Embed Size (px)
Citation preview
Ui
D
1
Clamfltrm�f
snudtivt
avmMU
2
se of a Spectralon panel to measure the downwellingrradiance signal: case studies and recommendations
avid Doxaran, Nagur C. Cherukuru, Samantha J. Lavender, and Gerald F. Moore
Field determinations of the remote sensing reflectance signal are necessary to validate ocean colorsatellite sensors. The measurement of the above-water downwelling irradiance signal Ed�0�� is com-monly made with a reference plaque of a known reflectance. The radiance reflected by the plaque �Ldspec�can be used to determine Ed�0�� if the plaque is assumed to be near Lambertian. To test this assump-tion, basic experiments were conducted on a boat under changing sky conditions �clear, cloudy, covered�and with different configurations for simultaneous measurements of Ldspec and Ed�0��. For all mea-surement configurations, results were satisfactory under a clear sky. Under cloudy or covered skies,shadow effects on the plaque induced errors up to 100% in the determination of Ed�0��. An appropriatemeasurement configuration was defined, which enabled Ed�0�� to be determined with an accuracy ofbetter than �15% regardless of the sky conditions. © 2004 Optical Society of America
OCIS codes: 000.2170, 010.4450, 120.4800.
cwahcapi
wrtTa
c
ttscswaEst
. Introduction
alibration and validation of remote sensing �satel-ite or airborne� ocean color data require regular andccurate field optical measurements. One of theain measured parameters is the water-leaving re-
ectance, commonly called remote sensing reflec-ance �Rrs, in sr�1�. This Rrs signal is defined as theatio between the water-leaving radiance �Lw, in W
�2 nm�1 sr�1� and the downwelling irradianceEd�0��, in W m�2 nm�1� just above the water sur-ace.1
Protocols have been defined concerning the mea-urement of these two parameters.2 Different tech-iques to determine Lw from above-water ornderwater measurements have been widelyiscussed.3–9 Most of these studies were focused onhe uncertainty in Lw determinations, depending onnstrumental characteristics and environmentalariability, with only a small discussion concerninghe techniques used to measure Ed�0�� and its asso-
D. Doxaran �[email protected]�, N. C. Cherukuru,nd S. J. Lavender are with the School of Earth, Ocean and En-ironmental Sciences, University of Plymouth, Drake Circus, Ply-outh, Devon PL4 8AA, UK. G. F. Moore is with the Plymoutharine Laboratory, Prospect Place, Plymouth, Devon PL1 3DH,K.Received 6 February 2004; revised manuscript received 2 August
004; accepted 19 August 2004.0003-6935�04�325981-06$15.00�0
h© 2004 Optical Society of America
1
iated accuracy.6 Ed�0�� can be measured directlyith an irradiance sensor or indirectly with a radi-nce sensor and a reference plaque �e.g., Spectralon�aving a known bidirectional reflectance. In thisase, the measured signal is the downwelling radi-nce �Ldspec� reflected by the plaque. Assuming thelaque to be near Lambertian,10,11 Ed�0�� is approx-mately determined as2
Ed�0�� ��
RgLdspec, (1)
here Rg �dimensionless� is the plaque’s bidirectionaleflectance function. The wavelength dependence ofhe parameters is omitted to simplify the notation.he Ldspec and Rg parameters also depend on the Sunnd viewing geometry.2This second technique is commonly adopted be-
ause it requires use of only one radiance sensor.12–22
Our purpose in this paper is to quantify the uncer-ainties in use of the reference plaque, depending onhe configuration of the measurement �respective po-itions of the plaque and sensor� and illuminationonditions �cloud cover�. Different cases have beentudied during fieldworks in the Plymouth coastalaters comparing Ed�0�� determined with a radi-nce sensor and a Spectralon plaque system withd�0�� measured with an irradiance sensor. Mea-urements were carried out from 350 to 950 nm �spec-ral resolution of 3.3 nm� with a Trios RAMSES-ARC
yperspectral radiance sensor �7° field of view� and a0 November 2004 � Vol. 43, No. 32 � APPLIED OPTICS 5981
Trpasp
2
A
TEamstm
rSwc�ntlfOr
ttcmwhbd�ioc
pdepcsttp
wssvrt�av
tpmsTss
aE
it
ttl
io
B
WbtcfcswrtdtA��opEailTota
�ssltdm
5
rios RAMSES-ACC-VIS irradiance sensor. Theeference plaque was a Labsphere white Spectralonanel �SRT-99-050� of dimensions 12.5 cm 12.5 cmnd 99% efficiency �i.e., Rg 99%�. Results are pre-ented and discussed, leading to some importantractical recommendations.
. Effects of Changing Sky Conditions
. Objective and Method
he preliminary objective is to know how accuratelyd�0�� can be determined with a radiance sensor andSpectralon panel system under quite different illu-ination conditions �clear, cloudy, and covered
kies�. Notably, a specific objective is to estimatehe errors committed during previous field opticaleasurements in the Gironde and Loire estuaries.14
To address this problem, measurements were car-ied out during two consecutive days in the Plymouthound and the mouth of the Tamar estuary �south-est UK�. On 3 September 2003, optical data were
ollected from 11:15 to 17:00 Greenwich Mean TimeGMT�. The sky was covered and almost homoge-eous; from time to time, the Sun was visible behindhe clouds. On 4 September 2003 data were col-ected from 09:15 to 15:45 GMT. The sky was per-ectly clear, i.e., blue without any apparent cloud.n both dates, the sea surface was almost plane,
esulting in slight movements of the boat.The irradiance sensor was fixed to a vertical bar on
he top of the boat, viewing the zenith. Concerninghe radiance sensor and the Spectralon panel, theonfiguration adopted during previous field opticaleasurements in the Gironde and Loire estuariesas reproduced. The Spectralon plaque was heldorizontally by a person on the back extremity of theoat, at a distance of 4 m from the cockpit. Theimensions of the cockpit are 4 m �width� 2 mheight� 6 m �length�. The same person was point-ng the radiance sensor vertically toward the middlef the Spectralon plaque. A minimum distance of 20m was respected between the sensor and the plaque.
On 4 September 2003, under a clear blue sky, thelaque was presented in front of the Sun to avoid anyirect shadow effect. On 3 September, under a cov-red sky, the plaque was presented to the clearestart of the sky, where the Sun was hidden by thelouds. At each station, the irradiance and radianceignals were recorded simultaneously five consecu-ive times. Ten stations were completed on 3 Sep-ember; the same ten stations �i.e., same geographicalositions� were completed on 4 September.The five consecutive radiance and irradiance scansere inspected and then averaged. Whatever the
ky conditions, the recorded irradiance spectrahowed slight variations �typically �4% of the meanalue, highest difference observed in the near infra-ed �700–900 nm��. The five recorded radiance spec-ra systematically showed higher variations �up to12% of the mean value, highest difference observedt low wavelengths �400–500 nm��. These higher
ariations certainly resulted from the movements of i982 APPLIED OPTICS � Vol. 43, No. 32 � 10 November 2004
he boat, inducing movements of the radiance sensorlus the Spectralon panel system. Because of theseovements, the sensor was not always viewing the
ame surface of the plaque with the same angle.hese first observations logically indicate a bettertability of the signal measured with the irradianceensor.Then the mean values of the radiance and irradi-
nce signals were considered, respectively noted asd and Ldspec. Their ratio was calculated as
rspec �Rg Ed
Ldspec(2)
n steridians. According to Eq. �1�, it was expectedo obtain � sr.
The difference between the measured Ed�0�� andhe Ed�0�� signal derived from the measurement onhe plaque, assumed to be Lambertian, was calcu-ated as
diffspec ��Rg
�1Ldspec � Ed
Ed� 100 (3)
n percent. The variations of rspec and diffspec werebserved as a function of wavelength.
. Results
e first present the results obtained under a clearlue sky on 4 September 2003 �Fig. 1�. As expected,he obtained rspec ratios �Fig. 1�a�� show values ratherlose to �. Low spectral variations are observedrom 400 to 900 nm, except at specific wavelengthsorresponding to gaseous absorption peaks �e.g., ab-orption by oxygen at 687 and 762 nm�.23 At theseavelengths, the observed slight variations certainly
esult from slight differences in the spectral calibra-ion of the two Trios sensors. However, the observedifferences between the measured Ed�0�� signal andhe one derived from Eq. �1� are significant �Fig. 1�b��.s can be seen, diffspec ranges between �20% and25% and is typically positive with a mean value of10%. Thus, assuming the reliability and accuracy
f the irradiance sensor, use of a Spectralon referencelaque leads to a slight overestimation of the actuald�0�� signal. Spectral variations are also observedround 762 nm for the reasons explained above andn the blue domain �400–430 nm�. These blue wave-ength variations may be due to the calibration of therios optical sensors.24 Under a clear blue sky andn the basis of the adopted configuration, results areherefore globally satisfactory although not highlyccurate.Under a covered sky, results are quite different
Fig. 2�. rspec is systematically higher than �, withignificant variations from 3.7 to approximately 6.7r �Fig. 2�a��. However, spectral variations remainow, excluding peaks around 762 nm �Fig. 2�b��. Inerms of a difference between the measured and theerived Ed�0��, diffspec is always negative, whicheans that Ed�0�� derived from Eq. �2� is systemat-
cally underestimated. diffspec ranges from �10% to
uaTitmE
Dttstpatsrabfds
cubs
b
ppt
m
3
A
TmaTpwe
tawcsbtittav
Fm09:15 to 15:45 GMT in the Plymouth Sound �southwest UK�.
Fm11:15 to 17:00 GMT in the Plymouth Sound �southwest UK�.
p to more than �60% and tends to increase �inbsolute values� from low to high wavelengths.hese results are definitely not satisfactory. They
ndicate that, under a covered sky and on the basis ofhe adopted configuration, errors up to a factor of 2ay be committed when Ed�0�� is determined fromq. �2� compared with the Rrs signal.This certainly explains the observations made byoxaran et al.14 These authors compared Rrs spec-
ra �400–1000 nm� recorded in the turbid Loire es-uarine waters during two consecutive days. Theky was blue the first day and covered homogeneoushe second day. At equivalent turbidity �i.e., sus-ended matter concentrations� they observed Rrsmplitude variations up to a factor 2 from one dayo the other, the spectra recorded under a coveredky that was systematically higher than the onesecorded under a clear sky. These variations werettributed to possible significant variations of theidirectional effects of the Rrs signal, namely, the�Q ratio.25–28 In fact, they were most probablyue to large errors committed when Ldspec was mea-ured under a covered sky.On the basis of these first experiments, under a
lear blue sky, a reference Spectralon plaque can besed to estimate Ed�0�� with an acceptable accuracy;ut this is not the case under a covered sky. At thistage, two explanations can be proposed:
�1� The Spectralon plaque that we used is not Lam-
ertian under diffuse incident light conditions. p1
�2� The measurement configuration is not appro-riate and induces shadow effects on the Spectralonlaque, especially under diffuse incident light condi-ions.
The second hypothesis can be easily verified byeans of improving the measurement configuration.
. Improved Measurement Configuration
. Objective and Method
he objective was to define an appropriate measure-ent configuration, minimizing as much as possible
ny shadow effect influencing the Spectralon plaque.he appropriate measurement configuration wouldermit us to obtain rspec ratios equal or close to � srhatever the illumination conditions �i.e., cloud cov-
r�.Possible origins of shadow effects on the plaque are
he cockpit of the boat, the person holding the plaque,nd the radiance sensor. Different configurationsere tested to minimize these influences under
loudy and covered skies. The plaque was placeduccessively on the back extremity of the boat, on theack of the boat close to the cockpit, and on the top ofhe boat �i.e., on the cockpit roof �. The person hold-ng the sensor was successively above and then belowhe plaque. Finally, the distance of the sensor abovehe plaque varied from 5 to 50 cm. The viewingngle �relative to nadir� also varied from 0° �nadiriewing� to 45°, with the plaque facing the clearest
ig. 1. Spectral variations of �a� rspec and �b� diffspec. Measure-ents were taken under a clear blue sky on 4 September 2003 from
ig. 2. Spectral variations of �a� rspec and �b� diffspec. Measure-ents were taken under a covered sky on 3 September 2003 from
art of the sky. Under stable illumination condi-
0 November 2004 � Vol. 43, No. 32 � APPLIED OPTICS 5983
ti
oupNsGcwi
s
u
m
B
Tsctwcosd11iett4�sw
s
ts
i
Oticcp
rcsti
c�
adTscto
embpSahepbad
4
Tcttd
FmN�
5
ions �Ed�0�� variations lower than 3%�, the variabil-ty of Ldspec was observed.
A second set of optical measurements were carriedut in the Plymouth Sound on 13 November 2003nder a dark heterogeneous sky �ten stations com-leted from 13:00 to 14:00 GMT� and then on 19ovember 2003 under a highly cloudy heterogeneous
ky �ten stations completed from 11:00 to 12:00MT�. On both dates, the swell resulted in signifi-
ant movements of the boat. The same procedureas followed concerning the acquisition and process-
ng of the recorded Ed�0�� and Ldspec spectra:
• acquisition of a minimum of five consecutivepectra,
• calculation of the Ed�0�� and Ldspec mean val-es,• quantification of the variations around theean value, and• calculation of the rspec and diffspec parameters.
. Results
he different configurations showed that significanthadow effects on the plaque were induced by theockpit when measurement took place on the back ofhe boat. These effects were obviously amplifiedith the proximity of the cockpit, and Ldspec de-
reased by up to 50%. Therefore placing the plaquen the cockpit roof appeared to be the appropriateolution. Significant shadow effects were also in-uced by the sensor placed at a distance lower than0 cm above the plaque �Ldspec decreasing by up to5%�. These shadow effects became almost insignif-cant when a distance higher than 20 cm was consid-red �Ldspec variations lower than 3% without regardo sign�. Considering a distance higher than 20 cm,he influence of the viewing angle �ranging from 0° to5° relative to nadir� appeared almost insignificantLdspec variations lower than 2% without regard toign�. For practical reasons, a nadir-viewing angleas adopted.The best results �lowest diffspec values� were ob-
erved with the following configuration:
• the Spectralon plaque placed horizontally onhe top of the boat �i.e., on the roof � to remove thehadow effects induced by the cockpit and
• the person holding the radiance sensor remain-ng below the plaque.
nly the radiance sensor was obviously placed abovehe plaque, at a distance of 30 cm, and may havenduced shadow effects. A distance higher than 30m would have been more appropriate but was diffi-ult to assess because of the limited dimensions of thelaque and the movements of the boat.Results are globally satisfactory with this configu-
ation. Regardless of the sky conditions, rspec valueslose to � are obtained, ranging between 2.7 and 3.6r from 400 to 900 nm �Fig. 3�a��. However, a spec-ral dependency is also observed, resulting in a slight
ncrease of rspec with increasing wavelength. Spe- o984 APPLIED OPTICS � Vol. 43, No. 32 � 10 November 2004
ific spectral variations are still observed in the blue400–450 nm� and notably around 762 nm.
The difference observed between the measurednd the derived Ed�0�� is also greatly minimized.iffspec ranges between �10% and �15% �Fig. 3�b��.ypically, diffspec is positive in the blue �leading to alight overestimation of Ed�0�� from 400 to 450 nm�,lose to zero from 500 to 700 nm, and then negative inhe near infrared �leading to a slight underestimationf Ed�0���.Consequently, it can be concluded that shadow
ffects were at the origin of the erroneous measure-ents carried out in September on the back of the
oat; the Lambertian properties of the Spectralonlaque are not dependent on the sky conditions.hadow effects, induced by the cockpit of the boatnd by the person holding the sensor, resulted in aigh inaccuracy of measurement. However, theseffects can be greatly minimized. Adopting an ap-ropriate measurement configuration, Ed�0�� cane accurately estimated with a radiance sensor andSpectralon plaque, whatever the illumination con-itions.
. Conclusions and Recommendations
he objective of this study was to know how Ed�0��an be accurately estimated from measurements ofhe downwelling radiance signal reflected by a Spec-ralon plaque �Ldspec�. Basic experiments were con-ucted, measuring Ed�0�� and Ldspec simultaneously
ig. 3. Spectral variations of �a� rspec and �b� diffspec. Measure-ents were taken under various cloudy and covered skies on 13ovember 2003 �13:00 to 14:00 GMT� and on 19 November 2003
11:00 to 12:00 GMT� in the Plymouth Sound �southwest UK�.
n a boat under different illumination conditions
�s
tcmLE
ecctcESasttLimpropSaaLctssbibat8tmtmtfmfaR
isteebt
C
5sps
R
1
1
1
1
1
1
1
clear, cloudy, covered skies� and with different mea-urement configurations.Under a clear sky, it was observed that presenting
he plaque horizontally in front of the Sun is suffi-ient to obtain a correct Ldspec signal. This signal,ultiplied by � sr �i.e., assuming that the plaque is aambertian reflector�, provides a good estimation ofd�0�� ��10% mean accuracy�.Under a diffuse incident light �i.e., cloudy or cov-
red sky�, obtaining a correct Ldspec signal is moreomplicated. In fact, shadow effects induced by theockpit of the boat, by the person or system holdinghe radiance sensor, and by the radiance sensor itselfan dramatically modify the measured Ldspec signal.rrors up to 100% can easily be committed when apectralon plaque is used on the deck of a boat undercovered sky. These errors related to Ldspec mea-
urements can induce errors up to 100% on the de-ermined Rrs signal. This certainly explains most ofhe Rrs variations observed by Doxaran et al.14 in theoire estuary with changing sky conditions. Thus,
n these conditions, special care must be taken toinimize the shadow effects. The Spectralon
laque must be placed on the top of the boat �on theoof �, above any object or holding system. It is rec-mmended that the radiance sensor, viewing thelaque, be placed as far as possible from the plaque.uch recommendations may be difficult to respect onboat with a rough sea surface, but are necessary to
void large errors. Numerous measurements ofdspec and Ed�0�� were carried out respecting thisonfiguration with a minimum plaque–sensor dis-ance of 30 cm under various cloudy and coveredkies. The measured Ldspec signal, multiplied by �r, provides a good estimation of Ed�0�� �accuracyetter than �15%�. This accuracy may logically bemproved by use of a plaque of large dimensions andy an increase in the distance between the plaquend the radiance sensor. It can be compared withhe uncertainty in Lw determinations: typically–18% in case 1 waters,6,8,9 considering different
echniques �above- and in-water optical measure-ents� and processing and correction methods. In
urbid coastal �case 2� waters where use of in-watereasurement techniques is limited and above-water
echniques are influenced by surface reflection ef-ects, imperfect corrections for the Sun and sky glintay result in significantly higher errors.6,7 There-
ore the uncertainty in Ed�0�� must be taken intoccount when the percentage of error associated withrs measurements is assessed.Regardless of the illumination conditions and the
nduced shadow effects, the ratio between the mea-ured Ldspec and Ed�0�� signals showed slight spec-ral variations from 400 to 900 nm. Consequently,rrors committed when a Spectralon plaque is used tostimate the Ed�0�� signal, then the Rrs signal, cane significantly minimized by use of a ratio betweenwo wavelengths.
This research was supported by a European Marie
urie fellowship �Framework 5, EVK3-CT-2002-1
0012� and a Natural Environment Research Councilmall grant �NER�B�S�2002�00555; S. Lavender,rincipal investigator�. The authors also thank thekippers of the Catfish boat.
eferences1. C. D. Mobley, Light and Water: Radiative Transfer in Natu-
ral Waters �Academic, San Diego, Calif., 1994�.2. J. L. Mueller, A. Morel, R. Frouin, C. Davis, R. Arnone, K. L.
Carder, Z. P. Lee, G. H. S. Hooker, C. D. Mobley, S. McLean, B.Holben, M. Miller, C. Pietras, K. D. Knobelpiesse, G. S. Far-gion, J. Porter, and K. Voss, Ocean Optics Protocols for SatelliteOcean Color Sensor Validation, Rev. 4, SeaWiFS Tech. Rep.Ser., NASA Tech. Memo. 2003-21621, Vol. 3 �NASA GoddardSpace Flight Center, Greenbelt, Md., 2003�.
3. A. Morel, “In water and remote measurements of ocean color,”Boundary-Layer Meteorol. 18, 177–201 �1980�.
4. C. D. Mobley, “Estimation of the remote-sensing reflectancefrom above-surface measurements,” Appl. Opt. 38, 7442–7455�1999�.
5. B. Fougnie, R. Frouin, P. Lecomte, and P. Y. Deschamps, “Re-duction of skylight reflection effects in the above-water mea-surements of diffuse marine reflectance,” Appl. Opt. 38, 3844–3856 �1999�.
6. D. A. Toole, D. A. Siegel, D. W. Menzies, M. J. Neumann, andR. C. Smith, “Remote-sensing reflectance determinations inthe coastal ocean environment: impact of instrumental char-acteristics and environmental variability,” Appl. Opt. 39, 456–468 �2000�.
7. R. W. Gould, R. A. Arnone, and M. Sydor, “Absorption, scat-tering, and remote sensing reflectance relationships in coastalwaters: testing a new inversion algorithm,” J. Coastal Res.17, 328–341 �2001�.
8. S. B. Hooker, G. Lazin, G. Zibordi, and S. McLean, “An eval-uation of above- and in-water methods for determining water-leaving radiances,” J. Atmos. Oceanic Technol. 19, 486–515�2002�.
9. S. B. Hooker and A. Morel, “Platform and environmental ef-fects on above-water determination of water-leaving radi-ances,” J. Atmos. Oceanic Technol. 20, 187–205 �2003�.
0. D. A. Haner, B. T. McCuckin, R. T. Menzies, C. J. Bruegge, andV. Duval, “Directional hemispherical reflectance for Spec-tralon by integration of its bidirectional reflectance,” Appl.Opt. 37, 3996–3999 �1998�.
1. D. A. Haner, B. T. McCuckin, and C. J. Bruegge, “Polarizationcharacteristics of Spectralon illuminated by coherent light,”Appl. Opt. 38, 6350–6356 �1999�.
2. D. Doxaran, J. M. Froidefond, S. J. Lavender, and P. Castaing,“Spectral signature of highly turbid waters. Application withSPOT data to quantify suspended particulate matter concen-trations,” Remote Sens. Environ. 81, 149–161 �2002�.
3. D. Doxaran, J. M. Froidefond, and P. Castaing, “A reflectanceband ratio used to estimate suspended matter concentrationsin sediment-dominated coastal waters,” Int. J. Remote Sens.23, 5079–5085 �2002�.
4. D. Doxaran, J. M. Froidefond, and P. Castaing, “Remote-sensing reflectance of turbid sediment-dominated waters.Reduction of sediment type variations and changing illumina-tion conditions effects by use of reflectance ratios,” Appl. Opt.42, 2623–2634 �2003�.
5. E. J. D’Sa and R. L. Miller, “Bio-optical properties in watersinfluenced by the Mississippi River during low flow condi-tions,” Remote Sens. Environ. 84, 538–549 �2003�.
6. J. M. Froidefond, L. Gardel, D. Guiral, M. Parra, and J. F.Ternon, “Spectral remote sensing reflectances of coastal wa-ters in French Guiana under the Amazon influence,” Remote
Sens. Environ. 80, 225–232 �2002�.0 November 2004 � Vol. 43, No. 32 � APPLIED OPTICS 5985
1
1
1
2
2
2
2
2
2
2
2
2
5
7. R. W. Gould and R. A. Arnone, “Remote sensing estimates ofinherent optical properties in a coastal environment,” RemoteSens. Environ. 61, 290–301 �1997�.
8. L. Han and D. C. Runquist, “The impact of a wind-roughenedwater surface on remote measurements of turbidity,” Int. J.Remote Sens. 19, 195–201 �1998�.
9. V. Lafon, J. M. Froidefond, F. Lahet, and P. Castaing, “SPOTshallow water bathymetry of a moderately turbid tidal inletbased on field measurements,” Remote Sens. Environ. 81,136–148 �2002�.
0. Z. P. Lee, K. L. Carder, T. G. Peacock, C. O. Davis, and J. L.Mueller, “Method to derive ocean absorption coefficientsfrom remote-sensing reflectance,” Appl. Opt. 35, 453–462�1996�.
1. Z. P. Lee, K. L. Carder, C. D. Mobley, R. G. Steward, and J. S.Patch, “Hyperspectral remote sensing for shallow waters. 2.Deriving bottom depths and waters properties by optimiza-tion,” Appl. Opt. 38, 3831–3843 �1999�.
2. M. Sydor and R. A. Arnone, “Effect of suspended particulate
986 APPLIED OPTICS � Vol. 43, No. 32 � 10 November 2004
and dissolved organic matter on remote sensing of coastal andriverine waters,” Appl. Opt. 36, 6905–6912 �1997�.
3. J. A. Curcio, L. F. Drummeter, and G. L. Knestrick, “An atlasof the absorption spectrum of the lower atmosphere from 5400Å to 8520 Å,” Appl. Opt. 3, 1401–1409 �1964�.
4. R. Heuermann, Managing Direction, Trios GmbH, Werftweg15, D-26135 Oldenburg, Germany �personal communication,2003�.
5. A. Morel and B. Gentili, “Diffuse reflectance of oceanic waters:its dependance on Sun angle as influenced by the molecularscattering contribution,” Appl. Opt. 30, 4427–4438 �1991�.
6. A. Morel and B. Gentili, “Diffuse reflectance of oceanic waters.II. Bidirectional aspects,” Appl. Opt. 32, 6864–6879 �1993�.
7. A. Morel and B. Gentili, “Diffuse reflectance of oceanic waters.III. Implication of bidirectionality for the remote-sensingproblem,” Appl. Opt. 35, 4850–4861 �1996�.
8. H. Loisel and A. Morel, “Non-isotropy of the upward radiancefield in typical coastal �Case 2� waters,” Int. J. Remote Sens.
22, 275–295 �2001�.
