Bio-optical properties in waters influenced by the...

Bio-optical properties in waters influenced by the Mississippi River

during low flow conditions

Eurico J. D’Saa,*, Richard L. Millerb

aGB Tech, NASA Earth Science Applications Directorate, Stennis Space Center, MS 39529, USAbNASA Earth Science Applications Directorate, Stennis Space Center, MS 39529, USA

Received 15 July 2002; received in revised form 26 September 2002; accepted 26 September 2002

Abstract

Spatial and temporal patterns of bio-optical properties were studied in the Northern Gulf of Mexico during cruises in April and October of

2000, a year during which the discharge volume from the Mississippi River was unusually low. Highly variable surface Chl a concentrations

(0.1 to 17 mg m� 3) and colored dissolved organic matter (CDOM) absorption (0.07 to 1.1 m� 1 at 412 nm) were observed in the study region

that generally decreased with increasing salinity waters, being highest nearshore and decreasing at offshore stations. The optical properties of

absorption, scattering, and diffuse attenuation coefficients reflected these distributions with phytoplankton particles and CDOM contributing

to most of the spatial, vertical, and seasonal variability. The diffuse attenuation coefficient Kd(k) and spectral remote sensing reflectance

Rrs(k) were linear functions of absorption and backscattering coefficients a(k) and bb(k) through the downward average cosine ld and the

ratio of variables f/Q at the SeaWiFS wavebands for waters with widely varying bio-optical conditions. Although various Rrs(k) ratio

combinations showed high correlation with surface Chl a concentrations and CDOM absorption at 412 nm, power law equations derived

using the Rrs(490)/Rrs(555) and Rrs(510)/Rrs(555) ratios provided the best retrievals of Chl a concentrations and CDOM absorption from

SeaWiFS reflectance data.

D 2002 Elsevier Science Inc. All rights reserved.

Keywords: Bio-optical properties; Mississippi River; SeaWiFS; Chlorophyll; CDOM

1. Introduction

The Mississippi River system ranks among the top 10

rivers in length, freshwater discharge, and sediment delivery

(Milliman & Meade, 1983) and thus exerts a strong influ-

ence on the biological, chemical, and physical properties of

the Northern Gulf of Mexico (Del Castillo et al., 2001;

Muller-Karger, Walsh, Evans, & Meyers, 1991). A season-

ally varying discharge of fresh water and nutrients from the

Mississippi River influences the spatial and temporal dis-

tribution of carbon fluxes and primary productivity in the

Northern Gulf of Mexico (Dinnel & Wiseman, 1986;

Lohrenz et al., 1999; Rabalais et al., 1996; Redalje, Loh-

renz, & Fahnenstiel, 1994). Recent outbreaks of hypoxia

over large areas of Louisiana and Texas coastal waters have

been attributed to high levels of primary productivity in

surface plume waters that are sustained by nutrients from the

Mississippi River (Rabalais & Turner, 2001). While satellite

remote sensing provides the potential to describe and

monitor the variability of bio-optical properties (e.g., phy-

toplankton biomass) and the oceanic carbon fluxes, there is

still a need to evaluate currently used or to develop regional

ocean color algorithms for coastal regions as those influ-

enced by the Mississippi River.

Ocean color algorithms developed to estimate biological

variables such as Chl a have been used in both case I and

case II waters (O’Reilly et al., 1998) and are based on the

premise that other in-water constituents such as colored

dissolved organic matter (CDOM) and detrital material co-

vary with Chl a, the primary pigment used to represent

phytoplankton concentrations. However, the ability of these

algorithms to accurately estimate bio-optical variables in

coastal waters is not very clear as in-water constituents such

as CDOM and suspended sediments may not co-vary with

Chl a (Carder et al., 1991; Tassan, 1994), although still

0034-4257/02/$ - see front matter D 2002 Elsevier Science Inc. All rights reserved.

PII: S0034 -4257 (02 )00163 -3

* Corresponding author. Tel.: +1-228-688-1724; fax: +1-228-688-

1774.

E-mail address: [email protected] (E.J. D’Sa).

www.elsevier.com/locate/rse

Remote Sensing of Environment 84 (2003) 538–549

influencing the water leaving radiance. The contribution of

particulate and dissolved constituents to the variability of

optical properties and ocean color in coastal waters requires

a better understanding of the linkages between the concen-

tration of these constituents, the inherent optical properties

(IOPs) of absorption and scattering coefficients, and the

apparent optical properties (AOPs) such as the spectral

attenuation for downward irradiance Kd(k) and remote

sensing reflectance Rrs(k). Knowledge of these relationshipsis important for characterizing the marine optical environ-

ment and developing remote sensing ocean color algorithms

for coastal waters.

In this study, we present results of bio-optical measure-

ments obtained during two research cruises conducted in the

coastal margin influenced by the Mississippi River in April

and October of 2000. While river discharge in spring and

fall generally corresponds to high/low flow periods, the

discharge for the study year was unusually low compared to

previous years. We evaluate the spatial and temporal vari-

ability of the bio-optical properties and the relationships

between the IOPs (absorption, scattering and backscattering

coefficients) and the AOPs (Kd(k) and Rrs(k)) within the

context of low discharge. This data may provide an under-

standing of the baseline effect of the Mississippi River on

the regional coastal waters. The properties of the two main

AOPs of importance to remote sensing, Kd(k) and Rrs(k), arediscussed and analyzed in relation to absorption and scatter-

ing in an attempt to examine closure between the AOPs and

IOPs based on field data. We also derive empirical relation-

ships between the ratios of Rrs(k) and in-water constituents

such as Chl a concentrations and CDOM absorption and

compare estimates obtained using these local algorithms

with the standard ocean chlorophyll OC4v4 algorithm

(O’Reilly et al., 1998) on a SeaWiFS image of the region.

2. Methods and data

2.1. Study site

The study site is located in the Northern Gulf of Mexico

(Fig. 1) in waters largely influenced by the Mississippi

River. Most of the River discharge flows westward onto

the Louisiana–Texas shelf (Dinnel & Wiseman, 1986). Two

research cruises using the RV Pelican were undertaken in

spring (5–13 April) and fall (24 October–1 November) of

2000. Stations were occupied in the plume waters of the

main flow of the Mississippi River (Southwest Pass) to the

outer shelf, the inner shelf, as well as some open ocean

locations in the Gulf of Mexico (station locations shown in

Fig. 1. Study site showing location of bio-optical stations occupied in the Northern Gulf of Mexico during research cruises in April and October of 2000. Bio-

optical properties of four representative stations (squares) during the April cruise and two stations (circles) in October are noted for further discussion in the text.

E.J. D’Sa, R.L. Miller / Remote Sensing of Environment 84 (2003) 538–549 539

Fig. 1). While a full suite of bio-optical measurements was

made during the April cruise, in-water bio-optical data were

obtained at only four stations during the October cruise.

However, remote sensing reflectance Rrs(k) measurements

using an above-water radiometer were made at all the

stations (Fig. 1) during both cruises. Although river flow

is generally high during spring, the average discharge

volume for the 2 months prior to the spring 2000 cruise

was about 45% lower than the previous years (discharge

data are from the U.S. Army Corps of Engineers monitoring

station at Tarbert Landing, Mississippi). The fall cruise was

conducted during a period usually considered as a low-flow

period of river discharge. Overall, river discharge was much

lower for the year, and therefore, the results presented here

may be considered as prevailing under such conditions.

Ocean color SeaWiFS satellite imagery was obtained for

one clear day in spring (e.g., April 8), while three images of

the region were obtained for the fall experiment. The

satellite data were obtained from the Goddard Distributed

Active Archive Center and processed using NASA’s Sea-

DAS (version 4.1) software.

2.2. Bio-optical measurements

Bio-optical measurements were obtained using an optical

profiling package that integrated a number of optical instru-

ments within a metal cage. Discrete water samples were

obtained immediately after the optical cast with a rosette

sampler equipped with 10-l Niskin bottles. The bio-optical

package contained sensors for measurements of conductivity

and temperature (SBE-19, Seabird), stimulated chlorophyll

fluorescence (Wetstar, WetLabs), absorption/attenuation at

nine wavebands (ac-9, WetLabs), downwelling spectral

irradiance and upwelling spectral radiance at seven channels

(OCI and OCR-200, Satlantic), and backscattering at two

spectral bands (Hydroscat-2, HOBILabs). A data acquisition

and power unit (Mpak3, WetLabs) was used to integrate and

archive the data collected from the different instruments in

the bio-optical package. Subsequently, upon transfer of the

data to a desktop computer, a separate software program was

used to unpack into separate files the data corresponding to

individual instruments. A program was then used to depth-

align the data based on time stamps referenced to the CTD

depth. The data were then separated into downcast and

upcast segments and averaged into 0.5- or 1-m bins for

further processing and analysis.

The ac-9 consists of dual, 25-cm pathlength flow tubes

for measurement of in-water absorption and attenuation at

nine 10-nm wide spectral wavebands (412, 440, 488, 510,

532, 555, 650, 676, and 715 nm). Optically clean water

from a Nanopure system (Infinity UV/UF, Barnstead) was

used to calibrate the ac-9 during the cruise (Twardowski,

Sullivan, Donaghay, & Zaneveld, 1999). Software was

written to implement three additional corrections (temper-

ature, salinity, and scattering) to the ac-9 data (Pegau, Gray,

& Zaneveld, 1997; Zaneveld, Kitchen, & Moore, 1994).

Particulate scattering coefficient bp(k) was calculated as the

difference between the corrected attenuation and absorp-

tion. The ac-9 data were generally binned at 1-m depth

intervals, but at stations with shallow depths ( < 20 m),

binning was set to 0.5 m. The deployment of the bio-

optical package at each station was repeated for dissolved

or CDOM absorption measurements using a 0.2-Am Supor-

Cap cartridge filter (Gelman Sciences) that was attached to

the intake tubes of the ac-9 meter. The difference between

unfiltered (particulate plus dissolved absorption or

[ap(k) + acdom(k)]) and filtered (dissolved absorption or

acdom(k)) profiles can be attributed to particulate absorp-

tion, ap(k). Scattering data from the Hydroscat-2 sensor

were obtained during the October cruise only and were

processed using software (Hydrolab, HOBILabs) that gen-

erated the uncorrected backscattering coefficients at two

wavelengths of 442 and 589 nm. The ac-9 derived absorp-

tion and scattering values were then used to correct the

backscattering data bb(k) (here, bb is the sum of back-

scattering due to particulates bbp and pure water bbw) to

account for losses due to attenuation of some backscattered

light in the water between the instrument and the detection

volume. The 440- and 442-nm central wavebands for the

ac-9 and the Hydroscat measurements were interpolated to

the 443-nm waveband to match the SeaWiFS waveband

(band 2). The change in values due to this shift was

insignificant for the absorption and scattering data.

Vertical profiles of spectral downwelling irradiance Ed(k)and upwelling radiance Lu(k) at seven 10-nm wavebands

(412, 443, 490, 510, 555, 665, and 683 nm) were processed

to calculate the diffuse attenuation coefficients for down-

welling irradiance Kd(k) and upwelling radiance Ku(k).These coefficients were determined as the slope of ln Ed(k)or Lu(k) versus depth. An 11-point moving linear regression

of ln Ed(k) versus depth was used to obtain vertical profiles

of Kd(k) and Ku(k). Values of Kd(k) and Ku(k) at the

shallowest depth were then used to estimate Ed(k,0�) and

Lu(k,0�) just beneath the sea surface. Based on known values

of Fresnel reflectance at the water–air interface for upwell-

ing radiance and downwelling irradiance, the subsurface

values Ed(k,0�) and Lu(k,0

�) were extrapolated to values

just above the sea surface (Mueller & Austin, 1995). Remote

sensing reflectance Rrs(k,0+) or Rrs(k) was then calculated as

RrsðkÞ ¼ Luðk; 0þÞ=Edðk; 0þÞ; ð1Þ

where Lu(k,0+) is the water-leaving radiance measured in the

nadir direction just above the sea surface and Ed(k,0+) is the

downwelling irradiance incident on the sea surface.

In addition to the in-water derived values of Rrs(k), wealso derived Rrs(k) from above-water measurements of

signals proportional to surface and sky radiance, and radi-

ance reflected from a horizontal plaque with known bi-

directional reflectance (Mobley, 1999). Above-water remote

sensing reflectance measurements of Rrs(k) from 400 to 825

nm at 1-nm spectral resolution were obtained using a GER

E.J. D’Sa, R.L. Miller / Remote Sensing of Environment 84 (2003) 538–549540

1500 (Geophysical and Environmental Research) 512-chan-

nel, fiber optic system adapted for field operations. Follow-

ing SeaWiFS protocols (Mueller & Austin, 1995), replicate

radiance spectra of the sea surface were measured with the

fiber-optic radiometer at 30j, followed by the sky radiance

and the radiance reflected from a 10% reference Spectralon

plaque (Labsphere), and the data processed for determina-

tion of above-water Rrs(k).

2.3. Discrete water sample analysis

Water samples at discrete depths were collected using a

CTD rosette multi-bottle array system (General Oceanics

Model 1015-12). Water samples collected from CTD casts

immediately following optical profiles were used for corre-

lation with optical measurements. Discrete water samples

were analyzed for: (1) Chl a and selected pigments using

high performance liquid chromatography or HPLC, (2)

particulate ap(k) and detrital ad(k) absorption coefficients

using the quantitative filter pad technique (Kishino, Taka-

hasi, Okami, & Ichimura, 1985; Mitchell & Kiefer, 1988),

and (3) CDOM absorption acdom(k) using a multiple path-

length capillary waveguide (Miller, Belz, Del Castillo, &

Trzaska, 2002). Water samples collected from Niskin bottles

were filtered through GF/F glass fiber filters and frozen/

stored in liquid nitrogen for laboratory analysis of pigments

and particle absorption. Chlorophyll a used in this study and

determined from HPLC represents chlorophyll a and deriv-

atives (chlorophyllide a, chlorophyll a, allomers and epi-

mers). Measurements of CDOM absorption were carried out

with a waveguide using pre-rinsed 0.2-Am Nuclepore nylon

filters and verified with spectrophotometric measurements

using a Perkin-Elmer ultraviolet/visible (UV/VIS) Lambda-

18 dual beam spectrophotometer.

3. Results and discussion

3.1. Spatial and seasonal variability in the bio-optical

properties

One of the consequences of the Mississippi River flow

into the Northern Gulf of Mexico is the density stratifica-

tion, with lower salinity waters overlying a more saline

oceanic water body. At most stations sampled during the

April cruise (Fig. 1), the overlying surface plume waters had

salinity values that varied spatially from lows in the 20 to 33

psu, while in the underlying oceanic waters, salinities varied

from 33 to 36.4 psu. Example stations close to the Mis-

sissippi River outflow into the Gulf of Mexico (st. 192, near

SW Pass), a mid-shelf station (st. 46) and an off-shore

station (st. 24), are shown as representing the range of bio-

optical properties at the study site during the spring cruise.

High Chl a concentrations were associated with lower

salinity surface plume waters with concentrations varying

from highs of about 17 mg m� 3 at station 192 (Fig. 2(a)) to

lows of < 0.2 mg m� 3 at the weakly stratified offshore

stations (st. 24 and 122). High surface Chl a concentrations

and CDOM absorption at stations with lower salinities (e.g.,

Fig. 2(a) and (b)) strongly attenuated the light field thus

potentially restricting primary production to near surface

waters. In comparison, a weakly stratified surface layer

(e.g., station 24, Fig. 2(c)) was associated with low surface

Chl a concentrations of about 0.2 mg m� 3, which increased

with depth to about 0.7 mg m� 3 at up to 25 m and then

again decreases. An examination of inorganic nutrient con-

centrations (Powell, LUMCON, personal communication)

indicated elevated levels of nutrient concentrations (>3 AMNO3) at most stations with high Chl a concentrations in the

river plume, while at the two offshore stations (st. 24 and

122), nitrate concentrations in surface waters were relatively

low (0.14 and 0.19 AM NO3, respectively).

Vertical profiles of particulate plus CDOM and CDOM

absorption and scattering coefficients (determined from ac-9

measurements) at the 440 nm waveband (Fig. 2(d)–(f)) at the

same three representative stations show a strong resemblance

to Chl a profiles, higher values in surface waters that decrease

with depth. These high values of absorption were all asso-

ciated with lower salinity waters and high surface Chl a

concentrations. At the nearshore station (st. 192), a near-

surface [ap(440) + acdom(440)] maximum of 1.5 m� 1

decreased rapidly to about 0.3 m� 1 at a depth of 10 m.

Spatially, surface values of [ap(440) + acdom(440)] further

decreased to 0.5 m� 1 at the mid-shelf station (Fig. 2(e))

and to about 0.13 m� 1 at the offshore station 24. Surface

acdom(440) varied from a high of about 0.4 m� 1 (nearshore

station 192) to about 0.01 m� 1 at station 24. Particulate

scattering bp(440) at stations 192 and 46 also showed surface

highs (4 and 1.6 m� 1), reflecting higher phytoplankton

concentrations at these two stations. In contrast, surface

scattering was relatively low (f 0.2 m� 1) at station 24

(surface Chl a concentration < 0.2 mg m� 3). Subsurface

and near-bottom highs in scattering unrelated to phytoplank-

ton concentrations were observed at all three stations. Spec-

trally, both the absorption and scattering typically decreased

with increasing wavelengths. In the case of CDOM, the

spectra showed an exponential decrease in absorption with

increasing wavelength, an expected trend (Miller et al., 2002)

with slopes of the log-linearized absorption spectra, S, in the

range 0.014 to 0.017.

Seasonality of the Mississippi River discharge is observed

to strongly influence the hydrographic and bio-optical prop-

erties of waters in the Northern Gulf of Mexico. This is

illustrated by the variability in bio-optical properties at

closely located stations (st. 192 and 34 near SW Pass) and

mid-shelf stations (st. 46 and 70) sampled in April and

October of 2000, respectively (Table 1). A higher volume

of river discharge in spring results in lower salinity and

temperature values observed at both the nearshore and mid-

shelf stations. Seasonal changes (April to October 2000) in

hydrographic conditions were greatest at the mid-shelf

location where surface salinity changed from 29.48 to 36.0


psu and temperatures from 22.5 to 25.8 jC. Biological andoptical properties were also directly influenced by the

seasonal flow of the Mississippi River. While at the near-

shore stations surface Chl a concentrations and absorption

and scattering coefficients were lower in October, the mid-

shelf stations (Table 1) experienced the greatest change.

Here, average surface Chl a concentration decreased from

5.57 mg m� 3 to a low of 0.43 mg m� 3, and surface

absorption (particulate plus CDOM) at 443 nm decreased

from 0.55 to 0.062 m� 1, respectively. Surface values of

other variables such as CDOM absorption, scattering, and

diffuse attenuation coefficient Kd(443) also decreased sig-

nificantly at the mid-shelf location.

3.2. Absorption and scattering characteristics

Particulate absorption due to phytoplankton cells and

detrital material generally dominated the absorption field.

Fig. 2. (a–c) Vertical profiles of temperature (dotted line), salinity (dashed line), and Chl a concentrations (solid line) for three representative stations (st. 192,

46, and 24) at the study site (Fig. 1) during the April 2000 cruise. (d– f) CDOM (acdom, solid line), particulate plus CDOM absorption (ap + acdom, dashed line),

and particulate scattering (bp, dotted line) at 440 nm for the same three stations.


However, CDOM absorption also formed an important

though variable component of the total absorption, varying

from less than 25% of the absorption field at the nearshore

stations to greater than 50% at the offshore stations. Also, at

the turbid nearshore stations, detrital material contributed a

large fraction (about 40%) to the particulate absorption

while being negligible at the offshore stations. The partic-

ulate scattering coefficient bp (where appropriate, we drop

the k argument for brevity), as derived from field measure-

ments with the ac-9, varied over a large range (e.g., bp(443)

varied from 3.55 m� 1 near SW Pass to about 0.2 m� 1 at the

offshore station 24). Spectrally, scattering decreased in a

similar fashion (Gould, Arnone, & Martinolich, 1999), with

increasing wavelength from 412 to 715 nm (not shown).

However, the backscattering coefficient bb(k) is the variablethat directly influences ocean color variability and is most

often used in inversion models. From simultaneous meas-

urements of bb(443) and bp(443) at various stations during

the October cruise, we obtained a relationship (Fig. 3(a))

between bb(443) and bp(443) given by

bbð443Þ ¼ 0:00069 þ 0:01799bpð443Þ ðr2 ¼ 0:98Þ: ð2Þ

A similar value for the slope (0.01829) was obtained by

Gould et al. (1999) using Petzold’s values of b and bb that

were applicable over a wide wavelength range (400 to 600

nm) and samples that were obtained from different oceanic

regions. Similarity in the relationship between scattering and

backscattering obtained by these two studies are interesting

in that these parameters are also dependent on particle size,

shape, and composition (i.e., index of refraction). However,

an examination of Fig. 3(a) suggests that a nonlinear fit may

also be applicable, especially for the mid-range data. A

general cubic-polynomial equation (bb = a0 + a1bp + a2bp2 +

a3bp3) showed a strong fit to the data (coefficient of de-

termination of 0.98) with values of coefficients being a0 =

0.00248, a1 = 0.01, a2 =� 0.0036, and a3 =� 0.00041. The

intercept, a0 = 0.00248, which represents the backscattering

by pure water bbw(443) is about the same value (0.002406

m� 1) as that obtained by Smith and Baker (1981). Based on

the assumption that the relationship between bb and bpwould also be valid temporally for the same waters, we

Table 1

Hydrographic and bio-optical properties of two closely located stations near Southwest Pass (st. 192 and 34) and two mid-shelf stations (st. 46 and 70) in April

and October of 2000, respectively

Surface properties (0–3 m) Spring 2000 (April 5–13) Fall 2000 (October 24–31)

SW Pass (st. 192) Mid-shelf (st. 46) SW Pass (st. 34) Mid-shelf (st. 70)

Salinity (psu) 23.35 29.48 25.66 36.0

Temperature (jC) 21.76 22.50 23.80 25.88

Chl a (mg m� 3) 17.34 5.57 8.4 0.43

TSM (mg/l) 7.0 13.5 3.12 0.34

CDOM abs. acdom(443) (m� 1) 0.45 0.26 0.33 0.031

[ap(443) + acdom(443)] (m� 1) 1.3 0.55 0.69 0.062

Scattering, bp(443) (m� 1) 3.55 1.63 1.81 0.18

Scattering albedo x0(443) 0.73 0.74 0.72 0.74

Backscattering bb(443) (m� 1) 0.065 0.031 0.028 0.004

Attenuation coeff. Kd(443) (m� 1) 1.18 0.89 0.14 0.08

Nitrate +Nitrite (AM) 11.52 4.29 15.27 1.25

Fig. 3. (a) Backscattering (bb) at 443 nm (m� 1) plotted as a function of

particulate scattering bp at the same waveband measured for three stations

during the October 2000 cruise. The solid line represents the least-squares

regression fit to the data; (b) particulate backscattering bbp (m� 1) as a

function of wavelength at four stations (Fig. 1) derived from scattering

measurements.


use the linear relationship of Eq. (2) to derive bbp(k) frombp(k) for the April cruise. Typical surface spectral bbp(k)values (Fig. 3(b)) for four stations (st. 192, 46, 24, and

207) varied by an order of magnitude from about 0.003

m� 1 at the offshore station (st. 24) to >0.06 m� 1 at the

two nearshore stations (st. 192 and 207). Under the low

flow regime of the Mississippi River, both suspended

sediments and phaeopigments were relatively low suggest-

ing phytoplankton biomass to be the dominant factor

determining backscattering. HPLC pigment analysis

showed higher levels of the carotenoid fucoxanthin, indi-

cative of diatoms at these stations.

Observed spatial and vertical changes in the magnitude

of the water column attenuation coefficient for downward

irradiance Kd(k) were primarily influenced by particulate

and CDOM absorption (D’Sa, Miller, McKee, & Trzaska,

2002). Estimated at the 10% light level, Kd(443) was

linearly related to total absorption (a= aw + ap + acdom) at

443 nm (Fig. 4) determined at the same optical depth. Here,

absorption due to pure water aw was obtained from Pope

and Fry (1997). The diffuse attenuation coefficient Kd has

been expressed as a function of absorption and backscatter-

ing coefficients a and bb using the two-stream irradiance

model (Aas, 1987) through the approximation

Kdcðaþ bbÞ=ld; ð3Þ

where ld is the cosine of the mean downwelling light field

and the absorption a and backscattering bb coefficients are

the sum of components representing the contribution of

various in-water constituents. The various components of

absorption are due to pure water, particulates, and dissolved

materials, while backscattering comprised of pure water

backscatter and particulate backscatter. From field determi-

nations of a, b, and Kd at the 10% light level, we also

observe that (a + bb) and Kd are linearly related (Fig. 4) for

stations comprising a variety of optical conditions. From Eq.

(3), the slope (0.65) in Fig. 4 represents the mean cosine ld,which on neglecting the effects of bb reduces ld to a value

of 0.61. As observed in Fig. 4, while the bb contribution to

this formulation appears to be negligible at the offshore

stations with low absorption values, it is quite significant at

the nearshore stations with higher absorption. The mean

cosine ld has been modeled for a variety of solar zenith

angles (0–90j), scattering to absorption ratios (1–20) and

water depths (Bannister, 1992). Modeled values of ld havebeen shown to vary from 0.8 at surface to an asymptotic

value of 0.65 at the 0.1% light level for a sun zenith angle of

30j and a b/a value of 4 (Bannister, 1992). For stations at

our study site, the b/a ratio at the 10% light level varied

from a high of about 9.7 at coastal stations (st. 157 and 91)

to a low of about 1.23 at the offshore station 24. A

comprehensive examination of ld as a function of depth

and the IOPs (Berwald, Stramski, Mobley, & Kiefer, 1995;

McCormick, 1995) and the relationship between Kd and Chl

a concentrations (Morel & Maritorena, 2001) will be useful

in further characterizing the bio-optical properties of these

waters.

3.3. Remote sensing reflectance relationship to absorption

and backscattering

Spectral remote sensing reflectances Rrs(k) determined

from measurements with an in-water radiometer were var-

iable in shape and magnitude (Fig. 5). Reflectance spectra of

the nearshore and the mid-shelf stations (st. 192, 207, and

46) show well-pronounced maxima in the green spectral

band, while the offshore station 24 shows maximal reflec-

tance at 490 nm. The relative role of absorption and

scattering appear to control the reflectance spectra at these

stations. While absorption due to phytoplankton and CDOM

is strong at the nearshore and mid-shelf stations, enhanced

Fig. 4. Absorption a(443) and absorption plus backscattering a(443) +

bb(443) as a function of Kd at 443 nm determined from field measurements

at the 10% light level. The lines are least-squares fit to the data (r2 = 0.91,

n= 9 in both cases).

Fig. 5. Typical remote sensing reflectance spectra Rrs(k) determined from

measurements using an in-water radiometer at four stations at the study site

in April of 2000.


backscattering (Fig. 3(b)) due to suspended particles (phy-

toplankton, detrital, and suspended sediments) contributed

to the large Rrs(k) values (>0.005) at the green wavebands.

We also observe the signature associated with solar-stimu-

lated Chl a at 683 nm at these stations. At the offshore

station 24, backscattering contribution by particulate mate-

rial is minimal (Chl a < 0.2 mg m� 3), while the relatively

higher CDOM levels (0.1 m� 1 at 443 nm) at this station

results in lower Rrs(k) values in the blue spectral region.

Ocean color is usually understood in terms of remote

sensing reflectance Rrs(k), which is defined as the ratio of

the upwelling radiance and downwelling irradiance (Eq.

(1)). Rrs(k) has been related to the irradiance reflectance just

beneath the surface R(0�) through the approximation given

by (Mobley, 1994),

RrsðkÞc0:54 Rð0�Þ=Q; ð4Þ

where Q is the ratio of upwelling irradiance to upwelling

radiance (Morel & Gentili, 1996). The subsurface irradiance

reflectance R(0�) defined as the ratio of upwelling irradi-

ance to downwelling irradiance has been related through

modeling studies to the absorption a(k) and backscattering

bb(k) coefficients of seawater through the approximation

(Gordon, et al., 1988; Morel & Gentili, 1996)

Rð0�Þcfbb=ðaþ bbÞ; ð5Þ

where f is a variable that depends upon the solar zenith

angle, the optical properties of seawater, and light wave-

length (Morel & Gentili, 1991). From Eqs. (4) and (5),

Rrs(k) relates to the IOPs of absorption and backscattering

through the approximation given by

RrsðkÞc0:54½ f =Q�½bb=ðaþ bbÞ� ð6Þ

The f/Q ratio above, although being dependent on the

geometric conditions (solar-zenith angle and the viewing

angle), wavelength, and the bio-optical state (Morel &

Gentili, 1996), varies to a lesser degree than f and Q

individually.

Using Eq. (6), we examine the spectral relationship

between Rrs(k) and concurrent surface (0–3 m average)

estimates of backscattering and absorption coefficients at the

SeaWiFS wavebands (Fig. 6). The criteria for choosing the

0–3 m surface average for the IOPs and Chl a were based

on the observation that the lower salinity surface mixed

layer in the study region, while variable, was generally

greater than 3 m. We observe that Rrs and bb/(a + bb) are

linearly related (r2>0.8) at the 490 and 555 nm wavebands

(Fig. 6) as well as at the 412, 510, and 683 nm wavebands.

From Eq. (6), the slope divided by 0.54 gives the ratio f/Q

and lies between 0.09 and 0.12 (Fig. 7) over the spectral

range of 412 to 683 nm. These f/Q values derived from our

field studies fall well within the range modeled for various

solar-zenith angles, the viewing angle, wavelength, and the

bio-optical state (Morel & Gentili, 1996). While the solar-

zenith angle has been shown to strongly affect value of f/Q,

this was not observed in our data since most of our field

measurements were done within 2 h of midday. Although

surface values of Chl a concentrations varied widely (0.2 to

>17 mg m� 3) at the study site, it did not contribute greatly

to the f/Q variability. A similar range in the field determi-

nations of f/Q ratio (Reynolds, Stramski, & Mitchell, 2001)

in the Southern Ocean suggests a narrow range in its

variability. High correlations between the independently

measured reflectances and the IOPs (backscattering and

absorption) and the low f/Q variability suggest a measure

of closure in our field determinations of AOPs and IOPs for

waters with considerable range in absorption and scattering

properties.

3.4. Remote sensing reflectance algorithms

Remote sensing reflectance Rrs(k) were determined

using both in-water and above-water radiometers during

the April 2000 cruise. A comparison of the ratios of

Fig. 6. Comparison of remote sensing reflectance Rrs with bb/(a+ bb) at

various stations during the April cruise corresponding to two SeaWiFS

wavebands at 490 and 555 nm. From Eq. (6) in text, we interpret the slope

divided by 0.54 as representing the average value of the ratio f/Q at each of

the wavebands.


Rrs(k) determined using the two techniques, indicate the

ratios to be highly correlated (Fig. 8) for the various

combinations of wavebands generally used in ocean color

algorithms. These results suggest that ratios of Rrs(k)determinations using the above-water radiometer also

closely relate to the ratios of IOPs (backscattering and

absorption) for any of the relevant combination wave-

bands. We use an extended set of these Rrs(k) and surface

measurements of Chl a and CDOM absorption to

evaluate appropriate empirical relationships between the

variables.

The application of color ratios of water-leaving radiances

or remote sensing reflectances to estimate Chl a concen-

trations is based on the absorption characteristics of Chl a

which absorbs more blue light than green. The operational

SeaWiFS ocean chlorophyll algorithm (OC4), for example,

uses a four-band (443, 490, 510, and 555 nm) maximum

band ratio formulation (O’Reilly et al., 1998). Regression

analysis of the log-transformed reflectance ratios and Chl a

for the four bands used in the OC4 algorithm indicates

high correlations between the Rrs(k) ratios and Chl a

(r2>0.8) for all the bands (Fig. 9). For the same range in

Chl a concentrations, however, the range in Rrs(k) ratio

values decreases with increasing wavebands (i.e., Rrs(443)/

Rrs(555)>Rrs(490)/Rrs(555) and >Rrs(510)/Rrs(555)). The

Rrs(443)/Rrs(555) ratio is expected to give the most precise

estimates at lower Chl a concentrations, the Rrs(510)/

Rrs(555) ratio at higher Chl a concentrations while the

Rrs(490)/Rrs(555) ratio in the Chl a range between the other

two ratios (O’Reilly et al., 1998). However, in the presence

of variable and relatively high CDOM absorption in the

region, the effect of shifts in the ratio selection is not clear,

although for our data set, it was observed that the OC4

algorithm would predominantly select the Rrs(510)/Rrs(555)

band ratio. With Rrs(490)/Rrs(555) ratio providing a higher

dynamic range for the same range in Chl a concentrations,

we propose the use of this single band ratio for estimating

Fig. 8. Comparison between the ratios of remote sensing reflectance Rrs(k) determined from measurements with above-water and in-water radiometers at same

locations (April 2000).

Fig. 7. Spectral variations of the ratio f/Q computed from regression

analysis (Fig. 6) and Eq. (6) in text for the SeaWiFS wavebands.


Chl a from remote sensing data for the study region. The

power law equation derived from field measurements (Fig.

9(b)) is given by

Chl a ¼ 1:629ðRrsð490Þ=Rrsð555ÞÞ�2:551: ð7Þ

We also derived empirical relationships between CDOM ab-

sorption at 412 nm and the ratios of reflectances Rrs(412)/

Rrs(510), Rrs(443)/Rrs(510), and Rrs(510)/Rrs(555) (Fig. 10)

and obtained high correlations in all cases (Miller & D’Sa,

2002).

Empirical relationships derived from field data were

applied to SeaWiFS imagery of April 8, 2000 to retrieve

Chl a concentrations and CDOM absorption. These re-

trieved values were compared to a limited set of in situ

measured values. Similar trends in Chl a retrievals (Fig.

11) were observed using the power law Eq. (7) and the

standard operational OC4v4 algorithm. Similarly, we eval-

uated empirical relationships between the reflectance band

ratios and acdom(412) and applied these relationships (Fig.

10) to retrieve CDOM absorption from SeaWiFS satellite

data. The closest retrievals of CDOM absorption to in situ

values were obtained using the Rrs(510)/Rrs(555) band ratio

(not shown). While the Rrs(443)/Rrs(510) band ratio was

found suitable in other regions (Kahru & Mitchell, 2001),

in terms of dynamic range (Fig. 10), the Rrs(412)/Rrs(510)

is the optimal band ratio for the application. However, we

observed larger CDOM absorption retrieval errors for these

ratios. This may be attributed to lower water-leaving

radiances and greater extrapolation errors due to atmos-

pheric correction (Gordon & Wang, 1994) at these wave-

bands (412 and 443 nm) than at the 510-nm waveband. A

Fig. 9. Surface Chl a concentrations (mg m� 3) plotted as a function of the

ratios of Rrs(k) for the wavebands (a) 443 and 555 nm, (b) 490 and 555 nm,

and (c) 510 and 555 nm; both the scales are logarithmic. The solid lines

represent the regression, with the coefficient of determination r2 = 0.80,

0.81, and 0.82, respectively. Closed circles represent measurements in April

and inverted triangles in October 2000.

Fig. 10. Surface CDOM absorption (m� 1) at 412 nm plotted as a function

of the ratios of Rrs(k) for wavebands (a) 412 and 510 nm, (b) 443 and 510

nm, and (c) 510 and 555 nm.


validation with a larger data set will help determine the

optimal waveband ratio.

4. Conclusions

River-impacted coastal margins influenced by large riv-

ers such as the Mississippi River are highly dynamic regions

where the use of global ocean color algorithms may not be

appropriate for accurate retrieval of bio-optical variables

such as phytoplankton biomass. In developing local ocean-

color remote sensing algorithms or parameterizations of bio-

optical models, there is a need to understand the processes

and mechanisms that influence the ocean color or the remote

sensing signal. In this work, we have examined the concen-

trations and distribution of optically active in-water constit-

uents such as Chl a and CDOM in relation to the IOPs

(absorption and scattering coefficients) and AOPs (diffuse

attenuation coefficient Kd and the remote sensing reflectance

Rrs(k)) under very low discharge conditions of the Missis-

sippi River. Surface stratification resulting from lower

salinity river plume waters overlying more saline oceanic

waters was observed to be a major factor influencing the

bio-optical properties in the region, varying both spatially

and temporally. CDOM absorption at 412 and 443 nm,

while variable, formed a significant component of the total

absorption field at these wavebands. The combination of

absorption and particle backscattering strongly influenced

the remote sensing reflectance signal. Semi-analytic algo-

rithms (Carder, Chen, Lee, & Hawes, 1999; Garver &

Siegel, 1997) are based on the formulation relating irradi-

ance reflectance or remote sensing reflectance to absorption

and backscattering (Gordon et al., 1988). The linear rela-

tionship between the AOPs (Kd and Rrs) and absorption and

backscattering for the SeaWiFS wavebands, and the derived

f/Q and ld values that lie within the modeled ranges,

suggests the potential for development of new or parameter-

ization of existing models for this region of study. These

results also formed the basis for the robust relationships

between Rrs(k) ratios and surface Chl a concentrations and

CDOM absorption that we obtained for a bio-optically

complex, river-impacted coastal region. The acquisition of

a larger data set of bio-optical variables under more typical

flow conditions will allow the refinement of the above

algorithms for the estimation of Chl a and CDOM absorp-

tion for this coastal region.

Acknowledgements

The authors are grateful to Carlos Del Castillo, Callie Hall,

and Jinchun Yuan (NASA Earth Science Applications

Directorate, Stennis Space Center), Brent McKee (Institute

for Earth and Ecosystem Sciences, Tulane University, New

Orleans) and the crew of RV Pelican for their assistance

during field data collection. We also wish to thank Rodney

Powell (Louisiana Universities Marine Consortium, Chau-

vin, LA) for the nutrient data. This work was funded by the

Mississippi River Interdisciplinary Research Program (MiR-

IR) established under ONR grant # BHRP 533565G1, and a

NASA Center Director’s Discretionary Fund grant from

Stennis Space Center.

References

Aas, E. (1987). Two-stream irradiance model for deep waters. Applied

Optics, 26, 2095–2101.

Bannister, T. T. (1992). Model of the mean cosine of underwater radiance

and estimation of underwater scalar irradiance. Limnology and Oceano-

graphy, 37, 773–780.

Berwald, J., Stramski, D., Mobley, C. D., & Kiefer, D. A. (1995). Influen-

ces of absorption and scattering on vertical changes in the average

cosine of the underwater light field. Limnology and Oceanography,

40, 1347–1357.

Carder, K. L., Chen, F. R., Lee, Z. P., & Hawes, S. K. (1999). Semianalytic

moderate-resolution imaging spectrometer algorithms for chlorophyll a

and absorption with bio-optical domains based on nitrate-depletion

temperatures. Journal of Geophysical Research, 104, 5403–5421.

Carder, K. L., Hawes, S. K., Baker, K. A., Smith, R. C., Steward, R. G., &

Mitchell, B. G. (1991). Reflectance model for quantifying chlorophyll a

in the presence of productivity degradation products. Journal of Geo-

physical Research, 96, 20599–20611.

Del Castillo, C. E., Coble, P. G., Conmy, R. N., Muller-Karger, F. E.,

Vanderbloemen, L., & Vargo, G. A. (2001). Multispectral in situ meas-

urements of organic matter and chlorophyll fluorescence in seawater:

documenting the intrusion of the Mississippi River plume in the West

Florida Shelf. Limnology and Oceanography, 46, 1836–1843..

D’Sa, E.J., Miller, R.L., McKee, B.A., & Trzaska, R. (2002). Apparent

optical properties in waters influenced by the Mississippi River. In:

Proceedings 7th International Conference on Remote Sensing for Ma-

rine and Coastal Environments, Miami, FL. Ann Arbor, MI: Veridian.

Dinnel, S. P., & Wiseman, W. J. (1986). Fresh-water on the Louisiana and

Texas shelf. Continental Shelf Research, 6, 765–784.

Garver, S. A., & Siegel, D. A. (1997). Inherent optical property inversion

of ocean color spectra and its biogeochemical interpretation: I. Time

Fig. 11. A comparison between Chl a concentrations estimated using the

power law Eq. (7) and the operational SeaWiFS algorithm (OC4v4) for a

SeaWiFS image of April 8, 2000 plotted against in situ measurements of

surface Chl a at six stations for the same day at the study site. The lines are

least squares fit to the data points (r2 = 0.88, n= 6).


series from the Sargasso Sea. Journal of Geophysical Research, 102,

18607–18625.

Gordon, H. R., Brown, O. B., Evans, R. H., Brown, J. W., Smith, R. C.,

Baker, K. S., & Clark, D. K. (1988). A semianalytic radiance model of

ocean color. Journal of Geophysical Research, 93, 10909–10924.

Gordon, H. R., & Wang, M. (1994). Retrieval of water-leaving radiance

and aerosol optical thickness over the oceans with SeaWiFS: a prelimi-

nary algorithm. Applied Optics, 33, 443–452.

Gould, R. W., Arnone, R. A., & Martinolich, P. M. (1999). Spectral de-

pendence of the scattering coefficient in case 1 and case 2 waters.

Applied Optics, 38, 2377–2383.

Kahru, M., & Mitchell, B. G. (2001). Seasonal and nonseasonal variability

of satellite-derived chlorophyll and colored dissolved organic matter

concentration in the California Current. Journal of Geophysical Re-

search, 106, 2517–2529.

Kishino, M., Takahasi, M., Okami, N., & Ichimura, S. (1985). Estimation

of the spectral absorption coefficients of phytoplankton in the sea.

Bulletin of Marine Science, 37, 634–642.

Lohrenz, S. E., Fahnenstiel, G. L., Redalje, D. G., Lang, G. A., Chen, X., &

Dagg, M. J. (1999). Variations in primary production of Northern Gulf

of Mexico continental shelf waters linked to nutrient inputs from the

Mississippi River. Marine Ecology Progress Series, 155, 435–454.

McCormick, N. J. (1995). Mathematical models for the mean cosine of irra-

diance and the diffuse attenuation coefficient. Limnology and Oceano-

graphy, 40, 1013–1018.

Miller, R. L., Belz, M., Del Castillo, C. E., & Trzaska, R. (2002). De-

termining CDOM absorption spectra in diverse aquatic environments

using a multiple pathlength, liquid core waveguide system. Continental

Shelf Research, 22, 1301–1310.

Miller, R. L., & D’Sa, E. J. (2002). Evaluating the influence of CDOM on

the remote sensing signal in the Mississippi River Bight. EOS Tran-

sactions AGU, Honolulu, HI (p. 171). AGU, Washington, DC.

Milliman, J. D., & Meade, R. H. (1983). World-wide delivery of river

sediment to the ocean. Journal of Geology, 91, 1–21.

Mitchell, B. G., & Kiefer, D. A. (1988). Variability in the pigment specific

fluorescence and absorption spectra in the northeastern Pacific Ocean.

Deep Sea Research, Part A, 35, 665–689.

Mobley, C. D. (1994). Light and water: radiative transfer in natural

waters. San Diego: Academic Press.

Mobley, C. D. (1999). Estimation of the remote-sensing reflectance from

above-surface measurements. Applied Optics, 38, 7442–7455.

Morel, A., & Gentili, B. (1991). Diffuse reflectance of oceanic waters: its

dependence on sun angle as influenced by the molecular scattering

contribution. Applied Optics, 30, 4427–4438.

Morel, A., & Gentili, B. (1996). Diffuse reflectance of oceanic waters: III.

Implication of bidirectionality for the remote-sensing problem. Applied

Optics, 35, 4850–4862.

Morel, A., &Maritorena, S. (2001). Bio-optical properties of oceanic waters:

a reappraisal. Journal of Geophysical Research, 106, 7163–7180.

Mueller, J. L., & Austin, R. W. (1995). Ocean optics protocols for SeaWiFS

validation, revision 1. In S. B. Hooker, & E. R. Firestone (Eds.), NASA

Tech. Memo. 104566, vol. 25 (pp. 1–67). Greenbelt, MD: NASA God-

dard Space Flight Center.

Muller-Karger, F. E., Walsh, J., Evans, R. H., & Meyers, M. B. (1991). On

the seasonal phytoplankton concentration and sea surface temperature

cycles of the Gulf of Mexico as determined by satellites. Journal of

Geophysical Research, 96, 12645–12665.

O’Reilly, J. E., Maritorena, S., Mitchell, G. G., Siegel, D. A., Carder,

K. L., Garver, S. A., Kahru, M., & McClain, C. (1998). Ocean color

algorithms for SeaWiFS. Journal of Geophysical Research, 103,

24937–24953.

Pegau, W. S., Gray, D., & Zaneveld, J. R. V. (1997). Absorption and

attenuation of visible and near-infrared light in water: dependence on

temperature and salinity. Applied Optics, 36, 6035–6046.

Pope, R. M., & Fry, E. S. (1997). Absorption spectrum (380–700 nm) of

pure water: II. Integrating cavity measurements. Applied Optics, 36,

8710–8723.

Rabalais, N. N., & Turner, R. E. (2001). Hypoxia in the Northern Gulf of

Mexico: description, causes and change. In N. N. Rabalais, & R. E.

Turner (Eds.), Coastal hypoxia: consequences for living resources and

ecosystems (pp. 1–36). Washington, DC: AGU.

Rabalais, N. N., Turner, R. E., Justic, D., Dortch, Q., Wiseman Jr., W. J.,

& Sen Gupta, B. K. (1996). Nutrient changes in the Mississippi River

and system responses on the adjacent continental shelf. Estuaries, 19,

386–407.

Redalje, D. G., Lohrenz, S. E., & Fahnenstiel, G. L. (1994). The relation-

ship between primary production and the vertical export of particulate

organic matter in a river-impacted coastal ecosystem. Estuaries, 17,

829–838.

Reynolds, R. A., Stramski, D., & Mitchell, B. G. (2001). A chlorophyll-

dependent semianalytical reflectance model derived from field measure-

ments of absorption and backscattering coefficients within the Southern

Ocean. Journal of Geophysical Research, 106, 7125–7138.

Smith, R. C., & Baker, K. S. (1981). Optical properties of the clearest

natural waters (200–800 nm). Applied Optics, 20, 177–184.

Tassan, S. (1994). Local algorithms using SeaWiFS data for the retrieval of

phytoplankton, pigments, suspended sediment, and yellow substance in

coastal waters. Applied Optics, 33, 2369–2377.

Twardowski, M. S., Sullivan, J. M., Donaghay, P. L., & Zaneveld, J. R. V.

(1999). Microscale quantification of the absorption by dissolved and

particulate material in coastal waters with an ac-9. Journal of Atmos-

pheric and Oceanic Technology, 16, 691–707.

Zaneveld, J.R.V., Kitchen, J.C., & Moore, C. (1994). The scattering error

correction of reflecting-tube absorption meters. In: J.S. Jaffe (Ed.), Pro-

ceedings of Ocean Optics XII ( pp. 44–55). SPIE.


Bio-optical properties in waters influenced by the...

Documents

Transcript of Bio-optical properties in waters influenced by the...