Full text of DOI:10.1080/01431160118022

International Journal of Remote Sensing
ISSN: 0143-1161 (Print) 1366-5901 (Online) Journal homepage: http://www.tandfonline.com/loi/tres20
Reflectance properties of hydrographically and
optically stratified fjords (Scottish sea lochs)
during the Spring diatom bloom
A. Cunningham , P. Wood & K. Jones
To cite this article: A. Cunningham , P. Wood & K. Jones (2001) Reflectance properties of
hydrographically and optically stratified fjords (Scottish sea lochs) during the Spring diatom
bloom, International Journal of Remote Sensing, 22:15, 2885-2897
To link to this article: http://dx.doi.org/10.1080/01431160118022
Published online: 25 Nov 2010.
Submit your article to this journal
Article views: 27
View related articles
Citing articles: 4 View citing articles
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=tres20
Download by: [Florida Atlantic University]
Date: 21 March 2016, At: 23:51

int. j. remote sensing, 2001, vol. 22, no. 15, 2885–2897
Re ectance properties of hydrographically and optically stratiŽ ed
fjords (Scottish sea lochs) during the Spring diatom bloom
A. CUNNINGHAM and P. WOOD
Department of Physics and Applied Physics, University of Strathclyde,
107 Rottenrow, Glasgow, G4 ONG, Scotland, UK
e-mail: a.cunningham@strath.ac.uk
p.wood@read.com
K. JONES
DunstaVnage Marine Laboratory, P.O. Box 3, Oban, Argyll, PA34 4AD,
Scotland, UK
kjj@dml.nerc.ac.uk
(Received 20 October 1999; in Ž nal form 17 April 2000)
Abstract. Re ectance spectra of Scottish sea lochs were measured using a spe-
cially constructed  oating spectroradiometer with a bandpass of 400–800 nm and
4 nm resolution over a period of 10 weeks during the development of a Spring
diatom bloom. Most stations showed marked salinity stratiŽ cation with high
1
surface gelbstoV concentrations and chlorophyll levels up to 80 mglÕ .
Characteristic features of the re ectance spectra included very low values in the
blue, a maximum around 600nm and distinct solar stimulated  uorescence peaks
at 685nm. The height of the  uorescence peak (scaled by dividing by the maximum
re ectance value) was well correlated with near-surface chlorophyll concentra-
tions. The ratio of re ectances at 400 and 650nm could be used as an indicator
of surface gelbstoV concentration. However the commonly observed correlation
/
between the blue green re ectance ratio (443nm:555 nm) and surface chlorophyll
concentration held only at stations where a surface layer of low-salinity, high
gelbstoV water was not present. Calculations of the diVuse attenuation coeYcient
at 440nm showed that layers with the highest chlorophyll concentrations were
frequently located below one optical depth in the water column, implying that
much of the phytoplankton biomass in these waters would not be detectable by
optical remote sensing.
1. Introduction
The sea lochs of the west coast of Scotland share the morphological and hydrolo-
gical characteristics of larger fjords on other glaciated coastlines. They consist of
one or more deep basins isolated by relatively shallow rock sills, and receive high
freshwater inputs from the surrounding uplands (Edwards and Edelsten 1977). These
uplands are often covered by peat, and the rivers which drain them are heavily
stained by coloured dissolved organic matter (gelbstoV ) which has an absorption
spectrum that is highest in the ultra-violet and decreases exponentially at longer
wavelengths (Bricaud et al. 1981, Green and Blough 1994). For most of the year,
International Journal of Remote Sensing
ISSN 0143-1161 print/ISSN 1366-5901 online © 2001 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/01431160010007006

2886
A. Cunningham et al.
sea lochs show strong stratiŽ cation in salinity and temperature (Gillibrand et al.
1996) accompanied by marked layering of their inherent optical properties (McKee
et al. 1999). These waters clearly fall within the Case 2 classiŽ cation of Morel and
Prieur (1977), and their high gelbstoV content and hydrographic structure renders
them optically much more complex than the Case 1 waters of ocean basins
(Sathyendranath et al. 1989). Sea lochs often contain high phytoplankton concentra-
tions, particularly during the diatom-dominated spring bloom. The timing of the
bloom appears to be linked to stratiŽ cation of the water column after winter mixing,
which is assisted by the input of freshwater to the surface layer (Tett and Wallis
1978, Tett et al. 1986).
The re ectance properties of fjords have not been well studied, partly because
their scale is often of the same order of magnitude as the 1 km spatial resolution of
existing ocean colour satellites. However the imminent launch of new sensors with
signiŽ cantly higher resolution (MODIS, MERIS) together with an increasing need
to monitor the ecological status of fjordic coastlines is likely to raise interest in this
topic. In addition, the presence of yellow-stained surface water has a strong visible
eVect on the re ectance of sea lochs and adjacent coastal seas and this eVect must
be quantiŽ ed in order to develop appropriate algorithms for optical remote sensing.
We therefore describe a study of the remote sensing re ectance of stratiŽ ed sea lochs
carried out oV the west coast of Scotland during the spring diatom bloom in 1998.
It is likely that the results obtained are applicable to other fjordic regions and also
to areas such as the Baltic Sea which have high surface gelbstoV concentrations.
2. Methodology
2.1. Hydrography and sample analysis
The work reported here was carried out from March to May of 1998 from the
RV Calanus in Lochs Linnhe, Creran and Etive and in the Lynn of Lorne at stations
indicated in Ž gure 1. The study period encompassed the development of the spring
bloom, which was heavily dominated by chain-forming diatoms. At each station,
casts were made with a SeaBird CTD system Ž tted with a SeaTech transmissometer
and  uorometer, and water samples were taken at three depths using 30 l Niskin
bottles. Photosynthetic pigments were measured by GF/F Ž ltration, acetone extrac-
tion and spectrophotometry (JeVrey et al. 1997). GelbstoV concentration was deter-
mined measuring the absorption at 440 nm of samples passed through 0.2 mm Ž lters
(Bricaud et al. 1981). Suspended particulate material was measured by weighing
material collected from 5 litres of seawater on oven-dried GF/F Ž lters (Parsons et al.
1984). The analytical methods did not discriminate between chlorophyll a and its
immediate degradation product phaeophytin a, nor did they allow any deductions
to be made about the chemical composition of the dissolved organic matter.
2.2. Spectroradiometry
Downward spectral irradiance in air (E ) and the upward radiance just beneath
d air
the surface (L )
were measured using a specially constructed spectroradiometer.
u water
The electronics and optical layout of the sensing head are described more fully in
Wood and Cunningham (2001). The instrument employed a thermoelectrically cooled
CCD array as the detecting element, and had a useful bandpass from 400 nm to
800 nm with 4 nm spectral resolution. Light from dual collectors (a diVusing disk for
downward irradiance and an achromatic lens and aperture for upward radiance)
was transmitted to the sensor by means of a bifurcated Ž bre optic bundle. The ends

Fjord spectral re ectance
2887
Figure 1. Map of the study area on the west coast of Scotland, with stations indicated as
Ž lled circles.
of the bundle adjacent to the collectors were Ž tted with electrically operated shutters,
so that sequential radiance and irradiance measurements could be made at roughly
one second intervals. The response of the irradiance collector was within 2% of a
true cosine for angles between 0 and 65 , and within 10% for angles up to 85 : the
°
radiance collector had an acceptance angle of 1 . For the measurements reported
°
°
here, the radiometer was enclosed in a waterproof housing and  oated on a raft
(Ž gure 2) so that the irradiance collector was approximately 25 cm above the surface
and the radiance collector penetrated to a depth of around 30 cm. The raft was
constructed from four air-Ž lled 200 mm diameter plastic pipes braced with galvanised
steel tubes, forming a rectangle roughly 3 m long and 2 m wide. The radiometer
sensing head was mounted at one end of the raft and a second waterproof housing
containing a computer and power supply positioned at the other. The computer was
controlled by a monitor and keyboard in the ship’s laboratory through a long-
distance communications link (Maxlink), enabling the raft to be  oated up to 35 m
from the ship. This minimised ship shadow eVects (Helliwell and Sullivan 1990),
though some residual self-shading of the instrument may have occurred. The raft
was operated successfully in the relatively sheltered waters of the sea lochs in wave
heights of up to approximately 35 cm, and around ten consecutive re ectance meas-
urements were averaged in order to reduce any eVect of wave-induced motions.
Since remote sensing re ectance (R ) is deŽ ned as the ratio of upward radiance to
downward irradiance just above the water surface ((L ) /(E ) ), water radiance
u air d air
rs

2888
A. Cunningham et al.
Figure 2. Schematic drawing of the spectroradiometer, approximately to scale: the radiance
collector which penetrates the sea surface is hidden behind the cylindrical  oats.
measurements were converted to air values using the relationship (L )
(L )
u water
0.55
=
u air
(Mobley 1994).
2.3. Radiative transfer modelling
Monte Carlo calculations of radiative transfer were carried out using a program
written in Turbo Pascal which was based on the Fortran code of Kirk (1981) with
the addition of depth-dependent inherent optical properties. Details of the structure
and testing of the program are given in Wood (1998): among other checks of validity,
the output of the code for simple unstratiŽ ed water columns was found to be
consistent with the semi-analytical model of Morel and Prieur (1977). Monte Carlo
calculations are capable of modelling underwater light Ž elds and re ectance spectra
in considerable detail, but in the context of this paper were used to estimate the
diVuse attenuation coeYcient for near-surface fjord water within the range of chloro-
phyll and gelbstoV concentrations observed at sea. For these calculations the absorp-
tion and scattering contributions of inorganic particles in surface waters was assumed
to be negligible compared to phytoplankton cells. Phytoplankton absorption was
estimated by taking the mean absorption spectrum from the multi-species survey of
Prieur and Sathyendranath (1981) and scaling it by the chlorophyll-speciŽ c absorp-
tion at 675 nm of the diatom Skeletonema costatum (Sathyendranath et al. 1987).
The chlorophyll-speciŽ c total scattering coeYcient was calculated using the expres-
sion derived by Gordon and Morel (1983), with an assumed backscattering ratio
(derived from more extensive modelling work) of 0.7%. GelbstoV spectral absorption
was measured by spectrophotometry.
3. Results
3.1. Relationships between in situ measurements and seawater composition
When samples from all the stations were aggregated, strong linear correlations
were found between the measured concentration of chlorophyll a (including any
phaeophytin a) and the SeaTech  uorometer signal (Ž gure 3) and between CDOM
and measured salinity (Ž gure 4). These relationships are unlikely to be universally

Fjord spectral re ectance
2889
Figure 3. Linear relationship between SeaTech  uorometer output and extracted chlorophyll
a (including phaeophytin a). All water samples from the study area are included. The
 uorometer reading was obtained at the same depth as the sample.
Figure 4. High correlation between gelbstoV concentration, measured as the absorption
coeYcient of Ž ltered samples at 440nm, and salinity for all samples from the study area.
valid, but they indicate that within the limited time frame and geographical area of
this study  uorescence and salinity can be used as proxy variables for chlorophyll
and gelbstoV concentrations respectively. This makes it possible to deduce depth
proŽ les of these substances at greater resolution than that provided by the limited
number of water samples taken at each station.
3.2. Water column stratiŽ cation
Figure 5 shows salinity and temperature proŽ les for a typical station in Loch
Etive measured at 13.20 hrs on 5 May 1998. A low salinity surface layer 6 m thick
lay over an intermediate layer whose salinity increased with depth down to 20 m.
Below 20 m a layer of rather uniform salinity and temperature extended to within a
few meters of the sea bed. SeaTech  uorescence and attenuation values were strongly
correlated at this station (Ž gure 6), indicating that phytoplankton cells were the

2890
A. Cunningham et al.
Figure 5. Salinity and temperature proŽ les for a typical station in Loch Etive, showing low
salinity surface water, an intermediate mixed layer, and a deep layer of higher salinity.
major source of beam attenuation at 650 nm and that suspended inorganic particles
were of much less signiŽ cance. The vertical variation in the  uorometer signal
indicated that the three hydrographic layers had diVerent phytoplankton concentra-
tions. A water sample taken near the surface at this station showed a CDOM
attenuation at 440 nm of 0.8 m and a chlorophyll a concentration of 6.2 mg l
Õ1 Õ1
The striking  uorescence maximum which occurred in the intermediate layer
.
corresponded to a chlorophyll a concentration of 20 mg l
.
Õ1
3.3. Re ectance measurements
Figure 7 shows typical irradiance (E ) and radiance (L ) spectra measured on the
d
u
Loch Etive station at the same time as the CTD proŽ les in Ž gure 5. Sea conditions
were calm, and the Sun was shining through a thin layer of strata. Most of the
Fraunhofer and atmospheric absorption bands in the solar irradiance spectrum are
lost at the 4 nm spectral resolution of the instrument, but the oxygen absorption
band at 760 nm is still evident. The remote sensing re ectance spectrum (Ž gure 8) is
low in the blue with a maximum around 600 nm, which is consistent with the brown
visual colour of the water. The general shape of the spectrum is remarkably similar
to that observed in Puget Sound, a Washington fjord, by Culver and Perry (1997).
The peak at 685 nm can be attributed to solar-stimulated chlorophyll  uorescence
(Gordon 1979), while the small but distinct bump in the re ectance spectrum at
760 nm may be caused by Ž lling of the oxygen absorption band by inelastic scattering
in a manner analogous to the Fraunhofer line Ž lling discussed by Ge et al. (1993 ).

Fjord spectral re ectance
2891
Figure 6. ProŽ les of SeaTech  uorescence and beam attenuation at 650nm for the
station shown in Ž gure 5. The optical signals (and by implication the phytoplankton
concentration) are highest in the intermediate layer extending from 6 to 20 m.
Figure 7. Downward irradiance (E ) and upward radiance (L ) spectra measured using the
 oating radiometer for the station illustrated in Ž gure 5. L Ž gures have been corrected
for refraction to give values in air just above the sea surface.
d
u
u
3.4. Solar-stimulated chlorophyll  uorescence
The possibility of sensing phytoplankton by solar-stimulated chlorophyll a  uo-
rescence was discussed by Neville and Gower (1977), and recent quantitative studies
include those of Garcia-Mendoza and Maske (1996) and Babin et al. (1996). The

2892
A. Cunningham et al.
/
Figure 8. Remote sensing re ectance (L E ) calculated from the data in Ž gure 7. The graph
has been scaled by dividing by the peak re ectance value. The geometrical method
u
d
ê
used to obtain the scaled solar-stimulated  uorescence peak height (F ) is also
illustrated.
latter authors point out that the  uorescence peak is a surprisingly distinct feature
in high resolution re ectance spectra, and suggest that that it can be used as an
indicator of the occurrence of near-surface chlorophyll at concentrations well below
1 mg l . For our sea loch data, the height of the re ectance peak at 685 nm was
Õ1
measured against a background obtained by linear extrapolation between 660 nm
and 745 nm (Culver and Perry 1997) and then divided by the maximum re ectance
value observed in the spectrum (as suggested by Gitelson 1992). The results obtained
show a clear linear correlation (r 0.94) with surface chlorophyll concentration
2=
(Ž gure 9), though without the normalisation step the correlation coeYcient was
reduced to r 0.75. It should be noted that the Spring bloom in sea lochs develops
2=
ê
Figure 9. Relationship between scaled solar-stimulated  uorescence (F in Ž gure 8) and
near-surface chlorophyll a concentration for all stations.

Fjord spectral re ectance
2893
when the sun angle is low, and conditions on the Scottish west coast are often
overcast. Consequently diurnal variations in non-photochemical quenching which
aVect the apparent chlorophyll-speciŽ c  uorescence yield are unlikely to be as marked
as those discussed by Marra (1992).
3.5. Assessment of wavelength-ratio algorithms
Most expressions used for extracting the concentration (C ) of a seawater constitu-
i
ent i from re ectance values measured at wavelengths l , l are of the form C
=
A(R /R ) where A and B are empirically derived constants (Rhea and Davis
1
2
i
B
1997). Figure 10 shows that this scheme works well for surface gelbstoV concentra-
tions in sea lochs when re ectances are measured at 400 nm and 650 nm and the
l1 l2
constants are set to A 0.22 and B
1.05 (r
0.88). However when the same
2=
form of algorithm (Morel 1978) is applied to chlorophyll concentrations, a more
=
=-
interesting situation arises. Figure 11 shows the performance of the algorithm at
SeaWiFs wavebands (443 and 555 nm) with the constants set to A 0.38 and B
=
3.4. The data, which represents surface samples from all the stations shown in
=
-
Ž gure 1, falls into two groups. One group follows the expected relationship while the
second diverges widely from it. The second group comes from stations with low
surface chlorophyll due to the presence of a layer of highly coloured surface water
of reduced salinity. If the band ratio algorithm was applied blindly to these stations
it would generate spuriously high chlorophyll values. This might have a bearing on
the observation by Gieskes (1990) that CZCS images persistently indicate high
concentrations of chlorophyll in the North Sea in winter, though suspended sediment
may also impair the performance of chlorophyll retrieval algorithms in shallow
waters.
3.6. Optical and physical depths
We have used a Monte Carlo model to calculate diVuse attenuation coeYcients
(K ) at 400 nm for the range of gelbstoV and chlorophyll concentrations actually
observed in surface waters. The results are displayed as contours of attenuation
d
Figure 10. Power law relationship between CDOM concentration (measured as the absorp-
tion of Ž ltered samples at 440nm) and the ratio of remote sensing re ectance values
measured at 400nm and 650nm.

2894
A. Cunningham et al.
Figure 11. Near-surface chlorophyll a concentration plotted against the ratio of remote
sensing re ectance values at 443 nm and 555nm. Filled triangles, which indicate
stations where signiŽ cant chlorophyll concentrations extend to the surface, show a
/
power law relationship. Open triangles, for stations with low chlorophyll high gelbstoV
surface water, form a distinct second grouping.
depth, the physical depth corresponding to the 1/(K ), on a plot of chlorophyll
against gelbstoV. This allows the attenuation depth for any combination of concentra-
d
tions of the two materials to be read oV (Ž gure 12). The points plotted on Ž gure 12
represent those stations which are indicated as Ž lled triangles in Ž gure 11, and whose
surface chlorophyll concentration can be predicted from the blue/green re ectance
ratio. For all but one of these stations the attenuation depth is less than 5 m, but
high chlorophyll concentrations were found at signiŽ cantly greater depths (see
Ž gure 4 for a typical example). At any given wavelength, approximately 90% of the
upwelling radiance (L ) originates from one attenuation depth (Apel 1987).The
u l
Figure 12. Contours showing the calculated loci at intervals of 1 m of physical depths
corresponding to one optical depth plotted on a graph of chlorophyll concentration
against CDOM concentration. The plotted points indicate positions occupied by the
stations represented by Ž lled triangles in Ž gure 11.

Fjord spectral re ectance
2895
obvious conclusion is that even where surface gelbstoV does not seriously aVect the
performance of the chlorophyll retrieval algorithm, much of the phytoplankton
biomass in sea lochs may be located too deep in the water column to have a
signiŽ cant aVect on remote sensing re ectance.
4. Discussion
The most notable features of sea loch re ectance spectra are low values at blue
wavelengths, a maximum around 600 nm, and a well-deŽ ned solar-stimulated  uo-
rescence peak at 685 nm. Similar spectral shapes are found in other high-gelbstoV
waters such as the Baltic Sea (Siegel et al. 1994) and in bays with high river water
inputs (Carder et al. 1986). The deleterious eVect of surface gelbstoV on the perform-
ance of standard remote sensing algorithms has been discussed by Witte et al. (1982 ),
Sathyendranath et al. (1989) and DoerVer and Schiller (1994) among others. It is
generally agreed that the use of deep blue re ectance (close to 400nm) to estimate
gelbstoV concentrations is a necessary preliminary step before attempting chlorophyll
retrieval, and Ž gures 10 and 11 bear this out. One unexpected outcome of the survey
was the observation that passive  uorescence was a more reliable indicator of surface
chlorophyll concentration in these waters than the blue-green re ectance ratio. This
may be of practical value, since the recently-launched MODIS sensor is suYciently
sensitive at red wavelengths to detect  uorescence from chlorophyll concentrations
as low as 0.5 mg l (Letelier and Abbot 1996). The high correlation between gelbstoV
Õ1
concentration and salinity, which has also been noted by other authors (e.g. Sørensen
and Aas 1994), suggests that gelbstoV may be used as an optical tracer for studying
mixing dynamics in fjordic systems. The observation of chlorophyll layers below the
level which in uences surface re ectance may limit the remote sensing of primary
production not only in fjords but in other high-gelbstoV regions. For example
Mitchell et al. (1991) found phytoplankton blooms embedded in the pycnocline in
the Barents Sea which were not detected by CZCS algorithms. In conclusion, the
results obtained suggest that sea lochs provide easily accessible examples of vertically
structured case II waters which are excellent sites for developing and validating
models of remote sensing re ectance. These models are required if we are to exploit
the opportunities oVered by the imminent launch of new ocean colour sensors with
sub-kilometre spatial resolution for studying physical and biological processes in
coastal waters.
Acknowledgments
This work was funded by the UK Natural Environment Research Council,
who also provided a studentship for Peter Wood. Optical calibration facilities were
kindly provided by Professor Jim Aiken and Gerald Moore at Plymouth Marine
Laboratory. Professor Andre Morel Ž rst suggested converting our ship-borne radio-
´
meter to a  oating instrument, and Iain MacCallum at the University of Strathclyde
wrote the Monte Carlo radiative transfer model.
References
Apel, J. R., 1987, Principles of Ocean Physics (London: Academic Press).
Babin, M., Morel, A., and Gentili B., 1996, Remote sensing of sea-surface sun-induced
chlorophyll  uorescence: consequences of natural variations in the optical character-
istics of phytoplankton and the quantum yield of chlorophyll a  uorescence.
International Journal of Remote Sensing, 17, 2417–2448.

2896
A. Cunningham et al.
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.
Carder, K. L., Steward, R. G., Paul, J. H., and Vargo, G. A., 1986, Relationships between
chlorophyll and ocean colour constituents as they aVect remote-sensing re ectance
models. L imnology and Oceanography, 31, 403–413.
Culver, M. E., and Perry, M. J., 1997, Calculation of solar-induced  uorescence in surface
and sub-surface waters. Journal of Geophysics Research, 102, 10 563–10 572.
Doerffer, R., and Schiller, H., 1994, Inverse modeling for retrieval of ocean colour para-
meters in case I coastal waters: an analysis of minimum error. SPIE, 2258 Ocean
Optics XII, 887–893.
Edwards, A., and Edelsten, D. J., 1977, Deep water renewal of Loch Etive: a three basin
Scottish fjord. Estuarine and Coastal Marine Science, 5, 575–595.
Garcia-Mendoza, E., and Maske, H., 1996, The relationship between solar-stimulated natural
 uorescence and primary productivity in Mexican PaciŽ c waters. L imnology and
Oceanography, 41, 1697–1710.
Ge, Y., Voss, K. J., and Gordon, H. R., 1993, In situ measurements of inelastic light scattering
in Monterey Bay using solar Fraunhofer lines. Journal of Geophysical Research, 100,
13 227–13 236.
Gillibrand, P. A., Turrell, W. R., Moore, D. C., and Adams, R. D., 1996, Bottom water
stagnation and oxygen depletion in a Scottish sea Loch. Estuarine, Coastal and Shelf
Science, 43, 217–235.
Gieskes, W. W. C., 1990, Algal pigment Ž ngerprints: clue to taxon-speciŽ c abundance, produc-
tivity and degradation of phytoplankton in seas and oceans. In Particle Analysis in
Oceanography, edited by S. Demers (Berlin: Springer Verlag), pp 61–99.
Gitelson, A. A., 1992, The peak near 700 nm on radiance spectra of algae in water: relationships
of its magnitude and position with chlorophyll concentration. International Journal of
Remote Sensing, 13, 3367–3373.
Gordon, H. R., 1979, DiVuse re ectance of the ocean: the theory of its augmentation by
chlorophyll a  uorescence at 685nm. Applied Optics, 18, 1161–1166.
Gordon, H. R., and Morel, A., 1983, Remote assessment of ocean colour for interpretation of
satellite visible imagery, a review. Lecture Notes on Coastal and Estuarine Studies,
Vol. 4 (New York: Springer Verlag).
Green, S. A., and Blough, N. V., 1994, Optical absorption and  uorescence properties
of chromophoric dissolved organic matter in natural waters. L imnology and
Oceanography, 19, 1903–1916.
Helliwell, W. S., and Sullivan, G. N., 1990, Ship shadowing: model and data comparisons.
SPIE, 1302 Ocean Optics X, 55–71.
Jeffrey, S. W., Mantoura, R. F. C., and Wright, S. W., 1997, Phytoplankton pigments in
oceanography (Paris: UNESCO).
Kirk, J. T. O., 1981, Monte Carlo procedure for simulating the penetration of light into
natural waters. CSIRO Australia, Division of Plant Industries T echnical Papers, 36, 1–16.
Letelier, R. M., and Abbot, M. R., 1996, An analysis of chlorophyll  uorescence algorithms
for the moderate resolution imaging spectrometer (MODIS). Remote Sensing of the
Environment, 58, 215–223.
Marra, J., 1992, Diurnal variability in chlorophyll  uorescence: observations and modelling.
SPIE 1750 Ocean Optics XI, 233–244.
McKee, D., Cunningham, A., and Jones, K., 1999, Simultaneous measurements of  uorescence
and beam attenuation: instrument characterisation and interpretation of signals from
stratiŽ ed coastal waters. Estuarine, Coastal and Shelf Science, 48, 51–58.
Mitchell, B. G., Brody, E. A., Eueng-Nan, Y., McClain, C., Comiso, J., and Maynard,
N. G., 1991, Meridional zonation of the Barents Sea ecosystem inferred from satellite
remote sensing and in situ bio-optical observations. Polar Research, 10, 147–162.
Mobley, C. D., 1994, L ight and water: radiative transfer in natural waters (San Diego:
Academic Press).
Morel, A., 1978, Remote measurement of ocean color. Boundary L ayer Meteorology, 18,
177–201.
Morel, A., and Prieur, L., 1977, Analysis of variations in ocean colour. L imnology and
Oceanography, 22, 709–722.

Fjord spectral re ectance
2897
Neville, R. A., and Gower, J. F. R., 1977, Passive remote sensing of phytoplankton via
chlorophyll a  uorescence. Journal of Geophysical Research, 82, 3487–3493.
Parsons, T. R., Maita, Y., and Lalli, C. M., 1984, A manual of chemical and biological
methods for seawater analysis (Oxford: Pergamon Press).
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,
26, 671–689.
Rhea, W. J., and Davis, C. O., 1997, A comparison of the SeaWiFS chlorophyll and CZCS
pigment algorithms using optical data from the 1992 JGOFS Equatorial PaciŽ c time
series. Deep-Sea Research II, 44, 1907–1925.
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.
Sathyendranath, S., Prieur, L., and Morel, A., 1989, A three-component model of ocean
colour and its application to remote sensing of phytoplankton pigments in coastal
waters. International Journal of Remote Sensing, 10, 1373–1394.
Siegel, H., Gerth, M., and Beckert, M., 1994, The variation of optical properties in the
Baltic Sea and algorithms for the application of remote sensing data. SPIE, 2258
Ocean Optics XII, 894–905.
Sørensen, K., and Aas, E., 1994, Remote sensing of coastal water quality. SPIE, 2258 Ocean
Optics XII, 332–341.
Tett, P., and Wallis, A., 1978, The general annual cycle of chlorophyll standing crop in Loch
Creran. Journal of Ecology, 66, 227–239.
Tett, P., Gowen, R., Grantham, B., and Jones, K., 1986, The phytoplankton ecology of the
Firth of Clyde sea lochs Striven and Fyne. Proceedings of the Royal Society of
Edinburgh, 90B, 223–238.
Witte, W. G., Whitlock, C. H., Harris, R. C., Usry, J. W., Poole, L. R., Houghton,
W. M., Morris, W. D., and Gurganus, E. A., 1982, In uence of dissolved organic
materials on turbid water optical properties and remote-sensing re ectance. Journal
of Geophysical Research, 87, 441–446.
Wood, P., 1998, Hyperspectral measurement and modelling of marine remote sensing
re ectance, PhD thesis, University of Strathclyde, Glasgow.
Wood, P., and Cunningham, A., 2001, Ship-borne measurement of sea colour: development
of a CCD-based re ectance radiometer and results from a meridional transect of the
Atlantic Ocean. International Journal of Remote Sensing, 22, 99–111.