Full text of DOI:10.1080/01431160120387

This article was downloaded by: [171.67.34.205]
On: 26 April 2013, At: 02:39
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK
International Journal of
Remote Sensing
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/tres20
Detecting photosynthetic
algal pigments in natural
populations using a
high-spectral-resolution
spectroradiometer
R. Aguirre-Gomez , S. R. Boxall & A. R. Weeks
Version of record first published: 25 Nov 2010.
To cite this article: R. Aguirre-Gomez , S. R. Boxall & A. R. Weeks (2001):
Detecting photosynthetic algal pigments in natural populations using a high-
spectral-resolution spectroradiometer, International Journal of Remote Sensing,
22:15, 2867-2884
To link to this article: http://dx.doi.org/10.1080/01431160120387
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/
terms-and-conditions
This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any
representation that the contents will be complete or accurate or up to
date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not
be liable for any loss, actions, claims, proceedings, demand, or costs or

damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.

int. j. remote sensing, 2001, vol. 22, no. 15, 2867–2884
Detecting photosynthetic algal pigments in natural populations using a
high-spectral-resolution spectroradiometer
R. AGUIRRE-GOMEZ*, S. R. BOXALL and A. R. WEEKS†
University of Southampton, Department of Oceanography, Southampton
Oceanography Centre, Waterfront Campus, European Way, Southampton,
England, SO14 3ZH, UK
(Received 20 November 1997; in Ž nal form 15 April 1999)
Abstract. Re ectance data from a high spectral resolution spectroradiometer
were obtained onboard a ship in Plymouth coastal waters. These data were
analysed to detect algal photosynthetic accessory pigments for comparison with
absorption spectra as measured in the laboratory by a spectrophotometer. The
overall spectral characteristics of Plymouth waters allowed identiŽ cation as to
population composition. Derivative analysis of the spectra was used to resolve
characteristic peaks of speciŽ c pigments. It was determined that chlorophyll
pigments, a speciŽ c carotenoid and sea water absorption bands were detect-
able in the re ectance data. Absorption bands of photosynthetic and accessory
pigments were assessed through chromatographic pigment analysis.
1. Introduction
The intrinsic changes of ocean colour are determined by the spectral variations
of the sea surface re ectance R(l), which is deŽ ned as the ratio between upwelling
irradiance E (l,0) and the downwelling irradiance E (l,0) at wavelength l at the
u
d
surface (depth Z 0) and is expressed as follows:
=
E (l,0)
u
R(l,0)
(1)
=
E (l,0)
d
The parameter E (l,0) carries information on the sea water and the dissolved and
particulate matter contained within it.
Many models of radiation transfer in the sea have served to establish relationships
u
between R(l) and the absorption (a) and backscattering (b ) coeYcients (e.g. Morel
and Prieur 1977, Gordon and Morel 1983). These coeYcients, in turn, are determined
b
by the optical characteristics of pure sea water and those of the particulate and
dissolved substances present in variable amounts in the sea. Therefore, re ectance
spectra are the result of absorption and backscattering processes due to water (w),
phytoplankton (phy), yellow substance (y) and non-chlorophyllian particles (x). Thus,
*Present address: Instituto de Geograf´Ìa, Universidad Nacional Autonoma de Mexico,
Circuito Exterior, 04510, Coyoacan, D.F. Mexico.
†Present address: Maritime Faculty, Southampton Institute, East Park Terrace,
Southampton, England, SO14 OYN, UK.
International Journal of Remote Sensing
ISSN 0143-1161 print/ISSN 1366-5901 online © 2001 Taylor & Francis Ltd
http://www.tandf.co.uk/journals

2868
R. Aguirre-Gomez et al.
any interpretation of the re ectance must be based on the knowledge of the optical
properties of these substances.
For oceanic waters, a simple relationship has been widely used (Morel and
Prieur 1977):
b
b
R
0.33
(2)
$
a
although variations of the value of 0.33 must be considered due to the zenith angle
of incident light (Kirk 1984) and the shape of the phase function (Gordon et al.
1988). This equation is valid for waters where b a, and it is assumed that the
%
b
coeYcients a and b obey the principle of superposition (Sathyendranath and Morel
b
1983). Experimental data and theory indicate that, under most conditions, values of
b (l) are relatively small compared to values of a(l) and decrease monotonically
b
with respect to wavelength (Gordon and Morel 1983). Direct measurements of b (l)
for phytoplankton cultures supports this notion of low backscatter coeYcients for
b
most algal groups (Ahn et al. 1992). The present paper will show that spectral
variations in R(l), and, hence, the water leaving radiance, primarily involves spectral
variations in the absorption coeYcient.
Correlating concentrations of the diVerent components present in a water body
with a number of possible combinations of sensor spectral band signal ratios and/or
diVerences have been widely used for determining water constituent concentrations
from remotely sensed data. Ratios and diVerences have traditionally been used in
ocean colour studies, but they are restricted to a small number of spectral bands.
The advent of high-spectral-resolution sensors requires the use of improved methods
of study such as the derivative analysis. The derivative method tackles many of the
problems of quantitative analysis in a more eVective way than ratios and diVerences
by taking into account a larger amount of data, which stands for more information
potentially available. Derivative analysis can be applied to spectra obtained by high-
spectral-resolution sensors. These sensors can be deŽ ned as those having a bandwidth
less than 5 nm and/or more than 100 spectral bands. Thus, high-spectral-resolution
sensors can provide information about smaller spectral variations than can coarse
bands. The derivative method is used for minimizing low-frequency background
noise and for resolving overlapping spectra (Butler and Hopkins 1970). This method
has successfully been applied in remote sensing studies of vegetation (Demetriades-
Shah et al. 1990) and suspended solids in water (Chen et al. 1992, Goodin et al.
1993) and can also be applied to marine remote sensing studies. As contemporary
optical sensors develop higher spectral resolution, there is potential for a greater
wealth of information relating to the absorption capabilities of phytoplankton which
could help in identifying the major diVerences in taxonomic groups.
Data provided by high-spectral-resolution sensors mounted either on an aircraft,
such as the Airborne Visible and Infrared Imaging Spectrometer (AVIRIS), the
Compact Airborne Spectro-graphic imager (CASI) and the Moderate Resolution
Imaging Spectroradiometer (MODIS), or on a satellite, such as the Medium
Resolution Imaging Spectroradiometer (MERIS), can be useful for testing new kinds
of algorithms designed to analyse the absorption by phytoplankton. This task can
also be carried out from an onboard ship spectroradiometer (e.g. SpectraScan). These
kinds of instruments are important because the desired information from a remotely
sensed signal of an area of surface water is the sub-surface irradiation re ectance

Detecting photosynthetic algal pigments
2869
R(0-). This magnitude can be useful for deriving multitemporally valid algorithms
in remote sensing studies (Dekker et al. 1991). Additionally, from R(0-) measurements
information can be extracted on those components in the water that interacts with
the natural light Ž eld. This interaction takes place through changes in the inherent
optical properties of absorption and scattering and is dependent on the volume
scattering function. R(0-) can be calculated by determining the ratio of the subsurface
upwelling (E ) to subsurface downwelling irradiance (E ), giving the subsurface
wu
wa
irradiance re ectance R(0-).
Data provided by shipborne spectroradiometers are important in remote sensing
studies because they may simulate those obtained through sensors mounted on
aircraft or satellites. Field-based spectroradiometric measurements are required for
determining R(0-) from the radiance L measured by a remote sensor. On the other
hand, spectroradiometric measurements are essential for determining the atmospheric
eVect between the airborne or spaceborne remote sensor and the ground, instead of
using models for removing atmospheric in uences. Finally, under certain circum-
stances, it could be better to use ship-based information because of higher conŽ dence
of in situ related data.
Consequently, the in situ optical properties, as calculated by equation (2), can
be related to ocean colour observations through re ectance values on the marine
surface as measured by equation (1).
The overall objective of the study described here was to assess the potential of
derivative analysis applied to high-spectral-resolution ship-based spectroradiometer
data for retrieving information about photosynthetic pigments.
2. Analysis of derivative re ectance spectra
According to equation (2), the solar radiation re ected from the sea surface is
dependent on absorption and backscattering processes. These processes are depend-
ent on the water itself, phytoplankton, inanimate particulate matter or ‘tripton’, and
dissolved yellow pigments (Kirk 1983). Thus, taking into account all of these compon-
ents, the derivative re ectance spectra can be expressed as the rate of change of
re ectance with wavelength dR(l)/dl (i.e. the slope of the measured re ectance
spectrum at wavelength l). Therefore, by taking the complete diVerentiation of
R
R(a , a , a , a , b , b
, b , l), we obtain
=
w
phy
x
y
bw bphy bx
dR(l)
R da R da
R da
R da
R db
ç
ç
ç
ç
ç
w
phy
x
y
+
bw
dl
=
+
+
+
+
dl
a
dl
a
dl
a dl
a dl
b
ç
ç
ç
ç
ç
w
phy
x
y
bw
R db
R db
ç
b
ç
bphy
bx
dl
(3)
+
dl
b
ç
ç
bphy
bx
where the partial derivatives are an implicit function of l.
Considering R(l) b (l)/a(l) and by calculating the partial derivative for each
#
component, and substituting in the equation (3), the following expression is obtained:
b
dR(l) b (l) da
da
da
da
1
db
db
db
b
w
phy
x
y
bw
phy
+
s
(4)
#
+
+
+
+
+
dl
a(l)2 Cdl
dl
dl
dl
D
a(l)
C
dl
dl
dl
D
This equation relates the slope of a directly measured re ectance spectrum at any
wavelength to each of the components. High-order derivatives can be calculated
iteratively. Thus, in principle, the presence of phytoplankton pigments, suspended
and organic matter within the sea water can be estimated by either of two alternative

2870
R. Aguirre-Gomez et al.
approaches: (a) by simultaneously solving equation (4) at three or more discrete
values of wavelength l; or (b) by considering re ectance spectrum components as
mutually independent. The second approach was used in this study for analysing
the spectral behaviour of each component of re ectance spectra.
3. Material and methods
Samples of a natural population of phytoplankton were collected during the
summer of 1995, 12 km oV the Plymouth coast (50 10 N, 4 10 W) (Ž gure 1). Two
°
ê
°
ê
litres of sea water were collected at depths of 0 m, 5 m, 10 m and 12 m. The sampling
was carried out onboard the research vessel RV Squilla (Marine Biological
Association) on 9 May, 14 June and 16 June. Water samples were analysed at
Plymouth Marine Laboratory (PML) to obtain absorption coeYcients and pigment
content of natural populations. Hydrographic measurements of temperature, depth,
and salinity were performed each time. Photosynthetic active radiation (2p) and
in situ upwelling radiance at 683 nm (L 683) were simultaneously measured using a
ProŽ ler Re ectance Radiometer (PRR-600, Biospherical Instruments Inc., San Diego,
CA, USA).
u
Upwelling and downwelling irradiances over the water surface oV Plymouth were
measured with a portable and fast scanning multi-spectral radiometer (SpectraScan
PR-650, Photo Research, Chatsworth, CA, USA).
3.1. Absorption measurements of natural populations
Absorption spectra of phytoplankton, yellow substance and non-chlorophyllian
particles were measured at PML using a double-beam Perkin Elmer lambda 2
spectrophotometer, from 350 to 750 nm, with a spectral resolution of 1 nm and an
associated instrumental error of 0.5 nm. A background correction was made before
±
m
Figure 1. Location map of sampling point near Plymouth coast ( ). Dotted line represents
the edge of the continental shelf (200m isobath).

Detecting photosynthetic algal pigments
2871
each measurement by running the instrument without samples in both cells and by
zeroing any remaining readout.
Water samples containing phytoplankton were Ž ltered onto a 24 mm glass-Ž bre
Ž lter (Whatman GF/F). The light absorbance of the particles retained on the Ž lter
was measured directly using the spectrophotometer, relative to a wetted glass-Ž bre
Ž lter for reference. After the measurement was made, the Ž lter paper was extracted
with 250 ml of methanol for 20–30 min. The bleached Ž lter paper was then measured
in the same manner as before, relative to a methanol-wetted Ž lter. The subtraction
of both spectra yields the light absorption by phytoplankton (Kishino et al. 1985).
The validity of the linear relationship of Beer’s law was assessed for every algal
group, and all measurements were within a linear range. All the absorption spectra
were corrected by subtracting the absorption value at 750 nm over all wavelengths.
Since phytoplankton pigments do not absorb at these wavelengths, it may represent
absorption by detritus and phytoplankton cells other than photosynthetic pigments
(Sathyendranath et al. 1987) or residual scattering (Duysens 1956). Thus, subtracting
these values over the entire wavelength implies that their absorption is neutral. The
package eVect present in the in vivo absorption measurements of algae was corrected
by using a theoretical approach proposed by Duysens (1956).
Yellow substance was measured by Ž ltering 500 ml of seawater through a
Whatman 0.2 mm nylon membrane Ž lter using a Millipore vacuum pump. The Ž ltered
liquid was then measured through a Perkin-Elmer Lambda 2 spectrophotometer
from 350 to 750 nm.
Absorption by non-chlorophyllian particles, measured by the Kishino et al. (1985)
method, is assumed to be due to the bleached particles, after methanol extraction
(Prieur and Sathyendranath 1981).
The absorption spectrum of pure sea water has been measured and presented
elsewhere (Morel 1974, Smith and Baker 1981).
Backscattering measurements of these components were not carried out in this
study, but their spectral characteristics were estimated from a model proposed by
Morel (1988). This model is a comprehensive one, as it includes diVerent situations.
The equation of this model can be expressed as follows:
1
2
1
4
550
b b 0.30C
2 · 10
2 · 10
logC
(5)
=
0.62
Õ3+
Õ2
-
b
C
A
B
l
D
The backscattering coeYcient of suspended particles is expressed as the product of
the scattering coeYcient b and the dimensionless backscattering ratio b , i.e.
b b(l). This equation includes the sum of (i) a constant term, wavelength and
concentration C independent, with a value of 0.2%, and (ii) a term varying with the
wavelength (l ) and the magnitude of C. This term has a value of 2% when
b
b
=
b
b
Õ1
l
550 nm and when C 10 mg m . It is allowed to decrease proportionally to
= Õ2 Õ3
=
log C and is zero for C 10 mg m . This model is regarded as a comprehensive
=
2
Õ3
one, because it includes a progressive change from an oligotrophic situation where
10
detritus are relatively more abundant (detrital particles, characterized by b 2%
=
scattering dependency) to an eutrophic situation where algal cells are
b
and a l
Õ1
the predominant particles, with low b (0.2%) and practically with spectral indepen-
dence. Thus, concentration values of backscattering from phytoplankton and non-
b
chlorophyllian particles were calculated by interpolating at 550 nm using the
Morel (1988) model.

2872
R. Aguirre-Gomez et al.
3.2. Pigment analysis
Pigments were analysed using high performance (pressure) liquid chromato-
graphy (HPLC). The method used in this study is a modiŽ cation of the reverse-
phase method described in Mantoura and Llewellyn (1983) and adapted by Barlow
et al. (1993). Samples were retained onto Whatman GF/F glass-Ž bre Ž lters (47 mm).
Filtration volume was 20 ml. The Ž lters were stored overnight in the freezer at
20 C prior to analysis. Frozen Ž lters were extracted in 2–6 ml in 90% acetone
°
-
aqueous solution. Sample manipulation was carried out before injection in faint light
to avoid any pigment photodegradation (Nelson 1993). The extraction volume was
5 ml, allowing for a suitable concentrated solution. Before injection, the extract
containing the Ž lter was homogenized and the solution clariŽ ed by Ž ltration through
a small Ž ltration unit containing a Whatman GF/F Ž bre Ž lter. An aliquot of 300 ml
of the sample was mixed with 300 ml of 1 m ammonium acetate prior to injection
(Zapata et al. 1987) and Ž nally 100 ml were injected into the column. Peak identiŽ ca-
tion was obtained by running extracts from natural populations. Pigments were
identiŽ ed by comparison with retention times of various pigments, which are well
documented in the literature (Mantoura and Llewellyn 1983, Wright et al. 1991).
3.3. Re ectance measurements
Re ectance measurements were carried out using the bi-directional re ectance
factor (BRF) described by Swain and Davis (1978). It is deŽ ned as the ratio of  ux
re ected by a target (r ) under speciŽ ed conditions of irradiation and viewing to
that re ected by a perfectly diVuse, completely re ecting surface (Lambertian surface)
as a reference (r ). The BRF can be expressed as follows:
ti
ci
r
ti
R
k
(6)
=
i
r
ci
where k is the panel correction factor which depends on the angular conŽ guration
and accounts for the non-Lambertian spectral re ectance of the panel.
Measurements of r were performed by placing the SpectraScan PR-650 over the
same side of the ship as the sun to avoid the direct shadow of the ship, which may
ti
introduce an error of up to 30% (Gordon 1985). The sensor was positioned to
provide a nadir view of the water surface reducing the unwanted irradiance of the
specularly re ected light, which is independent of the optical properties of the water
(Shifrin 1988). This guarantees that the light backscattered from the sea (providing
information on the elements present in the water) will be the major contributor of
the signal.
Measurements of r were carried out by looking at an halon (polytetra uoroethy-
lene) reference panel. It is a resistant and hydrophobic panel (i.e. washable and
ci
insensitive to humidity changes), properties that render it suitable for Ž eld-based
research (Schutt et al. 1981, Weidner and Hsia 1981).
Re ectance spectra were obtained by Ž rst scanning the water surface and then
by sequentially measuring over the calibration panel. These observations were carried
out while trying to keep the same viewing geometry by looking at nadir in both
cases. Since the calibration panel was onboard, the observations were performed
at diVerent heights. Then, target and panel scans were divided according to equa-
tion (6) and multiplied by a correction factor of k 0.99, in agreement to the
=
manufacturer’s instructions.

Detecting photosynthetic algal pigments
2873
Re ectance spectra were corrected for sun/sky glint error by subtracting the
re ectance measured at 750 nm from each re ectance spectrum. It is assumed that,
in open ocean, the water at this wavelength has absorbed all of the incoming
radiation (Shimwell 1996).
3.4. Derivative spectral analysis
In order to locate and identify ubiquitous peaks in the absorption, backscattering
and re ectance spectra, derivative curves were generated using a three-point central
diVerence numerical technique. This method estimates the derivative of a function
at both endpoints and the midpoint of any three points, as long as the points are
contiguous and evenly spaced. A custom computer program was implemented using
MATLAB language to perform the derivative routine on all data points. First-order
curves indicate more precisely the centre of each conspicuous peak. At the centre of
the peak the curve is horizontal, hence dR/dl shows an in ection point parallel to
the axis l. In ection points indicate the presence of either a maximum or a minimum
in the curve. The second derivative curve is useful for determining whether the
in ection points represent a minimum or a maximum. Maxima are found by looking
at the negative peaks in the second derivative spectrum, whilst the minima are
detectable as positive peaks in the curves. The third derivative provides similar
information to that of the Ž rst derivative, and, by looking at the in ection points
on this curve, a new set of maxima and minima becomes apparent. Thus, the fourth
derivative indicates the maxima or minima correspondent to the in ection points
marked by the third derivative.
First-order curves were obtained by applying the derivative algorithm over all
of the curves (zero-order). High-order derivative curves were obtained by iteratively
applying the same algorithm to the resultant curves, twice for absorption and
backscattering curves and four times in the case of re ectance curves. Because
diVerentiation tends to magnify the eVects of high-frequency noise in the spectra, a
polynomial Ž t smoothing Ž lter was applied over the data prior to calculating the
derivatives (Savitzky and Golay 1964).
Re ectance spectra were interpreted through the information provided by
absorption spectra of natural populations. Derivative peaks from absorption and
re ectance spectra were then visually associated to those of HPLC pigment stand-
ards to enable identiŽ cation of the pigment composition. The minima of the
fourth derivative re ectance spectra, corresponding to the absorption of pigments,
were compared with the local minima found in the second derivative spectra of
phytoplankton absorption curves.
4. Results and discussion
Derivative analyses of re ectance components were performed by examining the
absorption and scattering characteristics of every single case. The derivative re ection
spectra were dominated by phytoplankton and water absorption characteristics.
Figures 2–8 show absorption and backscattering spectra along with their derivatives
(magniŽ ed 10 times for a better visualization).
Figure 2 shows the spectra of pure sea water absorption and its derivative. It is
evident that the derivative curve shows no important features from 400 to 560 nm,
but a number of absorption peaks, revealed as local minima, are apparent at 600,
610 and 660 nm. These peaks are in good agreement with those found by Pegau and

2874
R. Aguirre-Gomez et al.
Figure 2. Absorption spectrum of pure water (after Smith and Baker 1981) and second
×
derivative curve ( 10), showing absorption peaks as local minima (dashed line).
Zaneveld (1993). They showed that light absorption by pure water depends on its
temperature, being apparent in absorption bands around 600, 660 and 740 nm.
Absorption spectrum of a natural population and the second derivative curve
are shown in Ž gure 3. Absorption peaks due to chlorophylls and carotenoids can
clearly be seen within the regions comprised between 400–500 nm (blue) and
600–700 nm (red) of the spectrum, which correspond to the local minima of the
derivative curve. Derivative analysis of phytoplankton absorption spectra has been
performed by a number of researchers, and nowadays it is an established technique
for pigment identiŽ cation (Gulyayev and Litvin 1970, Faust and Norris 1985, Bidigare
et al. 1989, Gieskes 1991, Aguirre-Gomez et al. 1995). Table 1 shows the natural
populations absorption peaks, found by the derivative technique for the three
sampling dates, with their associated pigments.
×
Figure 3. Absorption by phytoplankton and second derivative spectra ( 10). Local minima
show pigment absorption peaks (dashed line).

Detecting photosynthetic algal pigments
2875
Table 1. Absorption peaks (nm) found for natural populations through derivative analysis
=
at surface level (Z1 0 m) and their associated pigments.
9 May
1995
14 June
1995
16 June
1995
Band
Pigments
Pigments
Pigments
1
2
3
4
5
6
7
8
9
10
11
12
13
14
428
454
468
482
494
514
530
556
Chlorophyll a
Chlorophyll c
Fucoxanthin
Carotenes
Carotenes
Carotenes
426
442
460
476
494
Chlorophyll a
Chlorophyll a
412 Chlorophyll a
442 Chlorophyll a
456 Chlorophyll c
472 Carotenes
+
Carotenes Chl c
Carotenes
Carotenes
498 Carotenes
Carotenes
Prot. complex
540
568
594
Carotenes
530 Carotenes
556 Prot. complex
Prot. complex
Chlorophyll c
588
Chlorophyll a
+
598 Chl a Chl c
612 Chlorophyll comp.
630 Chlorophyll c
676 Chlorophyll a
+
630
674
Chl a Chl c
626
676
Chlorophyll c
Chlorophyll a
Chlorophyll a
Table 2 presents the pigment identiŽ ed by HPLC in the water samples. It is
evident that chlorophyll a was the predominant pigment on these dates, accounting
for nearly 50% of total pigment concentration. The subsequent identiŽ ed pigments
were fucoxanthin (19.4%), chlorophyll c
c
(8.50%), diadinoxanthin (6.1%),
+
1
2
hexoxyfucoxanthin (5.2%), chlorophyll c (2.2%) and small traces of b-carotene,
3
chlorophyll b, alloxanthin and butoxyfucoxanthin. These numbers indicate the dom-
inance of brown algae (e.g. diatoms, chrysophytes, dino agellates) which are mainly
characterized by the presence of chlorophyll a, fucoxanthin, chlorophyll c c ,
+
diadinoxanthin and b-carotene (Prezelin and Bo´czar 1986). Hexoxyfucoxanthin
1
2
´
indicates the presence of coccolithopores, alloxanthin the presence of cryptophytes
and chlorophyll b the presence of a type of green algae.
Absorption spectra of yellow substance and non-chlorophyllian particles along
with their derivatives are shown in Ž gures 4 and 5, respectively. The shape adopted
by these spectra is typically that of an exponential curve, as theoretically found by
Bricaud et al. (1981).
Table 2. HPLC analysis for natural populations with the retention time (RT) for each
1
pigment (min), the correspondent concentration (mg lÕ ) and the relative percentage (%).
Pigments
RT
9 May 1995
14 June 1995
16 June 1995 Average (%)
Chlorophyll c
1.4
2.2
3.3
5.1
5.6
6.4
7.6
8.4
11.2
12.1
14.0
—
0.041 (0.79)
0.407 (7.83)
0.006 (0.11)
0.038 (0.73)
0.888 (17.1)
0.194 (3.73)
0.199 (3.83)
0.180 (3.46)
0.091 (1.75)
3.089 (59.4)
0.063 (1.21)
5.196
0.030 (1.82)
0.122 (7.41)
0.012 (0.73)
0.008 (0.48)
0.320 (19.4)
0.126 (7.65)
0.160 (9.71)
0.060 (3.64)
0.037 (2.25)
0.728 (44.2)
0.044 (2.67)
1.647
0.123 (5.89)
0.233 (11.2)
0.031 (1.49)
0.001 (0.05)
0.525 (25.2)
0.146 (6.99)
0.187 (8.96)
0.028 (1.34)
0.039 (1.87)
0.708 (33.9)
0.065 (3.12)
2.086
2.2
8.5
0.5
0.5
19.4
5.2
6.1
3.0
3
Chlorophyll c c
Peridinin
1 2
Butoxyfucoxanthin
Fucoxanthin
Hexoxyfucoxanthin
Diadinoxanthin
Alloxanthin
Chlorophyll b
Chlorophyll a
b-Carotene
1.9
50.1
1.9
/
8.929 100%
Total

2876
R. Aguirre-Gomez et al.
Figure 4. Absorption spectrum of organic dissolved material (yellow substance) and its
×
second derivative curve ( 10). Exponential derivative curve (dashed line) shows no
absorption regions.
Figure 5. Absorption spectrum of non-chlorophyllian particles and its second derivative
×
curve ( 10). Exponential derivative curve (dashed line) shows no absorption regions.
Scattering spectrum of pure water is shown in Ž gure 6. It follows an exponential
decreasing path, and, therefore, the second derivative is an exponential curve, but
many times smaller than the original curve. Consequently, the eVect of scattering of
pure water is virtually eliminated through the derivative process. Thus, pure water
scattering eVects can be reduced by a Ž rst-order derivative curve as reported by a
number of authors (e.g. Philpot 1991, Goodin et al. 1993). Figures 7 and 8 show the
modelled backscattering spectra of non-chlorophyllian particles and phytoplankton,
respectively. The shapes of these spectra are similar, both of them smoothly decrease
within the visible range (400–700 nm).

Detecting photosynthetic algal pigments
2877
Figure 6. Scattering spectrum of pure water (after Smith and Baker 1981) and its second
×
derivative curve ( 10). Exponential derivative curve (dashed line) shows no absorp-
tion regions.
×
Figure 7. Backscattering by phytoplankton and second derivative spectra ( 10). The derivat-
ive spectrum is negligible (dashed line) and shows no absorption regions.
It must be noticed that even though these spectra have been taken following the
Morel (1988) model, their spectral shapes are rather smooth and their order of
magnitude is small (about 10 ) compared to the magnitude of absorption of
Õ3
phytoplankton and pure water. Thus, their derivative spectra are even smaller and
can be considered to be negligible in the derivative re ectance analysis. Additionally,
scattering spectral shapes can be linear or exponential. The linear case is eliminated
by the second derivative and the exponential case becomes smaller after a derivative
because of the small values in the argument. Through high-order derivative analysis
of refectance spectra, most of the terms in equation (4) can be ignored, leaving only
the in uence of phytoplankton and pure sea water.

2878
R. Aguirre-Gomez et al.
Figure 8. Backscattering by non-chlorophyllian particles and second derivative spectrum
×
(
regions.
10). The derivative spectrum is negligible (dashed line) and shows no absorption
Considering these results, it is possible to apply high-order derivatives for analys-
ing re ectance spectra. Figures 9–11 show the re ectance and derivative spectral
curves of water surface measured on 9 May, 14 June and 16 June, respectively. The
troughs of both the re ectance and the derivative spectra correspond to the
absorbance, which is primarily due to photosynthetic and accessory pigments and
pure sea water between 400 and 700 nm. The centres of the troughs, corresponding
to absorption maxima, are given in table 3. A comparison between absorption
maxima obtained with a spectrophotometer is also present in this table. Comparing
Table 3. Comparison between peaks found by the derivative technique when applied to
absorption and re ectance spectra as measured by a spectrophotometer (SP) and
spectroradiometer (SR), respectively. Particular pigments or water characteristics
associated with absorption peaks in re ectance spectra are also shown.
9 May 1995
14 June 1995
16 June 1995
Band
SP
SR
SP
SR
SP
SR
Associated pigments
1
2
3
4
5
6
7
8
428
454
468
482
494
514
530
556
416
440
472
426
442
460
476
494
412
444
464
412
442
456
472
498
530
416
440
468
Chlorophyll a
Chl c Chl a
/
+
Carotenes Chl c
516
552
516
552
516
552
Carotenes
540
556
Water slope
9
568
594
10
11
12
13
14
588
630
674
598
612
630
600
628
660
604
664
600
632
664
Water
/
626
676
Chl c Chl a
Water
675
Chlorophyll a

Detecting photosynthetic algal pigments
2879
the absorption peaks, it is possible to establish a good agreement between the two
sets within instrumental error margins ( 2 nm for the spectroradiometer and 1 nm
±
±
for the spectrophotometer). It must be noted that the number of spectrophotometric
absorption peaks is greater than those found in the spectroradiometric measurements.
This fact can be explained as a consequence of atmospheric and water eVects that
are present in Ž eld conditions and absent in controlled laboratory measurements.
Figure 9 shows the re ectance spectrum measured on 9 May, along with its derivative
spectrum. Chlorophyll a is identiŽ ed through the peaks appearing at 416, 440 and
628 nm. A mixture of chlorophyll c and some carotenoids absorption can be seen at
472 nm. The high concentration of fucoxanthin and other photosynthetic pigments
creates a strong absorption peak at 516 nm. Since absorption by pure sea water
begins to be important near 550 nm, its in uence is perceived at the slope between
568 and 600 nm and at the absorption peak at 660 nm. Re ectance peaks were also
apparent from the derivative analysis, particularly distinctive local maxima centred
at about 500 and 680 nm. The latter is due to the sum of irradiances produced
through natural  uorescence emission by chlorophyll a and elastic scattering (Kirk
1983 ).
The re ectance spectrum obtained on 14 June is presented in Ž gure 10. It shows
a strong absorbance in the blue and the red regions of the electromagnetic spectrum.
Absorption in the blue is largely due to carotenoids and chlorophylls. Chlorophyll
a absorption peaks are located at 412 and 444 nm, while carotenoids and chlorophyll
c are monitored together at 468 nm and chlorophyll c alone at 640 nm. Peaks located
between 516 and 560 nm indicate the presence of fucoxanthin derivatives and alloxan-
thin. Conspicuous re ectance peaks are apparent at 500, 568 and 684 nm. The Ž rst
one shows the dominant colour of the water, and the second peak indicates the
starting point of water absorption, which reaches a maximum at 600 nm. The last
re ectance peak can be associated to  uorescence by chlorophyll a.
Figure 11 shows the re ectance spectrum obtained on 16 June and its derivative
curve. This was dominated by chlorophyll a, chlorophyll c and fucoxanthin. Troughs
Figure 9. (Upper) Re ectance spectrum measured on 9 May 1995. (Lower) Fourth derivative
spectrum (dashed line). Absorption peaks are apparent as local minima.

2880
R. Aguirre-Gomez et al.
Figure 10. (Upper) Re ectance spectrum measured on 14 June 1995. (Lower) Fourth derivat-
ive spectrum (dashed line). Absorption peaks are apparent as local minima.
Figure 11. (Upper) Re ectance spectrum measured on 16 June 1995. (Lower) Fourth derivat-
ive spectrum (dashed line). Absorption peaks are apparent as local minima.
due to chlorophyll a are located at 416 and 440 nm. An absorption peak of fucoxan-
thin-like carotenoids is noticed at 516 nm. Carotenoids and chlorophyll c, combined
together, absorb at 468 nm, while chlorophyll c alone is absorbing at 632 nm.
The strong water absorption is noticed at the steep slope, located within the
range 572–600 nm. Conspicuous re ectance peaks are located at 500 nm and the
 uorescence peak detected at 688 nm.
It is important to mention that the chlorophyll a absorption peak around 675 nm
(red region) was not detected through the derivative analysis. Gitelson et al. (1993)
have reported that the absorption peak at 675 nm becomes apparent when chloro-
phyll a concentration is over 3 mg m . The cases analysed in the present study were
Õ3
on the limit (9 May) or below (14 June and 16 June) this concentration value, as
shown in table 3; therefore, this peak was small enough to not be detected.

Detecting photosynthetic algal pigments
2881
5. Conclusions
The results obtained show that derivative of the re ectance spectra of marine
waters, can be used to identify absorption bands of diVerent pigments present in
natural populations of algae.
Atmospheric and water absorption peaks often conceal chlorophyll a absorption
in the blue and red regions of the electromagnetic spectrum, respectively. The blue
peak of chlorophyll a, centred near 440 nm, is located in the area of carotenoid
spectral overlap, yet it could be identiŽ ed by the derivative method. This was not
the case for chlorophyll a absorption peak at 675 nm, which was not detected because
the concentrations measured were below of the empirical limit found by Gitelson
et al. (1993).
Even though it could be easier to detect pigment signatures from dense algal
communities, the results shown here are promising and can be applied to natural
sea waters. Therefore, remote sensing of accessory pigments may be successfully
applied in other environments (e.g. relatively coloured coastal areas and eutrophic
lakes or in oceanic waters).
Using spectroradiometer data as a bridge, hyperspectral airborne remote sensing
may be able to provide useful information on accessory pigments as a tool to study
marine, estuarine, riverine and lacustrine environments on a regional level (e.g.
Carder et al. 1993, George 1997).
The use of two radiometers may substantially improve re ectance measurements.
Changes in irradiation due to atmospheric eVects (e.g. cloudiness) can be removed
by arranging the target and reference panel to be sampled at exactly the same instant
(Duggin 1982). Milton (1987) has stated that neither the sequential nor the simultan-
eous methods are suitable for periods when clouds are passing in front of the sun.
A careful design of the instruments and an accurate inter-calibration between the
two sensors are essential to this task. However, the method does not account for all
of the consequences of variations in irradiance. Nevertheless, under clear sky condi-
tions, the simultaneous method is preferable for eliminating short-term temporal
changes, but at the expense of possible greater problems in instrument calibration.
The methodology presented here looks promising in using high-resolution remote
sensors mounted in aircraft or satellites. Successful application of the technique
presented here in airborne or satellites sensors requires consideration of (a) a proper
atmospheric correction for detecting the weak phytoplankton signal from air or
space and (b) a detailed model of the interaction between sub-surface and above-
water re ectance requires development. Fischer and Santer (1995) have recently
approached the atmospheric problem using MERIS capabilities. Derivative algo-
rithms have been used for either normalizing atmospheric eVects (Philpot 1991) or
Ž ltering the atmospheric signal out from radiance spectra (Campbell and Esaias
1983). Shimwell (1996), using in situ and airborne data, has recently tackled a
sub-surface model of above-water re ectance.
Acknowledgements
The authors acknowledge Dr J. Aiken from PML for allowing them to participate
at the weekly samplings over Plymouth waters. They are also grateful to the techni-
cians at PML for providing them with the pigment analysis using HPLC techniques,
and especially to Mr Guy Westbrook for his help in the sampling programme and

2882
R. Aguirre-Gomez et al.
absorption measurements. Thanks are also given to Mr AugustÌn Fernandez for
´
diagrams drawing. The Ž rst author acknowledges the Ž nancial support given by the
´
´
Universidad Nacional Autonoma de Me´xico during his studies at the University of
Southampton.
References
Aguirre-Go’mez, R., Boxall, S. R., and Weeks, A. R., 1995, IdentiŽ cation of algal pigments
using high order derivatives. Proceedings of the International Symposium on Quantitative
Remote Sensing for Science and Applications, Florence, Italy, 10–14 July 1995
(IGARSS’95) (Florence: IEEE), vol. 3, pp. 2084–2086.
Ahn, Y. H., Bricaud, A., and Morel, A., 1992, Light backscattering eYciency and related
properties of some phytoplankters. Deep-Sea Research, 39, 1835–1855.
Barlow, R. G., Mantoura, R. F. C., Gough, M. A., and Fileman, T. W., 1993, Pigment
signatures of the phytoplankton composition in the northeastern Atlantic during the
1990 spring bloom. Deep-Sea Research, 40, 459–477.
Bidigare, R. R., Morrow, J. H., and Kiefer, D. A., 1989, Derivative analysis of spectral
absorption by photosynthetic pigments in the western Sargasso Sea. Journal of Marine
Research, 47, 323–341.
Bricaud, A., Morel, A., and Prieur, L., 1981, Absorption by dissolved organic matter of
the sea (yellow substance) in the UV and visible domains. L imnology and Oceanography,
26, 43–53.
Butler, W. L., and Hopkins, D. W., 1970, Higher derivative analysis of complex absorption
spectra. Photochemistry and Photobiology, 12, 439–450.
Campbell, J. W., and Esaias, W. E., 1983, Basis for spectral curvature algorithms in remote
sensing of chlorophyll. Applied Optics, 22, 1084–1093.
Carder, K. L., Reinersman, P., Chen, R. F., Muller-Karger, F., Davis, C. O., and
Hamilton, M., 1993, AVIRIS calibration and application in coastal oceanic environ-
ments. Remote Sensing of Environment, 44, 205–216.
Chen, Z., Curran, P. J., and Hansom, J. D., 1992, Derivative re ectance spectroscopy to
estimate suspended sediment concentration. Remote Sensing of Environment, 40, 67–77.
Dekker, A. G., Malthus, T. J., and Seyhan, E., 1991, Quantitative modelling of inland
water quality for high resolution MSS systems. IEEE T ransactions on Geoscience and
Remote Sensing, 29, 89–95.
Demetriades-Shah, T. H., Steven, M. T., and Clark, J. A., 1990, High-resolution derivative
spectra in remote sensing. Remote Sensing of Environment, 33, 55–64.
Duggin, M. J., 1982, The need to use two radiometers simultaneously to make re ectance
measurements in Ž eld conditions. Photogrammetric Engineering and Remote Sensing,
48, 142–144.
Duysens, L. N. M., 1956, The  attening of the absorption spectrum of suspension, as compared
to that of solutions. Biochemical and Biophysica Acta, 19, 1–12.
Faust, M. A., and Norris, K. H., 1985, In vivo spectrophotometric analysis of photosynthetic
pigments in natural populations of phytoplankton. L imnology and Oceanography, 30,
1316–1322.
Fischer, J., and Santer, R., 1995, Remote sensing of atmospheric properties. A perspective
view of MERIS applications. Proceedings of the 21st Annual Conference of the Remote
Sensing Society, Southampton, 1995 (Nottingham: Remote Sensing Society),
pp. 149–156.
George, D. G., 1997, Bathymetric mapping using a compact airborne spectrographic imager
(CASI). International Journal of Remote Sensing, 18, 2067–2071.
Gieskes, W. W. C., 1991, Algal pigments Ž ngerprints: clue to taxon-speciŽ c abundance,
productivity and degradation of phytoplankton in sea and oceans. In Particle Analysis
in Oceanography, edited by S. Demers (Berlin: Springer-Verlag), pp. 61–99.
Gitelson, A., Szilagyi, F., and Mittenzwey, K. H., 1993, Improving quantitative remote
sensing for monitoring of inland water quality. Water Research, 27, 1185–1194.
Goodin, D. G., Han, L., Fraser, R. N., Rundquist, D. C., and Stebbins, W. A., 1993,
Analysis of suspended solids in water using remotely sensed high resolution derivative
spectra. Photogrammetric Engineering and Remote Sensing, 59, 505–510.

Detecting photosynthetic algal pigments
2883
Gordon, H. R., 1985, Ship perturbation of radiance measurements at sea. 1: Monte Carlo
simulations. Applied Optics, 24, 4172–4182.
Gordon, H. R., and Morel, A., 1983, Remote assessment of ocean color for interpretation
of satellite visible imagery, a review. In L ecture Notes on Coastal and Estuarine Studies
(New York: Springer-Verlag), vol. 4, pp. 1–114.
Gordon, H. R., Brown, O. B., Evans, R. H., Brown, J. W., Smith, R. C., Smith, K. S.,
Baker, K. S., and Clark, D. K., 1988, A semianalytical radiance model of ocean
color. Journal of Geophysical Research, 93, 10 909–10 924.
Gulyayev, S. A., and Litvin, F. F., 1970, First and second derivatives of the absorption
spectrum of chlorophyll and associated pigments in cells of higher plants and algae
°
at 20 C. BioŽ sika, 15, 701–712.
Kirk, J. T. O., 1983, L ight and Photosynthesis in Aquatic Ecosystems (Cambridge: Cambridge
University Press).
Kirk, J. T. O., 1984, Dependence of relationship between inherent and apparent optical
properties of water on solar altitude. L imnology and Oceanography, 29, 350–356.
Kishino, M., Takaheshi, M., Okami, N., and Ichimura, S., 1985, Estimation of the spectral
absorption coeYcients of phytoplankton in the sea. Bulletin of Marine Science, 37,
634–642.
Mantoura, R. F. C., and Llewellyn, C. A., 1983, The rapid determination of algal chlorophyll
and carotenoid pigments and their breakdown products in natural waters by reverse-
phase high-performance liquid chromatography. Analytica Chimica Acta, 151, 297–314.
Milton, E. J., 1987, Principles of Ž eld spectroscopy. International Journal of Remote Sensing,
8, 1807–1827.
Morel, A., 1974, Optical properties of pure water and pure seawater. In Optical Aspects of
/
Oceanography, edited by N. G. Jerlov and E. Steeman-Nielsen (New York London:
Academic Press), pp. 1–24.
Morel, A., 1988, Optical modelling of the upper ocean in relation to its biogenous matter
content (Case I waters). Journal of Geophysical Research, 91, 10 749–10 768.
Morel, A., and Prieur, L., 1977, Analysis of variations in ocean color. L imnology and
Oceanography, 22, 709–722.
Nelson, J. R., 1993, Rates and possibly mechanism of light dependent degradation of pigments
in detritus derived from phytoplankton. Journal of Marine Research, 51, 155–179.
Pegau, W. S., and Zaneveld, J. R. V., 1993, Temperature-dependent absorption of water in
the red and near infrared portion of the spectrum. L imnology and Oceanography,
38, 188–192.
Philpot, W. D., 1991, The derivative ratio algorithm: avoiding atmospheric eVects in remote
sensing. IEEE T ransactions on Geoscience and Remote Sensing, 29, 250–357.
Prezelin, B. B., and Boczar, B. A., 1986, Molecular bases of cell absorption and  uorescence
’
’
in phytoplankton: potential applications to studies in optical oceanography. In
Progress in Phycological Research, Vol. 4, edited by F. E. Round and D. J. Chapman
(Bristol: Biopress), pp. 349–464.
Prieur, L., and Sathyendranath, S., 1981, An optical classiŽ cation of coastal and oceanic
waters based on the speciŽ c spectral absorption curves of phytoplankton pigments,
dissolved organic matter, and other particulate materials. L imnology and Oceanography,
24, 671–689.
Sathyendranath, S., and Morel, A., 1983, Light emerging from the sea—interpretation and
uses in remote sensing. In Remote Sensing Applications in Marine Science and
T echnology, edited by A. P. Cracknell (Dordrecht: D. Reidel), pp. 323–357.
Sathyendranath, S., Lazzara, L., and Prieur, L., 1987, Variations in the spectral values of
speciŽ c absorption of phytoplankton. L imnology and Oceanography, 32, 403–415.
Savitzky, A., and Golay, M. J. E., 1964, Smoothing and diVerentiation of data by simpliŽ ed
least squares procedures. Analytical Chemistry, 36, 1627–1639.
Schutt, J. B., Holben, B. N., Shai, L. M., and Henninger, J. H., 1981, Re ectivity of TFE—
a washable surface—compared with that of BaSO4. Applied Optics, 20, 2033–2035.
Shifrin, K. S., 1988, Physical Optics of Ocean Water (New York: American Institute of Physics).
Shimwell, S. J., 1996, An investigation into the optical properties of North Sea coastal waters
in relation to ocean colour remote sensing. PhD thesis, University of Southampton,
UK.

2884
Detecting photosynthetic algal pigments
Smith, R. C., and Baker, K. S., 1981, Optical properties of the clearest natural waters. Applied
Optics, 20, 177–184.
Swain, P. H., and Davis, S. M., 1978, Remote Sensing: the quantitative approach (New York:
McGraw-Hill).
Weidner, D. L., and Hsia, J. J., 1981, Re ection properties of pressed polytetra uorethylene
powder. Journal of the Optical Society of America, 71, 856–861.
Wright, S. W., Jeffrey, S. W., Mantoura, R. F. C., Llewellyn, C. A., Bjornland, T.,
Repeta, D., and Welshmeyer, N., 1991, Improved HPLC method for the analysis of
chlorophylls and carotenoids from marine samples. Marine Ecology Progress Series,
77, 183–196.
Zapata, M., Ayala, A. M., Franco, J. M., and Garrido, J. L., 1987, Separation of chloro-
phylls and their degradation products in marine phytoplankton by reversed phase
high performance liquid chromatography. Chromatographia, 23, 26–30.