Content uploaded by Cédric G. Fichot
Author content
All content in this area was uploaded by Cédric G. Fichot
Content may be subject to copyright.
A novel method to estimate DOC concentrations from CDOM
absorption coefficients in coastal waters
Cédric G. Fichot
1
and Ronald Benner
1
Received 10 November 2010; revised 17 December 2010; accepted 28 December 2010; published 12 February 2011.
[1] A novel method to accurately retrieve DOC concentra-
tions (±4%) from CDOM absorption coefficients, a
g
(l), at
l= 275 and 295 nm is presented. By using these two wave-
lengths, the method exploits useful information about the
ratio of a
g
(l) to [DOC] contained in the 275–295 nm spectral
slope coefficient, S
275−295
. This approach was developed
using data (n= 222) collected on a seasonal basis in surface
waters of the Northern Gulf of Mexico. The approach is dem-
onstrated to accurately and consistently estimate [DOC] for
all seasons and for a broad [DOC] range of 63–611 mM.
Application to coastal waters of the Beaufort Sea (n= 33)
demonstrated that similar performance can be expected in
other river‐influenced ocean margins after parameterization
using local data. Applicability to other marine environments
remains to be tested but such assessment can be pursued
immediately where appropriate data have already been col-
lected. Citation: Fichot, C. G., and R. Benner (2011), A novel
method to estimate DOC concentrations from CDOM absorption
coefficients in coastal waters, Geophys. Res. Lett.,38, L03610,
doi:10.1029/2010GL046152.
1. Introduction
[2] Conventional methods for the analysis of Dissolved
Organic Carbon (DOC) are restricted to measurements of
discrete samples and are limited to providing synoptic
coverage on relatively small spatial scales. The estimation
of DOC concentrations, [DOC], through measurement of the
optical properties of dissolved organic matter (DOM)
(absorption and fluorescence) therefore represents a com-
pelling alternative. Under optimal conditions and with proper
instrumentation, the optical properties of DOM can be rapidly
and continuously acquired in situ [Vodacek et al., 1997;
Hitchcock et al., 2004].
[3] The relationship between [DOC] and DOM absorption
(Chromophoric Dissolved Organic Matter, CDOM) has
been investigated in a variety of coastal systems [Ferrari
et al., 1996; Del Vecchio and Blough, 2004a; Guéguen
et al., 2005]. Although strong positive correlations have
been observed between CDOM absorption coefficients,
a
g
(l), and [DOC], the relationship varies among geograph-
ical regions and seasons [Blough and Del Vecchio, 2002].
For example, the ratio of a
g
(l) to [DOC] varies seasonally
by more than 25‐fold in the Middle Atlantic Bight alone
[Del Vecchio and Blough, 2004a]. Such variability in this
ratio sets a limit on our capability to predict [DOC] from
simple linear relationships between DOC and CDOM.
[4] The spectral characteristics of a
g
(l) are representative
of the types of chromophores present in DOM [Del Vecchio
and Blough, 2004b]. A single exponential fit (equation 1) is
typically used to describe the spectral dependency of a
g
(l):
agðÞ¼ag0
ðÞeS0
ðÞ ð1Þ
where l
0
<land Sis the spectral slope coefficient in the
l
0
to lnm spectral range. Several studies have related the
spectral characteristics of DOM absorption to the chemical
and structural nature of DOM such as molecular weight and
aromaticity [Chin et al., 1994; Helms et al., 2008]. Other
studies utilized molar absorptivity or carbon‐specific UV
absorbance as indicators of the molecular weight and aro-
matic content of organic matter isolates [Chin et al., 1994;
Weishaar et al., 2003]. A linkage between the spectral
characteristics of a
g
(l) and the ratio a
g
(l)/[DOC] is there-
fore possible. If such a connection exists for DOM in the
marine environment it could be exploited to improve pre-
dictions of [DOC] from measurements of a
g
(l).
[5] In this study, we explore the relationship between
Sand the ratio a
g
(l)/[DOC] in surface waters of the
Northern Gulf of Mexico (NGoM), with the intent of testing
this hypothesis. Following recommendations by Helms et al.
[2008] on the use of Sin the 275−295 nm spectral range,
a strong relationship between S
275−295
and a
g
(l)/[DOC] was
discovered. This connection is exploited in a method to
accurately estimate [DOC] from simple in situ measure-
ments of a
g
(275) and a
g
(295) in coastal areas. Its applica-
bility in other marine systems is discussed.
2. Sampling and Methods
[6] Surface water from the NGoM was collected and fil-
tered for DOC analysis and CDOM absorbance measure-
ments. A total of 222 stations (n= 222) were sampled during
five research cruises (January, April, July, October/November
2009 and March 2010) as part of the GULFCARBON project.
About 50 stations were sampled per cruise (Figure 1a) with
the exception of January 2009 when 24 stations were sam-
pled. Representing a salinity range of 0–37 psu, these
samples include most water types typically encountered in
river‐dominated ocean margins. DOC analysis was done by
High Temperature Combustion (HTC) and CDOM absorbance,
A(l), was measured using a dual‐beam spectrophotometer.
Absorbances were converted to absorption coefficients, a
g
(l),
and spectral slope coefficients, S, were calculated using linear
fits of log‐linearized a
g
(l). The carbon‐specific absorption
coefficients of DOM were calculated as the ratio of a
g
(l)
to DOC concentration and are denoted here as a*
g
(l), with
units of m
−1
mM
−1
. The value of a*
g
(l)atl= 355 nm was
calculated for consistency with previous studies [Vodacek
1
Marine Science Program, University of South Carolina, Columbia,
South Carolina, USA.
Copyright 2011 by the American Geophysical Union.
0094‐8276/11/2010GL046152
GEOPHYSICAL RESEARCH LETTERS, VOL. 38, L03610, doi:10.1029/2010GL046152, 2011
L03610 1of5
et al., 1997; Del Vecchio and Blough, 2004a]. Detailed
sampling and methods are provided in the auxiliary material.
1
[7] Surface water from the Beaufort Sea was sampled in
August 2009 as part of the MALINA project. A total of
33 stations (n= 33) were sampled across a salinity gradient
of 0 to 30 psu and [DOC] and absorbance were measured as
described above.
3. Dynamics of DOC and CDOM in the Northern
Gulf of Mexico
[8] The measured range of DOC concentrations, ([DOC]:
63–611 mM) spanned an order of magnitude over the
salinity range of 0 to 37 psu. The strong relationship
between salinity and [DOC] (r
2
= 0.83) indicates that DOM
dynamics in surface waters were, to a first degree, domi-
Figure 1. (a) Station locations in the Northern Gulf of
Mexico. (b) CDOM absorption coefficient spectra, a
g
(l),
and corresponding spectral slope coefficient, S
275−295
,for
three contrasting water samples. The y‐axis for a
g
(l)isa
log scale. S
275−295
is typically low in river water and in-
creases in coastal and oligotrophic waters. A modeled,
downward plane irradiance spectrum just above the sea sur-
face, E
d
(0
+
,l), is overlaid and illustrates the absence of
photons at wavelengths (l) < 295 nm. E
d
(0
+
,l) was modeled
for June 21st, 12:00 p.m., at latitude 28°N, with 300 DU of
ozone, and a clear sky. These conditions correspond to the
maximal incident irradiance expected in the Northern Gulf
of Mexico.
Figure 2. Relationships between different DOM properties
in the Northern Gulf of Mexico: (a) [DOC] vs. salinity;
(b) a
g
(355) vs. [DOC]; (c) a*
g
(355) = a
g
(355)/[DOC] vs.
S
275−295
. In Figures 2a and 2b, the inset plots present linear
regressions for the entire range of the data whereas the main
plots magnify the 50–300 mM [DOC] range. In Figure 2c, the
inset plot shows the lack of relationship between S
350−400
and a*
g
(355).
1
Auxiliary materials are available in the HTML. doi:10.1029/
2010GL046152.
FICHOT AND BENNER: DOC ESTIMATES FROM CDOM ABSORPTION COEFFICIENTS L03610L03610
2of5
nated by terrigenous inputs (inset of Figure 2a). Within this
general view, however, a significant seasonality and devia-
tion from linear mixing at salinity extremes was apparent
(Figure 2a). A poor correlation observed between [DOC]
and salinity at salinities less than 20 psu (r
2
= 0.18) indicated
the presence of multiple riverine sources with varying
DOM properties. This observation is in agreement with
earlier studies demonstrating varying mixing behavior of
DOC as it transits from estuaries to the Gulf of Mexico [Guo
et al., 1998].
[9] The NGoM is a river‐dominated system in terms of
DOM optical properties [Chen and Gardner, 2004; Conmy
et al., 2004; D’Sa and DiMarco, 2009]. A strong, linear
relationship (r
2
= 0.90) was observed in the present study
between [DOC] and the CDOM absorption coefficient,
a
g
(355) (inset of Figure 2b). However, this strong relation-
ship is misleading for the purpose of retrieving [DOC] from
a
g
(355). The ratio a
g
(355)/[DOC] = a*
g
(355), varies by a
factor of 55, from a minimum value of 5.3*10
−4
m
−1
mM
−1
in the most oligotrophic sample to a value of 2.9*10
−2
m
−1
mM
−1
in freshwater. The trend, range and values of a*
g
(355)
observed in this study are in general agreement with those
observed by Del Vecchio and Blough [2004a] in the Middle
Atlantic Bight. Some of this variability can be attributed to the
conservative mixing of river water and seawater along the
salinity gradient and can be accounted for in a linear rela-
tionship of the form [DOC] = a+ba
g
(355), where aand b
are regression coefficients. However, a magnified view of
the relationship between [DOC] and a
g
(355) (Figure 2b)
also shows a strong seasonality and some non‐linearity
which can result from: 1) the presence of multiple riverine
sources with different a*
g
(355); 2) the decoupling between
the autochtonous sources and sinks of DOC with those of
CDOM; and 3) photobleaching, known to decrease a*
g
(355)
[Del Vecchio and Blough, 2004a]. This variability cannot be
constrained in the simple linear model and although sea-
sonal linear models can be derived, their implementation is
always difficult.
[10] Numerous studies have investigated the dynamics of
the spectral slope coefficient in aquatic environments and
have concluded that it is of limited utility as a biogeocheo-
mical indicator [Blough and Del Vecchio, 2002]. However,
the spectral range used among investigators has been incon-
sistent, generally broad (e.g., 290–700 nm) and restricted to
the UV‐A and visible domains. Helms et al. [2008] recently
suggested the use of S
275−295
(UV‐B domain, narrow
range) as an indicator of photochemical alterations and
DOM molecular weight and source in the marine environ-
ment. In light of their results, a remarkable finding of the
present study is a strong relationship between S
275−295
and
a*
g
(355) (Figure 2c), which is best approximated by an
exponential equation of the form: a*
g
(355) = e
(a−bS
275−295
)
+
e
(g−dS
275−295
)
, where a,b,gand dare regression coeffi-
cients. This relationship is remarkable because a strong
connection does not exist between a*
g
(355) and other spec-
tral slope coefficients such as S
300–350
,S
300–400
or S
350–400
in
this data set (inset of Figure 2c), thereby highlighting the
unique potential of the 275−295 nm spectral range to retain
information about DOM composition and the ratio of a
g
(l)
to [DOC]. The strong link between S
275−295
and a*
g
(355) is
indicative that the dynamics of these two DOM properties
are regulated by the same processes. Although gaining a full
understanding of the dynamics responsible for this rela-
tionship is beyond the scope of this manuscript, it can be
inferred from a few recent studies that the processes
responsible for the variabilities in S
275−295
and a*
g
(355) are
of the same nature [Del Vecchio and Blough, 2004a; Helms
et al., 2008; Ortega‐Retuerta et al., 2009].
[11] A unique and important aspect of S
275−295
not men-
tioned by Helms et al. [2008] is the unique position of the
275–295 nm spectral region on the outside edge of the
natural solar spectrum (Figure 1b). Even under optimal
conditions, very few photons of l< 295 nm are present in
the natural environment. According to the work of Del
Vecchio and Blough [2002] on the photobleaching of
a
g
(l) using monochromatic irradiations, the decrease in a
g
(l)
upon absorption of photons of wavelength l
ex
is maximum
at or near l=l
ex
and decreases exponentially towards other
wavelengths. It is therefore expected that any natural photon
absorbed would always lead to a greater change in a
g
(295)
than in a
g
(275), and consequently, to an increase in S
275−295
.
In contrast, other spectral regions used for the determination
of Stend to overlap with the photochemically‐active part of
the natural solar spectrum (typically 300–400 nm). A well
behaved response of Sto photobleaching is therefore
unlikely for spectral regions other than 275–295 nm. This
phenomenon can contribute to the erratic behavior of S
typically observed in the marine environment.
[12] An important implication of the relationship between
S
275−295
and a*
g
(l) in this system is the novel capability to
constrain the variability in the ratio a
g
(l)/[DOC] using
information contained in the spectral shape of a
g
(l). Because
a
g
(l)/[DOC] can vary by a large factor, exploiting this
information can considerably improve the accuracy of [DOC]
retrieved from a
g
(l).
4. Estimating DOC from CDOM
[13] DOC concentrations can be retrieved from the com-
bination of a
g
(l) and a non‐linear fit of a*
g
(l) versus S
275−295
.
However, we found through extensive testing that the most
accurate [DOC] were obtained by performing multiple linear
regressions (MLR) of log‐linearized [DOC] against log‐
linearized a
g
(275) and a
g
(295), as described in equation (2):
ln DOC½¼þln ag275ðÞ
þln ag295ðÞ
ð2Þ
where a,band gare regression coefficients.
[14] This method exploits all the useful information
contained in S
275−295
while being simpler, more direct and
accurate. The best wavelengths for prediction of [DOC]
were l= 275, 295 nm. The use of additional variables in the
MLR (e.g., a
g
(l)atl≠275, 295 nm, salinity, chlorophyll‐a
fluorescence) did not improve the predictive capability of
the model. A MLR against ln [a
g
(275)] and ln [a
g
(295)]
therefore represents an optimal model. In order to relieve the
constraint of using a single MLR for a broad range of
[DOC] (63–611 mM), the data were separated into two
subsets and a specific MLR was done on each subset. The
data were divided based on the cutoff value a
g
(275) = 3.5 m
−1
,
which corresponds to the median value in these data. The
regression coefficients are provided in Table 1 and the per-
formance of the model is evaluated in Figure 3.
[15] The performance of the model (Figure 3b) was
compared to that of a single regression of ln [DOC] versus
ln [a
g
(355)] model (Figure 3a). A large seasonal bias and
FICHOT AND BENNER: DOC ESTIMATES FROM CDOM ABSORPTION COEFFICIENTS L03610L03610
3of5
poor accuracy at low and high DOC concentrations re-
sulted from the use of the single regression, even after log‐
linearization of [DOC] and a
g
(355). An even larger bias and
lower accuracy was observed if these values were not log‐
linearized before regression. The new approach demon-
strates that the use of two carefully chosen wavelengths and
their use in MLR can considerably improve the accuracy of
estimated DOC concentrations. Overall, [DOC] estimated
using this approach were within ±4.2% of the measured
[DOC]. For comparison, the percent error associated with
replicates of [DOC] measurements was typically ±1%. A
sensitivity analysis of the model also revealed that about half
of the ±4.2% error could be attributed to errors in the
reproducibility of the absorbance measurements. The dis-
tribution of points around the 1‐to‐1 line indicated that the
percent error associated with the [DOC] retrieved using this
approach was consistent over the entire [DOC] range
(Figure 3b). The error associated with the estimate is typi-
cally ±2.4 mM for a measured [DOC] value of 60 mM, and
±12 mM for a measured [DOC] value of 300 mM.
5. Applicability of the Approach
[16] The results presented in the previous section are valid
in the NGoM, for which the model was parameterized.
Application of these parameters to other marine systems can
therefore lead to unpredictable results. In order to test the
applicability of the approach to a different coastal system,
we applied it to independent data (n= 33) acquired in
August 2009 in the Beaufort Sea, in a region influenced
by the Mackenzie river outflow. Our results indicate that
applying the approach to this region using the NGoM
parameters results in decreased accuracy (±18% of mea-
sured [DOC]) and biases in the derived [DOC]. However,
excellent accuracy (±4.7%) and a consistent percent error
over the full range of measured [DOC] (66–458 mM) is
obtained after the model was re‐parameterized using local
data (Figure 3c and Figure 4 in the Auxiliary Material). The
regression coefficients derived for the Beaufort Sea are
given in Table 1.
[17] Although the approach may be applied to other coastal
environments, differences in DOM sources and regulatory
processes make the re‐parameterization of the model using
local data necessary to achieve high performance. Suitable
data have already been collected in other coastal systems
such as the Middle Atlantic Bight [Del Vecchio and Blough,
2004a; Mannino et al., 2008] and are therefore readily
available for parameterizing the model. For other marine
systems where both the range in DOM properties and the
Table 1. Parameters a,band gDerived From the Multiple Linear
Regressions of ln [DOC] Against ln [a
g
(275)] and ln [a
g
(295)] for
the Northern Gulf of Mexico and the Beaufort Sea
a
ab gAdjusted r
2
Northern Gulf of Mexico
a
g
(275) ≤3.5 m
−1
3.4707 1.8591 −1.2421 0.92
a
g
(275) > 3.5 m
−1
2.9031 2.7703 −2.0400 0.95
Beaufort Sea
a
g
(275) ≤3.5 m
−1
4.2952 0.1153 0.3187 0.80
a
g
(275) > 3.5 m
−1
2.3603 3.0395 −2.1298 0.96
a
See equation (2).
Figure 3. (a) [DOC] estimated using a single linear regres-
sion of ln [DOC] against ln [a
g
(355)] in the Northern Gulf of
Mexico; (b) [DOC] estimated using the new approach in the
Northern Gulf of Mexico. (c) [DOC] estimated using the
new approach in the Beaufort Sea, after re‐parameterization
of the model using local data (see Figure 4 in the Auxiliary
Material for station locations). In all plots, “Estimated
DOC”indicates concentrations estimated from a
g
(l).
FICHOT AND BENNER: DOC ESTIMATES FROM CDOM ABSORPTION COEFFICIENTS L03610L03610
4of5
chemical composition of DOM is less influenced by ter-
rigenous inputs (e.g., Sargasso Sea), the validity of the
approach itself remains to be assessed.
[18] Besides its simplicity of implementation, this method
presents a number of advantages that make it suitable for
high‐resolution and long‐term in situ monitoring of [DOC]
in coastal environments. First, absorbance measurements at
only two wavelengths are required, which can be acquired at
a fast rate while storing minimal amounts of data. Second,
the values of a
g
(275) and a
g
(295) are high in most environ-
ments thereby making the approach less sensitive to limita-
tions in the precision of the instrument. Third, the relationship
between [DOC] and a
g
(275) or a
g
(295) should remain
minimally affected by changes in inorganic ion concentra-
tions (e.g., nitrate, nitrite, bromide and bisulfide) [Johnson
and Coletti, 2002]. Finally, if the model is parameter-
ized using representative data, the approach should be
applicable to a given region for all seasons using a single
parameterization.
[19]Acknowledgments. This work was supported by the National
Science Foundation (NGoM grants NSF 0850653 and 0713915 to R. Benner,
OCE‐0752254 to S. E. Lohrenz and OCE‐0752110 to W‐J. Cai; Arctic grant
NSF 0125301 to R. Benner). We thank S. E. Lohrenz, W‐J. Cai and K.
Gundersen for providing a berth on the GULFCARBON cruises, and L.
Powers and the crews of the R/V Cape Hatteras and the R/V Hugh Sharp
for their assistance with sample collection. We are grateful to M. Babin
and S. Bélanger for providing a berth on the MALINA cruise.Finally,we
thank Y. Shen and S. Walker for some sample measurements and two anon-
ymous reviewers for their constructive comments.
References
Blough, N. V., and R. Del Vecchio (2002), Chromophoric DOM in the
coastal environment, in Biogeochemistry of Marine Dissolved Organic
Matter, pp. 509–546, Academic, San Diego, Calif.
Chen, R. F., and G. B. Gardner (2004), High‐resolution measurements of
chromophoric dissolved organic matter in the Mississippi and Atchafalaya
River plume regions, Mar. Chem.,89(1–4), 103–125, doi:10.1016/j.
marchem.2004.02.026.
Chin, Y. P., G. Aiken, and E. OLoughlin (1994), Molecular‐weight, poly-
dispersity, and spectroscopic properties of aquatic humic substances,
Environ. Sci. Technol.,28(11), 1853–1858.
Conmy, R. N., P. G. Coble, R. F. Chen, and G. B. Gardner (2004), Optical
properties of colored dissolved organic matter in the northern Gulf of
Mexico, Mar. Chem.,89(1–4), 127–144, doi:10.1016/j.marchem.2004.
02.010.
Del Vecchio, R., and N. V. Blough (2002), Photobleaching of chromo-
phoric dissolved organic matter in natural waters: kinetics and modeling,
Mar. Chem.,78(4), 231–253.
Del Vecchio, R., and N. V. Blough (2004a), Spatial and seasonal distribu-
tion of chromophoric dissolved organic matter and dissolved organic car-
bon in the Middle Atlantic Bight, Mar. Chem.,89(1–4), 169–187,
doi:10.1016/j.marchem.2004.02.027.
Del Vecchio, R., and N. V. Blough (2004b), On the origin of the optical prop-
erties of humic substances, Environ. Sci. Technol.,38(14), 3885–3891,
doi:10.1021/es049912h.
D’Sa, E. J., and S. F. DiMarco (2009), Seasonal variability and controls on
chromophoric dissolved organic matter in a large river‐dominated
coastal margin, Limnol. Oceanogr.,54(6), 2233–2242.
Ferrari, G. M., M. D. Dowell, S. Grossi, and C. Targa (1996), Relationship
between the optical properties of chromophoric dissolved organic matter
and total concentration of dissolved organic carbon in the southern Baltic
Sea region, Mar. Chem.,55(3–4), 299–316.
Guéguen, C., L. Guo, and N. Tanaka (2005), Distributions and characteris-
tics of colored dissolved organic matter in the western Arctic Ocean,
Cont. Shelf Res.,25(10), 1195–1207.
Guo, L., P. H. Santschi, and T. S. Bianchi (1998), Dissolved Organic Matter
in Estuaries of the Gulf of Mexico, in Biogeochemistry of Gulf of Mexico
Estuaries, edited by T. S. Bianchi, J. R. Pennock, and R. R. Twilley,
pp. 269–299, John Wiley, New York.
Helms, J. R., A. Stubbins, J. D. Ritchie, E. C. Minor, D. J. Kieber, and
K. Mopper (2008), Absorption spectral slopes and slope ratios as indica-
tors of molecular weight, source, and photobleaching of chromophoric
dissolved organic matter, Limnol. Oceanogr.,53(3), 955–969.
Hitchcock, G. L., R. F. Chen, G. B. Gardner, and W. J. Wiseman (2004), A
Lagrangian view of fluorescent chromophoric dissolved organic matter
distributions in the Mississippi River plume, Mar. Chem.,89(1–4),
225–239, doi:10.1016/j.marchem.2004.03.018.
Johnson, K. S., and L. J. Coletti (2002), In situ ultraviolet spectrophotometry
for high resolution and long‐term monitoring of nitrate, bromide and bisul-
fide in the ocean, Deep Sea Res., Part I,49(7), 1291–1305, doi:10.1016/
S0967-0637(02)00020-1.
Mannino, A., M. E. Russ, and S. B. Hooker (2008), Algorithm development
and validation for satellite‐derived distributions of DOC and CDOM in the
U.S. Middle Atlantic Bight, J. Geophys. Res.,113, C07051, doi:10.1029/
2007JC004493.
Ortega‐Retuerta, E., T. K. Frazer, C. M. Duarte, S. Ruiz‐Halpern, A. Tovar‐
Sanchez, J. M. Arrieta, and I. Reche (2009), Biogeneration of chromo-
phoric dissolved organic matter by bacteria and krill in the Southern
Ocean, Limnol. Oceanogr.,54(6), 1941–1950.
Vodacek, A., N. V. Blough, M. D. DeGrandpre, E. T. Peltzer, and R. K.
Nelson (1997), Seasonal variation of CDOM and DOC in the Middle
Atlantic Bight: Terrestrial inputs and photooxidation, Limnol. Oceanogr.,
42(4), 674–686.
Weishaar, J. L., G. R. Aiken, B. A. Bergamaschi, M. S. Fram, R. Fujii, and
K. Mopper (2003), Evaluation of specific ultraviolet absorbance as an
indicator of the chemical composition and reactivity of dissolved organic
carbon, Environ. Sci. Technol.,37(20), 4702–4708, doi:10.1021/
es030360x.
R. Benner and C. G. Fichot, Marine Science Program, University of
South Carolina, Columbia, SC 29208, USA. (benner@mailbox.sc.edu;
cgfichot@gmail.com)
FICHOT AND BENNER: DOC ESTIMATES FROM CDOM ABSORPTION COEFFICIENTS L03610L03610
5of5