&EPA
United States
Environmental Protection
Agency
EPA/600/R-15/211 | August 2015
www. e pa. g o v/o rd
DATA REPORT FOR CALIBRATION
OF A BIO-OPTICAL MODEL
FOR NARRAGANSETT BAY
3.0 -i
0.0
400 450 500 550 600
Wavelength (nm)
650
700
GlenThursby, Steven Rego, Darryl Keith
Atlantic Ecology Division
National Health and Environmental Effects Research Laboratory
Office of Research and Development
US Environmental Protection Agency
Narragansett, Rl 02882
Augusts, 2015
Office of Research and Development
National Health and Environmental Effects Research Laboratory
-------
-------
EPA/600/R-15/211
Data Report for Calibration
of a Bio-Optical Model
for Narragansett Bay
by
Glen Thursby, Steven Rego and Darryl Keith
Atlantic Ecology Division
National Health and Environmental Effects Research Laboratory
Office of Research and Development
US Environmental Protection Agency
Narragansett, RI 02882
August 5, 2015
-------
NOTICE
The research described in this report has been funded wholly by the U.S Environmental
Protection Agency. This report is contribution number ORD-013449 of the Atlantic Ecology
Division, National Health and Environmental Effects Research Laboratory, Office of Research
and Development. This document has been subjected to USEPA's peer review process and has
been approved for publication. The mention of trade names of commercial products does not
constitute endorsement for use.
ABSTRACT
Bio-optical models describe the quality and quantity of the light field at various depths in the
water column. The absorption and scattering of light within the water column are wavelength
dependent. The behavior of light also varies depending on the specific dissolved and particulate
constituents in the water column, making local (or at least regional) calibration of such models
necessary. This report provides the calibration data specific for Narragansett Bay, Rhode Island,
relative to absorption by colored dissolved organic matter (CDOM), non-algal particles (NAP)
and phytoplankton, and total backscattering. With the calibration in place, information on the
concentration of CDOM, total suspended solids and chlorophyll a is all that is needed in order
to calculate the light for any depth for sites those parameters represent.
Keywords: bio-optical model, Narragansett Bay, diffuse attenuation coefficient, light absorption,
light scattering
ACKNOWLEDGEMENTS
Dr. Brenda Rashleigh, Dr. Peg Pelletier and Mr. Joseph LiVolsi reviewed the final draft of the
data report. Joseph Bishop assisted with the analyses. Dr. Charles Gallegos from the Smithonian
Environmental Research Center provided valuable insights during our early attempts to
understand bio-optical modeling.
- n -
-------
INTRODUCTION
The data contained in this report supports the development of calibration curves for the use of
a bio-optical model for Narragansett Bay, RI. The calibration will support research evaluating
the applicability of a bio-optical model for light quantity in establishing depth limits for seagrass
(Zostera marina) in Narragansett Bay, RI. To accomplish this, a site specific calibration of the
model was necessary. The model provides estimates of changes in light quantity and quality with
depth which can be used, to predict the maximum depth likely for seagrass. In the case of
Narragansett Bay, the seagrass is Zostera marina. As light travels through the water column
it interacts with water molecules, dissolved materials, suspended inorganic and organic
particulates and phytoplankton, resulting in a variety of absorption and scattering phenomena.
Because these phenomena (especially absorption) are wavelength dependent and vary with the
particular constituents of a given waterbody, it is necessary to calibrate the bio-optical model's
parameters for each region of the country — and maybe for each body of water of interest within
those regions. With calibrations completed, measures of colored dissolved organic matter
(CDOM), total suspended solids (TSS) and chlorophyll a (Chi a) for any given location allows
estimates of light penetration to be calculated.
The model used is based on Lee et al. (2005), as described by Kenworthy et al. (2014).
ffd(A) = (1 + 0.00500)at(A) + 4.18[1 - 0.52exp(-10.8at)]fc6(A) EQ 1
Where:
X = wavelength of light (nm) — Equation 1 is wavelength dependent,
Kd (X) = spectral diffuse attenuation coefficient (m"1),
$o = above-water solar angle of incidence (degrees) — zenith angle,
at (A) = total absorption coefficient (m"1) — note, X omitted when this is in the exponent
for simplicity,
bb (A) = backscattering coefficient (m"1).
By integrating over the visible wavelengths, the diffuse attenuation coefficient for
photosynthetically active radiation (PAR) can be calculated,
Where:
Kd (PAR) = diffuse attenuation coefficient, integrated over 400 to 700 nm,
z = depth in meters,
PARZ = light spectrum at depth z, and
PARo = light spectrum at surface.
The Equation 1 requires two site specific pieces of information, the absorption and back
scattering coefficients. The absorption coefficient (m"1) is an additive equation:
aparticlesW +
- 1 -
-------
Where total absorption is a function of four primary components, absorption by water itself,
by colored dissolved organic matter (CDOM), by particles (total suspended solids—TSS) and
by pigments (i.e., phytoplankton—usually related to chlorophyll a). Absorption by pure water
is a fixed function (wavelength dependent), and the relationship between wavelength and
absorption from Pope and Fry (1997) is usually used. The difference in the absorption of light
between pure water and seawater is minimal and often ignored. The other three components
have predictable relationships between absorption and wavelength of light; however, these
relationships need to be calibrated for each body of water.
Since we are interested ultimately in how much light makes it down to a given depth for
seagrass, absorption only gives a portion of the story. Light also is scattered within the water
column. While most of this light is scattered in the forward direction (e.g., continues downward),
a significant portion of the light entering the water column is scattered back out of the system.
The percentage of light backscattered is empirically derived for each area of interest. Total
scattering (m"1) is calculated as:
b(X) = bp(X) + bw(X) EQ4
Where bw(A~) is the total scattering associated with pure water (also referred to as molecular
scattering), and, light absorption, is considered a constant, wavelength dependent factor.
The other parameter, &p(/l), is total particulate scattering (in all directions). Unlike absorption,
there is no easy method to distinguish between pigmented and non-pigmented particle scattering.
The methods to calibrate each of the parameters associated with the bio-optical model are
presented below. All of the data represented by the calibration figures are presented in
Appendices.
METHODS
This calibration establishes the relationship between several easily measured water column
properties (Chi a, TSS and CDOM) and the components of absorption and scattering that
contribute to Equation 1. Samples from Narragansett Bay were collected from the upper,
middle and lower Bay (see Figure 1) from different times of the year, representing Zostera's
predominant growing season. The purpose of the sampling scheme was to establish how variable
the calibration parameters are with respect to time and space. One calibration has been
established for the entire Bay for all months within which the seagrass Zostera marina grows.
-2-
-------
Sampling Stations
• J.Grear Sites
A Bi o-Optic al M odeli ng S ites
0 25 5
Figure 1. Map of Narragansett Bay showing location of sampling stations
-3 -
-------
Absorption by water
There is no calibration associated with the absorption by pure water. It is considered a standard
relationship, and is plotted in Figure 2. In general, the additional absorption at seawater salinities
due to the salts is minimal in the 400 to 680 nm range (Kirk 2011), and is usually ignored in
coastal applications.
400
500 600
Wavelength (nm)
700
800
Figure 2. The relationship between wavelength and absorption coefficient in pure
water. Data from Pope and Fry, 1997
Absorption by CDOM—The relationship between absorption coefficient due to Colored
Dissolved Organic Matter (CDOM) and wavelength of light follows an exponential decay
function (Bricaud et al. 1981, Kirk 2011). The slope for the decay function (designated as
SCDOM) is generally consistent for a given body of water. Because the relationship between
absorbance and wavelength is assumed consistent, the curve can be normalized to a single
reference wavelength—usually 440 nm.
EQ5
n , —
aCDOM\.AJnorm ~
The actual absorbance of light at a given wavelength will be governed entirely by the
concentration of CDOM at the reference wavelength. Therefore, all that is needed is the
absorbance at 440 on a filtered sample in order to "reconstitute" the entire absorbance vs
wavelength curve. That is:
= aCDOM(440) * aCDOM(A~)norm EQ 6
Note : a(440) vs CDOM concentration is generally considered to be a linear relationship.
-4-
-------
Calibration method: water samples were processed as described in Keith et al. (2002). CDOM
measurements were made using a Perkin-Elmer Lambda 35 spectrophotometer using a 10 cm
path length cell with deionized water as the reference. Measured absorbance was recorded at
1 nm intervals from 400 to 750 nm. Final data scans were averaged1 every 5 nm.
Generally, when using a spectrometer to measure the concentration of a dissolved compound or
compounds the only interaction of the beam of light with the substance in the cuvette is either
absorption or transmission. However, even under the best of conditions some light will be
scattered out of the pathway (scattering due to water is accounted for with the blank, but not
scattering due to CDOM). Since this is neither absorbed nor transmitted, it is a source of error —
it did not get measured as part of the transmitted light. To correct for this, measured absorbance
recorded at 750 nm was subtracted from values measured at all of the other wavelengths.
Absorbance at 750 nm is usually assumed to be zero, therefore any recorded non-transmittance
at 750 nm was assumed to be a loss of light due to scattering rather that true absorbance. These
corrected measurements are in absorbance, which is the logio of the ratio of transmitted to
incident light. Because we are interested in the absorption coefficient (which is related
to exponential decay), we actually need the natural log of that ratio — which is the same as
multiplying the original Perkin-Elmer absorbance by 2.303 (the natural log of 10). Finally,
the absorption coefficient was divided by the path length in meters (0. 1 m) to convert the
units torn"1.
Absorption by non-pigmented particles (or non-algal particles (NAP) — As with CDOM, the
relationship between absorbance due to non-pigmented parti culates and wavelength of light
follows an exponential decay function (Gallegos 2001, Roesler et al. 1989). This decay
coefficient (SNAP) is fairly consistent for a given body of water. As with CDOM, the NAP curves
also were normalized to a reference wavelength — 440 nm.
aNAPWnorm = e^™440 EQ 7
The actual absorbance of light at a given wavelength will be governed largely by the
concentration of TSS at the reference wavelength. Once that relationship is established the entire
absorption curve can be recreated. That is:
= aw^P(440) * aNAP(X)norm EQ 8
Note: unlike CDOM, «NAP (440) vs TSS concentration is not necessarily a linear relationship.
Therefore, the calibration step needs to include establishing the site-specific relationship between
the absorbance at a reference wavelength (usually 440 nm) and TSS. Our data for ONAP vs TSS
was very scattered. However, the range of TSS values in Narragansett Bay was narrow compared
to other locations (e.g, Chesapeake Bay — Gallegos et al. 2006). Our data appeared to be
consistent with that presented by Babin et al. (2003), therefore in our final model we used their
relationship between ONAP and TSS.
1 For example, data for 400 nm is the average of 400 through 404 nm, for 405 the average of 405 through 409,
and so forth.
- 5 -
-------
Calibration method: Calibration was performed in two parts. We followed the procedure
outlined in Gallegos and Neale (2002) and Mitchell et al. (2003), except we used a 50 mm
integrating sphere coupled to the Perkin-Elmer Lambda 35. Sample water (200 mL) was filtered
using a glass fiber filter (GF/F, nominal pore size of 0.7 |im), and absorption on the filter
(from 350 to 750 nm) was measured before and after extraction with methanol (to remove algal
pigments—see below)2. The second measurement is the one needed for non-pigmented particles.
The difference between the two spectra is the absorption by phytoplankton pigments (see next
section). Just as with CDOM, the decay slope for absorption coefficient plotted against
wavelength was calculated by fitting the exponential equation to the data. The second calibration
step was to establish the relationship between TSS and absorption at 440 nm. TSS was
determined by filtering a volume of water (e.g., 500 mL) onto a pre-weighed glass fiber filter.
Just as with CDOM, we multiplied the absorbance data from the spectrophotometer by 2.303
to convert to absorption coefficients (see Kishino et al. 1985, Tassan and Ferrari 1995). In
addition, correction for volume filtered and area of the filter was included (see below). The
750 nm correction listed above for CDOM is even more important with the quantitative filter
technique. Scattering of the light away from the entrance of the integrating sphere may be
appreciable, and will be recorded as absorption. Generally, absorption at 750 nm is considered
to be negligible, so any values recorded at this wavelength were likely due to scattering.
In addition to the above scattering error, photons have a high probability of experiencing
multiple scattering events on their way through the filter. This increases the effective path length
of the light for which correction is also needed. Fortunately, Cleveland and Weidemann (1993)
have done this correction empirically:
ODsusp(X) = 0.378 ODfilt(X) + 0.523 ODfilt(X)2 EQ 9
Where ODsusp is the optical density of the suspended particles for a given wavelength, and ODfut
is the optical density of the particles on the filters. Remember, optical density is the ratio of the
logic of the absorbance of the sample to the logic of a reference. To adjust this to an absorption
coefficient—which accounts for the exponential decrease in light with depth—we have to
convert to natural logarithms (multiply by 2.303). Finally, we have to account for the volume
filtered (V) and the clearance area (A) of the filter. If we express the former in m3 and the latter
in m2, then the unit for the absorption coefficient will be in m"1:
2.303 ODsusp(X) T?rvin
o-NApW = vi EQ 10
IA
Methanol treatment—After each filter was measured for total particle absorption, each was
placed into a separate polystyrene petri dish (60 x 15 mm) and 10 mL of room temperature
methanol was gently added. Extraction in the methanol lasted a minimum of 1 hr. Filters were
then removed from the petri dish, replaced into the filter apparatus, and the methanol within the
petri dish run through the filter to remove any particulates that might have been dislodged during
2 Because the degree of wetness of the filter can affect the scan results, after filtering each sample, lOOuL of 0.22 um
filtered seawater was added back to each filter just prior to measurements.
-6-
-------
the extraction process. Afterwards, 10 mL of 0.22 um filtered seawater was used to rinse the
methanol from the filter.
Absorption by phytoplankton—The relationship between absorbance due to phytoplankton and
wavelength does not follow a set function (e.g., exponential) as did CDOM and non-algal
particulates. Therefore, an empirically derived absorption spectrum is used in the modeling
effort. We used two procedures to establish this average spectrum. The first is derived from the
most recent research on seagrass bio-optical models. (Biber et al. 2008, Gallegos 2005, Gallegos
and Neale 2003, Gallegos et al. 2006). Here the empirically derived spectra were normalized to
absorption at 675 nm and then averaged by wavelength.
EQ 11
UQ^UtJJ
Where:
a(f>(tynorm = the normalized absorption of pigmented particles for a given wavelength,
a0(l) = the absorption of pigmented particles for a given wavelength, and
a^(675) = the absorption of pigmented particles at 675 nm.
A relationship between chlorophyll a (Chl a) concentration and a^(675) was used to establish
a specific pigment absorption spectra for a given location. As with aNAp(440) vs TSS
concentration, the relationship between absorption at 675 nm and chl a concentration is not
necessarily linear. Therefore, the calibration step included establishing the site specific
relationship between the absorbance at 675 nm and chl a concentration.
a0(675) =A[Chla]B EQ 12
The coefficients A and B are empirically derived. The actual absorbance of light at a given
wavelength now can be estimated entirely by the concentration of chlorophyll a using the
reference wavelength. That is:
a^(A) = a^(675) * a^(X)norm EQ 13
In a second approach, we established relationships between Chl a and every wavelength between
400 and 700 nm in 5 nm steps. This is the approach used by Bricaud et al. (1995) and Matsuoka,
et al. (2007). This latter approach accounts for the fact that phytoplankton light absorption in the
blue region of light is much more variable than that in the red region. This is largely due to
packaging effect (the amount of pigment per unit cell and its arrangement within the cell), photo-
acclimation, or both. Photo-acclimation can result in a significant change in the concentration of
accessory photosynthetic pigments, as well as non-photosynthetic, photo-protective pigments.
Both of these types of pigments have their greatest absorption in the blue region.
Calibration method: As with absorption by non-pigmented particles, absorption by
phytoplankton pigments requires two steps. However, the first step—establishing the absorption
spectrum—was accomplished during the calibration of the non-pigmented particles by
subtracting from the total absorption spectrum of the filter the absorption spectrum of the
-7-
-------
particulates alone. The second calibration step needed was to establish the relationship between
the concentration of Chi a and absorption (from the particle filter method) at wavelengths of
interest. In general, the procedure for collecting and processing water samples followed those
used by NASA (Mitchell, et al. 2003)—which were consistent with the procedure used by
Gallegos and Neale (2002) for their work in Case 2 waters. Chlorophyll a analyses were similar
to those described in Keith et al. (2002). Samples (100 mL) were filtered onto 47 mm Whatman
GF/F filters using a hand vacuum pump. Filters were placed into 15 mL screw-capped
polystyrene tubes containing 10 mL of 90% acetone, and extracted in a freezer overnight
(approximately 18 hr) in the dark. Fluorescence was measured using a Turner Designs Model
AU-10 fluorometer equipped with the Non-Acidification Optical Kit (P/N 10-040R).
Scattering
Unlike absorption, whose parameters can be measured with typical instrumentation available
in most laboratories, scattering requires specialized equipment. We used an ac-s in situ
spectrophotometer from WET Labs to measure total scattering. This instrument simultaneously
measures total light attenuation and total absorbance in approximately 4nm increments between
the wavelengths of 400 and 730 nm. Scattering is calculated by the difference between total
attenuation and total absorption. Absorption was corrected for scattering using 739 nm as the
reference wavelength according to the procedure described by Slade et al. (2010). The ac-s
measurements were made in the laboratory using a gravity fed system (Slade et al. 2010).
The actual scattering of light at a given wavelength is governed largely by the concentration
of TSS. Scattering is assumed to be minimally influenced by CDOM (if at all) and Chi a has no
direct influence. Its influence on scattering is through the "particles" (phytoplankton) with which
it is associated. The relationship between scattering at a reference wavelength (555 nm) and total
suspended solids (TSS) is described by the equation (Kenworthy et al. 2014):
bp(555~) = ccTSS? EQ 14
Where:
bp(555) = total particle scattering at 555 nm,
TSS = total suspended solides (mg/L), and
a and P are empirically derived coefficients.
In 2014 backscattering was measured directly in situ Narragansett Bay using a Satlantic
Profiler II Ocean Profiler (Satlantic LP Canada), equipped with a backscattering sensor
(a 470 and 700 nm). We also made some direct backscattering measurements in 2013 using
a HydroScat (Hydro-Optics, Biology and Instrumentation Laboratories) (at 420 nm). The data
from both were combined3 to establish the relationship between TSS (mg/L) and backscattering
3 Backscattering data from the HydroScat at 420nm was adjusted downward using the relationship between
wavelength and particle backscattering established using the Satlantic profiler.
-8-
-------
coefficient (m"1). Wavelength specific backscattering was calculated assuming a linear
relationship between backscattering at 470 and 700 nm.
As with absorption, there is no calibration associated for scattering by pure water, it is
considered a standard relationship with wavelength of light (Buiteveld et al. 1994).
Backscattering is assumed to be half of the total scattering The backscattering absorption
coeffients are plotted in Figure 3.
Backscattering = 1/2 bw
C
_QJ
'
-------
0.0
CDOM-Narragansett Bay
400 450 500 550 600
Wavelength (nm)
650
700
Figure 4. CDOM absorption scans for samples from Narragansett Bay—summer 2013 and spring 2014.
Data for these curves are presented in Appendix A
CDOM-Normalized & SD
2.25
2.00 -
1.75 -
1.50 -
o'
S. 1.25
Vt
> 1.00
_a
<
0.75
0.50 -
0.25 -
0.00
400 450 500 550 600
Wavelength (nm)
650
700
Figure 5. Summary of CDOM calibration data for Narragansett Bay—normalized at 440 nm. Solid black
line is the mean of all 51 curves. Red dashed lines are plus and minus one standard deviation. Solid yellow
line is the best fit trend line (Equation 5). The open and closed markers are data from Gallegos and Neale
(2002) for the Rhode River, Maryland and the St. Johns River, Florida, respectively
- 10-
-------
NAP Calibration
Figure 6 shows the Non-algal particle (NAP) absorption scans for samples collected during the
summer of 2013. Figure 7 is a summary the data presented in Figure 6 normalized to 440 nm.
The overall average spectral slope (SNAP) was 0.0136 nm"1 (n = 37). As with CDOM, this also is
similar to the overall average (0.0123 nm"1) presented in Babin et al. (2003) for 348 NAP
samples for coastal waters around Europe. Although, it is greater than the average reported
(0.0088 nm"1) for recent work in the Chesapeake Bay (Gallegos et al. 2006). Figure 7 also shows
a slight depression in the absorption coefficient for wavelengths less than 425 nm. It is not clear
why this happened; however, a similar effect is shown in samples from the Baltic Sea in Babin et
al. (2003). For the purposes of the bio-optical model for Narragansett Bay, this slight shift in
absorption coefficient was ignored.
The absorption curves for NAP occasionally were not smooth, especially in the 400 to 450 and
650 to 700 nm ranges. This is not uncommon. We followed the procedure as presented by Babin
et al. (2003) and Matsuoka et al. (2011), whereby the non-linear regression was conducted on
each curve without the data in these ranges. This was done to "avoid any residual pigment
absorption that might still have been present."
1.6 n
0.0 -I-
400
NAP Absorption
450 500 550 600
Wavelength (nm)
650
700
Figure 6. Non-algal particle (NAP) absorption scans from Narragansett Bay—summer 2013 (July, August
and September). Data for these curves are presented in Appendix B
- 11 -
-------
Normalized Absorbance: NAP
1.80 n
1.60 -
.. _ 0-0.0136*(A-440)
y ~ K
0.00
400 450 500 550 600
Wavelength (nm)
650
700
Figure 7. Summary of NAP calibration data for Narragansett Bay—normalized at 440 nm. Blue markers
represent the mean of all 37 curves. Red dashed lines are plus and minus one standard deviation. Solid black
line is the best fit trend line (Equation 7)
Typically, a correlation between absorption by non-pigmented particles and TSS is used to adjust
the normalized calibration curve to in situ absorption coefficients (Gallegos et al. 2006, and
others—including Kenworthy et al. 2014). Figure 8 (top left) shows the relationship between
«NAP (440) and TSS for samples collected in Narragansett Bay. There does not appear to be a
relationship. However, the range of TSS values in our data set is narrow compared to those in
other data sets. Gallegos et al. (2006) show what appears to be a similar amount of scatter within
their Figure 9a. Our Figure 8 (top right) shows the Narragansett Bay data re-plotted with the
range on the axes the same as the range used in Gallegos et al. (2006). The degree of vertical
scatter is similar to that displayed in Gallegos et al. (2006) Figure 9c—although the absorption
coefficient in Narragansett Bay for a give TSS value is slightly less than that in their Figure 9c.
Similarly, we re-plotted the data along with the regression for «NAP vs TSS (SPM4 in their paper)
published in Babin et al. (2003), see Figure 8 (bottom). The degree of scatter within the range of
TSS for Narragansett Bay is similar to that depicted in Figure 15 of Babin et al. (2003). The
central tendency of our absorption coefficient data appears to line up fairly well with the
regression from Babin et al. (2003), so for the purpose of using TSS to establish NAP absorption
(m"1) within Narragansett Bay, we used their equation (0.31 * TSS in mg/L).
' Suspended Particle Matter
- 12-
-------
1.2 -
— °-8 '
]3 06
CL
<
z
0.2 -
0
1
0.1
£ o.oi
CL
<
z
0.001
0.0001
0
oNAP(440),(m-1
•
• •
« . •
• •
• •
•
•
• •
•
•
5 10 15 20 25
TSS (mg/L)
0NAP(440),(m-1
/
/
/
/
,;
A
/
<_
10.0 -
9.0 -
8.0 -
7.0 -
° 6.0 -
£ 5.0 -
CL
< 4.0 -
3.0 -
2.0 -
oNAp(440),(m-1)
./
0 50 100 150 200 250
TSS (mg/L)
Figure 8. The relationship between non-algal
particle absorption and total suspended solids for
Narragansett Bay. The data in the plot at the top
left are re-plotted in the top right plot using the
axes ranges from Gallegos et al. (2006). The data
are re-plotted again in final panel using the axes
from Babin et al. (2003). Data are in Appendix C.
01 0.1 1 10 100
TSS (mg/L)
Phytoplankton Calibration
Figure 9 shows the phytoplankton (pigmented particles) absorption scans for samples collected
during the summer of 2013 (July, August and September). Each curve is normalized to the
absorption at 675 nm. Figure 10 shows the average of all 31 normalized curves. This is the
empirical calibration curve for the bio-optical model. The final absorption curve is created based
on the chlorophyll a concentration for a given site (see Equation 3). Figure 11 shows the
relationship between chl a and phytoplankton particle absorption at 675 nm. A similar plot is
shown in Figure 12 for chl a and absorption at 440 nm. Table 1 contains all of the coefficients
for absorption at wavelengths ranging from 400 to 700 nm in 5 nm increments relative to chl a
concentration. These data can be used to establish a calibration curve whose shape will vary
depending on the concentration of chl a. This is the process used by Matsuoka et al. (2007) and
others to create their absorption curves. Figure 13 compares calculated absorption curves using
- 13 -
-------
both procedures. The "normalize" curves use Figure 10 and the relationship in Figure 11
to create the five phytoplankton absorption curves for chl a ranging from 0.5 to 20 ug/L.
The "A-B" curves used the data presented in Table 1 to create an additional five absorption
curves. The difference between the two calibration techniques is chl a concentration dependent
and confined largely to the blue end of the spectrum. This region of the curves are influenced
by photo-acclimation. During photo-acclimation cells growing in higher light (e.g, low
concentrations of Chl a in the water) create additional pigments (mostly non-photosynthetic
pigments) for protection against the higher energy wavelengths (the blue region). The clearer
the water (i.e., the lower the concentration of Chl a), the greater the expected concentration of
photo-protective pigments—which have their primary absorption in the higher energy (short
wavelengths) end of the spectrum.
Summer-2013
3.0 n
0.0
400 450 500 550 600
Wavelength (nm)
650
700
Figure 9. Phytoplankton particle absorption scans from Narragansett Bay—summer 2013 (July, August and
September). Each curve is normalized to its absorption at 675 nm. Original, non-normalized data are in
Appendix D
- 14-
-------
2013 AVG--normalized to 675
1.8 -i
1.6 -
o
I 0.8 H
_TD
Q.
2 0.6 -
°~ 0.4 -
0.2 -
0.0
400 450 500 550 600
Wavelength (nm)
650
700
Figure 10. Average of the normalized pigmented particle (phytoplankton) absorption scans from Figure 9
0.7 n
0.6 -
0.5 -
0.4
0.3
0.2 -
0.1 -
Abs (675)
0.0282 * Chi a0-855
0 5 10 15 20 25 30 35 40
Chi a (ug/L)
Figure 11. The relationship between chl a concentration and particle absorption at 675 nm. Data are in
Appendix E
- 15 -
-------
Abs (440)
0.8 -|
0.7 -
0.6 -
- 0.5
E
° 0.4
.Q
< 0.3
0.2 -
0.1 -
0
0.0695 * Chi a0-673
10 15 20 25 30 35
Chi a (ug/L)
40
Figure 12. The relationship between chl a concentration and particle absorption at 440 nm. Data are in
Appendix E
Phytoplankton
0.70
Normalized 20 ug/L
Normalized 10 ug/L
Normalized 5 ug/L
Normalized 2.5 ug/L
Normalized 0.5 ug/L
A-B 20 ug/L
A-B 10 ug/L
A-B 5 ug/L
A-B 2.5 ug/L
A-B 0.5 ug/L
0.00
400 450 500 550 600
Wavelength (nm)
650
700
Figure 13. A comparison of phytoplankton particle absorption spectra using two different methods of
calibration. See text for explanation. Top 2 plots are for 20 ug/L, next 2 are for 10 ug/L, and so on
- 16-
-------
Table 1. Parameters for power regression expressed as Abs(X) = A*Chl if. Units of absorption are m-1, units for
chl a are ug/L. These data are used to create absorption curves for phytoplankton particles—the shape of which is
chl a concentration dependent.
Wavelength
(nm)
400
405
410
415
420
425
430
435
440
445
450
455
460
465
470
475
480
485
490
495
500
505
510
515
520
525
530
535
540
545
550
A
0.0576
0.0617
0.0663
0.0703
0.0634
0.0641
0.0675
0.0702
0.0695
0.0675
0.0642
0.0632
0.0627
0.0611
0.0588
0.0557
0.0533
0.0506
0.0476
0.0436
0.0389
0.0344
0.0302
0.0266
0.0238
0.0213
0.0190
0.0172
0.0156
0.0140
0.0126
B
0.636
0.642
0.644
0.643
0.669
0.662
0.669
0.672
0.673
0.667
0.670
0.669
0.666
0.664
0.659
0.658
0.661
0.673
0.686
0.701
0.717
0.731
0.748
0.763
0.774
0.787
0.802
0.810
0.816
0.819
0.818
r2
0.647
0.678
0.700
0.715
0.713
0.674
0.723
0.751
0.764
0.770
0.779
0.787
0.792
0.795
0.786
0.782
0.780
0.786
0.796
0.805
0.810
0.809
0.810
0.806
0.802
0.805
0.809
0.808
0.808
0.800
0.792
Wavelength
(nm)
555
560
565
570
575
580
585
590
595
600
605
610
615
620
625
630
635
640
645
650
655
660
665
670
675
680
685
690
695
700
A
0.0111
0.0097
0.0094
0.0089
0.0081
0.0075
0.0074
0.0071
0.0068
0.0067
0.0071
0.0077
0.0085
0.0089
0.0093
0.0098
0.0102
0.0104
0.0104
0.0103
0.0116
0.0158
0.0220
0.0272
0.0282
0.0237
0.0162
0.0089
0.0050
0.0032
B
0.819
0.831
0.814
0.815
0.840
0.862
0.867
0.870
0.867
0.869
0.872
0.881
0.886
0.885
0.881
0.877
0.861
0.844
0.835
0.858
0.891
0.895
0.882
0.867
0.855
0.864
0.888
0.913
0.904
0.879
r2
0.772
0.735
0.756
0.757
0.761
0.769
0.766
0.753
0.726
0.716
0.728
0.760
0.788
0.806
0.815
0.830
0.839
0.839
0.847
0.865
0.898
0.921
0.929
0.933
0.931
0.928
0.916
0.871
0.776
0.626
Backscattering Calibration
Scattering is the most variable of the parameters measured for water clarity estimates using a
bio-optical model. Scattering data also are not as prevalent in the literature as are absorption
data, since scattering measurements require specialized instrumentation that has only recently
become more available. Calibration for scattering coefficients typically involves establishing
a correlation between TSS and total scattering for a reference wavelength (usually 555 nm).
Although relationships between a combination of parameters (e.g., TSS, Chl a and CDOM) have
- 17-
-------
been used occasionally (Gallegos 2005). No matter what parameter or combination of
parameters, the variability of scattering coefficients is usually fairly high, especially when
compared with that for absorption coefficients. One plausible explanation is that scattering is
highly dependent on not only the concentration of suspended matter, but also the specific make
up of that material with respect to size, shape and composition (especially inorganic versus
organic). And, unlike absorption measurements, there is not an easy method to segregate
measurements on the different components.
Figure 14 shows the relationship between total scattering at 555 nm and the concentration of TSS
for samples collected in Narragansett Bay. Clearly there is not a good relationship. There also
was no relationship between location or time of year, with respect to the relationship. Gallegos
(2005) explains some of the variation observed in scattering coefficient for water samples
collected in the lower St. Johns River, Florida by segregating his samples by salinity. He was
able to show better relationships among TSS and total scattering than we were; however, the
ranges of values for both TSS and measured scattering were greater for his measurements. When
our data are plotted with these expanded ranges (Figure 15), the degree of variability is very
similar to that presented in Figure 9c of Gallegos (2005). This is also true when we compare our
data using observation from Figure 1 la in Gallegos et al. (2006) for Chesapeake Bay. Figure 16
is our data re-plotted again, this time using the axes ranges from Gallegos et al. (2006).
b(555)
10 -i
9 -
^ 8 -
-§. 7 -
QJ C
._ D ~
-------
b(555)
^H" ,fi
_£
C
it
TO
y A -
2 .
n -
,
. %•
f- 1
* *
•
•
*• «
B
(
I
10 20 30 40 50
TSS (mg/L)
60
70
80
Figure 15. The same data as presented in Figure 14 except the axes are expanded to coincide with those
from Gallegos (2005)
b(555)
120 -i
1 1 n -
*-\ yo •
C Qf) _
-M
n\ 70 -
[u
M— en .
(U
o
U en .
COD
C /in .
l_
(U
±^ Qn .
TO
U
l/l nn
c
ft
} 2
1
5 5
0 7
5 1(
DO i;
TSS(r
!5 IE
Tlg/L)
>0 1"
'5 2C
JO 2^
.5 25
Figure 16. The same data as presented in Figure 14 except the axes are expanded to coincide with those
from Gallegos et al. (2006)
- 19-
-------
Our highly variable relationship between total scattering and TSS is consistent with what others
have experienced. We were able, however, to establish a better relationship between TSS and
particle backscattering. Data showing this are presented in Figure 17. Gallegos et al. (2006) also
had a relationship between backscattering and TSS. The range of their data was estimated from
their Figure 12c. The backscattering data from Gallegos et al. (2006) was measured at 532 nm.
We adjusted these data upward by about 10% to account for the slight influence of wavelength
on backscattering (see below). Even though backscattering is a better relationship to TSS, it still
is not great, but it is likely the best that can be done for now. The regression line is a non-linear
fit to a power function. The equation for the line is:
V(47°) = °-0095 * TSS0-379 EQ 15
Where:
fr&p(470)= particle backscattering at 470 nm (m"1), and
TSS = total suspended solids (mg/L).
A calibration curve for backscattering versus wavelength was established using the data we have
for backscattering at 470 nm and 700 nm from the Satlantic profiles. By necessity we assumed
that the relationship between wavelength and backscattering was linear. This line is:
= -0.041
0.0477
EQ 16
We then took these data and normalized them to bbp (470). This relationship is multiplied by the
bbp (470) calculated from the relationship in Figure 17 to achieve the final backscattering
coefficients for a given site.
0.050 -i
backscattering (m-1)
0.000
15 20
TSS (mg/L)
Figure 17. Relationship between backscattering at 470 nm and total suspended solids for Narragansett Bay. The red
markers are data collected using a Hydro Scat in 2013. The blue markers are data from the Satlantic Profiler in 2014.
The blue solid line is the regression (Equation 15) fit to the red and blue markers. The vertical dashed lines are
ranges estimated from Gallegos et al. (2006) for comparison. Data are in Appendix F.
-20-
-------
PUTTING IT ALL TOGETHER
Referring back to Equation 1, we need three pieces of information in order to calculate the
diffuse extinction coefficient (Kd): the solar zenith angle of incidence in degrees, total absorption
by wavelength, and total backscattering, also by wavelength. The zenith angle is easily
calculated using NOAA's solar calculator, which can be used directly on their webpage, or
spreadsheets can be downloaded (htt!)i//ww^
All that is needed is the latitude and longitude for a site, date, time zone and time of day. For
total absorption and backscattering all that is needed are measurements or modeled values for
CDOM, TSS and Chi a.
Total absorption (Equation 3) is wavelength dependent. We have chosen to make the calculation
in 5 nm increments from 400 to 700 nm. Absorption by water (awater) has a fixed relationship
with wavelength (see Figure 2). Absorption by CDOM for Narragansett Bay is calculated from:
Where: O-CDOM (440) is the directly measured (or modeled) value of CDOM absorption at 440 nm.
Absorption by non-algal particles is calculated by first deriving the absorption at 440 nm:
0^(440) = 0.31* TSS
Where: TSS is total suspended solids in mg/L. Absorption for each wavelength of interest is
then calculated using:
n m — r, fAA(\\ * ,,-0.0136*(A-440)
aNAP lyU — aNAP 144U ) * 6 ^ J
Finally, absorption by phytoplankton, a^(/l), is estimated by first deriving the absorption at
675 nm, using:
a0(675) = 0.0282[C/i/a]
0.855
Where: [Chi a] is the concentration of chlorophyll a in |ig/L. This value is then used to establish
the absorption at all of the wavelengths of interest, using:
Where: a0(A)norm is the wavelength dependent absorption curve, normalized to 675 nm
(see Figure 10). Alternately, we could derive the phytoplankton absorption curve by calculating
absorption for each wavelength as an independent relationship with the concentration of Chi a
(see Table 1).
Total backscattering is calculated as the sum of backscattering from pure water and
backscattering due to everything else. Backscattering from water is considered a fixed
relationship with wavelength (see Figure 3). Backscattering by the constituents within
-21 -
-------
Narragansett Bay water is calculated based on spectral backscattering coefficients normalized
to 470 nm. Backscattering at 470 nm is calculated using:
-379
V(47°) = °-0095 * TSS0-
Where: TSS is total suspended solids in mg/L. Final spectral backscattering coefficients
are calculated using:
bbp(X) = V(47°) * (1-678 - 0.00144 * 1)
The latter portion of the right side of the above equation is the Equation 16 normalized at 470 nm
(dividing Equation 16 by Equation 16 solved using X = 470).
The total spectral absorption and spectral backscattering coefficients are then plugged into
Equation 1 to calculate the spectral diffuse attenuation coefficients. These are used, in turn, in
Equation 2 to calculate the total diffuse attenuation coefficient for photosynthetically active
radiation (PAR). To do this you need the spectrum of light at the surface of the water. This
information can be calculated using readily available free software from the National Renewable
Energy Laboratory (htt]x//wwwjrf What is needed is the global horizontal
irradiance (the irradiance hitting a horizontal surface). This is a combination of direct, beam and
diffuse irradiance. The units used are not important (Wm~2 or photon flux are often used), the
attenuation coefficent is a ratio — the units cancel out.
-22-
-------
REFERENCES
Babin, M, D Stramske, GM Ferrari, H Claustre, A Bricaud, G Obolensky, and N Hoepffner.
2003. Variations in the light absorption coefficients of phytoplankton, nonalgal particles,
and dissolved organic matter in coastal waters around Europe. Journal of Geophysical
Research 108, C7, 3211, doi: 10.1029/2001JC000882.
Biber, PD, CL Gallegos, and WJ Kenworthy. 2008. Calibration of a bio-optical model in the
North River, North Carolina (Albemarle-Pamlico Sound): a tool to evaluate water quality
impacts on seagrass. Estuaries and Coasts. 31:177-191.
Bricaud, A, M Babin, A Morel, and H Claustre. 1995. Variability in the chlorophyll-specific
absorption coefficients of natural phytoplankton: Analysis and parameterization. Journal of
Geophysical Research. 100, C7:13,321-13,332.
Bricaud, A, A Morel, and L Prieur. 1981. Absorption by dissolved organic matter of the sea
(yellow substance) in the UV and visible domains. Limnology and Oceanography 26:43-53.
Buiteveld, H, JHH Hakvoort, and M Donze. 1994. The optical properties of pure water,
Ocean OpticsXIII, SPIE 2258, 174-183.
Cleveland, JS and AD Weidemann. 1993. Quantifying absorption by aquatic particles: a multiple
scattering correction for glass-fiber filters. Limnology and Oceanography 3 8(6): 1321-1327.
Gallegos, CL. 2001. Calculating Optical Water Quality Targets to Restore and Protect
Submersed Aquatic Vegetation: Overcoming Problems in Partitioning the Diffuse
Attenuation Coefficient for Photosynthetically Active Radiation. Estuaries. 24(3):381-397.
Gallegos, CL. 2005. Optical water quality of a blackwater river estuary: the Lower St. Johns
River, Florida, USA. Estuarine, Coastal and Shelf Science. 63:57-72.
Gallegos, CL, EA Lewis, and H-C Kim. 2006. Coupling suspended sediment dynamics and light
penetration in the upper Chesapeake Bay. Report from the Smithsonian Environmental
Research Center. http://www.chesapeakebay.net/content/publictions/cbp_13364.pdf.
Gallegos, CL and PJ Neale. 2002. Partitioning spectral absorption in case 2 waters:
discrimination of dissolved and particulate components. Applied Optics. 41(21):4220-4233.
Keith, DJ, JA Yoder, and SA Freeman. 2002. Spatial and temporal distribution of coloured
dissolved organic matter (CDOM) in Narragansett Bay, Rhode Island: Implications for
phytoplankton in coastal waters. Estuaries, Coastal and Shelf Science. 55:705-717.
Kenworthy, WJ, CL Gallegos, C Costello, D Field, and G di Carlo. 2014. Dependence of
eelgrass (Zostera marina) light requirements on sediment organic matter in Massachusetts
coastal bays: Implications for remediation and restoration. Marine Pollution Bulletin
83:446-457.
Kirk. JTO. 2011. Light and Photosynthesis in Aquatic Ecosystems. 3rd Ed. Cambridge University
Press. 649 pp.
-23 -
-------
Kishino, M, M Takahaski, N Okami, and S Ichimura. 1985. Estimation of the spectral absorption
coefficients of phytoplankton in the sea. Bulletin of Marine Science 37(2):634-642.
Lee, Z-P, K-P Du and R Arnone. 2005. A model for the diffuse attenuation coefficient of
downwelling irradiance. Journal of Geophysical Research. 110:C02016, doi.
10.1029/2004JC002275.
Matsuika, A, V Hill, Y Huot, M Babin, and A Bricaud. 2001. Seasonal variability in the light
absorption properties of western Arctic waters: Parameterization of the individual
components of absorption for ocean color applications. Journal of Geophysical Research.
116, C02007, doi:10.1029/2009JC005594.
Matsuoka, A, Y Huot, K Shimada, S-I Saitoh, and M Babin. 2007. Bio-optical characteristics of
the western Arctic Ocean: implications for ocean color algorithms. Can. J. Remote Sensing
33(6):503-518.
Mitchell, BG, M Kahru, J Wieland, and M Stramska. 2003. Determination of spectral absorption
coefficients of particles, dissolved material and phytoplankton for discrete water samples.
(in) JL Mueller, GS Fargion and CR McClain, editors. Ocean Optics Protocols for Satellite
Ocean Color Sensor Validation, Revision 4, Volume IV: Inherent Optical Properties:
Instruments, Characterization, Field Measurements and Data Analysis Protocols. Report #
NASA/TM-2003-211621/Rev4-Vol.IV. National Aeronautical and Space Administration.
Goddard Space Flight Space Center.
Pope, RM and ES Fry. 1997. Absorption spectrum (390-700 nm) of pure water. II. Integrating
cavity measurements. Applied Optics. 36:8710-8723.
Roesler, CS, MJ Perry, and KL Carder. 1989. Modeling in-situ phytoplankton absorption from
total absorption spectra in productive inland marine waters. Limnology and Oceanography
34:1510-1523.
Slade, WH, E Boss, G Dall'Olmo, MR Langner, J Loftin, MJ Behrenfeld, C Roesler, and TK
Westberry. 2010. Underway and moored methods for improving accuracy in measurement
of spectral particulate absorption and attenuation. Journal of Atmospheric and Oceanic
Technology. 27:1733-1746.
Tassan, S and GM Ferrari. 1995. An alternative approach to absorption measurement of aquatic
particles retained on filters. Limnology and Oceanography 40(8): 1358-1368.
-24-
-------
Appendices
Appendices A through F are in the attached Microsoft Excel file, Bio-optical model appendices—
data for Narragansett Bay.xlsx. To access the file, select the attachments icon within Adobe
Acrobat. It contains the data used to create the figures associated with the calibration for absorption
and scattering.
-25 -
------- |