is further evaluated with in situ measurements. The limited improvement observed with the field measurements highlights that the separation of and is highly dependent on the accuracy of the ocean color measurements and the estimated total absorption coefficient.
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Error Statistics between the Model-Estimated Absorption Coefficient and the Simulated Absorptiona
Band (nm)
RMSE ()
nRMSE (%)
Bias ()
()
380
500(500)
492(499)
0.83(0.92)
0.30(0.36)
18.7(21.9)
0.03(0.08)
410
500(500)
471(500)
0.87(0.93)
0.27(0.27)
20.5(20.0)
0.03(0.10)
440
500(500)
484(500)
0.88(0.93)
0.27(0.24)
21.4(19.1)
0.04(0.10)
490
500(500)
474(500)
0.83(0.90)
0.33(0.29)
22.5(19.8)
0.03(0.07)
550
500(500)
396(482)
0.80(0.84)
0.41(0.47)
22.2(23.5)
0.03(0.05)
()
380
500(500)
500(500)
1.0(1.0)
0.02(0.04)
5.2(9.3)
−0.02(−0.08)
410
500(500)
500(500)
1.0(1.0)
0.05(0.08)
7.2(12.6)
−0.03(−0.10)
440
500(500)
500(500)
0.99(0.99)
0.07(0.12)
8.8(15.0)
−0.03(−0.10)
490
500(500)
500(500)
0.98(0.97)
0.12(0.20)
10.2(17.6)
−0.03(−0.07)
550
500(500)
500(500)
0.96(0.93)
0.17(0.29)
11.1(19.6)
−0.02(−0.05)
The QAA derivations are presented within parentheses. is the total number of observations, and is the number of valid observations. The bold values indicate that the accuracy of model estimations has been improved by the QAA-UV algorithm.
Table 5.
Comparison of the Algorithm Uncertainty of QAA-UV and QAA Modela
A model II least square fit is applied to the log-transformed uncertainty.
Mean uncertainty for QAA-UV retrievals and that of QAA retrievals (within the parentheses)
Table 6.
Error Statistics between the Model-Estimated Absorption Coefficient and Field Measured Absorption Coefficienta
Band (nm)
RMSE ()
nRMSE (%)
Bias ()
()
380
80(80)
80(80)
0.16(0.29)
0.53(0.74)
30.0(40.1)
410
199(199)
199(199)
0.17(0.56)
0.50(0.33)
26.3(17.5)
−0.006(−0.008)
440
200(200)
200(200)
0.40(0.55)
0.37(0.31)
20.5(17.5)
490
199(199)
199(199)
0.44(0.51)
0.43(0.37)
21.7(18.8)
550
101(101)
101(101)
0.42(0.43)
0.75(0.75)
30.2(30.0)
()
380
206(206)
206(206)
0.55(0.54)
0.43(0.33)
35.2(26.8)
410
199(199)
199(199)
0.55(0.53)
0.43(0.35)
31.1(25.1)
440
207(207)
207(207)
0.54(0.52)
0.41(0.35)
25.9(22.3)
490
207(207)
207(207)
0.48(0.47)
0.41(0.40)
21.4(20.5)
550
187(187)
187(187)
0.44(0.46)
0.40(0.39)
17.4(17.2)
−0.003(−0.004)
The QAA derivations are presented within parentheses. is the total number of observations and is the number of valid observations. The bold values indicate that the accuracy of model estimations has been improved by QAA-UV algorithm.
Tables (6)
Table 1.
Symbols, Notations, and Abbreviations
Symbols
Definitions
Units
Total absorption coefficient
Absorption coefficient due to phytoplankton
Absorption coefficient due to gelbstoff and detritus
Absorption coefficient due to gelbstoff
Absorption coefficient due to pure seawater
Backscattering coefficient due to particles
Backscattering coefficient due to pure seawater
Normalized water-leaving radiance
Remote sensing reflectance just above water surface
Remote sensing reflectance just below water surface
Slope of absorption spectrum
dimensionless
Spectral ratio of phytoplankton absorption
dimensionless
Spectral ratio of CDM absorption
CDM
Colored detrital matter
CDOM
Colored dissolved organic matter
IOCCG
International Ocean Color Coordinating Group
MAPE
Mean absolute percentage error
PACE
Pre-Aerosol, Clouds, and Ocean Ecosystem
QAA
Quasi-analytical algorithm
RMSE
Root mean square error
SeaBASS
SeaWiFS Bio-optical Archive and Storage System
UV
Ultraviolet
Table 2.
Ranges of Main Parameters for the Synthetic Data and Field Data
No.
Parameter
Synthetic Data
Field Data
1
0.604–1.003
0.426–1.046
2
0.060–0.958
0.221–1.290
3
1.314–1.807
1.177–2.405
4
1.739–3.273
1.482–4.266
5
0.009–0.020
0.007–0.047
6
0.334–1.276
0.304–2.289
7
0.123–6.773
0.207–4.994
Table 3.
Absorption Decomposition Algorithms Used by QAA-UV and QAA Models
Error Statistics between the Model-Estimated Absorption Coefficient and the Simulated Absorptiona
Band (nm)
RMSE ()
nRMSE (%)
Bias ()
()
380
500(500)
492(499)
0.83(0.92)
0.30(0.36)
18.7(21.9)
0.03(0.08)
410
500(500)
471(500)
0.87(0.93)
0.27(0.27)
20.5(20.0)
0.03(0.10)
440
500(500)
484(500)
0.88(0.93)
0.27(0.24)
21.4(19.1)
0.04(0.10)
490
500(500)
474(500)
0.83(0.90)
0.33(0.29)
22.5(19.8)
0.03(0.07)
550
500(500)
396(482)
0.80(0.84)
0.41(0.47)
22.2(23.5)
0.03(0.05)
()
380
500(500)
500(500)
1.0(1.0)
0.02(0.04)
5.2(9.3)
−0.02(−0.08)
410
500(500)
500(500)
1.0(1.0)
0.05(0.08)
7.2(12.6)
−0.03(−0.10)
440
500(500)
500(500)
0.99(0.99)
0.07(0.12)
8.8(15.0)
−0.03(−0.10)
490
500(500)
500(500)
0.98(0.97)
0.12(0.20)
10.2(17.6)
−0.03(−0.07)
550
500(500)
500(500)
0.96(0.93)
0.17(0.29)
11.1(19.6)
−0.02(−0.05)
The QAA derivations are presented within parentheses. is the total number of observations, and is the number of valid observations. The bold values indicate that the accuracy of model estimations has been improved by the QAA-UV algorithm.
Table 5.
Comparison of the Algorithm Uncertainty of QAA-UV and QAA Modela
A model II least square fit is applied to the log-transformed uncertainty.
Mean uncertainty for QAA-UV retrievals and that of QAA retrievals (within the parentheses)
Table 6.
Error Statistics between the Model-Estimated Absorption Coefficient and Field Measured Absorption Coefficienta
Band (nm)
RMSE ()
nRMSE (%)
Bias ()
()
380
80(80)
80(80)
0.16(0.29)
0.53(0.74)
30.0(40.1)
410
199(199)
199(199)
0.17(0.56)
0.50(0.33)
26.3(17.5)
−0.006(−0.008)
440
200(200)
200(200)
0.40(0.55)
0.37(0.31)
20.5(17.5)
490
199(199)
199(199)
0.44(0.51)
0.43(0.37)
21.7(18.8)
550
101(101)
101(101)
0.42(0.43)
0.75(0.75)
30.2(30.0)
()
380
206(206)
206(206)
0.55(0.54)
0.43(0.33)
35.2(26.8)
410
199(199)
199(199)
0.55(0.53)
0.43(0.35)
31.1(25.1)
440
207(207)
207(207)
0.54(0.52)
0.41(0.35)
25.9(22.3)
490
207(207)
207(207)
0.48(0.47)
0.41(0.40)
21.4(20.5)
550
187(187)
187(187)
0.44(0.46)
0.40(0.39)
17.4(17.2)
−0.003(−0.004)
The QAA derivations are presented within parentheses. is the total number of observations and is the number of valid observations. The bold values indicate that the accuracy of model estimations has been improved by QAA-UV algorithm.