Some marine light sources and their effects on remote sensing reflectance models

Some marine light sources and their effects on remote sensing reflectance models

Material Information

Some marine light sources and their effects on remote sensing reflectance models
Peacock, Thomas G.
Place of Publication:
Tampa, Florida
University of South Florida
Publication Date:
Physical Description:
xii, 74 leaves : ill. ; 29 cm.


Subjects / Keywords:
Optical oceanography -- Pacific Ocean ( lcsh )
Remote sensing ( lcsh )
Bioluminescence ( lcsh )
Fluorescence ( lcsh )
Raman effect ( lcsh )
Dissertations, Academic -- Marine Science -- Masters -- USF ( FTS )


General Note:
Thesis (M.S.)--University of South Florida, 1992. Includes bibliographical references (leaves 51-59).

Record Information

Source Institution:
University of South Florida
Holding Location:
Universtity of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
029371159 ( ALEPH )
27929559 ( OCLC )
F51-00094 ( USFLDC DOI )
f51.94 ( USFLDC Handle )

Postcard Information



This item is only available as the following downloads:

Full Text


SOME MARINE LIGHT SOURCES AND THEIR EFFECTS ON REMOTE SENSING REFLECTANCE MODELS by Thomas G. Peacock A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Marine Science in the University of South Florida December 1992 Major Professor : Kendall L. Carder, Ph.D.


Graduate Council University of South Florida Tampa, Florida CERTIFICATE OF APPROVAL Master's Thesis This is to certify that the Master s thesis of Thomas G Peacock with a major in Marine Science has been approved by the Examining Committee on August 31, 1992 as satisfactory for the Thesis requirement for the Master of Science degree. Thesis Committee: ........... ------------Major Professor : Kendall L. Carder, Ph D. '"='= >='7"" =>r' ::= c ac z v = > c ---= Member: Ed>')afd S. Van Vleet, Ph D


ACKNOWLEDGEMENTS I would like to acknowledge the following people for their assistance : Bob Steward, for his help in data acquisition while on the CTZ cruise; Zhong-ping Lee for his CDOM fluorescence and water-Raman scattering parametric expressions; Steve Hawes, for fruitful discussions and fluorescence efficiency values; C.O. Davis, for the spectrally dependent Q-factor measurements; and Dennis Gruber (Scripps Institute of Oceanography), for the measured chlorophyll and phaeophytin concentrations on board the R/V Thomas Washington. 11


TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS AND ABBREVIATIONS ABSTRACT INTRODUCTION Historical Perspective Case 1 and Case 2 Waters The Second-Order Parameter Regime Remote Sensing Reflectance Models Closure THEORETICAL CONSIDERATIONS Reflectance and the Q Factor The Reflectance Model EXPERIMENTAL METHOD Specific Absorption Remote Sensing Reflectance RESULTS Bio optical Trends Comparison of Measured and Model Results DISCUSSION Potential Modeling Errors CONCLUSIONS LIST OF REFERENCES iii v Vl viii X 1 1 4 5 7 9 12 12 17 22 22 23 25 25 29 32 45 49 51


APPENDICES 60 APPENDIX 1. Water Raman Scattering 61 APPENDIX 2 Fluorescence of Colored Dissolved Organic Matter 65 APPENDIX 3 B i oluminescence of Marine Organisms 69 APPENDIX 4. Phytoplankton and Zooplankton in the Study Area 72 iv


LIST OF TABLES Table 1. [Chi a] [phaeo a], and spectral ratios of the specific absorption coefficients for the stations reported. Concentration units are 1-'g/1. Table 2 Model parameters for stations reported. v 26 32


LIST OF FIGURES Figure 1. Map of the CTZ study area. 11 Figure 2. Schematic diagram of measurement geometry for upwelling radiance. 13 Figure 3. Particulate scattering dependence on viewing angle. 14 Figure 4. Specific absorption coefficient curves for just below the surface at Stations 46 and 55. 27 Figure 5. Specific absorption coefficient curves for just below the surface at offshore Stations 39, 63, and 77. 28 Figure 6. Measured vs modeled Rn(A) at Station 46. 29 Figure 7. Measured vs modeled Rn(A) at Station 55. 30 Figure 8. Measured vs modeled Rn(A) at Stations 39, 63, and 77. 31 Figure 9. Two published absorption coefficient curves for water. 34 Figure 10. Derived vs measured specific absorption coefficient curves for surface waters at Station 55. 37 Figure 11. Comparison of the ratios of derived water-leaving radiance (L.,(field)/L.,(model) and measured particulate absorption (ap(10m)/ap(Om)), for Station 55. 40 Figure 12. Reflectance curves at Station 39, showing the effect of adding CDOM fluorescence and water-Raman scattering to the model. 43 Figure 13. Reflectance curves at Station 63. 44 Figure 14. Reflectance curves for Station 63. 45 Vl


Figure 15. Vibrational modes of a water molecule which produce Raman scattering. 63 Vll


Symbol () h v a(>-., z) a.(A,z) a(}.., z) ad(A,z) tzp(A,z) a,.(>-., z) LIST OF SYMBOLS AND ABBREVIATIONS Units degrees degrees gcmZSec em' m m2(mg Chi a)"' m m Description Index of refraction ; 11 = 1.334 for seawater The observation angle, measured between the vertical and the line of sight of the sensor The azimuth angle, measured anti-clockwise between the observation plane and the illumination (solar) plane Planck's constant, 6.63 x 1027 gcm2 sec. Wavenumber (reciprocal wavelength) Total absorption coefficient. Also referred to as the volume absorption coefficient. a(A,z) =tzp(A,z) +a,.(A,z) +llcoo M(A, z) Chlorophyll a mass-specific spectral absorption coefficient for photo-active pigmented particles only. Spectral absorption coefficient for photo-active pigmented particles (mainly phytoplankton) only. a.(A,z) = a.(}..,z)*[Chl a]. Spectral absorption coefficient for non photo-active colored particles (detritus and suspended particulates) Spectral absorption coefficient for colored dissolved organic matter (gelbstoff) only. Spectral absorption coefficient for all particles tzp(A,z) = a.(A z)+ad(}..,z) Spectral absorption coefficient for water only. vm


b(A,z) m1 Total spectral scattering coefficient for all directions. b(A,z) = b,(A.,z)+bb(A,z). Also referred to as the volume scattering coefficient. bbp(A.,z) m 1 Spectral backscattering coefficient for all particles only bbw(A,z) m 1 Spectral backscattering coefficient for water only. b b(A.,z) m 1 Spectral backscattering coefficient for particles and water. bb(A,z) = b,(A.,z) m 1 Spectral forward scattering coefficient for particles and water. c(A.,z) m 1 Total spectral attenuation coefficient. c(A.,z) = a(A.,z)+b(A.,z). Also referred to as the volume attenuation coefficient. EiA.,z) W m 2nm 1 Spectral downwelling irradiance at depth z Equals the integral over 2"X" of LiA.,O,z). E.(A.,z) W m2nm1 Spectral upwelling irradiance at depth z. Equals the integral over 2"X" of L .(A.,O,z). Ld(A,O,z) W m 2ster1nm 1 Spectral downwelling radiance within a unit solid angle L .(A.,O,z) W m 2ster1nm 1 Spectral upwelling radiance within a unit solid angle. R(A.,z) p(80), p(O) ----Spectral irradiance reflectance at depth z. This is the ratio of upwelling irradiance to downwelling irradiance R(A.,z) = E.(A.,z)/E d(A.,z). Spectral irradiance reflectance just beneath the sea surface Remote sensing reflectance, measured above the sea surface. This is the ratio of water-leav ing radiance to the above surface downwelling irradiance Rn(A.) = L.,(A.)/ E d(A.,O+). Fresnel reflectance factors for the air-sea interface for incident angles measured from the zenith (8, above surface) and the nadir ((Jo below surface) respectively. lX


SOME MARINE LIGHT SOURCES AND THEIR EFFECTS ON REMOTE SENSING REFLECTANCE MODELS by Thomas G. Peacock An Abstract Of a thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Marine Science in the University of South Florida December 1992 Major Professor: Kendall L. Carder, Ph D. X


Constituents in seawater that effect its spectral optical signature include photo-active pigments (e.g. chlorophylls, accessory pigments, and degraded pigments), suspended particulates, and dissolved organic matter (DOM) It is generally assumed that color constituents co-vary with at least in the open ocean (Morel Case I waters). The limited spectral information available from the Coastal Zone Color Scanner (CZCS) after atmospheric correction (only 4 spectral bands, each 20 nm wide) does not facilitate verification of this co-variance, and the error in estimation of chlorophyll concentration from a CZCS scene is generally high (+I40%) for concentrations between 0.02 and 20 mg/m3 (Gordon and Morel, 1983) High spectral resolution (5nm half-bandwidth) optical models of reflectance (or water-leaving radiance) employing field measurements or estimates of inherent optical properties (absorption, scattering) produce results which, when compared to a measured spectral signature, can provide estimates of pigment concentrations, and information about those constituents not actually measured in the field (e.g. DOM absorption, particle backscattering). These models normally exclude contributions due to DOM fluorescence, water-Raman scattering, or bioluminescence as these marine light sources are considered to be insignificant in most cases. However, accurate interpretation of data from future high spectral resolution satellite radiometers will depend in part on how well we understand these "second order" optical effects From data collected at several stations off the California coast, measured and modeled Xl


reflectance curve comparisons indicate the possibility of influence from one or more of the above-mentioned marine light sources. Sample handling and storage methods appear to be major factors in introducing bioluminescence, but the influences of DOM fluorescence and water-Raman scattering appear to be dependent upon the input light field, the concentrations of photo-active pigments, and the scattering regime. Specific tests and/ or additional field measurements can be employed to better characterize the influence of marine light sources on remote sensing reflectance measurements and models. Abstract approved:; 'P' = /Majof Professor: Kendaliftarder, Ph.D Professor, Department of Marine Science Date of Approval xu


1 INTRODUCTION Historical Perspective The existence of marine light sources has been recognized for many years The best known source is bioluminescence from marine organisms. Investigations of bioluminescence have centered around the chemical processes involved (Hastings 1983), its functions in reproduction, predation and predator avoidance (Morin, 1983), taxonomic diversity (Lapota and Losee, 1984), or its optical characteristics (Widder et al., 1983; Filimonov and Sadovskaya, 1986). Fluorescence produced by many of the absorbing materials in sea water has been studied with considerable i nterest employing both pass i ve and active stimulation methods (Cullen et al. (1988) and refs. cited) Fluorescence of chlorophyll has been studied rigorously for many years in efforts to relate it to biological processes and properties of marine organisms in the photic zone (e g., photosynthesis (Prezelin and Ley ,1980), cellular pigment concentration (Cleveland and Perry, 1987), physiolog i cal state (Kiefer, 1973), and the diffuse and remote sensing reflectance (Gordon 1979; Carder and Steward, 1985). Fluorescence of colored dissolved organic matter (CDOM, gelbstoft) has been studied extensively in an effort to understand its origin


(Fogg, 1966; Gagosian and Steurmer, 1977; Harvey et al., 1983), chemistry (Pocklington, 1977; Harvey et al., 1983; Jackson and Williams, 1985), or spectral characteristics (fraganza, 1969; Carder et al., 1989; Coble et al., 1990; Hawes et al., 1992). The contribution of chlorophyll fluorescence to the water-leaving radiance and to the submarine light field has also been studied extensively (Gordon, 1979; Kiefer et al., 1989). Fluorescence of CDOM is of interest to modelers of remote sensing reflectance (Lee et al., 1992; Hawes et al., 1992) for its potential impact on the retrieval of pigment concentrations from remotely-sensed ocean color. 2 Water Raman scattering (a quasi-fluorescent process) has likewise been studied extensively, and has been used to obtain information about the physical properties of the water body (Leonard et al. 1979; Hoge and Swift, 1981; Collins et al. 1984), or to normalize Chl a fluorescence (Bristow, 1978; Hoge and Swift, 1981). Only recently has this effect been seriously considered as a potentially significant contributor to the water leaving radiance field (Stavn and Weidemann, 1988a,b; Marshall and Smith, 1990). Although much is known about these marine light sources in terms of their relationships to other marine processes, little has been done until recently to incorporate them into the modeling of remote sensing reflectance (R..). For bioluminescence, the reason is obvious-it is a nocturnal phenomenon while R,. measurements are obtained during the day The reason usually given for excluding fluorescence and water Raman scatter was that they have only a small effect on the bulk measurement or on the derivation of parameters of interest (Gordon and Morel,


3 1983 ; Marshall and Smith, 1990) One benefit of accepting the exclusion of these parameters was that it greatly simplified the derivation of the photo-active (photosynthetic as distinguished from non-photosynthetic) pigment concentration from a remotely sensed ocean color sample spectrum. However, the error associated with satellite-derived pigment concentrations in the range 0.02 and 20. 0 p.g/1 was generally high (a factor of 2) (Gordon and Morel, 1983), and differences between measured and modeled reflectance were difficult to assess Fluorescence and water Raman scattering are examples of in situ light sources producing photons, which were not originally incident as solar or sky light, in response to the incident light field The above exclusions were accepted, in part, because of the apparently dominant role of Chl a and its accessory pigments in early ocean color data collected and analyzed, and because the statistical analysis of the data used to develop the pigment algorithms indicated that most color constituents covaried with Chl a concentration. However under certain conditions the contributions of fluorescence and Raman scatter could be falsely attributed to Chl a and/or its accessory pigments, thus affecting derived pigment concentrations The importance of the early studies of ocean color lay in the identification and characterization of major (first order) parameters, definition of fundamental methods of remote sensing analysis, and the establishment of a sound theoretical basis on which future investigations are built (see for example, Bricaud et al., 1983). As research continues in this field investigators employ more sensitive instrumentation and they move into more physically and optically complex study areas. Earlier


4 modeling methods may be insufficient to satisfactorally explain the measured spectral character of these new study regions, and established pigment algorithms may not produce pigment concentrations in agreement with those that are measured. Valid simplifying assumptions made earlier may no longer apply when attempting to model the higher resolution reflectance signatures of these more complex regions. Additional (second order) parameters may be necessary to adequately explain the difference between measured reflectance and the modeled curve generated by the earlier models. This paper presents evidence which, although not conclusive, indicates the need for just such second-order parameters. Examples are provided of the possible effect of adding certain parameters to the model, and recommendations are proposed for additional field studies to test the hypotheses and avoid contamination. Case 1 and Case 2 Waters Morel and Prieur (1977) described two ocean water types, and illustrated their very different spectral reflectance (color) signatures. "Case 1" waters were those whose optical properties were dominated by the pigments of living algal cells (phytoplankton), while "Case 2" waters were those whose optical properties were dominated by inorganic particles. Gordon and Morel (1983) expanded on these definitions, describing the optical properties of Case 1 waters as dominated by the pigments of algal cells, their associated debris, and the dissolved organic material released by both the living cells and the debris. In contrast, the optical properties of Case 2 waters were described as being dominated by one or more of the following


5 constituents: resuspended sediments, terrigenous particles, terrigenous d i ssolved organic material (from land runoff), or anthropogenic influx (whether dissolved or particulate) Further, Case 2 waters may contain significant quantities of any or all of the Case 1 materials but waters classified Case 1 may not contain significant amounts of the materials found to dominate Case 2 waters. The Second-Order Parameter Regime In general, Case 2 waters are representative of a more demanding modeling regime alluded to earlier A region of transition from Case 1 to Case 2 waters, or from oligotrophic to eutrophic Case 1 waters may also require "second-order" models. For example in a recently productive parcel of water where declining algal blooms leave behind elevated CDOM concentrations (a transition from eutrophic toward oligotrophic Case 1), the absorption at 440nm due to CDOM could become significant relative to that due to Chl a, thus affecting derived pigment calculations (e g see Carder et al. 1989, 1991). In addition the decreasing pigment absorption of the declining bloom would make available more photons for CDOM to absorb, presumably enhancing the fluorescence component. Where nutrient stress results in increased exudation of CDOM by viable phytoplankton (Jensen 1984), the CDOM fluorescence contribution might be discernible, if the Chl a concentration is not too high There remains considerable debate over the potential effect of this CDOM fluorescence on the derivation of pigment concentration from a remotely sensed ocean color spectrum.


6 Transpectral scattering by water (so-called water Raman scattering), as observed in the oceans, has also been studied recently (Sugihara et al., 1984; Stavn and Weidemann, 1988a,b; Stavn, 1990; Marshall and Smith, 1990) Above about 520 nm the light field at depth is found to be non-conservative (i.e., more light of a particular wavelength is present than can be accounted for by propagating the incident light to depth through pure water), and the contribution of water Raman scattering is quantitatively significant throughout the photic zone in clear ocean waters(Stavn and Weidemann, 1988a) In addition, water-leaving radiance of the shorter wavelengths (between about 430 and 500nm) is augmented by water Raman emissions which are stimulated by absorption of solar UV and blue photons between = 370 and 420 nm (Stavn, 1990; Peacock et al., 1990) Marshall and Smith (1990) concluded that Raman scattering effects must be considered when estimating inherent optical properties (e.g. the volume absorption coefficient, a(A), and the volume backscattering coefficient, bl'A)) from in situ measurements of the irradiance field, but that its influence on in-water ocean color algorithms is already accounted for by the empirical nature of those algorithms. From a modeling perspective, however, Stavn (1990) argues that since the algorithms to estimate pigment concentration from a remotely-sensed signal generally employ the reflectance values near 440nm and between 550 and 565nm (Gordon and Morel, 1983 and refs. cited), the contribution to water-leaving radiance by water Raman emissions due to solar UV excitation may have a significant effect on the results of applying these algorithms under conditions of clear water and high incident light levels.


The spectral signatures of the two optical phenomena discussed above may, at times, be sufficiently large relative to the photo-active pigment signal to significantly influence the observed ocean color. One other marine light source is bioluminescence by organisms in the water. To our knowledge, a study of the implications of bioluminescent effects on the measurement of particulate absorption has not been undertaken. Not usually a factor in daylight optical measurements, bioluminescence may nevertheless have to be considered if certain field measurement methods are employed Conversely, knowledge of the potential for contamination by bioluminescence of particulate absorption coefficient measurements can aid the investigator in avoiding that problem. Differences between measured and modeled remote sensing reflectance may be attributable to one or more of these phenomena, under certain conditions, indicating that the derivation of pigment concentration will also be impacted. Remote Sensint: Reflectance Models 7 Preisendorfer (1961) defined irradiance reflectance R(A,z) and remote sensing reflectance R,.(A) as apparent optical properties (properties dependent on the spatial radiance distribution), and defined the volume absorption coefficient a(A,z) and the volume backscattering coefficient bb(A,z) as inherent optical properties (properties independent of the spatial radiance distribution) Models of R(A,z), (e .g., Gordon et al . 1975, 1988; Morel and Prieur, 1977; Morel 1988) and Rn(A) (e.g., Carder and Steward, 1985; Carder et al . 1985) have been proposed which describe these


apparent optical properties in terms of the inherent optical properties a(A,z) and bl'A z). As generally applied, these models do not specifically address inelastic or transspectral scattering such as fluorescence or water Raman scattering, and bioluminescence has typically not been included as a model parameter, since it is considered to be a nocturnal phenomenon Applying the aforementioned models to waters containing any of these light sources will result in differences between the measured and modeled reflectance values at wavelengths where the source(s) provide a significant part of the measured signal. A major goal of ocean color remote sensing is to accurately estimate the concentrations of the various constituents in the water (e.g pigment concentrations, total suspended particulates, dissolved organic material) This capability, especially from high altitude, would greatly enhance the value and e f fectiven e ss of studies of synoptic-scale biological and physical processes such as primary productivity, the carbon cycle, sediment transport, ocean fronts, and ocean circulation Rn(A) model algorithms which include correctly applied, second-order optical parameters would improve the quality of those estimates by allowing, through model inver sion the removal of these second order effects where they are present. Understanding of second order parameters also provides for the possibility that these e f fects may be detectable in a remotely sensed signal independent of any in-situ measurements, by appropriate instrument design and data analysis techniques. 8


9 Closure "Closure" in the context of remote sensing reflectance modeling is the condition that exists when the model, using valid estimates or measurements of all relevant input parameters, produces a modeled reflectance curve which matches a synoptic, calibrated, and remotely measured reflectance curve. The key is to correctly identify and quantify all the "relevant'' parameters, and where a parametric estimate is necessary, to base it on sound knowledge of the effect of that parameter on ocean color. When measured from high altitude, the water-leaving radiance is convoluted with (and overshadowed by) the optical signature of the atmosphere. Approximately 90% of the signal received by a nadir-looking radiometer in orbit over the oceans can be due to scattering and absorption in the atmosphere. Corrections for atmospheric effects are rendered unnecessary, and the closure problem modified, by "remotely" measuring the reflectance from just above the sea surface. The tradeoff is in the increased sensitivity to sea-surface roughness and local reflections from nearby objects (e.g the ship or other platform), but the corrections for these can be minimized by careful sampling. Synoptic sampling of pigment concentration provides "ground truth" for the pigment algorithms, and measurement of apparent (above surface L.,, subsurface E., Ed, L. ) and inherent ( a, b., ) optical properties of the water provides partial input for the model (see List of Symbols for definitions). Closure is a reality when application of the inverted model to the measured reflectance yields reliably accurate estimates of all the individual parameters, and the in-situ measurements are not needed.


10 In June and July of 1988 we participated in a cruise to the California Coastal Transition Zone (CTZ, Figure 1) to study an offshore-advecting jet and an associated cold water filament, which were transporting cold, upwelled coastal waters into the warmer waters of the open ocean (Huyer et al., 1991). We recorded measurements of bio-optical properties that allowed us to measure and model remote-sensing reflectance and, subsequently, water-leaving radiance. We were hoping to obtain closure between modeled and measured parameters for these waters. Early attempts to obtain closure between the measured and modeled spectral curves of remote sensing reflectance for waters in the CTZ were frustrated in certain wavelength regions by what appeared to be radiant sources not included in the model: (1) fluorescence by phytoplankton pigments, (2) Raman scattering by water molecules, (3) bioluminescence on the filter pad used to measure particulate absorption, (4) and fluorescence by colored dissolved organic matter (CDOM). The relative contributions attributable to these phenomena varied with the transition from eutrophic, coastal upwelling stations to oligotrophic, offshore stations. Effects believed to be due to Raman scattering were evident in clear offshore waters for wavelengths from 415 to about 550 nm, while effects attributed to fluorescence and bioluminescence were observed in nearshore waters. Fluorescence due to chlorophyll a and phaeophytin a were important from 665 to 705 nm, while bioluminescence effects were observed in particle-absorption data, generally from 475 to 540 nm, for nearshore waters. Bioluminescence was also observed visually ("glowing" filter pad) at a night-time offshore station during the particle absorption measurement procedure.


40 20 I A446 .sl I I 0A5 IU I f 63 =-= r I \ COAS T AL TRANSITION ZONE E XPERIMENT I lh: J UN E / J UL Y 1988 (PT A R[NA 1 a 54 ",.-,, J'\ \ r< ) I L : '>J \ , /}'. { ,.,.; g \ M i l S A N \ I I \ \fU \ 1J HI. \ /I n ) :.0 N o l o 0 ,.., r ;' ---.......- ---....... --o;;;;;; -= ...... .._._. ....... -..__-b;; ljjjjjjjjjj ;--:-,;;: -----=:t== ---. -ljjjjjjjjjj l;;;iiiii;;:Z: ....... :.:: lli 127 0 0 4 0 2 0 126'0 0 40 20 125 ()0 co 20 124 '00 4 0 2 0 12J'OO 40 20 122' 0 0 Ill 4 0 Figure 1 Map of the CTZ study area The lines indicate the approximate edge of the offshore advec tin g jet during the study period. The filament occ upi es a broad nearshore-offshore swa th adjacent to the south and eastern edge of the j et.


12 THEORETICAL CONSIDERATIONS Reflectance and the Q Factor On the basis of an approximate radiative transfer model, Duntley (1942) developed a relationship for subsurface irradiance reflectance at depth z (R(A.,z)) that is dependent on the ratio of the backscattering coefficient to the sum of the absorption and backscattering coefficients (for defmitions of terms see List of Symbols). Gordon et al . (1975) and Morel and Prieur (1977), using different approaches employing both models and ocean field data, corroborated the general form of this equation The net result is an equation that to first order expresses R(A,z) in terms of inherent optical properties R(}.,z) = E,(')..,z) Ej).,z) [la] Morel and Prieur (1977) simplified Eq. 1a for waters that they studied, noting under most circumstances that backscattering was less than 5% of absorption, thus removing the backscatter term from the denominator. Remote sensing radiometers measure radiance, so relating this subsurface irradiance reflectance to the remote-sensing reflectance measured above the surface


13 involves the relationship between upwelling radiance (L. ) and diffuse upwelling irradiance (E ), as well as the changes which occur in the radiance field as light crosses the sea-air interface. The ratio of upwelling irradiance E.("A,z), to upwelling radiance L .("A,z,O' ,c/>') (both subsurface) is defined by Austin (1974b) as Q("A, z,O' ,c/>'). Throughout this paper A is wavelength (nm), 8 is the angle of observation or illumination relative to the vertical, cJ> is the horizontal angle between the solar and observational planes, and the prime denotes subsurface angles (Figure 2; see also List of Symbols). I I I I So l a r plane ---r--_ I I I I I I I Zenilh j lines -1 1 Satellit e radiometer : Lr( e ) I I I I ----+Observation plane I I Figure 2 S c hematic diagram of measurement geometry for upwelling radiance 1


14 The Q-factor is essential to correctly relate Eq. 1a (the inherent optical properties) to the remotely sensed reflectance measured above the ocean surface. Smith (1974) and references cited suggest that L,(A,z,O' ,') is only weakly dependent upon 0 and 4>'. Any dependence weakens further for observation angles relative to the direction of propagation for sunlight that are less than 165 (e.g for a nadirviewing radiometer and subsurface solar zenith angles greater than 15 ; see Figure 3a). This is likely due to the relative "flatness" of the phase function at backscattering angles away from the glory angles near 180 (Figure 3b). Sensor 1 01 102 1 E 10''l I 1 0 ' f 10_, Bahama Is 1 o ... b oo 2o 40" so eo 100" 120 140" 1so 1eo Scatter ing angle 8 Figure 3. Particulate scattering dependence on viewing angle. A) Schematic diagram illustrating geometry described in the text. The polar diagram represents a volume scattering function {3(8). B) Graphical representation of the volume scattering functions for typical clear (Bahamas) and moderately turbid waters (data of Petzold, (1972)). From Kirk, (1983)


15 Gordon (1986) has shown for such scattering angles that L.,('A,z), and the ratio R('A,z)/Q('A,z) are, to first order, proportional to the right side of Eq. la. His work suggests a Q value of about 3.5 if Q is taken to be spectrally constant. Earlier field measurements supported use of a spectrally constant value (Gordon and Morel, 1983), and our initial measurements determined Q to have a value of about 3.4 for St. 77. A spectrally constant Q value of 3.4 was initially used for the offshore stations, and 4. 7 was used for nearshore stations, a value within the range found by Austin (1974a)). Measured Q-factors at two stations were later obtained from C.O. Davis (Jet Propulsion Lab), and the model curves at those two stations were recalculated using the spectrally dependent Q factor data. Satellite, airborne and surface radiometers detect a radiance signal modified by the sea surface and by the atmosphere such that the total radiance at the sensor is [lb] where tis the diffuse transmittance of the atmosphere (dependent on Rayleigh, ozone, and aerosol optical thicknesses (TR, 7"03, and TA, respectively)), Lw is the water-leaving radiance L and LR are the radiances scattered into the field of view by aerosol and Rayleigh scattering respectively. Contributions due to Fresnel reflectance of light by the sea surface are also included in the second and third terms. Gordon (1976), through Monte Carlo simulations of radiative transfer in an ocean-atmosphere system, determined that the ocean and atmosphere may be de-coupled for the purpose of computing Lr('A,O,), in the case of a sensor above the atmosphere, and when the


16 aerosol optical thickness rA < 0 6 We measured reflectance from a platform (ship) just above the sea surface, eliminating the need for Rayleigh and aerosol scattering terms, and allowing t to equal one. Gordon (1976) also found that the net effect of ocean constituents can be simulated by placing a Lambertian reflector of albedo R just beneath the surface and defined R as in Eq. la, E (.A. o ) R(.A.,o-) = = EJ..A.,o) Q(.A.) L.,(.A.,o-) EJ..A.,o) with the osignifying the values are for just beneath the sea-surface. We can now write Eq. 1a in terms of water-leaving radiance : [lc] (1 p(6o)) (1-p(6)) bb(.A.,o) [21 3 Q(.A., o ) 112 (a(.A.,o-) + bb(.A.,O)) where o+' o -signify values for just above and just below the sea surface, respectively' I describes interfacial effects (transmission losses (1p) and radiance divergence (1/172 ) across the interface), p(Oo ) and p(O) are the Fresnel reflectance factors for the downwelling and upwelling photon streams, respectively, and 11 is the index of refraction for seawater ( = 1.334) Austin (1974b) has reported that in general, waves increase the reflectance factors, and that p(O) is more sensitive to observation angle and windspeed than is p(Oo).


17 The Reflectance Model Based on theoretical relationships between subsurface irradiance reflectance, remote sensing reflectance, and the inherent optical properties, Carder and Steward (1985) developed a remote sensing reflectance model for investigating red tide blooms of a single dinoflagellate species in the Gulf of Mexico. Application of this model to a heterogeneous oceanic regime is achieved by the removal of the species-specific parameters What remain are parameters for absorption and scattering due to water and particulates (both living and detrital), and absorption due to CDOM. The extent to which one can reliably partition total absorption and backscattering in practice is controlled by the sampling methods employed in the field. Expanding the absorption and backscattering coefficients, we can rewrite the right side of Eq. 2 as [3] where the subscripts w, p, and CDOM refer to water, total particulates, and colored dissolved organic matter, respectively. Absorption and backscattering of water are well known (Morel, 1974; Tam and Patel, 1979; Smith and Baker, 1981). Water absorption curves from both Smith & Baker (1981) and Tam & Patel (1979) were used in our initial modeling of field Rn("A) data, and the results using the two curves were compared. Absorption due to particulates was measured.


18 Field absorption measurements did not include absorption due to CDOM, but its effects are present in the R,. data. Bricaud et al . (1981), Carder et al .. (1989), and Roesler et al .. (1989) and references cited present a relation describing absorption due to CDOM as a spectrally dependent, logarithmic function (4) where C1 is the absorption at 400 nm and c; is the spectral slope, and provide a range of values for cl and c2 for a variety of locations. Equation 4 was used to describe the form of CDOM absorption, with adjustment of the values of C1 and c; in a predictorcorrector scheme to improve the agreement between the field and model curves. Validation of model values determined for C1 and c; was obtained through the application of the following relation, where subsurface downwelling irradiance measurements were available from the Spectral Transmissometer and Radiometer (STAR, see Carder et al., 1988b): where Jld is the average cosine of the downwelling underwater light field,


19 is the downwelling attenuation coefficient in a depth interval just beneath the surface, tzp is the measured total particulate absorption, and a, is the water absorption. The wavelength dependence is implicit, but the designation is dropped for convemence. Particulate backscattering has been modeled by Gordon and Morel (1983) for Morel Case 1 waters with a IIA dependence at low pigment concentrations (chlorophyll a + phaeophytin a < 0.1 mg/m3). At high concentrations (e.g. chlorophyll a + phaeophytin a = 20 mg/m3), they express backscattering as a generally non-spectral function that is, however, slightly depressed by the absorption due to pigments. For the California Coastal Transition Zone, where aging phytoplankton blooms and their associated degradation products are carried offshore by coastal jets, significant detritus and phaeopigments are found even in low-chlorophyll offshore waters (Peacock et al, 1988) These components absorb strongly in the blue, where the strongest backscatter also occurs. Also, assuming larger (equivalent diameter > 5 #Lm) particles predominate in the generally polydisperse size distribution of this highly productive region, the response of the backscattering efficiency factor to particle size is smoothed (Morel and Prieur, 1977) These combined factors tend to reduce the spectral character of the particulate backscattering coefficient. The general form for the backscattering of particles is taken to be


20 (5) where B1 is the particulate backscattering coefficient at 400 nm. We chose as a first-order estimate a non-spectral form = 0) for the particulate backscattering coefficient for all stations other than the oligotrophic St. 77, for which we set equal to 1. Subsequent modeling modified these initial values to some extent (see Discussion). To "test" the hypothesis that CDOM fluorescence may be a significant factor in some R.. modeling, we employed an expression developed by Lee et al. (1992), which calculates CDOM fluorescence using the the full spectra of the downwelling irradiance above the surface, the modeled volume absorption coefficient, and the modeled CDOM absorption coefficient. CDOM fluorescence is a broad band emission ( 100nm full width at half maximum, FWHM), and the emitted radiation is itself a candidate for re-absorption by the CDOM, or for absoq)tion by the medium or other constituents. The convolutions this presents in the modeling process are not accomodated as yet, so the introduction here of this term is for illustrative purposes only. Lee et al. (1992) also developed an expression to calculate water Raman scattering, and that expression is used here as well for the same purpose as the CDOM fluorescence term There is justification for this approach to illustrating the potential effect of these internal sources on the model. After modeling R,. to the fullest extent with measured


21 input parameters and reasonable estimates of parameters not measured, there typically remain differences between the model and the field R" curves If those differences are reduced through the application of a theoretically sound expression for a new model parameter the implication is that the new parameter may play a factual role in the physical process under study. The obvious next step would be to devise a more rigorous series of validation tests for the new parameter(s).


22 EXPERIMENTAL :METHOD Specific Absorption Specific absorption coefficients for particulates were obtained for surface waters following the method developed by Mitchell and Kiefer (1988a). The sample volume was gently poured into a 25mm fllter funnel and flltered at 5 10 in Hg vacuum onto 25mm Whatman GF/F glass fiber filter pads. The pad was removed from the fllter apparatus, back-illuminated by a tungsten source and the transmitted light measured by a spectral radiometer (Spectron Engineering SE-590) This instrument records a full spectrum (about 375 to 1050 nm) simultaneously A wetted blank fllter pad was si milarly measured. The optical density (OD) of the sample pad relative to the blank pad is equal to the -log1o of the ratio of the sample pad transmittance to the blank pad transmittance The total particulate absorption coefficient was obtained by applying correction factors for path elongation, flltrate volume, and effective fllter-pad area to the OD measurement (Mitchell and Kiefer, 1988a) Dividing the result by the measured chlorophyll a concentration yields the chlorophyll a mass-specific particulate absorption coefficient. These filtering and measurement operations were performed in an environment of low ambient light


23 A time delay of from 0.5 to 2 hours sometimes occurred between collection of a water sample by the Niskin bottles (deployed on a rosette with CTD, etc.) and the measurement of particulate absorption at these stations. During this time, the water sample resided in the Niskin bottle and then in opaque sample bottles at room temperature ( = 23 o C). Remote Sensing Reflectance Upwelling radiance above the sea surface (Lu(A.,O+)) was measured directly using the Spectron spectral radiometer, as was the downwelling sky radiance (see Carder and Steward, 1985). The measured upwelling radiance includes the water-leaving radiance and Fresnel reflectance (by the sea surface) of a fraction of the downwelling sky radiance. The nominal contribution of above-surface Fresnel reflectance is about 2.1% for observational zenith angles less than about 20 and a flat ocean (Austin, 1974b). Measured reflectance based upon the fraction of skylight reflected at 785 nm (assuming Lw(785) =0 ) were typically higher than this, varying between 2.4 and 5% due to observation angle, and surface and wind conditions The nominal observational zenith angle was about 40 in this study, but the effective look angle was highly variable at times, due to surface roughness and ship roll. The sea surface area viewed while measuring L,(A.,O+) was approximately 0. 75m x 0. 75m. The total downwelling irradiance above the surface (Ed(A.,O+)), reflected off a horizontal Kodak grey card, was also measured using the Spectron. Assuming the card is a good Lambertian reflector, the downwelling irradiance is obtained by


24 multiplying the measured radiance by ..(Ierlov, 1968), and dividing by the spectral reflectance of the card Laboratory measurements of grey card reflectance characteristics taken after the cruise show that the card is a good Lambertian reflector for instrument observational angles of 5< 8 < 30 from the card normal In the range 410 to 700nm the spectral reflectance curve for the card is relatively flat to within 5%. Below about 410nm reflectance falls off rapidly, and above 700nm some variability also occurs. Grey card observation angles were generally around 15 o and the measured spectral reflectance curve for the grey card reflector was applied to obtain the downwelling irradiance spectrum.


25 RESULTS Bio-optical Trends The June/July 1988 cruise of the R/V Thomas Washington to the region off Point Arena, northern California resulted in 10 stations with clear skies, ideal for developing remote sensing model relationships Th i s cruise was part of the Coastal Transition Zone (CTZ) Experiment investigating processes associated with an ad v ected jet of nutrient-rich upwelled water from close inshore to about 350 km offshore. The jet was located using A VHRR infrared satellite imagery (Abbott, 1988), and was also tracked by deploying drifters with drogues at 20m depths in the jet axis (Figure 1). The coastal upwelling regime during this cru i se period resulted in surface chlorophyll a values greater than 10 J..'g/1 inshore with a general decrease in chlorophyll a concentrations along the jet axis to about 24 at St.39 (Table 1). Oligotrophic waters of the region contained about 0 .10 1 of Chi a (St. 77) We expected there would be a change from phytoplankton-dominated optical properties for nearshore waters to CDOMand detritus-rich properties for the offshore jet. Waters adjacent to the jet offshore (California Current) were expected to have much less CDOM, pigments, and detritus than waters in the jet (Note : "detritus" as used here denotes non-autotrophic particles includ ing bacteria)


TABLE 1. [Chi a], [phaeo a], and spectral ratios of the specific absorption coefficients for the stations r eported Concentration units are J.Lg/1. Station# Location [Chi a] [phaeo a] ap(675) 39 3go 37.3N 0.243 0.073 5 03 126 27.0W 46 3go 05.9N 7 609 1 631 2 31 124 08.8W 55 38 44.7N 10 657 1 399 2 .22 123 39.8W 63 38 35.8N 0 278 0.081 8 .37 124 35.6W 77 38 19 .8N 0 103 0.028 8.23 126 44.5W Stations reported in this paper comprise end members (in terms of pigment concentration and location with respect to the jet) of the 10 non-cloudy stations appropriate for remote sensing. Station 46, located on the shelf-slope about 37 km 26 offshore and just north of Point Arena, was in the upwelling region near the origin of the jet. Station 55 was about 13 km offshore south of Point Arena. It represented a high-pigment, non-jet, upwelling regime Station 63 was located just outside of the jet about 90 km offshore. Station 39 was a jet-axis station about 300 km offshore, and Station 77 was an oligotrophic, non -jet station about 130 km north of Station 39. The chlorophyll a specific absorption curves (Fig ure 4) for nearshore Stations 46 and 55 have shapes similar to "detritus free" phytoplankton cultu res of coastal species : 1) ap(436):ap(675) ratios were 2 31 and 2 .22 for Stations 46 and 55,


27 --. I -St. 46 s 0 .024 -St. 55 !=: 0 0 .018 ...... +J 1-o 0 rJl ..c 0 .012 .. ... ttl C) ...... '+-< ...... C) 0 .006 Q) C/) 350 450 550 650 750 850 Wavelength (nm) Figure 4. Specific absorption coefficient curves for just below the surface at Stations 46 and 55. [Chl a] is 7.6 and 10.6 p.g/1, respectively. respectively; and 2) the absorption values below 410 nm were significantly lower than the value at 436 nm, indicating little contribution from detrital absorption (see Roesler et al., 1989, and Mitchell and Kiefer, 1988b). At St.46, a small relative maximum was apparent near 410 nm, the phaeophytin/phaeophorbide absorption maximum; St.55 exhibits about a 20nm blue shift of the blue absorption maximum Both these spectral shapes indicate the presence of some phaeophytin a and/or phaeophorbide a (degradation products of chlorophyll a). The small feature at Sta. 46 is probably the result of grazing by zooplankton (Jeffrey, 1 980), while the shift at St. 55 implies an older phytoplankton stock, and higher grazing pressure from zooplankton. For stations further offshore and in or adjacent to the jet (St.39 and 63,


Figure 5), the chlorophyll a specific absorption curves appear to contain significant effects due to the presence of detritus and/ or phaeopigments : 1 ) ap(436) :ap (675) ratios were 5.03 and 8.37 respectively ; and 2) a relative absorption maximum was apparent at about 410 nm at both stations 0 .200 1=1 0 :0 0.150 p., 0 Cll 0.100 () ..... 'H ..... 0 .050 p., (/) . St.39 St.63 -St.77 350 450 550 650 750 850 Wavelength (nrn) 2 8 Figure 5 Specific absorption coefficient curves for just below the surface at offshore Stations 39 63, and 77 [Chl a] is 0 24, 0 28 and 0 10 J.Lg/1, respectively. As significant as detritus appeared to be in the particulate absorption curves for the offshore stations in and near the jet, it was even more important in waters offshore and well clear of the jet. Figure 5 also presents the ap(A.,z) c urve for St.77 with [Chi a] = .103 mg/ m and the ap(436):ap(675) ratio = 8 .23. This value is not appreciably diff e rent than that of Sta. 63, but at this station with about 64 % lower


29 {Chl a] pigment absorption features are virtually obscured by the relatively strong detrital signature Comparison of Measured Rn()..) and Model Results Measured remote sensing reflectance curves for the inshore stations (solid curves in Figures 6 and 7) show clearly the dominance of chlorophyll a as the primary pigment, with reflectance minima near 440 nm and 665 nm (the expected reflectance 0.004 't1 1":1 .......... J 0 .003 II rt.l 0 .002 J.. 0.001 v .. \ Station 46 Measured Modeled {\_ 350 450 550 650 750 Wavelength (nm) Figure 6. Measured vs modeled R n(A) at Station 46 minimum at 675 nm due to chlorophyll a has been partially "filled by the fluorescence peak centered at 685 nm). 850


0 .004 0.003 0.002 0.001 450 550 650 Wavelength (nm) Figure 7. Measured vs modeled R,.(A) at Station 55. Station 55 -Measured ..... Modeled 750 850 Measured remote sensing reflectance curves (solid curves in Figure 8) for the offshore stations have shapes manifesting the relatively weaker influence of 30 phytoplankton pigments No significant reflectance minima is observed near 440 nm. The fluorescence peak at 685 nm is not significant for any of the curves, while reflectance in the blue is higher than that in the red by one to two orders of magnitude, indicating a predominance of backscattering over absorption at short wavelengths, especially for Station 77 Negative curvature together with reduced reflectance below about 430 nm at Station 39 suggests the presence of CDOM, especially considering that the particulate specific absorption curve for that station also reflects a decrease in the same spectral region


31 0.008 -Measured '0 Modeled (39, 63) Pil 0.006 '-..... St. 77 --Modeled (77) y \ \ \' \ II Ul 0 .004 '-< .. St. 39 0.002 ............ St. 63 350 450 550 650 750 850 Wavelength (nrn) Figure 8. Measured vs modeled Rn(A.) at Stations 39,63, and 77.


32 DISCUSSION Model curves of Rn(A.) were developed by substituting into Eq. 3 the measured curves for ap(A.,z), literature curves for btw(A.,z) and aw(A), and variable values for bbp and acooM Since B2 was initially assumed to be 0 for all stations except Station 77, B and the coefficients c. and c; in Eq.4 were adjusted in a predictor-corrector mode until close agreement was reached with measured R..(A.) curves. Backscattering at St. 77 was given a A, dependence initially due to its low pigment concentration. Model parameters for all stations are shown in Table 2. It was found that applying a non zero slope to the particulate backscattering at all the stations provided a much better TABLE 2. Model parameters for stations reported. [Chl a] coeffs bbp coeffs. Station# (mg/m3 ) c. c; B. B2 Q-factor 39 0.200 .036 .016 .0018 1.18 3 7 46 10.000 .120 -.018 0014 0.69 4 3 55 13.500 .164 -.018 0068 0.60 4.4 63 0 250 .040 015 0005 0.95 Q(A.) 77 0.090 .005 -.012 .0018 1.04 Q(A) btw(A.): Smith & Baker (1981) = c.*exp(G;(A.-400)) a,.(A.): Stas 46,55 Composite (see discussion) bbp(A.) ""' B1 *( 400/A.)" B2 Stas 39,63 77 Smith and Baker (1981)


33 fit to the measured curves, particularly in the higher pigment stations. As seen in Table 2, this slope was 0.69 and 0.60 for Stations 46 and 55, respectively Bricaud et al. (1983) measured backscatter for three axenic phytoplankton cultures (particle size range 6-15 #'m), and indicated a slight negative spectral slope for two of the three (their Figure 5). In addition, they suggested that even highly productive waters may contain significant amounts of detritus, with higher backscattering efficiencies than phytoplankton and > 0. In light of this, we introduced some spectral slope in the particulate backscatter expression, but due to the high absorption the slope remains considerably less than one at these two stations. Particulate backscatter slope at Stations 39 and 63 are higher (1.18 and 0.95, respectively), reflecting the reduced absorption effect of the relatively low pigment concentrations at these stations. Water absorption coefficient curves from both Smith and Baker (1981) and Tam and Patel (1979) were used in the modeling effort. Smith and Baker's curve provided closer agreement for wavelengths shorter than about 570 nm, while Tam and Patel's curve provided closer agreement for wavelengths longer than about 570nm, at the two in-shore, high pigment stations. The best agreement was obtained with the composite curve shown in Figure 9, where the dotted curve shown provided a transition between the candidate literature curves. This composite curve was derived using data from Sta. 46, and it was subsequently used for modeling data from Stations 46 and 55. The water absorption coefficient curve from Smith and Baker was used at the remaining stations. Our criteria for choosing Smith and Baker or the composite curve was based on the fit of the modeled reflectance curve to the measured curve in the 570 to about


0 .600 0 .500 .......... Smith & Baker (1981) ...-! I 0.400 Tam & Patel (1979) s ..__, Composite !=: 0.300 0 ..... P. f... 0 0 .200 C"l) ..0

35 regions: 1) above 655 nm where fluorescence due to chlorophyll a is significant, and 2) between 475 and 540 nm. The loose fit between 475 and 540nm was at first thought to be due to accessory pigment effects. However, the effects of such pigments should be manifest in both the specific absorption curve and the field reflectance curve unless there was significant small-scale patchiness, in which case, non-synoptic sampling could produce the observed differences. At this station Spectron measurements of L.,(A) were obtained, while holding station, within 5 minutes of the collection of water samples for pigment analysis, so it is unlikely that patchiness caused the observed differences. Since the model curve is higher in this region than the measured curve, we sought a source that might have reduced the specific absorption coefficient, thus increasing the modeled reflectance. Bioluminescence (filter pad glow) was observed on the ftlter pad used for llp(A.,z) measurements at one offshore, nighttime station so we investigated the possibility that bioluminescent organisms captured on the ftlter pad might have caused the required reduction of the absorption coefficient. Bioluminescence is suppressed by continual daylight so its effects would not be manifest in measured Rn(A.) curves, thus the model curve would appear higher in the source region (Figures 6 and 7) Peak emission wavelengths for bioluminescence are primarily in the blue green between 440 and 500 nm (T.Cowles, pers comm. ; Widder et al.., 1989 ; Latz et al.., 1988) Of the organisms identified as present in the region of Station 55 (Mackas et al., 1991), the adult euphausiids are a possible bioluminescent candidate, but it is unlikely we would have overlooked an animal of


this size on the filter pad when we transfered it from the filter apparatus to the measurement system. Dinoflagellates are a frequent source of bioluminescence, and although Chavez et al. (1991) concluded that dinoflagellates were never significant during our study, they could have been present in sufficient numbers to have been captured on the pad and then luminesced during our measurement procedure. Gelatinous zooplankton were abundant nearshore, and a number of these species luminesce (see below). 36 The mere presence of an appropriate luminescent organism is not sufficient to explain our result. The additional requirement is that the organism be in a "bioluminescent mood" not a normal condition in daytime sunlight. Several investigators (Hamman et al. 1981; Sadovskaya and Filimonov, 1985) have reported that numerous bioluminescent dinoflagellate species, when placed in the dark after extended exposure to normal levels of daylight regained most of their luminescent capability within about 30 minutes. Both of the cited investigations reported their results in terms of absolute quanta emitted or quanta emitted relative to a reference value. The recovery time was essentially the same for log phase cultures and for freshly caught natural populations. It is not unreasonable to assume similar recovery times in other neritic pelagic organisms. At Station 55 a time lapse o f 41 minutes between water collection and filtering occurred. The water samples were stored in the dark until filtered under low ambient light conditions and measurements were made s o the organisms could have become dark-adapted. Sudden exposure to the bright light of the spectral transmissometer during the measurement period may also have


37 contributed to triggering the bioluminescent response (Sweeney et al . 1983; see also Appendix 3). Using the mode l parameters derived for Sta. 55 (see Table 2), a p (A,z) was derived from Eq. 3, with the measured curve serving as R..(A). The "derived" curve is compared to the measured in Figure 10 The difference between the ......... ........ I E '-" 0 ....... 0.. I-< 0 [/J ...0 () ....... ...... ....... () Q) 0.. (/) 0.018 0.015 0.012 0.009 0.006 0.003 0.000 350 450 550 Station 55 -Measured Derived 650 Wavelength (nm) 750 850 Figure 10 Derived vs measured specific absorption coefficient curves for surface waters at Station 55. curves in the 475 to 540 nm region is clear and suggestive of a reduction in absorption for the measured curve. The spectral form of a bioluminescent output is typically a single peak emission wavelength with variable bandwidth (both apparently species dependen t ); this shape output, with a peak at about 510nm, would account for the difference in the two curves.


38 Of the organisms mentioned above and specifically identified at Stations 46 and 55, only Euphausia pacifi c a is known to be b i oluminescent but their peak wavelength is about 470nm and is not in keeping with the implied effect here. W i dder et al (1983) presents a lengthy list of bioluminescent species in nine Phyla, in which only some members of the phylum Cnidarea (jellyfish soft corals ) exhibit emission at about 510nm Benthic soft corals (e.g. Renilla kollikeri, an anthozooan medusa) reproduce through planula larvae in the water column, but it is not known if the larvae are luminescent (T Hopkins, pers comm ) The pos si bility that pigment changes might occur in th e interval between water collec ti on filtering and measurement can not be ruled out, but contamination by bioluminescence i s implied both by the visually glowing pad and the model-field curve differences. There are a number of ways to a v oid such contamination. The opportunity for the organisms to dark-adapt should not be provided; a reasonable approach would be to hold and filter the sample in clear bottles and in normal ambient light levels Agitation of the sample prior to flltering or measurement will exhaust those organ i sms ab l e to luminesce. Since the luminescent process requires oxygen (G. Vargo pers comm.; Hastings 1983) bubbling nitrogen throught the sample will strip the oxygen out and should prevent a luminescent display Agreement between derived and measured Rn(A) curves for Sta. 46 (F i gure 6) was similar to that for Sta 55 (see Figure 7) above about 450nm with an apparent "bioluminescence effect" between 475 and 540 nm. Below 450nm however the two stations differ markedly A t Station 46 the model curve exhibits a minimum at about


39 440nm, while the measured curve minimum is at about 432nm Station 55 shows the opposite relation, with the model curve minimum at about 418nm and the measured curve at 434nm. These variations are likely due to the highly stratified nature of the pigments at both these stations. Station 55 surface [Chi a] is 10.6 ,...g/1, while at 10m it is 17.6 1-'g/1. Station 46 has 7.6, 9.9, and 10.8 1-'g/1 at Om, 10m, and 19m, respectively. The water-leaving radiance contains the signature of pigments down to the first attenuation depth (that depth at which the downwelling irradiance is 1/e times the surface value) At St. 55 (see Figure 11), the ratio of particulate absorption at 10m to that at the surface shows a clear reduction of absorption from about 405 to 430, while the ratio between water-leaving radiance derived from the measured Rn and that from the model shows a clear increase in the same region Thus the measured Lw appears to contain the signature of the reduced absorption at depth, while the model, which used the surface specific absorption curve, does not. The first attenuation depth corresponds to 1/K,. (lithe downwelling attenuation coefficient) which is approximately the average cosine divided by the total absorption coefficient (p./a..). If we take a..(410}=.307 from the modeled surface absorption coefficient at St.55, the first attenuation depth could be no more than 3.25m, calling into question the above interpretation. No clear explanation is available for the observed difference, but the evidence suggests a localized intrusion of the high chlorophyll subsurface layer into the surface waters.


40 2 0 0 0 t-+---+--t---Jf---+--+--+---+--t--+--+--f--+--+--+---+--+--1-----i--+ 4 ,........._ ....... 1.500 Q) 3 "'d ....... 0 -.... .. s : : ,........._ -: s '-" J 0 .'-" 1.000 ...... 2 0. '-... : ... (!j -\ .. "'/ '-... "'d ....... ,........._ Q) s ...... ap ratio 0 '-" J 0.500 1 ...... -Lwratio '-" 0. 1.\l 0 .000 0 350 450 550 650 750 850 Wavelength (nm) Figure 11. Comparison of the ratios of derived water-leaving radiance (L ... (field)/L..,(model)) and measured particulate absorption (tZp(10m)/tZp(Om)), for Station 55 Indicates the apparent influence on water-leaving radiance by sub-surface pigment stratification All measured R,.(A) curves for inshore stations manifested a peak at 685 nm where c hlorophyll a fluorescence typically oc c urs. Since fluorescence was not accounted for by the model the difference between modeled and measured curves in this spectral region is attributed to chlorophyll a fluorescence. Similar differences in this spectral region have been noted and discussed by Carder et al. (1985) and references cited therein Model R,.('A) curves for the offshore stations (Figure 8) provided close agreement with measured curves for wavelengths longer than 550 nm and for wave l engths shorter than about 415 nm. Comparison of the differences between our measured and


41 modeled curves with similar differences caused by water Raman scattering in Monte Carlo simulations of oligotrophic environments (Stavn, 1990), suggests that our differences (8 to 13% at 470 nm) are at least in part due to water Raman scattering (not included in our model). Stavn found the Raman contribution to be 13. 5% of the total reflectance at 470 nm for a simulation of Sargasso Sea reflectances His calculations used the Marshall and Smith (1990) Raman cross sections for stimulation at 488 nm and assumed a X."' spectral dependence for Raman scatter at other wavelengths The sharp decline in incoming solar irradiance for excitation wavelengths shorter than 365 nm and the presence of increased absorbance by CDOM and detritus at short wavelengths could account for the apparent truncation of the Raman effect at wavelengths shorter than 410-415 nm Incident photons at 365 nm have a Raman scattering response at about 415 nm The modeling process for Station 77 included a spectral Q-factor obtained from Curt Dav i s of the Jet Propulsion Laboratory. Q-factor measurements obtained w i th a 0 degree viewing angle are often smaller at 410 nm than at 550 nm, and they are likely to behave in a similar spectral manner at other viewing angles, thus an application of spectral Q-factor to model our non-vertically measured reflectance curve is appropriate. The spectral character in the blue of the model curve for Sta. 77 is we believe, du e to low signal to noise ratios in the specific absorption coefficient of particles. After looking closely at the raw data we have concluded that the light source was operating at a reduced level at this station Since the source was a tungsten lamp


which emits a low radiance in the blue anyway, when the light output dropped the blue end of the spectrum was more heavily affected than the green or red portions. The log ratio of two signals near noise level may fluctuate dramatically, and the modeled reflectance curve will mirror that fluctuation. 42 Particulate backscattering at Sta. 77 was initially assumed to be dependent on liA, due to the very low pigment concentration, and subsequent modeling showed that the backscattering slope near one is necessary to bring the blue end of the model curve as high as the field curve values The Raman effect is not observed at the inshore stations due to the large CDOM absorption coefficients (C1 in Table 2) and the relatively large particulate backscattering coefficients Table 2). For offshore stations, however, optical model development activities will require that Raman scattering be considered as well as all sources of fluorescence (e g CDOM and pigments). Future remote sensing algorithms should also include terms that adequately address these transpectral scattering and fluorescence phenomena To illustrate the potential effect of CDOM fluorescence and water Raman scattering on the model, we calculated their respective contributions at two offshore stations after first modeling the other parameters, and added the results to the model curves. This is not a rigorous addition of these parameters to the modeling process, since the reabsorption of the emissions of both these sources by other constituents would mitigate their final contribution, but it does indicate their potential importance.


43 Figure 1 2 presents the measured and modeled reflectance curves at Station 39, and a second model curve which is the sum of the first model and the calculated source parameters Note that the difference in the 410 to 500 nm region is greatly 0 006 -r-+--+--t---11---t--+-t---+--+----+--+--+--+---+-+-+--+----+ 0 .005 -Measured "0 0.004 "-J 0 .003 II r:ll 0.002 0 001 .. .... .. . ....... .. \, ---.. ....... .\ "<,. . .. : .. . ----Modeled (modified) Model + fluor + Raman \ 0.000 350 450 550 650 750 850 Wavelength (nm) Figure 12. Reflectance curves at Station 39 showing the effect of adding CDOM fluorescence and water Raman scattering to the model. The difference between field and model curves is much reduced in the blue wavelengths. reduced and the curves nearly overlay along the steeply sloping r egion dominated by water absorption Adjustments to other model parameters would compensate where the model exceeds the measured curve As an example of adjusting other paramete r s Figure 13 presents reflectance curves as in Figure 12, but for Station 63. We can see that the model with CDOM fluorescence and water Raman added is much too high in the 400 to 430 nm region.


'd --.....__ l II IZI J.-. 0 .003 0 .002 0.001 :., -Measured ----Modeled Model + fluor + Raman .... ""' 350 450 550 650 750 850 Wavelength (nm) 44 Figure 13 Reflectance c urves at Station 63. The model is shown, and the model with CDOM fluorescence and water-Raman scattering added resulting in a large excess of modeled reflectance from 400 to 430 nm Figure 14 presents the same curves as in Figure 13, but the model was calculated using 10 % more CDOM ab s orption. The modified model curve is now below the measured curve at 420 nm where before it had met the curve. The modified model with fluorescence and Raman added is still too high in the blue, but the excess i s much reduced These over-simplified illustrations show that if these source parame t ers were integrated i nto the modeling process, we could interactively adjust all the parameters, and we would be much closer to resolving the difference s between model and field reflectance curv e s for clear-water stations such as 39 and 63


"'d '-... l II C/l 1-i 0 .003 0.002 0.001 Y M --... _ -\ \ .\ -Measured Modeled (modified) Model+ fluor+ Raman 0. 0 00 __ .+--. !:!J u::..--+--+--+---+---+---1-350 450 550 650 750 850 Wavelength (nm) 45 Figure 14. Reflectance curves for Station 63. The model curve is calculated using 10% more CDOM absorption than the model in Fig 13. Recalculated CDOM fluorescence and water-Raman scattering are added to the modified model, resulting in a much reduced excess of reflectance in the 400 to 430 nm region. Potential Modeling Errors There are five areas where potential errors may be introduced affecting the modeling process : 1) embedded errors in published values of the absorption coefficient of pure water; 2) method errors in measurement of filter pad optical density (for particulate absorption coefficients); 3) compensating errors in model parameterization; 4) errors associated with the calculation of the measured R,.('A) curves; and 5) physical perturbations in the sampling regime The best available data for absorption of the clearest natural waters is a weighted average of the results of numerous attempts, over many years and by diverse


methods, to measure this property (Smith and Baker, 1981). The high slope of the water absorption coefficient above about 570nm is sensitive to small changes in spectral calibration, and the absolute values of the region between 400 and 500 nm are difficult to attain without contamination 46 H0jerslev and Trabjerg (1990) have found that absorption by pure seawater can vary as much as .003 m o c in the temperature range 10 to 30 C, for wavelengths 400 to 600nm. In general the absorption increased as temperature increased. They calculated that the derived pigment concentration using CZCS algorithms could vary as much as 23% in the range 0.1 to 10 mg/nt, due to this variability alone A temperature correction factor applied to the water absorption curve prior to using it in the modeling effort might result in a different pigment concentration being required This correction could be particularly important in oligotrophic waters, where water absorption may be greater than that of pigments, but it would have a rather smooth spectral effect and would not drastically affect the shape of R,. curves. Time delays between water collection filtering, and optical density measurement may introduce bioluminescence from dark adapted organisms. This can be largely avoided by rapid processing. Incorrect choice of the spectral slope parameter for absorption by CDOM or for particulate backscatter in the absence of measured values, can contribute to modeling errors by requiring erroneously high or low backscatter or absorption by other constituents to compensate for the error. The use of an unweighted (with depth) specific absorption coefficient curve in an area where significant near-surface pigment


stratification exists will result in mismatches of the reflectance curves in specific spectral regions. 47 The absence of consideration of CDOM fluorescence and water Raman scattering in models will prevent closure in areas with low pigment and relatively high CDOM concentrations. The reflected skylight correction may be a factor in matching measured and modeled reflectance curves, since too little removal will result in an erroneously high measured curve. Skylight removal is dependent on numerous factors, any one of which could cause the removal of less (or perhaps more) skylight than necessary. When natural light is incident on the sea surface the reflected rays are partially polarized. Using a vertically polarized ftlter to minimize the horizontally polarized component significantly improves the correction for skylight reflectance in the upwelling signal, since the percent reflectance of the vertical component is easily computed from the relevant Fresnel reflection equation. We have tested this procedure and found it to greatly reduce the variability in reflected skylight correction, thus prompting us to use this procedure in future measurements. We are also now using observation zenith angles near 15 degrees, since the sea-surface reflectance is nearly invariant for angles less than 20 degrees, even for windspeeds up to about 16 m / s (36 mph) (Austin (1974a)) These changes should significantly reduce errors associated with removing reflected skylight from the upwelling radiance measurement. They also will permit us to compare our results more directly with subsurface irradiance reflectance measurements that are typically measured in the nadir direction.


48 A measurement of downwelling irradiance using a non-horizontal grey card will result in an erroneous El'A) curve due to the cosine effect on reflectance of the direct sunlight. In general, this is only a significant problem when ship motion is erratic and vigorous, as small, slow ship rolls can be adjusted to by the operator. Small-scale horizontal patchiness or near-surface stratification of pigments may result in modeling difficulties if water samples are collected from other than the water viewed by the radiometer or stratification is ignored in the modeling effort.


49 CONCLUSIONS Recent studies indicate a Raman contribution to water leaving radiance may approach 13% in low pigment waters ([Chl a < 0.3 mg/m3), particularly when UV absorption is low. Peacock et al. (1990) suggested that water-Raman scatter could be important in waters of the California Coastal Transition Zone, and this paper provides evidence from additional stations in that region in support of their hypothesis. Peacock et al. (1990) also suggested that CDOM fluorescence could play a role in R,. modeling, and Carder et al. (1991) came to a similar conclusion for the area they studied In reflectance model development, care must be taken to consider the possible effects of these two internal light sources Time required for the dark-adaption of bioluminescent organisms is on the order of 30 minutes, when measuring particulate absorption via the filter pad method of Mitchell and Kiefer (1988a). Potential errors due to bioluminescence may be avoided by rapid processing In waters containing high pigment concentrations and significant pigment stratification near the surface, a weighted (with depth) lip* curve might be required to accurately model the spectral effects of the absorption regime


50 Methods of measuring water-leaving radiance which include a vertical polarizing fllter and small observation zenith angles will help to minimize the reflected skylight component and errors associated with removing it from the upwelling radiance measurement. The small zenith viewing angles will also provide more consistency with subsurface reflectance values Since satellite-derived water-leaving radiance values will be acquired for observational zenith angles as large as 40, it is incumbent upon field researchers to begin to gather data for Q factors that are relevant for all possible angles. Radiance distribution data gathered as a function of wavelength will be of great value in answering questions about the spectral behavior of the Q factor as a function of the variable solar and observational angles practical for satellite observation of ocean color. Further study of the absorption properties of pure water may be needed to resolve the diversity of measured values for this important parameter, and to bring the standard curve up to date.


LIST OF REFERENCES Abbott, M R., 1988. Satellite observations of SST during the Coastal Transition Zone Experiment in 1987 and 1988 (Abstract) Eos Trans. AGU. 69(44) : 1259 Altabet, M.A., 1990 Organic C, N, and stable isotopic composition of particulate matter collected on glass-fiber and aluminum oxide filters Limnol. Oceanogr 35(4): 902-909. 51 Andrews, C C., D.M. Karl, L.F. Small, and S W. Fowler, 1984. Metabolic activity and bioluminescence of oceanic faecal pellets and sediment trap particles. Nature. 307: 539-541. Austin, R.W., 1974a. Inherent spectral radiance signatures of the ocean surface. Ocean Color Analysis, SIO Ref. 7410, Scripps Inst. Oceanogr. Austin, R.W 1974b The remote sensing of spectral radiance from below the ocean surface, p.317-344 In N G. Jerlov and E S. Nielsen, [eds ] Optical Aspects of Oceanography. Academic. Austin, R.W., 1979. Coastal Zone Color Scanner radiometry. Ocean Optics VI, Proc. SPIE 208 : 170-178. Batchelder H.P., and E. Swift, 1989. Estimated near-surface mesoplanktonic bioluminescence in the western North Atlantic during July 1986. Limnol. Oceanogr 34(1): 113128. Bozin, S A., and V S. Filimonov, 1985. Spontaneous bioluminescence of dinoflagellates in Vostok Bay, Sea of Japan Oceanology. 25(3): 395 397. Bricaud, A., A. Morel, and L. Prieur, 1981. Absorption by dissolved organic matter of the sea (yellow substance) in the UV and visible domains. Limnol. Oceanogr. 26: 43-53. Bricaud, A., A Morel and L. Prieur 1983 Optical efficiency factors of some phytoplankters. Limnol. Oceanogr. 28(5) : 816-832.


Bristow, M.P.F., 1978 Airborne monitoring of surface water pollutants by fluorescence spectroscopy. Remote Sensing Environ. 7: 105. Buskey, J.E., and D.E. Steams, 1991. The effects of starvation on bioluminescence potential and egg release of the copepod Metridia long a. J. Plankt Res. 13(4): 885-893 52 Carder, K.L., and R.G. Steward, 1985. A remote sensing reflectance model of a red tide dinoflagellate off West Florida. Limnol. Oceanogr 30(2): 286-298. Carder, K.L., R.G. Steward and P R. Payne, 1985 Solid state spectral transmissometer and radiometer Opt. Engr 24(5): 863-868 Carder, K.L., R G Steward, T.G. Peacock, P R Payne, and W. Peck, 1988. Spectral transmissometer and radiometer : design and initial results. Ocean Optics IX, Proc SPIE. 925 : 189-195. Carder, K.L., R G. Steward, G.R. Harvey, and P.B. Ortner, 1989 Marine humic and fulvic acids: Their effects on remote sensing of ocean chlorophyll. Limnol. Oceanogr. 34(1): 68 81. Carder K .L., S.K. Hawes, K.A. Baker, R.C. Smith, R.G. Steward, and B.G. Mitchell, 1991. Reflectance model for quantifying chlorophyll a in the presence of productivity degradation products Journ. Geophys Rsrch. 96(Cll): 20599 20611. Chavez F.P., R .T. Barber, P M. Kosro, A Huyer S.R. Ramp, T P. Stanton and B R deMendiola, 1991. Hori z ontal transport and the distribution of nutrients in the Coastal Transition Zone off Northern California: Effects on primary production phytoplankton biomass and species composition. Journ Geophys. Resrch. 96(C8): 14833 14848. Cleveland, J.S., and M J. Perry, 1987. Quantum yield, relative specific absorption and fluorescence in nitrogen -limited Chaetoceros gracilis. Mar Biol. 94: 489-499. Coble, P.G., S A Green N.V. Blough, and R.B. Gagosian 1990. Characterization of dissolved organic matter in the Black Sea by fluoresc e nce spectroscopy. Nature. 348: 432-435. Collins, D.J., J.A. Bell, R. Zanoni, I.S. McDermid, J.B. Breckinridge and C.A. Sepulveda, 1984. Recent progress in the measurement of temperature and salinity by optical scattering. Ocean Optics VII, Proc SPIE. 489: 247-269.


Cullen, J .J., C M. Yentsch, T.L. Cucci, and H.L. Macintyre, 1988. Autofluorescence and other optical properties as tools in biological oceanography Ocean Optics IX, Proc. SPIE. 925: 149-156. 53 Donard, O.F.X., M. Lamotte, C. Belin, and M. Ewald, 1989. High sensitivity fluorescence spectroscopy of Mediterranean waters using a conventional or a pulsed laser excitation source. Mar. Chern 27: 117-136. Duntley, S.Q., 1942 Optical properties of diffusing materials. J. Opt. Soc. Am. 32: 61-70. Eisenberg, D., and W. Kauzmann, 1969. The structure and properties of water, Oxford University Press Esaias, W.E., and H.C. Curl Jr., 1972. Effect of dinoflagellate bioluminescence on copepod ingestion rates. Limnol. Oceanogr. 17(6): 901906. Ewald, M. H H. Stabel, and C. Belin, 1986. Composes organiques d'origine biogenique dans l'eau de Ia zone euphotique en Antarctique etudies directement par spectrofluorimetrie. C.R. Acad. Sci. Paris. 302(II): 883 -886. Filimonov, V.A.,and G.M. Sadovskaya, 1986. Photoinhibition of phytoplankton bioluminescence. Oceanology. 26(5): 621-622. Fogg, G.E 1966. The extracellular products of algae. Oceanogr Mar. Bioi. Ann. Rev. 4: 195-212. Gagosian, R B., and D.H. Stuermer, 1977. The cycling of biogenic compounds and their diagenetically transformed products in seawater Mar. Chern. 5 : 605-632 Gordon, H.R., 1976. Radiative transfer: a technique for simulating the ocean in satellite remote sensing calculations. Appl. Opt. 15: 197 4-1979. Gordon H R 1979. Diffuse reflectance of the ocean: The theory of its augmentation by chlorophyll a fluorescence at 685 nm. Appl. Opt. 18: 1161-1166. Gordon, H R., 1986. Ocean color remote sensing : Influence of the particle phase function and the solar zenith angle. (Abstract) Eos Trans. AGU. 67(44) : 1055.


54 Gordon, H.R., O.B Brown, and M.M. Jacobs, 1975. Computed relationships between the inherent and apparent optical properties of a flat homogeneous ocean. Appl. Opt. 14: 417-427. Gordon, H.R., and A.Y. Morel, 1983. Remote assessment of ocean color for interpretation of sattelite visible imagery: a review. Springer Verlag. Gordon, H.R. O.B. Brown, R .H Evans, J.W. Brown, R.C. Smith, K.S. Baker and D.K. Clark, 1988. A semianalytic radiance model of ocean color. Journ. Geophys. Resrch. 93(09): 10909-10924. Hamman, J.P., W.H. Biggley, and H H. Seliger, 1981. Photoinhibition of stimulable bioluminescence in marine dinoflagellates. Photochem. Photobiol. 33: 909-914. Harvey, G.R., D.A Boran, L.A. Chesal, and J .M. Tokar, 1983. The structure of marine fulvic and humic acids. Mar. Chern. 12: 119-132 Hastings, J.W., 1983. Chemistry and control of luminescence in marine organisms. Bull. Mar. Sci. 33(4): 818-828. Hawes, S.K., K.L. Carder, and G.R. Harvey, 1992. Quantum efficiencies of fulvic and humic acids: effects on ocean color and fluorimetric detection. Ocean Optics XI, Proc SPIE. 1750: in-press. Hayase, K and H. Tsubota, 1985. Sedimentary humic and fulvic acid as fluorescent organic materials. Geochem. et Cosmochem. Acta. 49: 159 163. Hoge, F.E., and R.N. Swift, 1981. Airborne simultaneous spectroscopic detection of laser-induced water Raman backscatter and fluorescence from chlorophyll a and other naturally occuring pigments Appl. Opt. 20(18): 3197-3205. H0jerslev, N.K., 1975. A spectral light absorption meter for measurements in the sea. Limnol. Oceanogr. 20(6): 1024 1034. H0jerslev, N.K and I. Trabjerg, 1990. A new perspective for remote measurements of plankton pigments and water quality. Report no.51, Geophysical Institute, University of Copenhagen. Huyer, A., P.M. Kosro, J. Fleischbein, S.R. Ramp, T Stanton, L. Washburn, F.P. Chavez, T .J. Cowles S D Pierce, and R.L. Pierce, 1991. Currents and water masses of the Coastal Transition Zone off Northern California, June to August 1988. Journ. Geophys. Rsrch. 96(C8): 14809 14831.


Ivanoff, A., N Jerlov, and T.H. Waterman, 1961. A comparative study of irradiance, beam transmittance, and scattering in the sea near Bermuda," Limnol. Oceanogr. 6: 129-148. Jackson, G A and P.M. Williams, 1985. Importance of dissolved organic nitrogen and phosphorus to biological nutrient cycling. Deep Sea Rsrch 32(2) : 223-235. Jeffrey, S W., 1980 Algal pigment systems, p 3358 In P.G. Falkowski [ed.], Primary Productivity in the Sea, Environ. Sci. Resrch. Ser. 19. Plenum Jensen, A ., 1984 Excretion of organic carbon as a function of nutrient stress, p. 61 -72. In 0 Holm-Hansen, L. Bolls, and R. Gilles [eds.], Marine Phytoplankton and Productivity, Lecture Notes on Coastal and Estuarine Studies 8 Springer Verlag. Jerlov N.G., 1968 Optical Oceanography, Elsev ier Oceanography Series, 5 Elsevier. 55 Kadko, D C L. Washburn and B. Jones, 1991. Evidence of subduction within cold fllaments of the Northern California Coastal Transition Zone Journ. Geophys Resrch 96(C8): 14909-14926. Kalle, K 1949 Fluoreszenz und gelbstoff im Bottnischen und Finnischen Meerbusen. Dtsch. Hydrogr Z 2 : 117124. Kiefer, D A., 1973a Chlorophyll a fluorescence in marine centric diatoms: Responses of chloroplasts to light and nutrient stress. Mar Biol. 23: 39 46 Kiefer, D.A. W S. Chamberlain, and C.R Booth 1989. Natural fluorescence of chlorophyll a : Relationsh i p to photosynthesis and chlorophyll concentration in the western South Pacific gyre Limnol. Oceanogr 34(5): 868-881. Kirk J .T.O. 1983. Light and photosynthesis in aquatic ecosys t ems Cambridge Univ. Press Lapota D ., and J R Losee 1984 Observations of bioluminescence in marine plankton from the Sea of Cortez J Exp Mar Biol. Ecol. 77: 209-240. Latz, M.I. T M Frank, and J F Case 1988 Spectral composition of bioluminescence of epipelagic organisms from the Sargasso Sea Mar Biol. 98(3): 441-446.


Lee Z., K.L. Carder, S.K Hawes, R G Steward, T G Peacock, and C.O. Davis, 1992. An interpretation of high spectral resolution remote sensing reflectance. Ocean Optics XI, Proc SPIE. 1705: in press. Leonard, D.A., B. Caputo, and F.E. Hoge, 1979. Remote sensing of subsurface water temperature by Raman scattering. Appl. Opt. 18: 1732. Long, D A., 1977. Raman Spectroscopy. McGraw-Hill. Lorenzen, C .J., 1966. A method for the continuous measurement of in-vivo chlorophyll concentration. Deep Sea Resch. 13: 223-227 56 Mackas, D.L. L. Washburn, and S.L. Smith, 1991. Zooplankton community pattern associated with a California Current cold filament. Joum Geophys. Resrch. 96(C8): 14781-14797. Marshall B.R and R C Smith, 1990 Raman scattering and in-water ocean optical properties. Appl. Opt. 29(1): 71-84. Mitchell B.G., and D.A. Kiefer, 1988a. Chlorophyll g specific absorption and fluorescence excitation spectra for light limited phytoplankton Deep Sea Res. 35(5): 635-663. Mitchell, B.G., and D A Kiefer, 1988b. Variability in pigment specific particulate fluorescence and absorption spectra in the N.E.Pacific Ocean Deep Sea Res. 35(5): 665 689. Morel A. 1974. Optical properties of pure water and pure sea water p 1-24 In N.G. Jerlov and E.S. Nielsen [eds .], Optical Aspects of Oceanography. Academic. Morel, A., and L. Prieur, 1977. Analysis of variations in ocean color. Limnol. Oceanogr 22(4): 709-722. Morel, A., 1988. Optical modeling of the upper ocean in relation to its biogenous matter content (Case I waters). Journ Geophys. Resrch. 93(C9): 1 074910768. Morin, J.G. 1983. Coastal bioluminescence : Patterns and functions. Bull Mar. Sci. 33(4): 787-817. Nissenbaum A and I.R. Kaplan, 1972. Chemical and isotopic evidence for the in situ origin of marine humic substances. Limnol. Oceanogr. 17(4) : 570 582.


Olson, R.J., S.W. Chisholm, E.R. Zettler, and E.V. Armbrust, 1988 Analysis of Synechococcus pigment types in the sea using single and dual beam flow cytometry. Deep Sea Res. 35(3): 425-440. Orzech J.K., and K H. Nealson, 1984. Bioluminescence of marine snow: its effect on the optical properties of the sea. Ocean Optics VII. Proc. SPIE Peacock T.G., K L. Carder, and R.G Steward, 1988. Components of spectral attenuation for an offshore jet in the Coastal Transition Zone. (Abstract) Eos Trans AGU. 69(44): 1125. Peacock, T.G., K.L. Carder, C.O Davis and R.G. Steward, 1990. Effects of fluorescence and water Raman scattering on models of remote sensing reflectance. Ocean Optics X Proc SPIE. 1302: 303-319. Petzold, T.J., 1972 Volume scattering functions for selected ocean waters. SIO Ref 72-78. Scripps Inst. Oceanogr. 57 Pocklington, R., 1977. Chemical processes and i nteractions involving marine organic matter. Mar Chern 5 : 479-496. Preisendorfer, R.W., 1961. Application of radiative transfer theory to light measurements in the sea. UGGI Monogr no 10 Symp. Rad Energy in the Sea. 11-30. Prezelin B.B, and A C Ley, 1980. Photosynthesis and chlorophyll a fluorescence rhythms of marine phytoplankton. Mar. B i ol. 55 : 295-307. Roesler, C S., M J. Perry, and K.L. Carder, 1989 Modeling In Situ Phytoplankton Absorption from Total Absorption Spectra. Limnol. Oceanogr. 34(8) : 1510-1523. Sadovskaya, G M and V.S. Filimonov 1985. Factors determining the diurnal dynamics of phytoplankton bioluminescence. Oceanology 25(5): 642-646. Seliger, H H., J P. Hamman, and W.H. Biggley, 1981. Photoinhibition of stimulable bioluminescence in marine dinoflagellates. Photochem. Photobiol. 33(6) : 909-914. Sieburth J.McN ., and A. Jensen, 1968 Production and transformation of extracellular organic matter from littoral marine algae: A resume p.203 -223 In D W. Hood [ed ], Organic matter in natural waters. Univ. Alaska In st. Mar. Sci.


58 Sieburth, J. MeN., and A. Jensen, 1969 Studies on algal substances in the sea II: The formation of gelbstoff (humic material) by exudates of Phaeophyta. J Expl. Mar Biol. Ecol. 3: 275-289. Smith, R.C., 1974. Structure of solar radiation in the upper layers of the sea p.95 -119. In N.G. Jerlov and B S. Nielsen, [eds.], Optical Aspects of Oceanography. Academic. Smith, R.C. and K.S. Baker, 1981. Optical properties of the clearest natural waters (200-800 nm). Appl. Opt. 20(2) : 177-184. Stavn, R .H., and A.D. Weidemann, 1988a. Optical modeling of clear ocean light fields: Raman scattering effects Appl. Opt. 27(19): 4002-4011. Stavn, R .H., and A.D. Weidemann, 1988b Raman scattering effects in ocean optics. Ocean Optics IX. Proc SPIE. 925: 131-139. Stavn, R.H., 1990. Raman scattering effects at the shorter visible wavelengths in clear ocean waters. Ocean Optics X. Proc. SPIE. 1302: 94-100 Stuermer, D.H., 1975. The characterization of humic substances in sea water. Ph D. Thesis, MIT and WHOI 188pp. Sugihara, S., M Kishino, and M. Okami, 1984. Contribution of Raman scattering to upward irradiance in the sea J Oceanogr. Soc. Japan. 40: 397-404. Sweeney, B.M., D C Fork, and K. Satoh 1983. Stimulation of bioluminescence in dinoflagellates by red light. Photochem Photobiol. 37(4): 457 465. Swift, E ., E J. Lessard, and W.H. Biggley 1985. Organisms associated with stimulated epipelagic bioluminescence in the Sargasso Sea and the Gulf Stream. J Plankt. Rsch. 7(6): 831-848. Tam, A. C. and C.K.N Patel 1979. Optical absorptions of light and heavy water by laser optoacoustic spectroscopy. Appl. Opt. 18(19): 3348-3358. Traganza, B.D., 1969. Fluorescence excitation and emission spectra for dissolved organic matter in sea water Bull. Mar. Sci. 19: 897-904. Washburn L. D C. Kadko B.H. Jones T. Hayward, P.M. Kosro, T.P. Stanton S. Ramp, and T Cowles, 1991. Water mass subduction and the mass transport of phytoplankton in a coastal upwelling system. Joum Geophys. Resrch. 96(C8): 14927 14945


Widder B A. M I. Latz and J F. Case, 1983 Marine bioluminescnece spectra measured with an optical multichannel detection system. Bull. Mar Sci. 33(4): 791-810 Widder, B A. S.A Bernstein D.F. Bracher J F Case, K.R. Reisenbichler J J. Torres, and B.H. Robison, 1989. Bioluminescence in the Monterey Submarine Canyon: Image analysis of video recordings from a midwater submersible Mar. Bioi. 100(4) : 541 551. Wilson, B.B. Jr. J C Dec i us, and P C. Cross, 1980. Molecular vibrations: The theory of infrared and Raman vibrational spectra Dover. 59 Zepp R G., and P F Schlotzhauer, 1981. Comparison of photochem i cal behaviour of various humic substances in water: ill Spectroscopic properties of humic substances Chemosphere. 10(5) : 479-486




61 APPENDIX 1 WATER RAMAN SCATTERING The theory of Raman scattering in liquids has been detailed by a number of authors (Eisenberg and Kauzmann, 1969; Long, (1977); Wilson et al . (1980)) The essential points of the theory are paraphrased here without the rigorous mathematical detail to provide a basic understanding of the process and to illustrate its potential for affecting remote sensing reflectance models and measurements. Raman scattering is, like fluorescence, an inelastic scattering phenomenon. That is, radiant energy is absorbed at one wavelength and a portion of the energy is emitted at a different (usually longer) wavelength. Unlike fluorescence, which is dependent on the presence of fluorescent materials or organisms, Raman scattering is a product of interaction between photons and the molecules of the illuminated medium. The photon-molecule interaction occurs in two very distinct ways. A freely rotating molecule can be induced to change its rotational energy level, and the Raman shift (the change in wavelength of the emitted photon with respect to the absorbed photon, expressed in wavenumber (reciprocal wavelength)) of the scattered radiation is a function of the difference in energies of the two levels. These energy levels are generally quite close together, and the Raman shift is therefore small (order of 10'


62 em -). Interaction can also induce a change in the vibrational state of molecular bonds. The vibrational energy levels are generally further apart than are those for rotation, and the Raman shift associated with changes in vibrational energy level are 1 to 2 orders of magnitude larger than those due to rotational changes. The Raman scattering process may be visualized as a transfer of energy between incident radiation and a scattering system (the molecule), raising the energy of the molecule above its ground state, E., to a higher level (in this instance, "ground state" refers to the energy of the molecule associated with its normal or fundamental modes of vibration, rather than to its electronic state). This energy (ilE = -E.) may be expressed in terms of a wavenumber (vM) associated with the change in levels, as ilE = licvM (see List of Symbols for definitions). An incident photon of energy licvo is annihilated and an emission photon of lower energy lieu. (where v. = v o -vM) is simultaneously created, so the scattered radiation is of longer wavelength, in keeping with Stoke's Law. The energy change is determined by the shift in wavenumber vM, and this corresponds to a wavelength shift in the observed scattered light. In some cases, the system may already be in an elevated energy state and the incident radiation may cause a downward transition to a lower energy level. In that case the system gives up the additional energy licvM, and the emission radiation is of a higher energy than the incident radiation, since v. = v o + vM The scattered light is then of a shorter wavelength than the incident light, the so-called anti-Stokes radiation effect. The intensity of this effect decreases rapidly with increasing wavenumber shift,


63 The symmetry of a polyatomic molecule and its normal vibrational modes are important in determining its Raman activity. The presence of hydrogen bonds between molecules also contributes to the Raman scattering behavior of the system by modulating the molecular bond vibrations, and by inhibiting free rotation of the molecules so bonded. The H -0-H water molecule is nonlinear with a net electric dipole, and exhibits three fundamental vibrational modes (see Figure 15). In addition, the geometry of liquid water is that of "flickering clusters" or polymeric groups of \ \ Symmetric stretch Bending AsymDnetric stretch Figure 15. Vibrational modes of a water molecule which produce Raman scattering molecules loosely bonded together by up to four hydrogen bonds per molecule (Eisenberg and Kauzmann, 1969). The Raman scattering due to rotational changes is therefore assumed to be insignificant in water due to the rotational inhibition of these bonds (Long, 1977). Water Raman scattering is dominated by wavelength shifts associated with the symmetric and asymmetric stretching (vibrational) modes ( v1 =


64 3215 to 3400 em', and v, = 3455 em', respectively) of the molecule and the net effect of scattering through these modes produces a spectral shift of .dv -3350 em '. Natural light is unpolarized before crossing the air-sea interface, and becomes partially polarized upon entering the sea. The degree of polarization of the incident radiation is a mitigating factor in water Raman scattering, but the analysis of its effect is non-trivial, and I do not address it here


65 APPENDIX 2 FLUORESCENCE OF COWRED DISSOLVED ORGANIC MATTER Numerous studies of the nature of CDOM have been undertaken since Kalle (1949) defmed "gelbstoff'' (yellow substance, CDOM) as that portion of DOM whose absorption of light increased (monotonically) with decreasing wavelength, and laid the groundwork for characterizing this important optical component in the oceans Most investigations were aimed at the origin of CDOM found in the marine environment (Fogg, 1966; Sieburth and Jensen, 1968, 1969; Nissenbaum and Kaplan, 1972; Gagosian and Steurmer, 1977; Harvey et al., 1983) and/or its chemistry (Pocklington, 1977; Harvey et al., 1983; Jackson and Williams, 1985). Some studies of optical characteristics dealt with identification of peak excitation and emission wavelengths (Traganza, 1969; Hayase and Tsubota, 1985, Ewald et al., 1986), development of more sensitive instrumentation (Donard et al., 1989), or testing of methods for detecting multiple chromophores and/or fluorophores (Coble et al., 1990). While this knowledge is essential to our understanding of the CDOM absorption and fluorescence processes, the results obtained did not provide spectral data which is usable as input to a reflectance model. Other optical studies addressed the need for quantitative understanding of the absorption coefficient (Wheeler, 1976; Bricaud et al., 1981),


66 specific absorption coefficient (Zepp and Schlotzhauer, 1981; Carder et al 1989), and fluorescence quantum yield or fluorescence efficiency (Donard et al., 1989; Hawes et al., 1992). The results of these studies are readily adaptable to use in a reflectance model, and have begun to elucidate the significance of CDOM absorption and fluorescence, relative to that of Chl a and the accessory pigments, in the study of ocean color. Fluorescence by CDOM is a very broadband emission ( = 100 nm emission bandwidth) and when stimulated by the full solar spectrum the fluorescent emission from excitation in the UV may be partially reabsorbed as an excitation at a higher wavelength. This cascading excitation/emission process further broadens the net emission spectrum, but is mitigated by the spectral slope of the absorption spectrum (see below). The absorption spectrum (acooM(A)) of CDOM is well documented as a smooth exponential function, increasing as excitation wavelength decreases (Bricaud et al., 1981; Zepp and Schlotzhauer, 1981; Hayase and Tsubota, 1985). [10] where Ao may be any specified wavelength (usually in the range 375 to 450nm). This general spectral shape is representative of the material found in both aquatic and marine environments (Zepp and Schlotzhauer, 1981; Bricaud et al. 1981). The two parameters which describe the precise shape of a given curve are the value of acooM(A) at 400 nm (or, in some cases, at 375nm) and the spectral slope (C. and c; respectively). These parameters vary considerably as a function of the organic


material in question, and may be useful in interpreting the nature and origin of the CDOM present, although this hypothesis is not yet fully supported. Values for c were found to lie between 0 .06 and 0.3 m (Bricaud et al. (1981),A. o = 375nm) for various marine waters; found to lie between 012 and -. 018 by Bricaud tal (1981), and between -.014 and -.023 by Roesler et al. (1989). Using a high resolution spectrofluorometer and a three dimensional emission matrix display, Coble et al (1990) found evidence of at least three distinct fluorophores in waters of the Black Sea. The ability to resolve discrete fluorescent signatures for multiple fluorophores in a single sample is an important step in characterizing the DOM which is present. 67 A characteristic of CDOM of importance to the remote sensing modeler is the quantum fluorescence efficiency also known as the fluorescence quantum yield (Hawes et al., 1992). Knowledge of this spectral property, and of the fluorescence emission spectra of various CDOM substances, allows one to generate a spectral fluorescence efficiency function 17(A..,A,)), which can then be used with the spectral absorption coefficient of CDOM to estimate the fluorescence reponse o f the CDOM in a water sample. Hawes et al. (1992) developed an expression for 17(A.,A,), which they used to characterize the fluorescence in their studies of humus from various marine environments Lee (1992) developed an expression based on in-elastic scattering theory and the work of Hawes et al. (1992), and using inputs of the absorption coefficient of CDOM the volume absorption coefficient and the incident


irradiance, to model CDOM spectral fluorescence as a paremeter in remote sensing reflectance models. 68 Donard, et al.. ,(1989) found DOM in coastal waters (with the influence of terrigenous runoff implied by lower salinity) to have about three times higher apparent fluorescence efficiency at 370nm than that from "marine" (offshore) waters. Stuermer (1975) (cited in Donard et al 1989) produced results in agreement with this, from samples in the Sargasso Sea and adjacent coastal areas. Their results imply that terrestrial DOM (transported into coastal waters by runoft) fluoresces more efficiently than that of offshore marine waters. Hawes et al. (1992) also found highest fluorescence efficiencies in coastal waters on the West Florida Shelf, but with high salinities implying little input from terrestrial sources, and they postulated that the CDOM in this region was from recent marine productivity. Intermediate efficiency values were found in waters influenced by terrestrial runoff, and lowest efficiencies were found in offshore waters where the material was thought to be older "background" marine CDOM leading them to suggest that variations in 4>(Ax,A...) may follow the pattern "new marine CDOM" > "old marine CDOM" > "terrestrial CDOM". This is in apparent contrast to the results of Donard et al., (1989), indicating the need for more research into the nature of this very important parameter.


APPENDIX 3 BIOLUMINESCENCE OF MARINE ORGANISMS 69 In natural settings marine bioluminescence is stimulated by mechanical or chemical means. Mechanically stimulated luminescence (MSL) is that induced by shear stresses on the organism due to water motion or by physical contact between organisms, and is manifested by flashes of usually blue-green light. Chemically stimulated luminescence (CSL), such as that seen in some species of bacteria, is perhaps due to changes in pH, oxygen, or other chemical parameters of the water. Of interest to remote sensing investigations are the effects of bioluminescence on the measurement of absorption and specific absorption coefficients of particles collected from the euphotic rone. These particles include bacteria, phytoplankton (dinoflagellates), zooplankton, and detritus (faecal pellets). CSL is common among free-living marine bacteria and those found in detritus (e.g faecal pellets (Andrews et al., 1984) and marine snow (Orzech and Nealson, 1984)), and is characterized by a continuous soft glow. MSL is frequently found in phytoplankton (Esaias and Curl, 1972; ; Bozin and Filimonov, 1985; Swift et al., 1985), and is also common in many species of zooplankton (Lapota and Losee, 1984 ; Batchelder and Swift, 1989; Buskey and Stearns, 1991).


70 Mitchell and Kiefer (1988a) describes the method used here to obtain absorption coefficients for particulates The method consists of flltering a water sample, illuminating the fllter pad, measuring the transmitted light, and performing some computations on the results The potential for contamination exists in the light measurement step. Relevant characteristics of bioluminescence which would mitigate this potential contamination are 1) the stimulus required to initiate an emission, 2) the rise time of the emission, 3) it's duration relative to the sampling time, and 4) time required for the organism to adapt to a darkness luminescent mode from a daylight photoinhibited mode (to scotophase from photophase) Intensity of the bioluminescence is assumed not to be a limiting factor in this relatively low light level measurement technique Mechanically stimulated luminescence requires contact with, or displacement by an external (relative to the luminescent organism) body Vigorously stirring a solution, or bubbling air through the solution are two examples of this type of stimulus (Hamman et al., 1981). In three species of dinoflagellates exposure to intense light is sufficient to trigger a luminescent response (Sweeney et al., 1983) Whether a similar response is found in other bioluminescent organisms is not known but Sweeney et al., (1983) conclude that the phenomenon they observed is of general occurence, since rise time, latency, and response to a series of stimulating flashes was the same in all three species Sweeney et al., (1983) found that the rise time of the bioluminescent response to light stimulation was dependent on light intensity, with faster response (shorter rise


71 time) being associated with a higher intensity of stimulating light. Under all continuous incident light durations, they found an exponential decay from maximum emission, but did not characterize this decay. Their incident light intensities ranged from 0.17 to 68 W/m2 with corresponding rise times of 60 to 0.19 seconds. Incident light intensity in our measurement setup was estimated to be about 40 W/m\ corresponding to a rise time of about 0.1 second in Sweeney et al 's data. Time between illumination of the sample filter pad and recording of the spectra was typically about a second, so the bioluminescent response would be underway when the sample was recorded. Hamman et al., (1981) studied the photoinhibition of bioluminescence in eight species of dinoflagellates They found that when any of these species were placed in the dark after an extended photoperiod, full bioluminescent potential was attained within about thirty minutes. Sadovskaya and Filimonov (1985) obtained similar results with three other dinoflagellate species.


72 APPENDIX 4 PHYTOPLANKTON AND ZOOPLANKTON IN THE STUDY AREA Phytoplankton species for the CTZ study area and period were identified by Chavez et al., (1991). Two distinct phytoplankton communities were encountered: 1) coastal diatom-dominated, and 2) oceanic, small, solitary phytoplankton-dominated. Chavez et al., (1991) found that for areas of the ftlament, its associated offshore jet, and the region immediately to the southeast which had high concentrations of chlorophyll a, neritic, centric diatoms (principally Chaetoceras debilis and C concavicomis) were dominant populations of phytoplankton. This was true wherever the [Chl a] concentration exceeded about 0.5 J,tgll. Below this value the relative concentrations of coccolithophores and small flagellates increased. Chavez et al., (1991) determined that at no time during the study period were dinoflagellates significant contributors to the phytoplankton biomass. Several stations were occupied just to the northwest of the jet border, and tens of kilometers to the west and north, in warmer, fresher water exhibiting low pigment concentrations. These stations with low [Chl a] ( < 0.5 J,tg/1) were dominated by small, solitary phytoplankton (principally Synechococcus and prochlorophytes) that are characteristic of ocean waters. Among the stations reported here, Stations 46 and


73 55 fall into the diatom-dominated community, while Stations 63 and 77 are examples of the oceanic, low-[Chl a] phytoplankton regime. Station 39 is an intermediate station well downstream of the upwelling region but near the core of the offshore jet, and therefore exhibits nutrient-depleted waters (Kadko, et al. (1991)) with an aging (and slowly sinking) phytoplankton stock. The length of the jet, from the inshore upwelling region (e.g. Station 46) to its approximate offshore limit (about 55 km W of Station 39), was about 345 km. Mean horizontal velocity for a surface water body entrained in the jet was 0 33 m/s (Washburn et al., 1991), making the transit time from Station 46 to Station 39 about 11 days. In general, a southeast to northwest transect across the fllamentljet system produced a "standard sequence" of zooplankton species (Mackas et al 1991). This sequence was consistent for both inshore and offshore transects. Areas to the south and east of the filament were dominated by gelatinous species, particularly Dolioletta gegenbauri, with moderate concentrations of adult euphausiids. Euphausia pacifica was prevalent along the southeast margin of the filament. Eucalanus califomicus and euphausiid larvae dominated the zooplankton community in the cool filament core, and extended partially across the seaward moving jet on the NW margin of the filament. The warm side of the jet and the oceanic waters to the north and west were occupied mostly by heteropod larvae, chaetognaths, Dolioletta sp., and a mixture of small copepods. Station 46, an upwelling station near the origin of the jet, was dominated by adult euphausiids, with euphausiid larvae and copepods. Station 55, inshore and


74 southeast of the jet/filament system, was probably dominated by gelatinous Dolioletta and with adult euphausiids. Station 63 was just outside the jet in warmer, oceanic water, and euphausiid larvae are the greatest portion of zooplankton biomass at this station Station 39 was well offshore in the jet, with E. pacifica the dominant species. Finally, Station 77 was a warm, oceanic water station well removed from the filament system, favoring small copepods, chaetognaths, and heteropod larvae, all at low density.


Download Options

No images are available for this item.
Cite this item close


Cras ut cursus ante, a fringilla nunc. Mauris lorem nunc, cursus sit amet enim ac, vehicula vestibulum mi. Mauris viverra nisl vel enim faucibus porta. Praesent sit amet ornare diam, non finibus nulla.


Cras efficitur magna et sapien varius, luctus ullamcorper dolor convallis. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Fusce sit amet justo ut erat laoreet congue sed a ante.


Phasellus ornare in augue eu imperdiet. Donec malesuada sapien ante, at vehicula orci tempor molestie. Proin vitae urna elit. Pellentesque vitae nisi et diam euismod malesuada aliquet non erat.


Nunc fringilla dolor ut dictum placerat. Proin ac neque rutrum, consectetur ligula id, laoreet ligula. Nulla lorem massa, consectetur vitae consequat in, lobortis at dolor. Nunc sed leo odio.