Interpretation of the Coastal Zone Color Scanner (CZCS) signature of the Orinoco River

Interpretation of the Coastal Zone Color Scanner (CZCS) signature of the Orinoco River

Material Information

Interpretation of the Coastal Zone Color Scanner (CZCS) signature of the Orinoco River
Hochman, Herschel T.
Place of Publication:
Tampa, Florida
University of South Florida
Publication Date:
Physical Description:
x, 81 leaves : ill. ; 29 cm.


Subjects / Keywords:
Remote sensing -- Orinoco River (Venezuela and Columbia) ( lcsh )
Optical oceanography -- Remote sensing ( lcsh )
Colors -- Analysis ( lcsh )
Orinoco River (Venezuela and Columbia) ( lcsh )
Dissertations, Academic -- Marine Science -- Masters -- USF ( FTS )


General Note:
Thesis (M.S.)--University of South Florida, 1992. Includes bibliographical references (leaves 75-81).

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:
029043228 ( ALEPH )
27152457 ( OCLC )
F51-00093 ( USFLDC DOI )
f51.93 ( USFLDC Handle )

Postcard Information



This item is only available as the following downloads:

Full Text


INTERPRETATION OF THE COASTAL ZONE COLOR SCANNER (CZCS) SIGNATURE OF THE ORINOCO RIVER by Herschel T. Hochman 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 August 1992 Major Professor: Frank E. Muller-Karger, 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 Herschel T. Hochman with a major in Marine Science has been approved by the Examining Committeee on April 28, 1992 as satisfactory for the Thesis requirement for the Master of Science degree Thesis Committee: =-:M:a j,.....o r---:P=-r o-==f=-e s s o r : Member: Ph.D. Member:


Herschel T. Hochman 1992 All Rights Reserved


ACKNOWLEDGMENTS I am indebted to Dr. Frank E. Muller-Karger who not only assisted me in selecting this project but guided my activities and research to completion. His expertise and understanding of computers is unparalleled, anc working under his direction has been an enlightening experience. I also want to express sincere appreciation to Dr. John J. Walsh for sharing his invaluable insight into the biological aspects of this project and for his willingness to patiently clarify many topics for me. I wish to acknowledge the assistance of Dr. Gabriel A. Vargo whose expertise in phytoplankton ecology was indispensable. All three have offered encouragement throughout the research project, and I thank them for serving as committee members. Many others on the faculty of the Marine Science Department assisted me when help was requested and were generous in sharing their time and advice, for which I am most appreciative. I would like to particularly thank Dr. Kendall L. Carder for his patience and help with my work. My colleagues, Ray Pribble, Mark Meyers, Dave Morrison, Paul Bissett and Dwight Dieterle were a supportive group and willing sources of information. They were very helpful in critiquing my work as it progressed. ii


This acknowledgement would not be complete without a special thank you to my wife, Doris, who has not only encouraged m e during this period of research and but also throughout similar times in the past.


LIST OF TABLES LIST OF FIGURES ABSTRACT INTRODUCTION Goals Background Amazon Orinoco CZCS ALGORITHMS TABLE OF CONTENTS Atmospheric Algorithm Water-Leaving Radiance Absorption and Scattering Optical Algorithms RESULTS Aerosol Correction Water-Leaving Radiance Calculations BioOp tical Analysis of czcs Images Clear Water Model Intermediate/High Concentration CDOC Evaluation BIO-OPTICAL MODELS: APPLICATION Mathematical Treatment Method I : Calculations of Conversion Method II: Calculatio n s from Op tical DATA INTERPRETATION CONCLUSION LIST OF REFERENCES iv v v i viii 1 1 2 10 12 1 5 1 5 17 22 25 30 30 32 3 9 46 47 47 51 5 1 Factor 52 Algorithms 5 3 54 70 75


LIST OF TABLES TABLE 1 Dissolved and particulate organic carbon and nitrogen in the Orinoco River. 14 TABLE 2 Location of regions studied and pigment concentrations. 39 TABLE 3 Variables used in the algorithms to calculate [Lw(l) ]N. 45 TABLE 4 Locations of stations used as test sites of GORDON et al 's model. 4 9 TABLE 5 Specific absorption coefficient values for different oceanic regions. 67 v


LIST OF FIGURES Figure 1 Study region: The Caribbean Sea with names of locations mentioned and transects discussed in text. 3 Figure 2 Section of nitrate concentration along the 440 series transect (Figure 1). Nitrate concentration in 1"1 (KETCHUM and RYTHER, Cruise #14, ATLANTIS II). 5 Figure 3 Section of salinity (psu) along the 440 series transect. (KETCHUM and RYTHER, Cruise #14, ATLANTIS II) 6 Figure 4 CZCS weekly composites of the Caribbean Sea. a) September 2 -7, 1979 b) October 3 -9, 1979. 7 Figure 5 Pigment spectra of living phytoplankton Figure 6 Absorption by phytoplankton chlorophyll and pure sea water. Figure 7 Light attenuation versus wavelength. Figure 8 Aerosol radiance (670nm) vs. pigment concentration in the Orinoco River plume 20 21 28 (9 October 1979). 31 Figure 9 Water-leaving radiance, versus pigment concentration derived from b1o-optical algorithms (GORDON et al. 1988). 37 Figure 10 Water-leaving radiance, [Lw(A)]N, versus pigment and yellow substance (YS) concentration derived from bio-optical algorithms (GORDON et al. 1988). 38 Figure 11 HIGH, INTERMEDIATE, AND CLEAR water regions of the typical plume as indicated by the pigment concentration values given by the bio-optical model. 40 Figure 12 Normalized water-leaving radiance [Lw(l)]N at 443nm with varying concentrations of CDOC. 42 vi


Figure 13 Normalized water-leaving radiance [Lw(l)]N at 520nm with varying concentrations of CDOC. 43 Figure 14 Normalized water-leaving radiance [Lw(l)]N at 550nm with varying concentrations of CDOC. 44 Figure 15 The study location and the in situ data; ( [CDOC] (mg 1"1), [PIG] (mg m-3), and salinity (psu). 50 Figure 16 Station M. Values of [CDOC] added to the GORDON et al. (1988) bio-optical model (equations (5) and (6)) to match the czcs water-leaving radiance at 520nm at in situ measured [PIG] (BIDIGARE et al. 1992). 55 Figure 17 Station H. Values of [CDOC] added to the GORDON et al. (1988) bio-optical model (equations (5) and (6)) to match the czcs water-leaving radiance at 443nm at in situ measured [PIG] (BIDIGARE et al. 1992). 56 Figure 18 station R5. Effects of high reflectance sediment on the GORDON et al. (1988) bio-optical model. 57 Figure 19 Application of different measured specific absorption coefficients, a*9 from the Orinoco River and Mississippi River plumes with equal [CDOC]. 65 Figure 20 Application of different measured specific absorption coefficients, a*9 from the Orinoco River and Mississippi River plumes with unequal [CDOC]. 66 Figure 21 Dilution effects of [CDOC] following the salinity change from Station R5 nearshore to Station H offshore. 72 vii


INTERPRETATION OF THE COASTAL ZONE COLOR SCANNER (CZCS) SIGNATURE OF THE ORINOCO RIVER by Herschel T. Hochman 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 August 1992 Major Professor: Frank E. Muller-Karger, Ph.D. viii


Satellite imagery from the Coastal Zone Color Scanner (CZCS) is used to assess the influence of the Orinoco River discharge on the bio-optical characteristics of the Caribbean Sea. During boreal fall, the Orinoco plume extends from the mouth of the river northward to Puerto Rico over a distance of approximately 1200 km. Using recently obtained shipboard data from this region, an attempt is made to deconvolve colored dissolved organic carbon (CDOC) from remotely sensed pigment concentration. No changes were necessary to the atmospheric correction algorithm which suggests that it can be utilized for Case II waters. In situ pigment concentration measurements were compared to values from representative CZCS i mages processed using standard algorithms. It was then assumed that differences between the measured and observed values were due to the presence of excess CDOC. A bio-optical model was applied to derive water-leaving radiances that should be detected by the CZCS. By adjustment of model backscattering characteristics of phytoplankton and CDOC absorption, CDOC concentrations were obtained which were then compared to in situ CDOC measurements. This study shows that CDOC is a major component of the remotely sensed p igment concentration. The results also ix


indicate a possible relationship between salinity and CDOC in the Orinoco plume. The causes for the variations in the modeled water-leaving radiances require further study. It is clear that there is a need to strengthen our database and make full use of optical models. Further investigation is needed to understand CDOC composition and its spectral absorption (i.e. specific absorption coefficients). Because of the compositional changes that are phytoplankton communities undergo, data on their varying optical characteristics are important to our understanding of satellite imagery. Bio-optical models open the door to a better understanding of optical properties of substances, particularly where we have high human impact. Abstract approved: __ Major Professor, Frank E. Mliller-Karger, Ph.D. Assistant Professor Department of Marine Science Date 6f Approval X


1 INTRODUCTION The Caribbean Sea is an ideal laboratory to examine the effects of particulate and dissolved matter entering the sea via rivers. This is an area that traditionally has been considered oligotrophic. However, the combined discharge of the Amazon and Orinoco Rivers makes up nearly 20% of the world's annual river outflow (MULLER-KARGER et al. 1989) These two rivers input large quantities of water and nutrients into the Caribbean (EDMOND et al. 1981; FROELICH et al. 1978), but little is known about their impact. Here, I have studied the bio-optical characteristics of the Orinoco River plume to determine the usefulness of Coastal Zone Color Scanner (CZCS) satellite data in assessing the role of river water in the biological productivity regime of the region. Goals The primary goal of this study was to deconvolve the signatures of colored dissolved organic carbon concentration [CDOC] and phytoplankton pigment concentrations [PIG] in CZCS images of the Caribbean Sea. An optical model was used to evaluate their effect on remote sensing data. The results are relevant to studies of carbon budgets above continental margins because it is here that over 20% of global marine


2 productivity occurs, and because it is here that it is most difficult to interpret ocean color satellite data (WALSH et al. 1992). The focus of this analysis was the plume that dominates the Caribbean during the wet season as a result of the discharge of the Orinoco River. czcs data, recent shipboard measurements (BLOUGH et al. 1992; BIDIGARE et al. 1992) and historical data (KETCHUM and RYTHER, 1966) of the southeastern Caribbean were used to develop and initialize the bio-optical model. Background This study concentrates on a region bound by 7N to 21N and 56w to 76w. Because the influence of the Amazon on this region is limited during the rainy season of August -November (MULLER-KARGER et al. 1988), the Orinoco River is an important source of nutrients affecting the pigment distribution (MULLER-KARGER et al. 1989) This plume leads to diatomdominated phytoplankton communi ties, particularly near the continental margin (BIDIGARE et al. 1992). Figure 1 shows the study region and transects used. The 400 series transect was occupied between October 25 -November 1, 1964, (ATLANTIS II, Cruise 14, KETCHUM and RYTHER, 1966). The stations identified with letters were taken September 25-29 and October 10-14, 1988, during the cruise of the R/V COLUMBUS ISELIN (BLOUGH et al. 1992; BIDIGARE et al. 1992).


22 20 18 16 14 12 10 8 6 431 -78 -74 -72 -70 -68 -66 -64 -62 -58 o' Puerto Rico E (Blough et ol.) 0.:: (Bidigare et al.} Caribbean Sea L. Maracaibo Colombia Venezuela Southern Sargasso Sea Transect__. (Ketchum & Ryther} (> -.:,o d> 0 {j &_H "-oL -78 -74 -72 -70 -68 -66 -64 -62 -60 20 18 16 14 12 10 8 6 -58 Figure 1: study region: The caribbean Sea with names of locations mentioned and transects discussed in text. w


4 The 400 series stations provided nitrate and salinity sections (Figures 2 and 3). Nitrate (N03 ) measured along the easternmost transect (Figure 2) suggests that, even during the period of reduced wind stress (MULLER-KARGER et al. 1989), coastal upwelling may contribute nutrients to nearshore regions. For ease of interpretation, Station 441 (Figure 1) was not included in the section. The data showed that similar conditions existed at stations 440 and 441, both being parallel to the shore. It is clear that near the shore (i.e. stations 439, 440, 441) a strong upwelling event occurs. During the wet season (August-November) sufficient nutrients and colored dissolved organic carbon are added by riverine and upwelling sources to sustain growth, senescence and decay of phytoplankton, seen by czcs imagery as an apparent plume of pigment extending up to 1200 km northwestward from its source. This can clearly be seen in Figure 4 which are two composite images for the weeks of September 2-7, 1979, and October 3-9, 1979. During the dry season (February-May) the extent of the plume is much reduced (MULLER-KARGER et al. 1989). The seasonal effects of the Amazon and Orinoco river plumes on the caribbean Sea result from the migration of the Intertropical Convergence Zone (ITCZ) which regulates the wet and dry seasonal patterns. Rainfall has a much smaller direct effect on seasonal salinity changes than these rivers (FROELICH et al. 1978). Migration of the ITCZ affects the wind and resulting wind stress patterns (HOREL et al. 1986; KATZ,


20 ..---... 40 E ......... I 1-a_ w 0 60 80 100 431 Figure 2 4.32 433 434 435 436 437 STATION ( DATA POINTS) 438 439 440 (Oct 25 -Nov 1, 1964) Section of nitrate concentration along the 440 series transect (Figure 1). Nitrate concentration in N03-N 1"1 (KETCHUM and RYTHER, Cruise #14, ATLANTIS II). U1


"'"" E ........... I I-0.. w 0 SALINI1Y (PSU) ORINOCO I T *\ * 7 \ * 20 40 60 100 431 * 432 433 * 434 435 STATION \ * 436 437 438 ( DATA POINTS) I 439 440 (Oct 25 Nov 1, 1964) Figure 3 Section of salinity (psu) along the 440 series transect. (KETCHUM and RYTHER, Cruise #14, ATLANTIS II).


a b Figure 4 czcs weekly composites of the Caribbean Sea. The plume covers the eastern Caribbean and flows northward past Puerto Rico. Warm colors (yellow, orange, red) represent high pigment and DOC concentrations. Blue color represents clear water. a) September 2 -7, 1979 b) October 3 -9, 1979.


8 1987), and the tropical ocean circulation (RICHARDSON and McKEE, 1984; MORRISON and NOWLIN, 1982; KINDER et al. 1985). These changes affect the distribution of phytoplankton biomass, river water, and resultant color of the Caribbean Sea as seen by the czcs. For this region, there are few shipboard measurements available for validation of remote sensing signals from the czcs. Also, limited spectral response of remote optical sensors has restricted the interpretation of the images because the standard bio-optical algorithms cannot discriminate between colored dissolved organic carbon (CDOC) and phytoplankton pigments (MOREL and PRIEUR, 1977; SMITH and BAKER, 1978a and 1978b). In fact, the remote sensing bio-optical algorithms used for processing of czcs data relate all colored material to chlorophyll-like pigment concentration. The problem in coastal and shelf areas or where river water is present, is that phaeopigment and accessory pigments, as well as CDOC absorb at 443nm (BAKER and SMITH, 1982) leading to overestimates of pigment concentration (CARDER et al. 1989). Near the coast, especially off river deltas, large amounts of sediment cause additional problems in proper interpretation of czcs images. However, the impact of sediment reflectance on pigment concentration is small farther offshore (at distances >50-100 km; MULLER-KARGER et al. 1989) where most sediment has settled out or is sinking. Shipboard samples were collected in September, 1988, during the cruise of the R/V COLUMBUS ISELIN (BLOUGH et al.


1992; BIDIGARE et al. germane to this study 9 1992) Figure 1 shows the stations (R1 to R5 and P1 to D) Shipboard samples were analyzed using state-of-the-art techniques for pigments and CDOC; pigment concentrations were determined by high-performance liquid chromatography (HPLC) methods (BIDIGARE et al. 1992), and CDOC by high-temperature catalytic combustion (BLOUGH et al. 1992). With these new data, we can investigate the bio-optical characteristics of the Orinoco River plume. The model discussed here is based on the following: (a) The fall season is the time of maximum river discharge and plume development. (b) The Orinoco River discharge is the main source of nutrients to the Caribbean Sea. c) Figures 2 and 3, developed from data of October 25 -November 1, 1964 (KETCHUM and RYTHER, 1966) indicate the possibility of upwelling contributing nutrients in the nearshore region of the plume. During the time period of this study (Fall season) when the river discharge is maximum, winds are weak and wind-driven upwelling would not normally be a factor. However, active coastal upwelling as in Figures 2 and 3 may be due primarily to geostrophic western boundary currents or upwelling around the periphery of anticyclonic rings that break off the area of the retroflection of the North Brazil Current and which then travel northward, interacting with the continental margin (JOHNS et al. 1990; MULLER-KARGER et al. 1988). While these eddies are clearly


10 seen in satellite imagery (i.e. November, 1979), there is still uncertainty as to the permanence of these features, how the eddies are formed, the role they play in recirculation, and their relationship to the North Brazil and Guiana Current systems. d) This nutrient supply enhances the growth of phytoplankton where turbidity by sediments decreases. This then contributes to the remotely sensed pigment concentration [Chl a + phaeopigments +accessory pigments and CDOC]. e) CDOC contributes to the total pigment concentration estimated by the satellite. The sources include river discharge (terrestrial CDOC) and excretion by phytoplankton (marine CDOC) Amazon Because the Amazon River is the largest river system in the world, discharging 16% of the annual outflow into the world's oceans, a logical assumption might be that it is the dominant riverine influence on the Caribbean Sea. However, CZCS imagery and drifting buoy trajectories have shown that during June-January, a large fraction of the Amazon River water is transported offshore around the North Brazil Current and east into the North Equatorial Countercurrent (MULLER-KARGER et al. 1988; JOHNS et al. 1990). Therefore, it would seem to have a minimal effect on the Caribbean Sea during this time of year. In contrast, during February-May, Amazon water moves with the northward-flowing western boundary


11 current of the North Brazil-Guiana system along the coast and into the Caribbean Sea (MULLER-KARGER et al. 1988; RYTHER et al. 1967; EDMOND et al. 1981). RYTHER et al. (1967) reports low concentrations of essential nitrogen and phosphorous in the Amazon discharge, and "extremely low" concentrations by the time the water has traveled 600 miles from the mouth of the Amazon. He notes that concentrations may be even lower than in typical tropical surface water. Furthermore, he found no observable difference in nutrient levels between spring and fall in the Amazon river drainage system. Finally, RYTHER et al. (1967) suggests that an over-all effect of the Amazon River is to decrease the fertility of the ocean into which it flows. On the other hand, HULBURT and CORWIN'S ( 1969) phytoplankton studies of the Amazon plume, done simultaneously with the RYTHER et al. (1966) study, show an increase in number of cells in plume water relative to oceanic waters. CALEF and GRICE {1967) also found that the plume contains upwards of three times the biomass of zooplankton relative to oceanic water. While lenses of Amazon water have been found to be rich in silicate {FROELICH et al. 1978), the limitation of other available nutrients could lead to low new productivity and fast recycling. Even under circumstances where significant concentrations of phytoplankton may exist, there is still a question how organisms remain viable within the deficient nutrient conditions that exist over the considerable distances covered by the plume (MULLER-KARGER et al. 1988; MULLER-KARGER


12 et al. 1989). Sufficient evidence exists to indicate only a minor influence due to the Amazon River discharge on the processes of interest in the southeastern Caribbean during boreal fall. Therefore, its effect is neglected in this study. Orinoco The Orinoco is the third largest river in the world in terms of average flow into the ocean and is the other major river in the region. The river flows mostly to the northwest through the Gulf of Paria and into the Caribbean. It is located north of the Equator, and the region has a distinct dry (February-May) and wet (August-November) season modulated by the migration of the ITCZ. The seasonal rainfall strongly affects the discharge which ranges from 6 x 103 m3 s-1 in March to approximately 6 x 104 m3 s-1 in August (MEADE et al. 1983) Studies of the terrestrially-derived nutrients in the Orinoco River output indicate that inorganic suspended solids, soluble silica, major ionic solids, and phosphorus are low by comparison with concentrations that are typical of other large rivers of the world (LEWIS and SAUNDERS, 1988). This reflects the relatively pristine nature of the Orinoco. During the period of highest discharge, mean NH4-N is approximately 35 J..Lg 1-1 and N03-N is 80 J..Lg r1 (LEWIS and SAUNDERS, 1988, their Table 1 and Figure 10). Phosphorus as P04-P ranges from 8 J..Lg 1-1 to 10 J..Lg 1-1 (LEWIS and SAUNDERS, 1988) The


13 discharge-weighted mean concentration1 of dissolved inorganic solids (defined as the colloidal portion in the dissolved fraction) for the Orinoco is 25 mg 1-1 as compared to 50 mg 1-1 for worldwide continental waters and 53 mg 1-1 for the Amazon. This is explained by a) the low rates of mechanical/chemical erosion and, b) the high rate of runoff from the orinoco drainage causing dilution of such compounds (LEWIS and SAUNDERS, 1988). However, the high rate of runoff (annual mean = 33 1 sec-1 km-2 ) from the Orinoco drainage is compensated for by high specific transport rates (the rate at which nutrient is transported into the system; LEWIS and SAUNDERS, 1988). In contrast, concentrations of biologically-derived substances, such as dissolved organic nitrogen (DON) and dissolved organic carbon (DOC), are high. For typical unpolluted rivers, the average DOC is given as 4.4 mg 1-1 and the median DOC/DON ratio about 20 (LEWIS and SAUNDERS, 1988; MEYBECK, 1982). Table 1 gives the dissolved organic constituent values for the Orinoco River. 1Discharge-weighted mean concentration is the sum of the products of discharges and concentrations over all days of the year divided by the annual discharge.


14 TABLE 1 Dissolved and particulate organic carbon and nitrogen in the Orinoco River. Spring Fall Maximum *POC 0.84 4.02 17.79(May) (PAOLINI et al. 1983) DOC 6.02 6.51 10.80(July) (PAOLINI et al. 1983) All units in mg r1 *POC = Particulate Organic Carbon Discharge Weighted Spring Fall Mean Concentration *PON 300 220 184.86 (LEWIS and SAUNDERS, 1988) DON 100 180 160.11 (LEWIS and SAUNDERS, 1988) All units in jJg r1 *PON = Particulate Organic Nitrogen


15 CZCS ALGORITHMS Atmospheric czcs images obtained between November 1978 and December 1986 were used to map pigment concentrations in the Caribbean Sea. These were screened using the BROWSE quick-look system developed at the Goddard Space Flight Center by G. Feldman and N. Kuring. Subsampling of the original images to 1/16 of their resolution produced 4km spatial resolution images of pigment concentration. The atmospheric correction and bio-optical algorithms of GORDON et al. (1983) and GORDON et al. (1988) were used to correct the sensor radiance for retrieval of water-leaving radiance. This is then related to the pigment concentration in the water. The images were corrected for aerosols, cloud cover, and sun glint. Pigment concentrations were derived from ratios of the blue (443nm) or blue-green (520nm) water-leaving radiances to the green radiance (550nm). The resulting pigment values were then converted to binary (grey) values, (G), using equation (1). G = (Log [PIG] + 1.4)/0.012 (1) where (Pig] = total chlorophyll-like pigment concentration (mg pig m-3 )


16 The data were then mapped to an Equidistant Cylindrical projection and binned into weekly composites having the same spatial resolution as the input images. Pixels having pigment concentration below 0.04 mg m-3 were excluded as well as those having extremely high pigment values (>30 mg m-3). The value of a composite pixel was an average obtained from the binning total divided by the number of pixels available (arithmetic average). Concentrations were coded as shown in the color bar on Figure 4. The algorithm relating atmospheric contribution to the total radiance, Lt, measured by the remote sensor at each wavelength (A) is given by GORDON et al. (1983): Lt (A ) = Lr ( A) + L8 (A) + t (A) Lw (A ) ( 2 ) This partitions the radiance into the components responsible for the total radiance seen by the remote sensor and is the sum of the radiances due to air molecules or Rayleigh scattering (Lr(A)), suspended particle or aerosol scattering (L8(A)), and scattering due to water (Lw(A)). Equation (2) decouples the ocean and atmosphere radiance. The photons in t(A) Lw(A) are the only ones which have passed through the airjwater interface, been subjected to absorption and scattering within the water, and backscattered out of the ocean to the satellite sensor. Therefore, they contain information on the dissolved and suspended material in the water. The other photons making up the balance of the equation will cause errors. Atmospheric correction uses band 4 (670 nm) for calculating L8 since it is assumed that water absorbs all


17 light at this wavelength, and in waters with low pigment concentration, Lw(670) = o. Since sediments in nearshore river plume environments may scatter at this wavelength, it is possible that L8 ( l) is overestimated in the region of the plume. Without further correction, Lw at 443, 520, and 550nm would thus be underestimated. To address this problem, separate regions within the core of the plume, as well as outside the plume, were analyzed for anomalous atmospheric effects by examining the 670nm band. Another factor that could cause reflectance at L8(670) is backscatter due to coccolithophores. However, their concentration has been found to be insignificant in the river plume (BETZER et al. 1977). While BIDIGARE et al. (1992) does indicate the presence of coccolithophores (i.e. prymensiophytes), it is not clear that their abundance is sufficient to affect the color of water at Stations M and H. Water-Leaving Radiance The water-leaving radiance, Lw, is dependent on the concentration of colored constituents within it, including CDOC and pigments. The color of these constituents is a function of backscattering (bb) and absorption (a). Most algorithms relate the average pigment concentration, , to the water leaving radiance, Lw, at two different wavelengths (i.e. Lw ( 4 4 3) Lw (52 0) ) where one wavelength serves as a control point (i.e. Lw(550)), and the second records variations in . By using ratios, which are more sensitive


18 to small changes in absorption of light, pigment concentration can be estimated (HOVIS and LEUNG, 1977). The use of ratios depends upon covariance between phytoplankton and detritus and colored dissolved organic matter for accurate remote sensing of chlorophyll. For Case I and mixtures of Case I and Case II waters, algorithms derived from statistical analysis of logtransformed data show that the ratio of Lw at different wavelengths is related to in the following manner (GORDON and CLARK, 1981; GORDON et al. 1983): where Log = Log A(ij) + B(ij)Log R(ij) (4) = optically weighted sea surface pigment concentration within 1 attenuation length A(ij), B(ij) =regression coefficients for waterleaving radiance, wavelength dependent constants R ( i j ) = Lw ( i ) / Lw ( j ) (ij) = (443,550) for C < 1.5 mgjm3 ; (520,550) for C > 1.5 mgjm3 The ratio Lw(443)/Lw(550) is used at lower concentrations (approximately <1. 5 mg m-3). Figure 5 shows the pigment absorption spectra which form the basis for this ratio; these are wavelengths of maximum and minimum absorption of light by phytoplankton (YENTSCH, 1960). However, at higher pigment concentrations (approximately >1. 5 mg m-3 ) Lw ( 443) becomes too


19 small to be retrieved with sufficient accuracy to be useful. Figure 6 shows that increasing the concentration of phytoplankton (0.5-10 mg m -3 ) causes the wavelength of maximum transmission to shift from the blue (440nm) into the green (550nm). This shift is due to the absorption by the shortwavelength bands by chlorophyll plus carotenoids. The curves also indicate that, at low concentrations of phytoplankton, small increases (e.g. 0.5-1.0) change color much more rapidly than at higher concentrations (e. g 3.0-5.0). It also indicates how strong the absorption coefficient is in the blue region for even moderate values of pigment concentration, leaving little light for scattering back upwards (YENTSCH, 1960)


10,------------------------------------, J 9 7 t 6 Ill z Ul 0 ...J ct .. 3 2 I 400 500 WAVE LENOTH IN MilLIMICRONS 20 Figure 5 Pigment spectra of living phytoplankton A. Diatom, Cyclotella sp., B. Dinoflagellate, Amphdium sp., C. green Flagellate, Chlamydomonas, D. Natural population sampled from Woods Hole waters. Adapted from YENTSCH, (1960).


:II .. ... + 400 21 soo eoo 700 WAVE LENGTH IN MILLIMICRONS Figure 6 Absorption by phytoplankton chlorophyll and pure sea water. The curves represent the combined attenuation coefficients for pure water and phytoplankton pigments. Numbers adjacent the curves represent the chlorophyll concentration in mg m3 Adapted from YENTSCH (1960).


22 At higher pigment concentrations ( >1.5 mg m-3), the ratio Lw(520)/Lw(550) is used in CZCS data processing (GORDON et al. 1983). The problem with the 520/550 ratio is the inability to distinguish between different absorbing constituents (MOREL and PRIEUR, 1977) since phaeopigments, accessory pigments, and yellow substances have nearly the same absorption characteristics as chlorophyll-a in the czcs bands (BAKER and SMITH, 1982). Furthermore, since 520nm is on the "shoulder" of the absorption spectrum of chlorophyll, it does not have the sensitivity of the blue (443) channel. Absorption and Scattering Photons within water can only be absorbed or scattered. Thus we need some measure of the extent to which the water absorbs and scatters light. Algorithms developed to extract pigment concentration from the water-leaving radiance are based on the dependency of the optical signal on absorption and backscattering. Since the remote sensor views reflected light, early investigators (SMITH and BAKER, 1978a, 1978b) converted attenuation of light by dissolved and particulate matter from absorption to a spectral reflectance, R(A). This is defined as the ratio of the upwelling irradiance just below the surface, Eu(A), to the downwelling irradianc e just abov e the surface, Ed(A). R(A) is governed by the ratio of the backscattering coefficient (bb), to the absorption coefficient (a) (MOREL and PRIEUR, 1977). The relationship between irradiance, E (radiant flux per unit area of surface) and sea surface


23 radiance, Lw, is developed in GORDON and CLARK (1981). In this paper, a normalized water-leaving radiance, [Lw]N, is used to remove the influence of the atmosphere variability and changes in the solar zenith angle from Lw. This makes the irradiance reflectance, R, dependent only on the pigment concentration. Experimental data compares favorably with the [Lw]N model, particularly for concentrations representative of Case I waters (GORDON et al. 1988) as will be discussed in more detail later. Case I waters are defined as those waters for which phytoplankton and their covarying detrital products are dominant when determining the ocean optical properties. Lw is proportional to the ratio of the backscattering coefficient, bb, and the absorption coefficient, a, of the water plus constituents (MOREL and PRIEUR, 1977, GORDON et al. 1983). Both a and bb are also proportional to in waters where phytoplankton dominate the color of the water. However, where high concentrations of CDOC do not covary with chlorophyll, absorption a is decoupled from . While pigments strongly influence a, it has been commonly assumed that they have little effect on bb. However, GORDON et al. (1988) show that: where ( 3) bb = total backscattering (bb)p = backscattering due to phytoplankton and their associated detrital material, m-1 (bb) w = pure seawater, m -1 (MOREL, 197 4)


24 While the coefficient for pure water is known, (bb) P must be determined in order to evaluate the diffuse reflectance of the sea. While bb may seem to be a potentially important factor, measurements of oceanic waters indicate that bb represents only 1% -2% of the total scattering coefficient, b. This implies that for most Case I waters, at moderate pigment concentrations, a >> bb (GORDON et al. 1988). The net result is that pigment concentrations obtained using the ratio of the water-leaving radiance at two wavelengths will be proportional to the inverse ratio of the absorption coefficients (GORDON et al. 1983). For Case II waters, such as river plumes, the situation is more complicated. For example, backscattering has a greater effect due to the larger amount of sediments near shore. Further offshore, sediments may settle out and absorption by CDOC may dominate (MOREL and PRIEUR, 1977). At low pigment concentrations, most of the particle backscattering results from detrital material. As phytoplankton concentration increases, the ratio of viable phytoplankton to detrital material also increases, so at higher concentrations, phytoplankton have a relatively greater effect on the optical properties of the through their absorption characteristics. Therefore, the relationship between and Lw(.A.2)/Lw(.A1 ) is non-linear (GORDON et al. 1983; GORDON et al. 1988).


25 optical Algorithms SMITH and BAKER (1978a, 1978b), and BAKER and SMITH ( 1982), developed a "bio-optical" model for ocean waters, thereby providing a relationship between physical measurements and the biological status of the ocean. The total diffuse attenuation coefficient for irradiance, KT(l), was chosen as the optical parameter of preference for several reasons. It relates the spectral irradiance just beneath the ocean surface, Ed(O,l), to the downwelling spectral irradiance at depth, Ed(z,l), thereby physically describing the bio-optical state of the ocean waters. PREISENDORFER (1976) had also shown that since KT is the apparent optical property relating irradiance just beneath the surface to irradiance at depth, it can be linked to radiant energy reflected from the upper layer of the ocean and seen by a remote sensor. This upper layer, called the upper attenuation length, KT-1 is where 90% of the diffusely reflected irradiance originates from (GORDON and McCLUNEY, 1975). Of particular interest to this study is the presence of other materials in the water such as colored dissolved organic carbon (CDOC), and suspended materials which mask and complicate the relationship of Lw to . The semiempirical model developed by SMITH and BAKER (1978a, 1978b), BAKER and SMITH (1982) and incorporated into GORDON et al. 's (1988) model. Kr(l) = !

26 Kc ( A ) = kc ( A ) CK { exp [ kc 2 ( A) ( 1 og 10 ( CK/ c0 ) ] 2 } + 0 0 0 1 c/ and I\-5(A) = D0[ay5(375) 0.06]*exp[-0.014(A -375)] KT total diffuse attenuation coefficient for irradiance (m-1 ) Kw diffuse attenuation coefficient for clear ocean waters (m-1 ) Kc = contribution to the diffuse attenuation directly attributable to all chlorophyll-like pigments that covary with chlorophyll (m-1 ) 1\-s = contribution to diffuse attenuation by the optically relevant colored dissolved organic carbon (CDOC) (m-1). (Note: BAKER and SMITH, 1982) kc = specific attenuation coefficient for irradiance due to plankton (chlorophyll-like) pigments [m-1 (mg Chl m-3 ) 1 ] c0 = surface concentration of chlorophyll-a + phaeopigments in the water column (mg pig m-3 ) CK = average chlorophyll concentration in water column to a depth of 1 attenuation length, Kr -1 (mg pig m-3 ) ay s = absorption coefficient dependence of yellow substances (CDOC, m-1 ) =a* [CDOC], where a*y is the specific ys s absorption coefficient Do = 1/COS eOw' downwelling distribution function


dependent on the solar zenith angle measured beneath the sea surface; nominally set to 1.1. In this equation kc, kc' and c0 are spectral fit parameters with values given in BAKER and SMITH (1982). 27 With the calculation of Ky, the upper attenuation length (Ky-1), of the euphotic zone is determined. If we compare the GORDON et al. (1988) algorithms shown in equation (5) to the analogous equation shown in BAKER and SMITH (1982), we see that where (5A) D0 and o:ys are defined in equation ( 5) kd(A0 ) = specific attenuation coefficient for irradiance due to colored dissolved organic carbon at a fixed wavelength o. 565 m1 (mg DOC li ter"1 ) 1 for Case I waters, (BAKER and SMITH, 1982) D = DOC (mg DOC liter-1). These relationships will be used to evaluate the contribution of CDOC and pigment to the color of the Orinoco plume. SMITH and BAKER (1978b), and BAKER and SMITH (1982), find a strong correlation between Kr-1 and CK only for waters whose dissolved and suspended material largely covaries with phytoplankton. Using equation (5) we can vary the CDOC concentration to mimic excess input of allochthonous or autochthonous DOC. Figure 7, based on the equations in Table 2 of BAKER and SMITH (1982), shows the additive contribution


,--, I E ..__, :::,c ..... c Q) '() :.;:: ..... Q) 0 (.) c 0 :;::; 0 :::l c Q) o+J < Figure 7 1 .6 1 5 1.4 I \ KT = Kw+KChi+Kd 1.3 1.2 1 1 \ \ 0.9 0.8 \ \ \ 0 7 \ \ 0.6 \ 0.5 04 r -"" __ 0.3 ..... .. 0. 2 ...... .. . ---0.1 I 0 300 4 00 w = water Chi = ch l orophy l l d =DOC T = total / / II . I ./.: / .... : / / .. ----___ :: .. : : _______ -.. --' 500 Wavelength [nm] 600 700 Light attenuation versus wavelength. KT, the total attenuation, is the sum of the attenuation of water, chlorophyll-like substances, and dissolved yellow substances (DOM) Maximum transmission occurs between 450nm-550nm (blue-green to green). (\) co


29 of each parameter on the total diffuse attenuation coefficient, Kr, in waters containing little particulate terrigenous material. As a first order estimate, a CDOC concentration of 0.5 mg l-1 and chlorophyll-a + phaeopigment concentration of 1 mg m-3 were added. Since CDOC has a low scattering to absorption ratio, we may increase CDOC and still use the models of GORDON et al. (1988). At the lower wavelengths (300nm-375nm) in Figure 7, only the organic parameters (chlorophyll plus detritus and DOC) have a significant attenuation effect. Bacteria may also have an effect here (STRAMSKI and MOREL, 1990). At intermediate wavelengths (520nm-550nm), chlorophyll-like pigments absorb very little, which results in enhanced blue-green and green water-leaving radiance. Deviations from theoretical models of the chlorophyll absorption model alone at wavelengths <500nm are primarily the result of increasing amounts of CDOC andjor suspended material. Most of the attenuation at the upper-end of the spectrum (red) is due to the strong absorption of water and a minimal amount due to pigments. DOC absorption in this range is negligible.


30 RESULTS Aerosol Correction In order to assess the possibility of overcorrection of the blue and green Lw in the river plume, due to misclassification of sediment reflectance at L8(670), several images were examined for covariation between pigment concentration and aerosol patterns. Figure 8 shows the aerosol radiance extracted from an image collected on October 9, 1979. This was one of the images that made up the weekly composite shown in Figure 4b. The results were typical of all images examined from this group. As seen in Figure 7, water absorbs strongly at 670nm. Essentially all the radiance measured can be attributed to scattering within the atmosphere over clear water. Any abnormal conditions, such as dust, sediment, or coccolithophore concentrations would become apparent and be seen as an increase in the L8 due to increased scattering. For the analysis of the October 9 image, transects were examined in the core as well as the outer edges along the trajectory of the plume. The edge of the plume was chosen where the pigment concentration measures <0.2 mg m-3 Figure 8 indicates no correlation between pigment concentration, [PIG], and average aerosol radiance at 670nm, < L8(670)> The coefficient of correlation, r2 was 0 .04 for n = 351. This


Figure 8 Aerosol radiance (670nm) vs. pigment concentration in the Orinoco River plume (9 October 1979). No discernible contribution can be seen due to atmospheric scattering. w f-l


32 comparison suggests that there is no further aerosol correction needed, at least in offshore areas of the plume. Closer to shore, large amounts of sediment complicate the interpretation of CZCS imagery. Here, reflectance from sands and clays is high and a land mask is applied by the processing algorithms based on high radiance in CZCS band 5. Water-Leaving Radiance Calculations The water-leaving radiance detected by the CZCS can be modeled by (GORDON et al. 1988): [Lw]N = [ (1 -p) (1 p)F0Rj(m2Q(1 -Rr))] (6) where p = Fresnel reflectance of the sea surface for normal incidence, = 0.021 p = Fresnel reflectance albedo of the sea surface for irradiance from the Sun and sky, mean value = 0.043 F0 = mean extraterrestrial solar irradiance, dependent on wavelength (GORDON et al. 1983) mW cm-2 J.Lm-1 R = irradiance reflectance just beneath the sea surface = EufEd where Eu and Ed are the upwelling and downwelling irradiances just beneath the surface respectively m = index of refraction of water= 1.3 Q = ster (for a totally diffuse radiance distribution, a value between 4 and 5 is used)


r = water-air reflectance for totally diffuse irradiance 33 The term (1-rR) is about 0.48 and accounts for the effect of internal reflectance of the upwelling radiance field by the sea surface. Lw was normalized to eliminate atmospheric variability and changes in the solar zenith angle. The clear water radiance concept (GORDON and CLARK, 1981; GORDON et al. 1983) assumes an accurate estimate of the radiance scattered by the atmosphere and sea surface. Since, in practice, the aerosol properties are highly variable in both space and time, and in general unknown, estimations are made from the satellite observation. The approach implemented in the NASA CZCS processing system use a default atmospheric correction which fixes the ratio of the aerosol radiance at two wavelengths to a constant. All images processed for this study used an epsilon coefficient, = 1. 0 ( relates the aerosol radiance at two wavelengths as a function of aerosol optical thickness and scattering albedo). This value was used since the czcs calibration was derived in a vicarious manner, and manipulation of without a more accurate basis may lead to unpredictable results. Clearly, errors in normalized waterleaving radiance can occur by using the default atmospheric correction values. Here, I assume that derived water-leaving radiances are correct, and develop a methodology to explore the abundance of optical constituents using a bio-optical model (GORDON et al. 1988). Equation ( 6) shows the relationship of [ Lw] N to the


34 optical properties of the water and R/Q. The spectral reflectance R, detected by the sensor, is dependent upon absorption and scattering. A relationship can be found between the absorption and scattering properties and R/Q This relationship links equations (5) and (6). GORDON et al. (1988) derive an expression relating the backscattering coefficient (bb) to R/Q as: R/Q = 0.110 (7) KT is calculated using equation (5). It then remains to relate the backscattering coefficient, bb, to the phytoplankton pigment concentration. Equation ( 3) shows total backscattering as a function of two components, water and phytoplankton, but the water contribution is known. We first note that backscattering, bb, of equation (3) is related to scattering, b, through: bb = bbb where bb is the dimensionless backscattering ratio or probability (GORDON and MOREL, 1983). An empirical relationship between scattering, b, and pigment concentration has been studied (GORDON and MOREL, 1983) and given as: where b ( 550) = b0c0 62 ( 8) b ( 550) = scattering coefficient at 550nm, m-1 b0 = regression analysis factor for scattering empirically measured at 550nm = o. 3 m-1 c = pigment concentration, mg m-3


35 This value for b0 was used by GORDON et al. (19SS) for [PIG) in the range of 0. 1 mg m3 to 20 mg m3 It is further assumed (GORDON and MOREL, 19S3) that (bb)p varies with wavelength and pigment concentration according to the relationship, (SA) where: (bb)p = backscattering coefficient due to particulate scattering, m1 A(A), B(A) = wavelength-dependent constants resulting from regression analysis, A(l) in m1 c = pigment concentration, mg m3 The link between equations (S) and (SA) was based on A(l) being proportional to b0 (GORDON and MOREL, 19SS). Therefore, the choice of A(A) will effect the value of water-leaving radiance. GORDON et al. (19SS) found that a single value of b0 measured at 550nm did not fit observed [Lw]N. This is evident comparing the theoretical [ Lw (A) ] N versus C curves for the 443nm, 520nm, and 550nm czcs bands in Figure 9 (GORDON et al. 19SS). By choosing different values of b0 (i.e. A(A)), GORDON et al. (19SS) were able to define an envelope for the data. For this study, I modified A(l) to fit the GORDON et al. (19SS) curve to the CZCS data points by changing [Lw(A)]N. The effect of changing wavelength can clearly be seen in Figure 9. [Lw(A) )N is reduced towards longer wavelengths because of the increased absorption by water (YENTSCH, 1960).


36 Pigment concentration variations effect the greatest changes in Lw at 443nm. As concentration increases, absorption has a proportionately greater effect on [Lw(l) ]N and the effects of A(l) are reduced. This is because the ratio of viable phytoplankton to detrital particles increases, with phytoplankton having a greater effect on the optical properties. A consequence of increased concentrations is the change of KT in equation ( 7) This causes a corresponding change in R/Q and a resultant decrease in [Lw(l)]N. This can be seen in Figure 9a (GORDON et al. 1988). Figures 9b and 9c show the effects at 520nm and 550nm. At these longer wavelengths absorption by phytoplankton and CDOC is at a minimum (Figure 7) and backscattering by particulate matter dominates. Therefore, [Lw(A) ]N is more stable over a wide pigment concentration range. For the curves in Figure 9, the effect of background CDOC is included in the Kc term of equation ( 5) When CDOC increases, as seen in Figure 10 (GORDON et al. 1988}, yet remains within a concentration representative of oceanic areas, a pronounced effect in the blue is seen because of the exponential spectral absorption curve of CDOC ( of Figure 7). In environments in which CDOC concentrations are very high (e.g. coastal regions where CDOC may reach an order of magnitude above the maximum concentrations of GORDON et al. (1988}), [Lw(A)]N becomes less sensitive to changes in pigment concentration, making the retrieval of [PIG] from radiance measurements difficult.


...:- e .. e u e ....... i E ... e u ........ ....... z ..... ..... 0 N "' J ._. ........ .. e ... E u ........ ....... ....! ........ 0 "' "' 1 0.01 0 .01 1 .00 0 .01 37 0 0 .10 1.00 10. 0 b 0 .10 1.00 10. 0 c O .lO 1 .00 10.0 Figure 9 Water-leaving radiance, [Lw(A)]N, versus pigment concentration derived from bio-optical algorithms (GORDON et al. 1988). [CDOC] = Background. This model shows the upper curve corresponding to b0 = 0.45 m-1 middle curve to b0 = 0.30 m -1 and the lower curve to b0 = 0.12 m -1 a) 443nm band b) 520nm band c) 550nm band. (GORDON et al. 1988).


...... ... ;; E .. :t E II E ....., ...... ... ;; E .,.:t E II E ....., 38 J .OO 0 b o .oo 0 01 0 .10 1.00 10. 0 c 0 0 1 0 .10 1.00 10.0 F igure 10 Water-leaving radiance, [Lw(A)]N, versus pigment and yellow substance (YS) concentration derived from bio-optical algorithms (GORDON et al. 1988). These are the results of the model, with b0 = o. 30 m1 and different curves correspond to different concentrations of YS. From top to bottom, the curves correspond to C y s = a s ( 3 7 5) 0.06 m1 of 0 .012, 0.024, 0.036, 0.048, 0.06 m 1 (GORDON et al. 1988).


39 Bio-Optical Analysis of CZCS Images The models of BAKER and SMITH (1982) and GORDON et al. (1988) were used to analyze the bio-optical properties of the Orinoco plume. Two images, which were representative of the many images available, were chosen for analysis. These images were collected on September 3, 1979 and October 9, 1979, with plume coverage from the mouth of the Orinoco to Puerto Rico, a distance of approximately 1200 krn. These images were part of the composites shown in Figure 4. Two regions were analyzed for which in situ CDOC was available (BLOUGH et al. 1992). One additional site was selected which was assumed to have negligible CDOC. These are designated as HIGH, INTERMEDIATE and CLEAR WATER, based on the observed CZCS-derived pigment concentrations. Within each of these, approximately 8-10 data points were extracted from the digital CZCS images of pigtnents and the three normalized water-leaving radiance bands: 443nm, 520nm and 550nm. Table 2 gives the range of concentrations and the geographical coordinates (see also Figure 11). TABLE 2 Location of regions studied and pigment concentrations Region [PIG_\ Lon Lat mg m (W) (N) HIGH 1.8 7.0 63391 121 INTERMEDIATE 0.3 0.8 67' 1218' CLEAR WATER 0.08 0.1 601 131


22 20 18 1 6 14 12 10 8 6 -78 -74 -72 -70 -68 -66 -64 -62 -60 58 Caribbean Sea Southern Sargasso Sea Puerto Rico L::) PLUME REGION 0 .CLEAR _.-HIGH. a Barbados L. Maracaibo Colombia Coria co T r e n c h Venezuela /(;j ( O r i n oc o R....... :__./..( Del t a 20 18 16 1 4 12 10 8 -78 -74 -72 -70 -68 -66 -64 -62 -60 58 Figure 11 HIGH, INTERMEDIATE, AND CLEAR water regions of the typical plume as indicated by the pigment concentration values given by the bio-optical model. The coordinates are given in Table 2. ""' 0


41 The INTERMEDIATE values were taken in the plume while the CLEAR values were taken outside the plume. A program was written to implement the algorithms given in equation (5) and {6), and run to generate the curves shown in Figures 12, 13, and 14. Table 3 shows the value of A(A) and B(A) used in the algorithms. A{A) was varied as a correction to the GORDON et al. {1988) model. The B(A) of GORDON et al. (1988) was used for each wavelength. It is assumed that the changes required in A(A) to model the CZCS data were due to backscatter. One explanation for this variation could be a change in the phytoplankton community composition. Pigment profiles and pigment composition have been reported to be distinctly different in the fall compared to the spring for this region (BIDIGARE et al. 1992). Since A(A) is proportional to b0 changes in A(A) have the greatest effect at low concentrations. Furthermore, equations (6) and (7) predict a shift in the GORDON et al. {1988) [Lw(A)]N curve in the same direction as the change in A(A), primarily for the blue band. Figures 12, 13, and 14 show this trend.


2 15 2 0 -1 .15 1.0 0 .15 f.-0 0 2 .!1 ,.......... L (!) ....... U> 2 0 E :J 1 .15 N E ..... -............ 0 "'-... 1 0 :3: E '--" 0 !5 c ,......., ,.......... t<) 0 0 ...:;!-2 .15 ....__,-3: _J L.......J --a b


o.e o.e I-o + I-0.2 I-0 0 o.a .........._ L Q) +' (I) E :::J N E u ........__ 3: E ...___, c ,.---, .........._ 0.0 0 N o.a L() ....__, _J L......J 0.2 Figure 13 .... a HIGH {Plume) cooc-o.s mg 1-1 A(szo)-3.3/1 000 m-1 .... .... ........ + b INTERMEDIAT (Plume) CDOC=0.2 mg 1-1 A(520)=3.3/1000 m 1 .. ----(Gordon's origina l value A=3.J) c ClfAR COOCBockoround A(520)=2.2 /1000 m 1 ------.0...... .....__ __ .0. ------0 1 0 1.00 Pigment (mg m-3) ,0.00 Normalized water-leaving radiance [Lw(A)]N at 520nm with varying concentrations of CDOC. 43


L.......J o.e I-0 .4-1c o.e f.(Gord on's o riginal value, A=3.3) CLEAR CDOC-Bockground ,6,(550)=1 5 / 1 000 m-1 0.<4-1\ ---..0.. --. --=------.0. Pig ment (mg m-3) Figure 14 Normalized water-leaving radiance [Lw(A)]N at 550nm with varying concentrations of CDOC. 44


TABLE 3 Variables used in algorithms to calculate (Lw(A)]N Wavelength 443 520 550 Region HIGH INTR CLEAR HIGH INTR CLEAR HIGH INTR A(A) 1 m ---3.0 ----3.3 ----A (A) 2 m 3.0 4.0 0.8 3.3 3.3 2.2 2.0 2.5 B(A) 1,2 0.22 0.22 0.22 0.35 0.35 0.35 0.36 0.36 cooc1 mg 1-1 ---0.0 ---0.0 --coocl mg 1 1 0.6 0.2 0.0 0.6 0.2 o.o 0.6 0 2 Key: A(A) (X 1000) 1GORDON et al. (1988) values used in Figure 9 curves 2values used in this study for curve fit 3BLOUGH et al. (1992) values 45 CLEAR 3. 3 1.5 0.36 0.0 o.o


46 INTERMEDIATE and HIGH concentration cases used values corresponding to in situ measurements of CDOC = 0. 2 mg 11 and CDOC = 0.6 mg 11 respectively (BLOUGH et al. 1992). For the CLEAR WATER case, it was assumed that excess CDOC = Background (GORDON et al. 1988). Clear Water Model Figure 12 compares the water-leaving radiance versus pigment concentration at 443nm. The theoretical GORDON et al. (1988) curve for Case I waters is also shown with the CZCS data (CDOC is considered to be negligible; Figure 12c) In order to fit the czcs data, A(A) was reduced from 3 0 x 103 m1 to 0. 8 x 103 m1 The result was a decrease in the water-leaving radiance, primarily at the low pigment concentrations, where backscattering has an increased effect due to the influence of detrital material (Figure 12c). At 0.10 mg 3 m the calculated [Lw(A) ]N decreased from approximately 1. 8 to 1. 2 mW cm2 1J.m1 ster1 The best fit for this curve is near the theoretical bottom curve given by GORDON et al. (1988) for b0 = 0.12 m 1 In Figure 13c, for the CLEAR WATER 520nm band, a value of A(520) = 2.2 x 103 m1 modeled the CLEAR WATER data well. Again, comparing this to Figure 9b, (GOR DON et al. 1988) the calculated curve fits between the theoretical mean and minimum curves whic h correspond respectively to b0 = 0 3 rn1 and bo = o. 12 m 1 For the 550nrn band in Figure 14c, a value of A ( 550) = 1. 5 x 103 rn1 was used and a similar trend is seen (GORDON et al. 1988; see their Figure 9c).


47 The results for CLEAR WATER czcs regions indicated that 1) it was necessary to decrease the backscattering coefficient for each band in order to model the czcs data and, 2) the magnitude of the change in A(l) was approximately 30% for the 443nm and 520nm band, and 50% for the 550nm band. These deviations may be due to errors in sensor calibration or not choosing the best clear water area for estimating Lw. However, in most cases, the magnitude of the value change remains within the guidelines given by GORDON et al. {1988). Intermediate/High Concentration The analysis of [Lw(l) ]N for the INTERMEDIATE and HIGH CZCS sites in Figures 12, 13, and 14, incorporated shipboard CDOC data (BLOUGH et al. 1992) taken at similar locations. CDOC was assumed not to zero since a) these regions showed higher CZCS-derived pigment concentrations than the values expected by BIDIGARE et al. (1992) estimates and b) in situ measurements of CDOC were available and showed values higher than background. The CDOC values were applied in equations (5) and (6). A(l) was then adjusted to fit the curves to the czcs data, reflecting a change in backscattering for the new conditions. CDOC Evaluation Two methods were used to deconvolve CDOC from pigment concentration in the CZCS imagery of the Orinoco River plume. The first method uses the ratio of measured CDOC to apparent


48 excess pigment concentration, [PIG] to obtain a CDOC conversion factor. The apparent excess [PIG] is the difference between the czcs-derived and measured in situ pigment concentration. Application of this conversion factor computes an estimate of the total CDOC observed by the satellite. The second method utilizes the GORDON et al. (1988) and BAKER and SMITH (1982) models, and in situ pigment measurements (BIDIGARE et al. 1992). Here, it is determined how much CDOC must be added to water containing only chlorophyll and phaeopigments to match the remotely-sensed pigment concentration. This added CDOC is then compared to in situ CDOC data (BLOUGH et al. 1992). Some of the locations analyzed are shown in Figure 15, with further details provided in Table 4. Values for colored dissolved organic carbon (CDOC) shown in Table 4 were obtained from high-temperature catalytic combustion shipboard data (BLOUGH et al. 1992). While the measurements of Table 4 correspond closely in location and time of year to those chosen from the satellite image, BLOUGH et al. and BIDIGARE et al's. data were collected in 1988, and the CZCS data in 1979.


49 TABLE 4 Locations of stations used as test sites of GORDON et al's (1988) model. Values of relevant variables are shown. STA CRUISE LAT LON [CDOC] [PIG] SAL [PIG] (N) (W) meas me as czcs R5 BL/BID 81 601 2.5 0.48 7.5 3.17 M BL/BID 111 621 0.6 0.75 30.5 1. 33 H BL/BID 141 641 0.2 0.30 33. 6 0.61 Key: [ CDOC] in mg 1-1 [PIG] in mg m-3 SAL = Salinity in psu (practical salinity units) BL/BID =BLOUGH et al. (1992), BID= BIDIGARE et al. VARIABLES [PIG] czcs [PIG] meas [CDOC]meas [AEP] [ CDOC] cf [CDOC] tot = = = = (1992) czcs pigment concentration (chlorophyll + phaeopigments accessory pigments + CDOC) In situ pigment value (chlorophyll + phaeopigments + accessory pigments) In situ CDOC from ship cruises apparent excess CZCS pigment concentration that is the difference between observed (CZCS) and measured pigment concentrations. This excess represents an optically sensed substance assumed to be CDOC = CDOC conversion factor representing 1 mg m -3 of excess CZCS pigment concentration = total CDOC concentration after applying the conversion factor to the apparent excess CZCS pigment concentration


Figure 15 The study location and the in situ data; ( [CDOC] (mg 1 -1 ) 1 [PIG] (mg m -3 ) 1 and salinity (psu) are shown in the plume region. Salinity and [PIG] trends suggest the riverine influence throughout the length of the plume. lJl 0


51 BIO-OPTICAL MODELS: APPLICATION Mathematical Treatment To illustrate this method, an average [CDOC] conversion factor is calculated based on measured data for Station M and Station H (Figure 15 and Table 4). This factor is then applied to another station, RS, to calculate the 'apparent' (CDOC] Stations were chosen for which complete information was available to model the bio-optical state of the plume. This information included measured and observed [PIG] and measured (CDOC] Pigment concentration was obtained from the t w o representative images mentioned previously for the months of September and October, 1979 (Figure 4). A data set consisting of approximately 4 x 4 pixels was extracted, corresponding to Stations RS, M, and H Surface shipboard measurements, made during a Fall cruise at these stations (BIDIGARE et al. 1992) included various types of chlorophyll (i. e Chl a, Chl b, Chl c) plus other dominant accessory pigments such as fucoxanthin, diadinoxanthin, zeaxanthin and For purposes of this analysis, I used only the values of chlorophyll-a + phaeopigment.


Method I: CDOC Conversion Factor ([CDOC]cf) STATION M [PIG] czcs [PIG] meas = [AEP] where [AEP] = apparent excess pigment (9) 1.33 mg m"3 -0.75 mg m3 = 0.58 mg m 3 apparent excess pigment [CDOC]meas 52 [CDOC] cf = (10) [AEP] 0. 6 mg 11 1. 034 mg 1"1 CDOC = -----------= -------------------0. 58 mg m3 STATION H [PIG] czcs [PIG] meas = [AEP] mg m "3 apparent excess pigment 0 61 mg m-3 0 3 0 mg m-3 = 0 31 mg m-3 o. 2 mg 1"1 [ CDOC] cf = -----------= 0. 31 mg m3 0.645 mg 1"1 CDOC mg m3 apparent excess pigment AVERAGE = [CDOC]cf PER 1 MG M "3 APPARENT EXCESS PIGMENT = 0. 840 mg CDOC 1"1 (mg pig m"3 ) 1 The difference in [CDOC]cf at these stations may be due to variability associated with the CZCS calibration or in situ and satellite data collected in different years and therefore in different water masses. Differences may also occur relative to the types of CDOC or phytoplankton community present. The average conversion factor was applied to the AEP of Station R5 to obtain a [CDOC] value for comparison to the measured value shown in Table 4. The total observed CDOC was


53 obtained for Station R5 was: [PIG]czcs -[PIG]mea = [AEP] 3 .17 mg m-3 -o. 48 mg m-3 = 2. 69 mg m-3 [AEP] X [CDOC]cf = [CDOC)t t 2 69 mg m-3 x o. 840 mg r1/mg pig m-3 = 2. 26 cooc mg r1 The 2. 2 6 mg r1 observed value of 'apparent' [CDOC] compares favorably with BLOUGH et al.'s (1992) measured value of 2. 5 mg -1 When concurrent ship and satellite data are available, we will have an opportunity to more accurately use this simple procedure. The basic concept can best be applied to any region providing that at least one of the optical components in the water (i.e. pigment or CDOC) are measured. METHOD II: Calculations from Optical Algorithms The other approach used the bio-optical models of GORDON et al. (1983) and BAKER and SMITH (1982). First, I chose a pigment concentration from GORDON et al 's. ( 1988) models (Figures 9 and 10) corresponding to the measured concentration of BIDIGARE et al. (1992). Second, I extracted the and [Lw(520) ]N values from the CZCS images at the same location corresponding to the in situ and CZCS [PIG] data collection. [Lw(443) ]N is used for CZCS [PIG] <1.5 mg m-3 and [Lw(520) h for [PIG] >1.5 mg m-3 (GORDON et al. 1983). Finally, I increased the CDOC in the GORDON et al. (1988) algorithm (equations (5) and (6)) to lower the calculated [Lw(A)]N to match the CZCS [Lw(A) JN. value. This CDOC was then compared to the in situ measurements of BLOUGH et al. (1992).


54 DATA INTERPRETATION The difference in the CZCS-derived [PIG] and the measured [PIG] is considerably higher for Station R5 than for either Station M or H. This is expected since R5 is located near the mouth of the Orinoco River (Figure 15}, where both phytoplankton and CDOC should be high due to undiluted nutrient and CDOC runoff. Figures 16, 17 and 18 show the results of applying the bio-optical model at Stations M, H, and R5. At Station M (Figure 16}, an amount of CDOC = 0.65 mg 1"1 was added to GORDON et al. 's (1988) model to lower [Lw(520) ]N from point "X" to point "Y". This compared favorably to the results using a measured 0. 6 mg 1"1 of BLOUGH et al. ( 1992) Therefore, the GORDON et al. ( 1988) model seems to explain the data from Station M. In contrast, at Station H (Figure 17), an amount of CDOC = 0.9 mg 1"1 was added to lower [Lw(443)]N to point "Y". The effect of adding BLOUGH et al. 's (1992} measured concentration of 0.2 mg 1"1 CDOC is also shown for comparison. Clearly, this was insufficient to match the CZCS [Lw ( 443)] N but this may be attributable to patchiness in the distribution of river water off the continental margin or to a different ratio of terrigenous:marine CDOC.


-'2' Cl) ....-0) E ::J N E ? E ..._., ,.......... 0 N I.{) ..._ 3: _J 0 8 .....-.., _,.... _,.... GORDON et ol. (1988) CDOC Background A(520) = 3 .371000m-1 ---............. Station M ""'-..._X "'-------B LOUGH ot ol. (1992) cpoc 0.6 ms. 1-1 A(520) 3.3(1000 m-1 -!-----y _:----I CDOC 0.65 mo 1-1 A(520)-3.3/1000m-1 BIDIGARE e t ol. (1992) 0 0 0 01 0.10 751.00 Pigmen t (mg m-3) 10.00 Figure 16 Station M. Values of [CDOC] added to the GORDON et al. (1988) bio-optical model (equations (5) and (6)) to match the czcs water-leaving radiance at 520nm at in s itu measured [PIG] (BIDIGARE et al. 1992). The resultant curv e with BLOUGH et al.'s (1992) in situ CDOC is shown for comparison. U1 U1


Figur e 17 ,......._ >.. 2 (/) E :J N E E ,.-...., n ........., 3 _j 2 51 -........... \ 2 0 1-1.5 BLOUCH et a l. ( 1992 ) COOC = 0.2 1 1 A { 443 } = 4 0 / 000 m-1 -. . .... 1 0 0 5 1--I \ GORDON et a l. (1988) COOC B

2.0 czcs Q) 't; 1 .5 E :J N E 0 ..s 0 1 0 N lD '-"' _J BLOUGH et o l (1992) CDOC 0.6 "'9 1 1 ..... t .. Station R5 0.01 0. 1 0 0.48 1.00 10.00 Pigment (mg m -3) Figure 18 Station RS. Effects of high reflectance sediment on the GORDON et (1988) bio-optical model. Ul


58 Figure 18 shows the results for Station R5. The [Lw(520)]N value of 1. 53 mW J,m-1 cm -2 ster-1 from the czcs (point "Y") exceeds the predicted value (point "X", GORDON et al. 1988) by over 1 mW J,m-1 cm-2 ster-1 under the condition of BIDIGARE et al. 's (1992) pigment concentration measurement of 0.48 mg m-3 It is assumed that this is due to excessive backscatter by nearshore sediment or bottom reflection (water depths here are less than 50 m) pointing out one of the problems that occur in coastal waters and near river mouths. The interpretation of results shown in Figures 16 and 17 requires an evaluation of the CDOC component. The absorption spectra of pure marine waters and those waters containing yellow substances differs markedly (MOREL and PRIEUR, 1977; KIRK, 1983). Furthermore, CDOC can be of terrigenous or marine origin, or a mixture of these. Much of the literature is not clear in terms of the values to be assigned to the specific absorption coefficient of each type nor the proportion of eac h that would be expected in various oceanic regions (i.e coastal versus offshore waters or Case I versus Case II waters). One investigation does suggest that marine DOC contribution due to phytoplankton excretion can average about 23% whether shelf, slope or basin (WALSH et al. 1992) However, if terrigenous and marine CDOC are different, and if their ratios differ depending on location from shore, the value of (CDOC] may vary. There also seems to be a difference between terrestrial and marine DOC absorption spectra (CARDER et al. 1989). If


59 [Lw(A)]N is dependent on the type of CDOC, then clearly it is necessary to investigate what the appropriate spectral absorption characteristics are for terrigenous and marine humic substances present in the plume. One postulated difference between marine and terrigenous CDOC lies in their specific absorption coefficients, which depend on molecular weight differences in the humus fraction. Marine humus tends to have a lower molecular weight than terrigenous humus (CARDER et al. 1989). Furthermore, the slope of the spectral absorption curve provides a measure of the relative fraction of fulvic and humic acid present (CARDER et al. 1989). CARDER et al. (1989) suggested that it is reasonable to assume a 90% fulvic and 10% humic composition for typical marine CDOC. The total specific absorption coefficient a*9(A), is then calculated by adding the individual humic and fulvic specific absorption values, a*h and a*f' each multiplied by its respective fraction of the total humic and fulvic acid present. BLOUGH et al. ( 1992) obtained a specific absorption coefficient for CDOC (a*9(300nm)) of 9.1 1 mg-1 m-1 at the mouth of the Orinoco River. At Station R5, approximately 60 km from the mouth, BLOUGH et al. (1992) finds a value of 2.5 1 mg-1 m-1 These values are much higher than CARDER et al's. (1989) 0.29 1 mg-1 m-1 measured in the Gulf of Mexico's Loop Current. This disparity led BLOUGH et al. (1992) to the conclusion that the CDOC in the Orinoco River plume was terrestrially-derived. This trend was also seen at 450nm with an a*9(450nm) value of


60 approximately 1.17 1 mg-1 m-1 (BLOUGH et al. 1992; their Table 1) which again is higher than 0. 034 1 mg-1 m-1 measured by CARDER et al. (1989). The qualitative and quantitative relationship between terrestrial and marine CDOC is unclear. However, it is assumed that terrestrial and marine CDOC occur as a mixture in all oceanic environments. Even at the mouth of the Orinoco River, where the terrigenous CDOC is most likely predominant, the phytoplankton bloom that occurs may introduce marine CDOC as a result of excretion and death. However, these processes of CDOC formation, mixing and dilution have never been quantified. Further complications arise with the fulvic:humic ratios. In contrast to the 90:10 fulvi c :humic ratio used for the marine specific absorption coefficients, a 50:50 or 60:40 fulvic:humic ratio may be applicable for the terrigenous case (CARDER, pers. comm .). Furthermore, the ratios probably vary with oceanic region, thereby allowing for the possibility of finding marine locations where a*9 is similar to terrigenous conditions. Much work is needed to accurately define the true nature of yellow substances and quantitatively determine their light absorbing characteristics. Another interpretation of the variation between CZCS and measured CDOC values at Station H (Table 4, Figure 17) could be the phytoplankton community composition. A certain composition unique to the Orinoco plume in the Caribbean (c.f. BIDIGARE et al. 1992) may suggest a characteristic spectral


61 response based on a community specific absorption coefficient, a*. This a* is probably different from the a* used to derive the bio-optical models. Further nearshore, another factor that might contribute to differences in absorption spectrum is the diversity of light-absorbing particles within the water column. Specifically, while it was assumed that sediment is not a significant component in waters some 20-50 km from the coast, low concentrations of red clay might affect [Lw(443)]N. Bacteria (<3 may also play a role. STRAMSKI and MOREL (1990) cite the phycocyanin-rich cyanobacterium, Synechocystis, as an example of an efficient absorber of blue light. In general, cyanobacteria exhibit specific optical properties because they are small and contain biliproteinabsorbing pigments. STRAMSKI and MOREL (1990) also argue that picoplankton should exhibit an increased efficiency (per unit of pigment) in capturing light. Even though picoplankton such as Synechocystis do scatter light, absorption diminishes the effects of scattering in the absorption bands (STRAMSKI and MOREL, 1990; MOREL and BRICAUD, 1986). The possible significance of these small organisms can be appreciated by noting that picop1ankton can represent 25%-90% of chlorophyll biomass and 20%-80% of the primary production in the open ocean (Li et al. 1983). In the caribbean Sea, >85% of chlorophyll-a is associated with the nanoplanktonic size fraction except when colonial cyanobacteria (Trichodesmium spp.) are abundant in surface


62 waters (BIDIGARE et al. 1992). BIDIGARE et al. (1992), using pigments (diatoms) as primary taxonomic markers, found fucoxanthin in the upper 40 meters of the water column at Station H with the upper 10 meters dominated by zeaxanthin (cyanobacteria). In contrast, near the mouth of the Orinoco River, (Station M), fucoxanthin clearly dominated throughout the entire upper 40 meters of the water column (Figure 3 and 4 of BIDIGARE et al. 1992). An examination of those data shows that zeaxanthin could represent approximately 20% of the total pigment concentration measured near Station H. Furthermore, BIDIGARE et al. (1992} discusses the role of bacteria as a mechanism for altering the CDOC spectral signature. Early studies concluded that dissolved organic carbon released from marine phytoplankton appears to be only about 3.2% of total productivity, and concentrations of DOC were not found to be significantly changed by microzooplankton grazing or otherwise associated with zooplankton abundance (SELLNER, 1981; WILLIAMS and YENTSCH, 1976). This suggests a minimal effect of marine DOC on bacterial growth. However, more recent studies linking DOC to the existence of bacteria, suggest that extra-cellular release (ER) by phytoplankton increases linearly with productivity and provides up to 50% of the high quality carbon required to support bacterial growth in natural systems, with the balance of the bacterial needs provided through sources such as sloppy feeding, zooplankton excretion and phytoplankton senescence (BAINES and PACE, 1991}. Another factor in the variation of [Lw(443)]N is


63 "patchiness". BLOUGH et al. (1992), noted the significantly different absorption characteristics between two stations in close proximity. Patchiness problems may be addressed with the availability of a new ocean color sensor. One other factor affecting [ Lw ( >..) ] N may be phaeopigments. Phaeopigments are an important part of total pigment in oligotrophic waters (BALCH et al. 1992), and are an integral part of the CZCS algorithms. The dominant source of phaeopigments is microzooplankton grazing, in the form of fecal debris suspended in the euphotic zone due to its slow sinking rate (WELSCHMEYER and LORENZEN, 1985). They may also play an important role in the Orinoco River plume. Quantifying this parameter in real-time can be difficult since photodegradation has been invoked as a significant cause for loss of phaeopigments from the euphotic zone with upwards of 92% disappearing exponentially with depth over one day under typical sunny conditions (WELSCHMEYER and LORENZEN, 1985). Clearly, it is difficult to ascertain the exact reasons for the variations in Figures 16-18 without concurrent measurements related to the specific water constituents. Certainly, the assumed pigment concentration, CDOC concentration, and CDOC specific absorption coefficient play major roles in the ability to model water-leaving radiances. Below I explore how changes in the specific absorption coefficient for CDOC, a* in different oceanic regions would g affect the radiance model of GORDON et al. ( 1988) Three variables were considered to generate the curves of Figures 19


64 and 20 (see Table 5). a*9(l) = specific absorption coefficient for CDOC, 1 mg -1 m-1 S = spectral slope for the wavelength dependence of the CDOC absorption coefficient, nm-1 [CDOC] = CDOC concentration, mg 1 -1


2 5 1-" Blough Corder Gordon G B c G og(375) 3.7 0.15 0.565 _.._ \ ( I mg 1 m-1) ..-S (nm 1) 0.014 0.0194 0.014 I 2 0 L [CDOC] (mg 1-1) 0.6 0 6 0.106 Q) +-' Sal i n i t y ( p s u ) 31 >24 >35 (f) .....c I E 1 5 :::l N I E (.) 3: E '-../ z ,......., ,---... t') -.;!-.;!-0.5 '-../ 3: _j '--' t 8 0.01 0.10 1.00 10.00 Pigmen t (mg m -3) Figure 19 Application of different measured specific absorption coeffi c ients, a*,, from the Orinoco River and Mississippi River plumes with equal [CDOC). 0'1 L11


......---, ..--I L Q) ........ (/) ..-2.5 1-2.or I E ::.! 1.5 N I E (.) 3: E ....._., z ,..--, ......---, t") ._...... _j L......J I \ \ B Blough Corder Gordon B c G 0*9(375) (I mg1 m-1) 3.7 0.15 0.565 S (nm-1) 0.014 0.0194 0.014 [ cooc] (mg 1 1) 0.6 0 3 0.106 Salinity (psu) 31 >24 >35 0. 0 0.01 0.10 1.00 10.00 P igment (mg m-3) Figure 20 Application of different measured specific absorption coefficients, a*g from the Orinoco River and Mississippi River plumes with unequal [CDOC). 0\ 0\


67 TABLE 5 Specific absorption coefficient values for different oceanic regions. GORDON CARDER BLOUGH et al. et al. et al. (1988) (1989) (1992) a*111(375) 0.565 0 .15<1> 3. 7(2 ) 1 mg1 m"1 s nm1 0 014 0. 0194<1> 0. 014<2> [CDOC] 0.106 0. 6<3> 0. 6<2> mg 1 1 The GORDON et al. (1988) bio-optical model value of A(440) = 3. 0/1000 m 1 is used for all curves. (1) CARDER et al. (1989) value, Mississippi River plume (2) BLOUGH et al. {1992) value, Nearshore plume (Station M) (3) HARVEY et al. (1983) This assumes CDOC = DOC/2 For this analysis: The CARDER et al. (1989) Mississippi River plume water (Salinity = >24 psu) will be considered to be similar to BLOUGH et al. (1992) nearshore plume (Station M) water (Salinity= 30.5 psu).


68 Equation (5A} shows that the absorption coefficient, ays' is a function of the product of (CDOCJ and the specific absorption coefficient, a* () .. 0 ) Once a* () .. 0 ) and the (CDOC] have been defined for any oceanic region, (equation 5) then affects the value of (Lw(443) ]N. The bio-optical model (GORDON et al. 1988} which generates the curve shown in Figure 9a assumes a background absorption coefficient, ay5(375) = 0.06 m-1 (corresponding to (CDOCJ = 0.106 mg r1), thus making = o.o m-1 This means that the (Lw(443} ]N obtained is maximum for these waters given the specified scattering properties. Any change in the product of (CDOC] and a*(A0 ) making ay5(375) greater than zero will result in a decrease in [Lw(443)]N. Furthermore, changes in (CDOC] or a*(A0 ) causing ay5(375) to decrease from its theoretical background value of 0.06 m-1 will cause the model to give invalid results. Figure 19 (Table 5) compares what should be similar water types (i.e. Orinoco River and Mississippi River plumes). It can be seen that the curve based on the parameters from BLOUGH et al. show a considerable difference in [Lw(443)]N compared to the curve based on CARDER et al. (1989} parameters. The specific absorption coefficient is the determining factor. This suggests that the constituents of the water are different. The GORDON et al. (1988} curve is also shown for comparison and again it seems obvious that the default model cannot be used for either the Mississippi or Caribbean region, at least during the season when the studies were undertaken.


69 Figure 20 shows what happens when the absorption due to (CDOC) goes below this critical background value (i.e o. 06 mg 1-1 ) for the GORDON et al. (1988) model. The same scenario was used as in Figure 19 except that (CDOCJ was reduced to 0.3 mg 1-1 This made 1\.s ( 443) negative thereby causing (Lw ( 443)) N to exceed the predicted maximum value (curve C) allowed by the GORDON et al. (1988) bio-optical model (curve G). While not shown, this same trend occurs when using the specific absorption coefficient of GORDON et al. (1988) and that measured in oligotrophic waters such as the Gulf Loop Current (CARDER et al. 1989). Of course, a different phytoplankton community composition in each case cannot be ruled out. While the hypotheses of change in spectral characteristics due either to a) species composition shift or b) DOC composition differences cannot be tested at this time, future efforts in obtaining concurrent measurements for the variables stated above should provide a higher confidence level in the interpretation of ocean color satellite images.


70 CONCLUSION The data examined (KETCHUM and RYTHER, 1966) shows strong evidence of coastal upwelling along the nearshore portion of the Orinoco River delta (Figures 2 and 3). This together with the river discharge adds sufficient nutrients to generate a significant bloom during the Fall season which is optically detected by the czcs. An examination of the plume region indicated that atmospheric correction models may be used for offshore Case II waters in this region. Furthermore, even though almost ten years had transpired between the observed and measured data, the data used in this study was considered to be representative of the plume. Absorption and scattering are the primary parameters determining the water-leaving radiance, Lw. The blue region of the spectrum is more sensitive to backscattering at low concentrations, while pigments at higher concentrations strongly influence absorption. However, the presence of CDOC confounds the relationship between pigment concentration and Lw at shorter wavelengths. corrections were necessary to the CZCS bio-optical model in order to make it applicable for this region. The backscatter values (i.e. A(l)) of the bio-optical model (GORDON et al. 1988) were modified for the three CZCS bands


71 (i.e. 443nm, 520nm, and 550nm). This was done for regions of the plume representing INTERMEDIATE, and HIGH pigment concentrations and a CLEAR WATER region. Generally, the A(l) values were reduced which lowered (Lw(443)]N (Figures 12, 13, and 14). It is clear that CDOC is an important constituent in the water. To take full advantage of the remote sensing capability, CDOC should be defined according to its type (i.e. terrigenous or marine) and its spectral characteristics quantified. While the proportion of terrigenous to marine CDOC is not clear, the values in Table 4 show the Orinoco River influence. As the plume disperses offshore, the dilution effects on CDOC concentration are evident (Figure 21) Two methods are shown to evaluate the amount of excess CDOC. A ratio method (equation (10)), resulting in an effective CDOC conversion factor, gave good results for the single station for which data were available. A second method adjusted the [Lw(l) ]N derived from GORDON et al. 's (1988) model to the [Lw(l)]N obtained from CZCS data, using in situ pigment and CDOC concentrations. Here, mixed results were achieved. For Station M (Figure 16), the observed excess [CDOC] matched the in situ measurement well. In the case of Station H (Figure 17), the observed [CDOC] exceeded the in situ measurement by more than a factor of 4. For Station R5 (Figure 18), it was concluded that the sediment in this nearshore region scattered the light, thereby invalidating the model results. This latter


3 R5 I ,...... .....-I 2 O'l E .............. r---'1 0 0 0 0 L.......J H 0 0 10 20 30 SALINITY (PSU) Figure 21 Dilution effects of [CDOC] following the salinity change from nearshore Station RS to Station H offshore. N


73 condition may not be uncommon for remotely sensed images of nearshore environments. These results suggest that additional corrections may be necessary to obtain more consistent results for all areas under study. Because absorption plays a dominant role in remote sensing, an attempt was made to determine the effect that changes in CDOC specific absorption coefficient, a* would 9 have on water-leaving radiance, Lw. Values of a*9 obtained from studies of other oceanic waters (BLOUGH et al. 1992; CARDER et al. 1989) were compared to the CZCS model (GORDON et al. 1988) which uses an a* 9 value of 0. 565 1 mg-1 m-1 and a "background 11 CDOC concentration (i.e. equal to 0.106 mg 1-1 ) corresponding to ay5(375) = 0.06 m-1 It was clearly evident that even if two supposedly similar water masses contain comparable [ CDOC] but differing a*9 values, significant variations can occur in [Lw(A) h (Figure 19). Furthermore, regardless of the [CDOC] value, if the product of a*9 and [CDOC] decreases below the bio-optical model value of a% = 0.06 m-1 the model will give erroneous results (Figure 20). This study concludes that the default parameters of the bio-optical model are not applicable in all oceanic regions. Further research is needed to characterize a*9 and spectral slopes. More intense utilization of current models using available data sets is encouraged, to help point out the limits within which the model parameters can operate. This study points out a number of problems that need to be addressed if satellite imaging is to become a viable tool


74 for the biogeochemical study of the ocean waters. It seems reasonable to conclude that current bio-optical models with appropriate corrections may be used for Case II waters provided concurrent data is available for chlorophyll or CDOC or both. Without this, satellite data cannot be interpreted. At the present time, satellite data provides estimates with an accuracy of % of in situ pigment, and accounts for only % of the total biomass in the water column (BALCH et al. 1992). To obtain more accurate results, particularly in Case II environments, it will be necessary to have concurrent in situ data on CDOC, accessory pigments, patchiness, dust, sediment, bacteria, photodegradation, and species composition.


LIST OF REFERENCES BAINES, S.B. and M.L. PACE. 1991. The production of dissolved organic matter by phytoplankton and its importance to bacteria: Patterns across marine and freshwater systems. Limnology and oceanography, 36:1078-1090. 75 BAKER, K.S. and R.C. SMITH. 1982. Bio-optical classification and model of natural waters. Limnology and Oceanography, 27:500-509. BALCH, W., R. EVANS, J. BROWN, G. FELDMAN, C. McCLAIN, and W. ESAIAS. 1992. The remote sensing of ocean primary productivity: Use of a new data compilation to test satellite algorithms. Journal of Geophysical Research, 97 C2:2279-2293. BETZER, P.R., D.W. EGGIMAN, K.L. CARDER, D.R. KESTER, and S.B. BETZER. 1977. Seasonal patterns in suspended calcium carbonate concentration during the dry and wet seasons in the eastern Caribbean. In: The Fate of Fossil Fuel C02 in the Oceans, N.R. Andersen and A. Malahoff, editors, Plenum Press, 63-79. BIDIGARE, R.R., M.E. ONDRUSEK, and J.M. BROOKS. 1992. Influence of the Orinoco River outflow on distributions of algal pigments in the Caribbean Sea. Journal of Geophysical Research, (In press). S. SWEET, and M.E. ONDRUSEK. 1989. R/V ----------COLUMBUS ISELIN CRUISE CI-8816, Preliminary Cruise Report, Geochemical and Environmental Research Group, Department of Oceanography, Texas A & M University, 83-100. BLOUGH, N.V. and O.C. ZAFIRIOU. 1992. Optical absorption spectra of waters from the Orinoco River outflow: terrestrial input of colored organic matter to the caribbean. Journal of Geophysical Research, (In press). CALEF, G.W. and G.D. GRICE. 1967. The influence of the Amazon River outflow on the ecology of the western tropical Atlantic. II. Zooplankton abundance, copepod distribution, with remarks on the fauna of


low-salinity areas. Journal of Marine Research, 25:84-94. 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. Limnology and Oceanography, 34:61-68. 76 EDMOND, J.M., E.A. BOYLE, B. GRANT, and R .F. STALLARD. 1981. The chemical mass balance in the Amazon plume I: The nutrients. Deep-Sea Research, 28A:1339-1374. FROELICH, P.N., O.K. ATWOOD, and G.S. GIESE. 1978. The influence of Amazon River water on the surface salinity and dissolved silicate concentration in the Caribbean Sea. Deep-Sea Research, 25:735-744. GORDON, H.R., O.B. BROWN, R.H. EVANS, J.W. BROWN, R.C. SMITH, K.S. BAKER, and O.K. CLARK. 1988. A Semianalytic Radiance Model of Ocean Color. Journal of Geophysical Research, 93:10909-10924. -----------, and D.K. CLARK. 1981. Clear water radiances for atmospheric correction of Coastal Zone Color Scanner Imagery. Applied Optics, 20:4175-4180. -----------' J.W. BROWN, O.B. BROWN, R.H. EVANS, and W.W. BRINKS. 1983. Phytoplankton pigment concentrations in the Middle Atlantic Bight: Comparison of ship determinations and CZCS estimates. Applied Optics, 22:20-35. , J.L. MUELLER, and W.A. HOVIS. ----------1980. Phytoplankton pigments derived from the Nirnbus-7 CZCS: Comparisons with surface measurements. Science, 210:63-66. and W.R. McCLUNEY. 1975. Estimation of the depth of sunlight penetration in the sea for remote sensing. Applied Optics, 14:413-416. and A. MOREL. 1983. Remote assessment of ocean for interpretation of satellite visible imagery: A review. Springer-Verlag, 114pp. HARVEY G.R., D.A. BORAN, L.A. CHESAL, and J.M. TOKAR. 1983. The structure of marine fulvic and humic acids. Marine Chemistry, 12:119-132. HOREL, J.D., V.E. KOUSKY, and M. KAGANO. 1986. Atmospheric conditions in the Atlantic sector during 1983 and 1984. Nature, 322:248-251.


HOVIS, W.A. and K.C. LEUNG. 1977. Remote sensing of ocean color. Optical Engineering, 16:158-164. HULBURT, E.M. and N.CORWIN. 1969. Influence of the Amazon River outflow on the ecology of the western tropical Atlantic. III. The planktonic flora between the Amazon River and the Windward Islands. Journal of Marine Research, 27:55-72. JOHNS, W.E., T.N. LEE, F.A. SCHOTT, R.J. ZANTOPP, and R.H. EVANS. 1990. The North Brazil Current retroflection: Seasonal structure and eddy variability. Journal of Geophysical Research, 95:22103-22120. 77 KATZ, E.J. 1987. Seasonal response of the sea surface to the wind in the equatorial Atlantic. Journal of Geophysical Research, 92 C2:1885-1893. KETCHUM, B.H. and J.H. RYTHER. 1966. Biological, chemical, and radiochemical studies of marine plankton. A.E.C. Report No. NYO 1918-138, Unpublished manuscript. KINDER, T.H., G.W. HEBURN, and A.W. GREEN. 1985. Some aspects of the Caribbean circulation. Marine Geology, 68:25-52. KIRK, J.T.O. 1983. Light and Photosynthesis in Aquatic Ecosystems. Cambridge University Press, 357pp. LEWIS, JR., W.M. and J.F. SAUNDERS. 1989. Concentration and transport of dissolved and suspended substances in the Orinoco River. Biogeochemistry, 7:203-240. LI, W.K.W., D.V. SUBBA-RAO, W.G. HARRISON, J.C. SMITH, J.J. CULLEN, B. IRWIN, and T. PLATT. 1983. Autotrophic picoplankton in the tropical ocean. Science, 219:292-295. MEADE, R.H., C.F. NORDIN, JR., D. PEREZ HERNANDEZ, A. MEJIA B., and J.M. PEREZ GODOY. 1983. Sediment and water discharge in Rio Orinoco, Venezuela and Colombia. In: Proceedings of the Second International Symposium on River Sedimentation, 11-16 October, 1983, Nanjing, China. Water Resources and Electric Power Press, Beijing, China, 1134-1144. MEYBECK, M. 1982. Carbon, nitrogen and phosphorus transport by world rivers. American Journal of Science, 282:401-450.


78 MOREL A. 1974. Optical properties of pure water and pure sea water. In: Optical Aspects of Oceanography, edited by N.G. Jerlov and E.S. Nielson, Academic, 1-24. ___________ and A. BRICAUD. 1986. Inherent optical properties of algal cells including picoplankton: Theoretical and Experimental Results. In: Photosynthetic Picoplankton, T. Platt and W.K.W. Li editors, Optical properties of pure water and pure sea water. In: Optical Aspects of Oceanography, Canadian Bulletin of Fisheries and Aquatic Sciences, ottawa, 214:521-559. ___________ and L. PRIEUR. 1977. Analysis of variations in ocean color. Limnology and Oceanography, 22:709-722. MORRISON, J.M. and W.O. NOWLIN, JR. 1982. General distributions of water masses within the eastern Caribbean Sea during winter of 1972 and fall of 1983. Journal of Geophysical Research, 87 C6:4207-4229. MULLER-KARGER, F.E., C.R. MCCLAIN, T.R. FISHER, W.E. ESAIAS, and R. VARELA. 1989. Pigment distribution in the Caribbean Sea: Observations from space. Prog. Oceanography, 23:23-64. , and P.L. RICHARDSON. 1988. The dispersal of the Amazon's water. Nature, 333:56-59. PAOLINI, J., R. HERRERA, and A. NEMETH. 1983. Hydrochemistry of the Orinoco and Caroni Rivers In: Transport of carbon and minerals in major world rivers, 2:223-234. PARSONS, T.R., M.TAKAHASHI, and B. HARGRAVE. 1977. Biological Oceanographic Processes. 2nd Edition, Pergamon International Library, 332pp. PREISENDORFER, R.W. 1976. Hydrologic optics. V.1,2,3. NOAA. REDFIELD A.C., B.H. KETCHUM, and F.A. RICHARDS. 1963. The I , influence of organisms on the compos1t1on of sea water. In: The Sea. M.N Hill, editor, Chapter II, Interscience, New York, 26-77. RICHARDSON, P.L. and T.K. MCKEE. 1984. Average Seasonal variation of the Atlantic Equatorial currents from historic ship drifts. Journal of Physical oceanography, 14:1226-1238.


79 RYTHER, J.H., D.W. MENZEL, and N. CORWIN. 1967. Influence of the Amazon River outflow on the ecology of the western tropical Atlantic. I. Hydrography and nutrient chemistry. Journal of Marine Research, 25:69-83. SELLNER, A.G. 1981. Primary productivity and the flux of dissolved organic matter in several marine environments. Marine Biology, 65:101-112. SMITH, R.C. and K.S. BAKER. 1978a. Optical classification for natural waters. Limnology and Oceanography, 23:260-267. ___________ 1978b. The bio-optical state of ocean waters and remote sensing. Limnology and Oceanography, 23:247-259. STRAMSKI, D. and A. MOREL. 1990. Optical properties of photosynthetic picoplankton in different physiological states as affected by growth irradiance. Deep Sea Research, 37:245-266. STROM, S.L. and N.A. WELSCHMEYER. 1991. Pigment-specific rates of phytoplankton growth and microzooplankton grazing in the open subarctic Pacific Ocean. Limnology and Oceanography, 36:50-63. WALSH, J.J. 1988. On the Nature of Continental Shelves. Academic Press, 508pp. K.L. CARDER, and F.E. MULLER-KARGER. 1992. Meridional fluxes of dissolved organic matter in the North Atlantic ocean, Journal of Geophysical Research, (In press) WELSCHMEYER N.A. and C.J. LORENZEN (1985) Chlorophyll budget: Zooplankton grazing and phytoplankton growth in a temperate fjord and the Central Pacific Gyres. Limnology and Oceanography, 30:1-21. WILLAMS, P.J. Le B. and C.S. YENTSCH. 1976. An examination of photosynthetic production, excretion of photosynthetic produts, and heterotrophic utilization of dissolved organic compounds with reference to results from a coastal subtropical sea. Marine Biology, 35:31-40. YENTSCH, c.s. 1960. The influence of phytoplankton pigments on the colour of sea water. Deep-Sea Research, 7:1-9.


80 GENERAL REFERENCES BANSE, K. 1974 On the interpretation of data for the carbon-to-nitrogen ratio of phytoplankton. Limnology and Oceanography, 19:695-699. CARDER, K.L., R.G. STEWARD, J.H. PAUL, and G.A. VARGO. 1986. between chlorophyll and ocean color Relationships constituents reflectance 31:403-413. as they affect remote-sensing models. Limnology and Oceanography, CLARK, D.K. 1981. Phytoplankton algorithms for the Nimbus 7 CZCS, in Oceanography From Space, J .R.F. Gower, editor 227-238. CORREDOR, J .E. 1976. Aspects of phytoplankton dynamics in the Caribbean Sea and adjacent regions. In: Cooperative Investigations of the Caribbean and Adjacent Regions (CICAR) II. Symposium on Proaress in Marine Research in the Caribbean and Adjacent Regions, Caracas, 12-16 July 1976, FAO Fisheries Report No. 200, 101-114. EPPLEY, R .W. and B.J. PETERSON. 1979. Particulate organic matter flux and planktonic new production in the deep ocean. Nature, 282:677-680. E. STEWART, M.R. ABBOTT, and U. HEYMAN. 1985. -------,,..-,-Estimating ocean primary production from satellite chlorophyll. Introduction to regional differences and statistics for the Southern California Bight. Journal of Plankton Research, 7 :57-70. GORDON, A.L. 1967. Circulation of the Caribbean Sea. Journal of Geophysical Research, 72:6207-6223. GORDON, H.R. and D.K. CLARK. 1980. Remote sensing optical properties of a stratified ocean. Applied Optics, 19:3428-3430. KINDER T.H. 1983. Shallow currents in the Caribbean Sea and . Gulf of Mexico as observed w1th satelllte-tracked drifters. Bulletin of Marine Science, 33, 239-246.


81 MORRIS, I., A.E. SMITH, and H.E. GLOVER. 1981. Products of photosynthesis in phytoplankton off the Orinoco River and in the Caribbean Sea. Limnology and Oceanography, 26:1034-1044. ROEMMICH, D. 1981. Circulation of the Caribbean Sea: A well resolved inverse problem. Journal of Geophysical Research, 86 C9:7993-8005. SMITH, R.C. 1981. Remote sensing and depth distribution of ocean chlorophyll. Marine Ecology, Progress series, 5:359-361. WALSH, J .J. 1974. A spatial simulation model of the Peru upwelling ecosystem. Deep-Sea Research, 22:201-236. WORTHINGTON, L.V. 1976. On the North Atlantic circulation. Johns Hopkins University Press, Baltimore, MD, 120pp. YOSHIOKA, P.M., G.P. OWEN, and D. PESANTE. 1985. Spatial and temporal variations in Caribbean zooplankton near Puerto Rico. Journal of Plankton Research, 7:733-751.


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.