raman scattering in ocean optics: quantitative assessment of internal radiant emission

Raman scattering in ocean optics: quantitativeassessment of internal radiant emission

Robert Hans Stavn and Alan Dean Weidemann

Raman-scattering activity in clear ocean waters is further documented from Monte Carlo simulations andoptical data that are collected in the Sargasso Sea. A method is proposed, based on the anomalousabsorption coefficient for a nonconservative irradiance field, to assess the percentile composition ofinternal radiant emission for the irradiance field at any depth.

Key words: Raman scattering, ocean optics, anomalous ocean optical properties, photon budgets.

Introduction

Internal radiant emission in clear oceans has recentlybeen implicated as a possible cause of anomalousocean optical properties that are determined from theirradiance field at longer wavelengths."~ Variousreports indicate that at longer visible wavelengths theabsorption coefficient a calculated from the irradi-ance field is less than the accepted value for molecularwater. 9 In addition the diffuse attenuation coeffi-cient K (both downwelling and upwelling) in certaincases is smaller than the coefficient that is acceptedfor molecular water.'," 2 Furthermore recent at-tempts to parameterize the penetration of clear oceanwaters by sunlight by using diffuse attenuation coeffi-cients derived from the surface layers of the clearestocean waters 3 have not been successful. The resultsof these parameterizations indicate more photons atdepth than would be expected if the solar photonswere being absorbed by molecular water.4" 5 Thus inclear ocean waters we have evidence of internalradiant emission, and water Raman scattering hasbeen invoked to explain many of the effects notedabove. 5 Other sources of internal radiant emissionare, to be sure, fluorescence from dissolved organicmatter" 2 0 and suspended living algae with specificpigment emission peaks.2. 23 However, the water mol-ecule is a fundamental source of internal emission in

R. Stavn is with the Department of Biology, University of NorthCarolina at Greensboro, Greensboro, North Carolina 27412. A.Wiedemann is with the National Oceanographic and AtmosphericResearch Laboratory, Code 331, Stennis Space Center, Mississippi39529.

Received 24 August 1990.0003-6935/92/091294-10$05.00/0.0 1992 Optical Society of America.

clear oceans, and it must be accounted for in additionto other added components in the marine hydrosol.The other components become more significant as weprogress from clear ocean waters to coastal andinland waters. Even in coastal waters, however, Carderet al.2 4' 25 have reported what appear to be waterRaman emission effects at 650 nm.

We therefore propose continuing with the resultsthat are generated from the National Oceanographicand Atmospheric Research Laboratory (NOARL) op-tical model, a Monte Carlo solution of the radiativetransfer equation, which is applied to clear oceanwaters. These simulations of the light field in clearocean water will then be compared with the field datacollected by NOARL on the cruise of the R/V En-deavor (EN-166) as part of the Biowatt-NOARLcruise to the Sargasso Sea. We then describe a methodto assess quantitatively the amount of internallyemitted radiation at any point in the marine hydrosol.

Theory

The proposed explanation of the anomalous opticalproperties reported in the literature is the presence ofinternal radiant emission. In clear ocean waters witha minimum of dissolved and suspended material theprimary source of internal radiant emission is waterRaman scattering. The effect of this emission on theinherent absorption coefficient calculated from theirradiance field is simple and quantitative. In thisdiscussion we assume that the irradiance field is afunction of wavelength without an explicit statementin the equations. The calculation of the absorptioncoefficient from the ambient irradiance field is basedon Gershun's equation,26 and the potential of internalradiant emission in producing an anomalous determi-nation of this coefficient was pointed out by H0jer-slev.27 The three-parameter model,28 which is an exact

1294 APPLIED OPTICS / Vol. 31, No. 9 / 20 March 1992

solution to the radiative transfer equation, allows thecalculation of the absorption coefficient from theambient irradiance field even in the presence ofmultiple scattering. This is the calculation that weuse to determine the absorption coefficient from fieldirradiance data and as a check on the output of theMonte Carlo simulations. This light field formulationrequires the scalar irradiance Eo, the downwellingirradiance E, and the upwelling irradiance E.. Fromthis we obtain the net downwelling irradiance ordownwelling vector irradiance E, = Ed - E,, and alsodetermine the average cosine of the mean photonpath of the light field.29 The average cosine parameterproves to be important in assessing the nature of thelight field, i.e., whether the light field seems to bedominated by the penetration of solar photons only orare there significant effects caused by internal radiantemission? The average cosine is determined by thesimple ratio

interval assuming that a is constant over the interval,as is common practice. We calculate the absorptioncoefficient in Eq. (5) where the integral in the denom-inator accounts for changes in the photon path (asmeasured by the average cosine) that are caused bymultiple scattering:

a(z) = -In [E,(z2) - In E,(z,)]

(5)

fI, z

For the quantitative assessment of the effects ofinternally emitted radiation we take advantage of thefact that this calculation requires a conservative lightfield.

When the irradiance field is not conservative, i.e.,internal radiant emission (water Raman scattering,fluorescence, or bioluminescence) is taking place atthe wavelength of interest, we must add new terms toEq. (3):

(1)EII=--

This light field parameter" (also termed by Preisen-dorfer3' an apparent optical parameter) varies inabsolute value from 0.0 to 1.0, a low value indicatingnearly equal upwelling and downwelling radiantfluxes, while a value approaching 1.0 indicates thatnearly all the radiant flux is propagated in a verticallydownward direction. Therefore the average cosine is asensitive indicator of the radiance distribution andstructure of the submarine light field.

The fundamental Gershun equation that is usedfor determining the absorption coefficient from theambient irradiance field is defined as

V E(z) = -a(z)EoW, (2)

where E is the vector irradiance of the light field and zis the geometric depth in meters. Preisendorfer3'pointed out that, when there is no horizontal diver-gence of the light field, the left-hand side of Eq. (2)becomes a function of only the irradiant flux of thevertical axis (z axis). This vertical component of thevector irradiance is the downwelling vector irradi-ance, also termed the net downwelling irradiancewith reference to its being the difference between thedownwelling irradiance and the upwelling irradiance,and

dE(z) - -(z)E(z). (3)dz

Solving Eq. (3) by substituting Eq. (1) into it, weobtain a general radiative transfer equation 2 that canbe recast in several forms:

Eo(z) = E(O) 11(o) exp - a(z) dzJ. (4)

From this equation comes a general relationship forcalculating the absorption coefficient for a depth

-~z) = - 1 d[Ez(z) + E.*(z)]a z) [E,(z) + Eo*(z)] dz (6)

where E*(z) is the net irradiant flux caused by theinternal radiant emission, E*(z) is the scalar irradi-ance caused by internal radiant emission, and d(z) isthe absorption coefficient at depth z calculated fromthe nonconservative irradiance field. In clear oceanwater we assume that the major source of internalradiant emission is water Raman scattering. With analgebraic rearrangement of Eq. (6) and timely substi-tution of Eq. (3),

[Eo(z) + Eo*(z)]d(z) + dE *(z) = dE.(z)dz =-dz

Eo(z)a(z) = [Eo(z) + E*(z)]a(z) +

ha(z) = a(z) - (z),

where a(z) is the inherent absorption coefficient of thehydrosol measured spectrophotometrically and

AaWz = a(z) - EO(z) _az) a) E,(z) + Eo*(z)] a(z)

+ E.*z' (7)[EO(z) + Eo*(z)] dzj

The Ez*(z) term, the difference between the upwellingand downwelling fluxes from internal emission, isrelatively small because the two fluxes are nearlyequal. If there were no contribution of flux fromdepths above and below depth z, the E*(z) termwould be exactly zero. The relative effects of multiplescattering and differential absorption, which aregreater at shorter wavelengths, will alter this relation-ship somewhat. The derivative of E,*(z) with depthwill be an even smaller number. Thus we assume thatthe last term inside the braces on the right-hand side

20 March 1992 / Vol. 31, No. 9 / APPLIED OPTICS 1295

of Eq. (7) is approximately zero. Then

Aa(z) E0 (Z)

a(z) 1[E(z) + Eo*(z)]

Aa(z) Eo*(z)

a(z) [E0 (z) + Eo*(z)]

The output of the Monte Carlo simulations of bothdirect solar and Raman photon fluxes permits us tocalculate Aa by using both the exact relation from Eq.(7) and the approximate relation from Eq. (8) as acheck on this approximation. The deviation betweenthe two estimates varies from 0.01 to 0.1%. Ourderivation involves an absorption coefficient a(z),which is determined from the ambient nonconserva-tive light field, and the inherent absorption coefficientof the hydrosol a(z), which is determined spectropho-tometrically. The ratio of the deviation of the noncon-servative absorption coefficient [a(z)] from the inher-ent absorption coefficient [a(z)] to the inherentabsorption coefficient is a quantitative estimate of thepercentile composition of the emission photons in thelight field. Therefore we propose that this ratio beused in the decomposition of the ambient light fieldinto directly penetrating solar photons and internallyemitted photons. This determination includes emis-sion from all sources, such as fluorescence. This ratio,which is illustrated on the left-hand sides of approxi-mations (8), may be called the relative absorptionanomaly.

Methods

The NOARL optical model is a Monte Carlo simula-tion of the radiative transfer equation that describesthe penetration of solar photons into ocean watersand accounts for the transpectral water Raman scat-tering. The Monte Carlo simulation is derived fromthe methods of Plass and Kattawar,33 Gordon andBrown,34 and Kirk.35 The required inputs for simulat-ing the penetration and fate of solar photons in oceanwaters are the absorption coefficient, the total-scattering coefficient, and the volume-scattering func-tion. The volume-scattering functions of each scatter-ing component (molecular water, quartzlikeparticulates, algae, organic detritus) are treated sepa-rately rather than being used as an average volume-scattering function. The NOARL Blue Water modelused in this simulation contains the absorption coeffi-cients for molecular water of up to 570 nm reportedby Smith and Baker'3 ; the coefficients reported byTam and Patel3" are used for coefficients at wave-lengths that are > 570 nm. Recent investigations ofthe backscattered upwelling radiance from oceanwaters indicate that this is the most efficient combina-tion of absorption coefficients.37 The total-scatteringcoefficient and the volume-scattering function of pureseawater are incorporated from Morel,38 and thetotal-scattering coefficient and volume-scatteringfunction for quartzlike material are incorporatedfrom Kullenberg 9 and Gordon et al.40 The Raman-scattering coefficient for molecular water was deter-

mined from the molecular Raman-scattering crosssection reported by Chang and Young.4 The Raman-scattering coefficient was then truncated to generateonly emissions in the 10-nm wave band of the emis-sion wavelength studied, as described earlier.2 Thesimulations were performed at 520, 550, and 589 nmto correspond with the wavelengths that are availablefor study on the NOARL instrument called theparticle-optical sampling system.

For a simulation run photons are transportedthrough a clear marine aerosol for both a Ramansource wavelength and a Raman emission wave-length, the determination being made on the basis ofa 3357-cm-' frequency shift from the source wave-length to the emission wavelength. The relative num-bers of photons propagated at the two wavelengthsare determined from the extraterrestrial solar irradi-ance spectrum reported by Iqbal.42 The atmosphericmodel of Iqbal is then used to propagate solar andskylight photons through the marine aerosol to sealevel. The angle of entry of a skylight photon isdetermined from the skylight radiance distribution ofHarrison and Coombes,43 while the entry angle of asolar photon is at 11° from the zenith. The sea surfaceis flat. At the air-water interface the photon eitherenters the ocean and is refracted or is reflected backinto the air based on the laws of Fresnel and Snell.These laws are used to determine the probabilities ofthese events that are chosen by a random numbergenerator. On entry into the ocean a photon can beabsorbed, scattered, or transpectrally scattered. Theprobabilities of these events are determined from therespective coefficients of the model, and the individ-ual events are chosen by a random number generator.When a photon is finally absorbed, a new one isgenerated and propagated through the system until2.5 x 106 photons have been propagated. Duplicateruns are made for the wavelength that is studied.

The downwelling and upwelling irradiances andscalar irradiances collected during the Endeavor cruisewere measured with a Biospherical Instruments, Inc.(San Diego, Calif.) Model MER 1048 attached to theparticle-optical sampling system. This instrumentwas equipped with thirteen downwelling irradiancechannels, eight upwelling irradiance channels, fourdownwelling hemispherical scalar irradiance chan-nels, and four upwelling hemispherical scalar irradi-ance channels. The instrument package was attachedto a rosette to permit the collection of water samplesat selected depths, and the sensors for transmittance,conductivity, temperature, pressure, and fluores-cence were also attached to the rosette. The darkcurrent for the irradiance sensors was determinedperiodically by lowering the instrument until randomfluctuations in output were observed (usually below200 m). The downcast rate vaired from 0.2 to 0.8 m/s.

Ship motion and wave focusing resulted in largefluctuations for downwelling irradiances and down-welling hemispherical scalar irradiances that re-quired some filtering for the estimation of the irradi-ances used in the optical calculations. After


Table . Optical Coefficients Determined for the R/V Endeavor Cruise

Yellow MolecularDepth Substance Particulates Water Total

(in) a, (520) m-1 a, (520) m-1 a. (520) m-1 a(520) m-l

Station 627 < 0.00la ± 0.0006 0.003 ± 0.0006 0.0477 0.0507 ± 0.0009

80 0.004 ± 0.0022 0.009 ± 0.0018 0.0477 0.0607 ± 0.0028

Station 8

34 0.006 ± 0.003 0.003 ± 0.0006 0.0477 0.0567 ± 0.003

100 0.007 ± 0.004 0.008 ± 0.0016 0.0477 0.0627 + 0.0043

aError calculated by assuming that the absorption coefficient is 0.001 m-l.

normalizing to surface irradiance, exponential regres-sion was performed on 2-m bins with a 1-m incrementfor Station 6, while Station 8 required this procedureplus two passes of a smoothing filter.

The absorption coefficients determined directly forwater samples collected by the particle-optical sam-pling system were obtained by the following methods:

Water absorption: The absorption coefficient ofpure seawater at 520 nm was determined from thetables of Smith and Baker.'3

Gelbstoff absorption: We define dissolved sub-stances as those that pass through a 0.22-,um Nucleo-pore filter.44 The absorption was measured on thefiltrate in 10-cm cuvettes from 320 to 420 nm. Thespectra showed an exponential rate of decline withincreasing wavelength, the rate constants varyingfrom 0.014 to 0.024 with a mean of 0.017. These ratesare in line with those reported by other investiga-tors.73045 Absorption at wavelengths that are > 420nm was determined by extrapolation of the exponen-tial curve. The variation in the rate constant was usedto estimate the error in the absorption coefficientsreported in Table I.

Particulate absorption: The contribution by partic-ulates was determined spectrophotometrically follow-ing the procedures of Mitchell and Kiefer.22 Two tofour liters of water were filtered through GF/C(Whatman) filters, and the spectral absorption wasmeasured from 400 to 750 nm on a Kontron dual-

beam (model Uvikon 860) spectrophotometer with ablank saturated GF/C filter. Absorption between 740and 750 nm was used to correct the spectrum forbackscattered light. The absorption coefficient wasdetermined by employing the equation of Mitchelland Kiefer.2 2 This procedure may underestimate par-ticulate absorption between the size classes of 0.22jim (dissolved substance limit) and 0.8 m (GF/Cnominal retention). Recently Mitchell46 reported onthe precision and accuracy of this absorption measure-ment technique. The use of GF/C filters may result inan underestimate of particulate absorption by asmuch as 25%. Extensive experimental tests of theMitchell/Kiefer algorithm indicated an overall preci-sion of ± 10%. The variation between filter lots canadd another error of ± 10%. We therefore estimate anerror of ±20% for the particulate absorptions re-ported in Table I. It should also be kept in mind thatthe values in Table I are likely to be underestimates ofthe particulate absorption. The total laboratory ab-sorption coefficient was calculated as the sum of thecontributions from water, gelbstoff, and particulates.

esults

The simulation from the NOARL Blue Water modelyielded the following results for clear ocean waterirradiated by a, clear sky composed of a marineaerosol; the coefficients used for the simulation arereported in Table II. The average cosines for theseparate Raman and solar photon streams are plotted

Table 11 Optical Coefficients Used in Monte Carlo Simulation

Coefficients Emission(m-2 ) Source Wavelengths (nm) Wavelength (nm)

433 443 453 520

Absorption 0.0145 0.0145 0.0147 0.0477

Scattering 0.0402 0.0389 0.0377 0.0314Raman scattering 0.0000967 0.000400 0.000060

454 464 474 550

Absorption 0.015 0.0156 0.0165 0.0638


482 492 502 589

Absorption 0.0180 0.0207 0.0275 0.1250



Average Cosine

.s 600

8' 0

90 589-m

1 00

11 0

120

Fig. 1. Average cosine of the separate photon streams: directlytransmitted solar photons and Raman emission photons at variouswavelengths. The standard errors of the estimates are indicated byeither vertical bars or the width of the dot.

in Fig. 1. The average cosine for the solar photonsvaries from 0.94 to 0.98 at the surface to 0.86-0.91 atthe maximum depth for all wavelengths considered.By contrast the Raman photons have an averagecosine of (-0.2) at the surface, 0.0 in the region of10-15-m depth, and 0.09-0.14 at the greatest depths.In Fig. 2 the average cosine of the combined streamsis plotted for each wavelength. In Figs. 3-5 areplotted the nonconservative absorption coefficients ofthe Raman-influenced irradiance fields at 520, 550,and 589 nm, respectively. The nonconservative absorp-tion coefficients deviate increasingly with depth fromthe inherent absorption coefficients and then deviatemarkedly from the inherent absorption coefficients at

Average Cosine0.0 0.2 0.4 0.6 0.8 1.0

0*

10-

20

30

40'

50,

60'

70 70

.~80-

100

1 10-

120.

130 50n 2 nm

140-

150

160-

Fig. 2. Average cosine of the combined solar and Raman photonstreams as would be measured in the ocean. The standard errorsare indicated by either vertical bars or the width of the dot.

0.00 0.02 0.04

E

Q

10

20

30

40

5060-

70 -

00-

90

100-

110-

120-

130-

140-

150-

a(520) m-1

0.06 0.08 0.10 0.12 0.14

Fig. 3. Absorption coefficients at 520 nm. The inherent absorp-tion coefficient for water molecules is indicated by a straight line,and the nonconservative absorption coefficient is indicated by thecurve and data points. The standard errors are indicated by eithervertical bars or the width of the dot.

140, 100, and 60 m, respectively. The relative absorp-tion anomalies that are determined from the modeloutput are plotted against the actual ratios of Ramanphotons to the total photons of the model output(Figs. 6-8). There is a 1-1 correspondence betweenthe two ratios for all the wavelengths that wereinvestigated.

Figures 9-12 show plots of the data that werecollected during the Biowatt-NOARL cruise of Aug.1987 to the Sargasso Sea. The stations were located inthe region of 340N, 69-70 0W. The field results provide

a(550) m'0.02 0.04 0.06 0.08 0.10 0.12 0.14

0-

10

20

30

40

50

60

80

100 (550)

110

120

130

aa(550)140 -0.0638

150

160~

Fig. 4. Absorption coefficients at 550 nm. The inherent absorp-tion coefficient for the water molecules is indicated by a straightline, and the nonconservative absorption coefficient is indicated bya curve and data points. The standard errors are indicated by eithervertical bars or the width of the dot.


i (520)

a,,(520)- 0.0477

U u TE 'nl | I 4 S , . | . . . . . .A . . . .

rnX Ho,4 ,^

a(589) m-1

70

80-

0

100 a. (589)-0.125

110.

120'

130

140

150

160

Fig. 5. Absorption coefficients at 589 nm. The inherent absorp-tion coefficient for water molecules is indicated by a straight line,and the nonconservative absorption coefficient is indicated by acurve and data points. Standard errors are indicated by eithervertical bars or the width of the dot.

further evidence of water Raman emission at thethree wavelengths studied in the green-yellow regionof the visible spectrum. The stations chosen forfurther analysis were the ones in which there weredata on both the ambient light field and on theinherent absorption coefficients of the water samplescollected at the stations. A complete suite of irradi-ance measurements was possible at 520 nm to allowfor the determination of the average cosine parame-ter and the calculation of the absorption coefficientfrom the irradiance field.

Upwelling and downwelling irradiances were mea-sured at 550 nm, and only the downwelling irradiancewas measured at 589 nm. The average cosine for thelight field at 520 nm from Station 4 is plotted in Fig. 9along with the average cosine generated by theNOARL Monte Carlo simulation. The values of theaverage cosine near the surface indicate probable

1.00.8

0.8

0.7-

0.6-

E(520) .s -

[E 0 (520) + Eo(520)] 0.4 -

0.3-

0.2-

0.1 ,

0.00.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Aa(520)a(520)

Fig. 6. Fraction of Raman emission photons in total scalarirradiance at 520 nm compared to the relative absorption anomalyfrom the NOARL Blue Water model output.

0.6-E o(550) _ "

[E (5 5 0) + E (5 5 0)] 0.4-

0.0 0.1 0.2 0.3 0.4 0.0 0.6 0.7 0.8 0.9 1.0

Aa(550)a(550)

Fig. 7. Fraction of Raman emission photons in total scalarirradiance at 550 nm compared to the relative absorption anomalyfrom the NOARL Blue Water model output.

wave-focusing effects, which increases the mean pho-ton path and thus decreases the average cosine. Atgreater depths agreement is within 8% down to 80m where a sharp decrease in the average cosine at75-80 m from the prediction based on water Ramanscattering may indicate particulate fluorescent activ-ity. In Fig. 10 is plotted the irradiance ratio (EU/Ed)for 550 nm at Station 6. The trend of the ratio followsthat predicted by the NOARL Monte Carlo simula-tion. The logarithm of the downwelling irradiance Edat 589 nm for Station 6 is plotted in Fig. 11 along withthe predicted penetration of solar photons at thiswavelength and the predicted solar plus Raman pho-tons at this wavelength. At a 40-m depth the photonsstart to deviate from the level predicted only for solarphotons; i.e., there are more photons present belowthese depths than can be accounted for by the trans-mission of solar photons through a medium consist-ing of only molecular water. The ocean containsdissolved and suspended materials in addition tomolecular water that would also absorb photons.

The nonconservative absorption coefficients andthe spectrophotometrically determined absorption co-efficients for Endeavor Stations 6 and 8 are plotted inFig. 12. The individual components of the spectropho-

1.0X

0.9-

0.8 -

0.7

0.6

E(589) 0.5

[E0 (589) + E;(58 9)] 04 -

0.3-

0.2

0.1

0.0- 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Aa(589)a(589)

Fig. 8. Fraction of Raman emission photons in the total scalarirradiance at 589 nm compared to the relative absorption anomalyfrom NOARL Blue Water model output.


0

10

20

30

0.:E

C)0

40

50

60

70

80

90

100

PL(520)

0.65 0.70 0.75 0.80 0.85 0.90 0.95

Fig. 9. Average cosine at 520 nm for Station 4, Endeavor cruiseEN-166 and from the output of the NOARL Blue Water model.

tometrically determined absorption coefficients atthose stations are in Table I. The tendency for theabsorption coefficient anomaly to increase with depthis evident at the two stations; there is also thepossibility that the layering of absorbent or fluores-cent material at Station 6 is superimposed on thedepth trend.

Oceanic Radiant Flux Field: QuantitativeCharacterization

We have demonstrated that various troublesome andanomalous phenomena associated with measured clearocean irradiance fields and the optical parameterscalculated from them can be attributed to waterRaman scattering. The average cosine that was re-corded from the data of the Biowatt-NOARL cruisecollected at 520 nm indicates the presence of water

Irradiance Ratio R

0.01 0.02 0.03 0.04 0.05

40

0.

IC)013

60

80

100

Fig. 10. Irradiance ratio at 550 nm for Station 6, Endeavor cruiseEN-166, and from the output of NOARL Blue Water model.

0

20

40

s0.

60

80

100

In of Ed(589)

-8 -6 .4 -2 0 2 4 6

Fig. 11. Natural logarithm of the downwelling irradiance at 589nm. The solid curves with a low variance represent the output fromthe NOARL Blue Water model normalized to the mean irradiancemeasured at Station 6 at the surface. The solid curve with arelatively high variance is a natural logarithm of measured down-welling irradiance in I.LW/cm/nm.

Raman scattering when compared with the results ofthe NOARL Blue Water model simulation (Fig. 9).This was reported earlier,2 and we confirm thoseresults here with the data from a different station.The irradiance ratio that is plotted for 550 nm in Fig.10 shows an increase in this ratio over that expected

- for clear ocean waters, as is predicted from theNOARL Monte Carlo simulation.

The effects of water Raman scattering on down-welling irradiance, which is the most common irradi-

Endeavor Station 8 Endeavor Station 6a(520) m ' a(520) m

1

0.05 0.1 0.15 0.2 0 0.05 0.1 8.15

Fig. 12. Absorption coefficients for two stations. The straightlines represent inherent absorption coefficients of hydrosol col-lected at depths that are indicated by black dots. The experimentalerrors of inherent absorption coefficients are indicated by errorbars. The curves represent the nonconservative absorption coeffi-cient calculated from the ambient irradiance field.


4) - /

l _ , .............. .................

............. ...........................

.... .......... .. ,........................ ...... i

20

ance measurement, are subtle at shorter wavelengthsand in the predominantly green region of the irradi-ance spectrum. The effects begin to show up stronglyin the downwelling irradiance as we progress towardthe longer yellow wavelengths (589 nm) as shown inFig. 11 and confirmed by Marshall and Smith.4 Thusthe effects of water Raman scattering tended to gounnoticed in previous field investigations, especiallythose that were limited to downwelling irradiancemeasurements. The effects that were noticed wereoften attributed to light leakage or spectral crosstalk,4"0" 4 a relatively wide filter bandwidth,4 or similarinstrumentation effects. The average cosine parame-ter accounts for the possibility of light leakage andsimilar problems, because if the anomalies in theirradiance readings and average cosines etc. are dueto instrumentation problems, the average cosinewould be identical to that of the source wavelengthfor leakage or spectral cross talk. The average cosinerecorded at depth from the Sargasso Sea at 520 nm isless than that of the possible source wavelengths. Wedo not have data on the average cosine at otherwavelengths, but we do have the predictions for thepattern of the average cosine at other wavelengthswith the presence of water Raman scattering (Fig. 2).Marshall and Smith4 have pointed out that the 589-nmregion is one where the absorption coefficient of thewater molecule is changing rapidly, and the fact thatthe light filters for the MER 1048 instrument have afinite bandwith (10 nm) will cause problems with theestimation of photons at the nominal 589-nm wave-length. The Gaussian-type shape of the filter's trans-mission curve will allow photons that are adjacent tothe nominal passband to be recorded as photons of589 nm. This phenomenon could explain the appar-ent increase in photons at 589 nm with the penetra-tion of the solar light field as recorded in Fig. 11.However, Marshall and Smith4 have calculated theprobable effect of the finite bandwidth of the MERfilter at 589 nm on the apparent number of photons of589 nm that were recorded at various depths in clearocean water. At a depth of 60 m the effect of Ramanemission, an increase in photons over those predictedby transmission of solar photons, is easily detectable.The calculated Kd(5 8 9 ) coefficient is less than theKd(5 8 9 ) coefficient that was predicted by Marshalland Smith for the penetration of photons at 589 nmsupplemented by photons not excluded by the opticalfilter. Thus the curve for the downwelling irradianceat 589 nm in Fig. 11 cannot be explained as an effectof the finite optical filter bandwidth, although it maybe making a contribution.

Recent attempts to parameterize the penetration ofsolar photons into clear ocean waters by using diffuseattenuation coefficients measured over the visiblespectrum in the surface layers are possibly hamperedby the effects of internal radiant emission. Tests ofthese parameterizations in clear ocean waters indi-cate that the models consistently underestimate thetotal photons present at depths that are > 80 m by50% or more.'4 5 The absorption coefficient anomalies

derived from the NOARL Monte Carlo output indi-cate that a significant admixture of water Ramanphotons to the longer (green-yellow) wavelengths ofthe clear ocean irradiance occurs at depths of 60-100m (Figs. 3-5). The assumption that all the photonsthat are lost at a given wavelength are converted toheat does not allow for accurate predictions as onegoes deeper into clear ocean waters. Internal radiantemission, in the form of water Raman scattering andpossibly fluorescence at certain wave bands, allowsfor the conversion of shorter-wavelength photonsinto longer-wavelength photons on penetration of thesolar beam. Thus a percentage of the photons that arepresumed removed from the downward photon streamin these parameterizations is not removed from thestream but is simply changed in the wavelength.Since this percentage of photons remains in thephoton stream, the anomaly associated with the twoattempted parameterizations is easily explained. Theseexcess photons at depth then explain the anomalouslylow absorption and diffuse attenuation coefficientsreported at longer wavelengths (in the green-yellowregion) in clear ocean waters.

The presence of water Raman scattering in clearocean waters alters the submarine light field to thepoint that quantitative estimations of the penetrationand fate of the total solar photons have not beenpossible heretofore. It is necessary to assess theproportion of a measured light field that is due tointernal emission and the proportion that comes fromdirectly transmitted, unaltered solar photons. Thenwe will be able to construct energy-photon budgetsfor the study of the penetration and fate of solarradiant energy in the world ocean. We have reportedtwo parameters that help in the quantitative estima-tion of water Raman emission: the average cosine andthe absorption coefficient anomaly.

The quantitative estimation of the percentage ofphotons due to internal radiant emission at any pointof the marine hydrosol requires two different measure-ments: the irradiance field (the net downwellingirradiance and the scalar irradiance) and a spectropho-tometric measurement of the inherent absorptioncoefficient of a water sample that is taken from thedepth of interest. This was done on the R/V Endeavorwith the spectrophotometry performed on board shipsoon after the samples were collected. Many investiga-tors are now working on improved instruments fordirectly determining the inherent absorption coeffi-cient more accurately on board ship and in situ.47 50

The use of approximation (8) permits a simple deter-mination of the percentage of photons caused byinternal radiant emission at any point in the marinehydrosol; the equation is an approximation, and theassumptions that are used to derive it were demon-strated to be robust in Figs. 6-8. The relative absorp-tion anomaly that is used in approximation (8) wasshown to be an accurate predictor of the percentilecomposition of emission photons in the ambientphoton field. The reason we are able to indirectly tagthe ambient photons in this method is the fact that


the solar photon stream and the Raman photonstream each have different mean photon paths asmeasured by the average cosine parameter (see Fig.1). Thus the contributions of the two photon streamsto the terms of Eq. (6) are not equivalent becauseinternal emission makes almost no contribution tothe net downwelling irradiance of the total photonstream. The upwelling and downwelling irradiancesthat originate from internal emission are nearlyequal, and the net downwelling irradiance, the differ-ence between the two photon streams, becomes verysmall, approaching zero as a limit. The contributionof internally emitted photons to the total scalarirradiance is relatively small near the surface com-pared with the solar photon stream but increaseswith depth as the solar photons are removed byabsorption. The discrepancy between the two contri-butions of internal emission to the terms of Eq. (6)creates the absorption coefficient anomaly, which isthe difference between the calculated nonconserva-tive absorption coefficient and the inherent absorp-tion coefficient of the hydrosol. When we compare theabsorption coefficient anomaly that is predicted forwater Raman scattering alone in Fig. 3 with theanomalies that are determined for Stations 6 and 8 inFig. 11, we begin to gain insights into the radiantenergy dynamics of the two stations. Both near thesurface (at a 30-m depth) and at greater depths (80 mor more) the absorption anomalies are somewhatgreater than that predicted for water Raman scatter-ing alone. Small amounts of yellow substance andalgal-type particles are present (Table I) and can thusbe contributing fluorescence to the basic water Ra-man scattering. The layering of the absorption anom-aly for Station 6 is evidence for fluorescence fromalgal-type particles. Both stations show an overalltendency to have the absorption anomaly increasewith depth as would be expected for a fundamentalbackground of water Raman scattering for the inter-nal radiant emission. With greater knowledge ofquantum yields for the fluorescence of dissolvedorganics 5 and the algal component,22 ,2 5 3 it will bepossible to partition the internal emission among itscomponents. For the present we have a method forassessing the quantitative contribution of all internalradiant emission sources to the submarine light field.

This paper was produced as part of the OpticalOceanography Program, NOARL, under the directionof Rudolph Hollman, NOARL contribution 331:013:91. We acknowledge gratefully the contribution ofLonzie Lewis, Jackson State University, Jackson,Miss., to the atmospheric analysis in the NOARLoptical model. At the University of North Carolina/Greensboro we have the support of the StatisticalConsulting Center and Computing and InformationSystems. A. D. Weideman acknowledges the supportof his contribution to this work by the NOARL BasicResearch Program under PE 61153N of the U.S.Office of Naval Research. We are thankful for thenumerous insights that were derived from discus-sions and the data resources that were provided by

Rudolph Hollman. Albert W. Green, Jr., Head of theOceanography Division, NOARL, read the manu-script and provided many valuable comments. R. H.Stavn is pleased to acknowledge the continuing sup-port of the U.S. Office of Naval Research, grantN00014-89-J-3137.

References1. R. H. Stavn and A. D. Weidemann, "Raman scattering effects

in ocean optics," in Ocean Optics IX, M. A. Blizard, ed., Proc.Soc. Photo-Opt. Instrum. Eng. 925, 131-139 (1988).

2. R. H. Stavn and A. D. Weidemann, "Optical modeling of clearocean light fields: Raman scattering effects," Appl. Opt. 27,4002-4010 (1988).

3. B. R. Marshall, "Raman scattering in ocean water," M.S.thesis. Department of Geography, University of California(Santa Barbara, Santa Barbara, Calif., 1989).

4. B. R. Marshall and R. C. Smith, "Raman scattering andin-water optical properties," Appl. Opt. 29, 71-84 (1990).

5. S. Sugihara, M. Kishino, and M. Okami, "Contribution ofRaman scattering to upward irradiance in the sea," J. Ocean-ogr. Soc. J. 40, 397-404 (1984).

6. E. Aas, "The absorption coefficient of clear ocean water," Rep.67 (December Institutt for Geofysikk, University of Oslo,1987).

7. L. Prieur and S. Sathyendranath, "An optical classification ofcoastal and oceanic waters based on the specific spectralabsorption curves of phytoplankton pigments, dissolved or-ganic matter, and other particulate materials," Limnol. Ocean-ogr. 26, 671-689 (1981).

8. D. Spitzer and M. R. Wernand, "In situ measurements ofabsorption spectra in the sea," Deep Sea Res. 28A, 165-174(1981).

9. D. Spitzer, M. R. Wernand, and G. C. Cadde, "Optical measure-ments in the Gulf of Guinea. Some aspects of remote sensing,"Oceanol. Acta 5, 41-47 (1982).

10. D. A. Siegel and T. D. Dickey, "Observations of the verticalstructure of the diffuse attenuation coefficient spectrum,"Deep Sea Res. 34A, 547-563 (1987).

11. J. C. Kitchen and H. Pak, "Observations of natural fluores-cence with an underwater radiometer," J. Oceanogr. Soc. Jpn.43, 356-362 (1987).

12. B. J. Topliss, "Optical measurements in the Sargasso Sea:solar stimulated chlorophyll fluorescence," Oceanol. Acta 8,263-270 (1985).

13. R. C. Smith and K. S. Baker, "Optical properties of the clearestnatural waters (200-800 nm)," Appl. Opt. 20, 177-184 (1981).

14. D. A. Siegel and T. D. Dickey, "On the parameterization ofirradiance for open ocean photoprocesses," J. Geophys. Res.92, 14,648-14,662 (1987).

15. S. Sathyendranath and T. Platt, "The spectral irradiance fieldat the surface and in the interior of the ocean: a model forapplications in oceanography and remote sensing," J. Geo-phys. Res. 93, 9270-9280 (1988).

16. D. Spitzer and R. W. J. Dirks, "Contamination of the reflec-tance of natural waters by solar-induced fluorescence of dis-solved organic matter," Appl. Opt. 24,444-445 (1985).

17. M. Bristow, D. Nielsen, D. Bundy, and R. Furtek, "Use ofwater Raman emission to correct airborne laser fluorosensordata for effects of water optical attenuation," Appl. Opt. 20,2889-2906 (1981).

18. F. E. Hoge and R. N. Swift, "Delineation of estuarine fronts inthe German bight using airborne laser-induced water Ramanbackscatter and fluorescence of water column constituents,"Int. J. Remote Sensing 3,475-495 (1982).

19. E. D. Traganza, "Fluorescence excitation and emission spectra


of dissolved organic matter in sea water," Bull. Mar. Sci. 19,897-904 (1969).

20. K. Carder, R. G. Steward, G. R. Harvey, and P. B. Ortner,"Marine humic and fulvic acids: their effects on remotesensing of ocean chlorophyll," Limnol. Oceanogr. 34, 68-81(1989).

21. C. M. Yentsch and C. S. Yentsch, "Emergence of opticalinstrumentation for measuring biological properties," Ocean-ogr. Mar. Biol. Ann. Rev. 22, 55-98 (1984).

22. B. G. Mitchell and D. A. Kiefer, "Determination of absorptionand fluorescence excitation spectra for phytoplankton," inFixation and Cycling of Carbon in the Marine Environment, 0.Holm-Hansen, ed. (Springer-Verlag, Berlin, 1984).

23. H. R. Gordon, "Diffuse reflectance of the ocean: the theory ofits augmentation of chlorophyll a fluorescence at 685 nm,"Appl. Opt. 18, 1161-1166 (1979).

24. K. L. Carder, R. G. Steward, and T. G. Peacock, "Spectraltransmissometer and radiometer (STAR): preliminary results,"Eos Trans. Am. Geophys. Union 68, 1683 (1987).

25. K. L. Carder, R. G. Steward, T. G. Peacock, P. R. Payne, and W.Peck, "Spectral transmissometer and radiometer: design andinitial results," in Ocean Optics IX, M. A. Blizard, ed., Proc.Soc. Photo-Opt. Instrum. Eng. 925, 189-195 (1988).

26. A. Gershun, "The light field," J. Math. Phys. 18, 51-151(1939).

27. N. K. H0jerslev, "A spectral light absorption meter for measure-ments in the sea," Limnol. Oceanogr. 20, 1024-1034 (1975).

28. R. H. Stavn, "The three-parameter model of the submarinelight field: radiant energy absorption and trapping in nephe-loid layers recalculated," J. Geophys. Res. 92, 1934-1936(1987).

29. G. Schellenberger, "Uber Zusammenhinge zwischen optis-chen Parametern von Gerdssern," Acta Hydrophys. 10, 79-105 (1965).

30. J. T. 0. Kirk, Light and Photosynthesis in Aquatic Ecosystems(Cambridge U. Press, Cambridge, 1983).

31. R. W. Preisendorfer, Hydrologic Optics (U.S. GovernmentPrinting Office, Washington, D.C., 1976).

32. R. H. Stavn, "Light attenuation in natural waters: Gershun'slaw. Lambert-Beer law, and the mean light path," Appl. Opt.20, 2326-2327(1981).

33. G. N. Plass and G. W. Kattawar, "Monte Carlo calculations ofradiative transfer in the Earth's atmosphere-ocean system: 1.Flux in the atmosphere and ocean," J. Phys. Oceanogr. 2,139-145 (1972).

34. H. R. Gordon and 0. B. Brown, "Irradiance reflectivity of a flatocean as a function of its optical properties," Appl. Opt. 12,1549-1551 (1973).

35. J. T. 0. Kirk, "Monte Carlo procedure for simulating thepenetration of light into natural waters," Division of PlantIndustry Tech. Paper 36 (Commonwealth Scientific and Indus-trial Research Organization, Canberra, Australia, 1981).

36. A. C. Tam and C. K. N. Patel, "Optical absorption of light andheavy water by laser optoacoustic spectroscopy," Appl. Opt.18,3348-3358 (1979).

37. T. G. Peacock, K. L. Carder, C. 0. Davis, and R. G. Steward,"Effects of fluorescence and water Raman scattering on mod-els of remote sensing reflectance," in Ocean Optics X R. W.

Spinrad, ed., Proc. Soc. Photo-Opt. Instrum. Eng. 1302,303-319 (1990).

38. A. Morel, "Optical properties of pure water and pure seawater," in Optical Aspects of Oceanography, N. G. Jerlov andE. S. Steemann Nielsen, eds. (Academic, London, 1974).

39. G. Kullenberg, "Scattering of light by Sargasso Sea water,"Deep Sea Res. 15, 423-432 (1968).

40. H. R. Gordon, 0. B. Brown, and M. M. Jacobs, "Computedrelationships between the inherent and apparent optical prop-erties of a flat homogeneous ocean," Appl. Opt. 14, 417-427(1975).

41. C. H. Chang and L. A. Young, "Seawater temperature measure-ment from Raman spectra," Research Note 920, N62269-72-0204, ARPA Order 1911 (Advanced Research Projects Agency,Washington, D.C., July 1972).

42. M. Iqbal, Solar Radiation (Academic, New York, 1984).43. A. W. Harrison and C. A. Coombes, "Angular distribution of

clear sky short wavelength radiance," Sol. Energy 40, 57-63(1988).

44. J. T. 0. Kirk, "Yellow substance (gelbstoff) and its contribu-tion to the attenuation of photosynthetically active radiationin some inland and coastal south-eastern Australian waters,"Aust. J. Mar. Freshwater Res. 27, 61-71 (1976).

45. 0. V. Kopelevich and V. I. Burenkov, "Relation between thespectral values of the light absorption coefficients of sea water,phytoplanktonic pigments, and the yellow substance," Ocean-ology 17, 278-282 (1977).

46. B. G. Mitchell, "Algorithms for determining the absorptioncoefficient of aquatic particulates using the quantitative filtertechnique (QFT)," in Ocean OpticsX, R. W. Spinrad, ed., Proc.Soc. Photo-Opt. Instrum. Eng. 1302, 137-148 (1990).

47. R. M. Pope, E. S. Fry, R. L. Montgomery, and F. M. Sogan-dares, "Integrating cavity absorption meter: measurementresults, in Ocean Optics X, R. W. Spinrad, ed., Proc. Soc.Photo-Opt. Instrum. Eng. 1302, 165-175 (1990).

48. C. C. Trees and K. J. Voss, "Optoacoustic spectroscopy and itsapplication to molecular and particle absorption," in OceanOptics X, R. W. Spinrad, ed., Proc. Soc. Photo-Opt. Instrum.Eng. 1302, 149-156 (1990).

49. J. R. V. Zaneveld, R. Bartz, and J. C. Kitchen, "Calibration anderror correction for a reflective tube absorption meter," inOcean Optics X, R. W. Spinard, ed., Proc. Soc. Photo-Opt.Instrum. Eng. 1302, 124-136 (1990).

50. J. S. Cleveland, "Spectral absorption coefficients measuredwith an integrating cavity absorption meter," in Ocean OpticsX, R. W. Spinrad, ed., Proc. Soc. Photo-Opt. Instrum. Eng.1302, 176-186 (1990).

51. S. K. Hawes and K. L. Carder, "Fluorescence quantum yieldsof marine humic and fulvic acids: application for in situdetermination," Eos Trans. Am. Geophys. Union 71, 136(1990).

52. C. H. Mazel, "Spectral transformation of downwelling radia-tion by autofluorescent organisms in the sea," in Ocean OpticsX, R. W. Spinrad, ed., Proc. Soc. Photo-Opt. Instrum. Eng.1302,302-327 (1990).

53. L. P. Shapiro, E. M. Haugen, and E. J. Carpenter, "Occurrenceand abundance of green-fluorescing dinoflagellates in surfacewaters of the Northwest Atlantic and Northeast PacificOceans," J. Phycol. 25, 189-191 (1989).


raman scattering in ocean optics: quantitative assessment of internal radiant emission

Documents