Near-surface phytoplankton pigment concentration in the ... · surface phytoplankton pigment...

4
McPherson and Rony Hermanto for helping with the data pro- cessing and model simulations. References ]Brody, E., 0. Holm-Hansen, G. B. Mitchell, and M. Vernet. 1992. Species- dependent variations of the absorption coefficient in the Gerlache Strait. Antarctic Journal of the U.S., this issue. Cox, C. S. and W. H. Munk. 1954. Measurements of the roughness of the sea from photographs of the sun's glitter. Journal of the Optical Society of America, 44:838-50. Frouin, R. and R. Hermanto. 1992. Analytical modeling of the specific intensity of sunlight backscattered by the ocean. Antarctic Journal of the U.S., this issue. Frouin, R., J. Y. Balois, P. Y. Deschamps, C. Verwaerde, M. Herman, M. Panouse, and J. Priddle. 1992. Aircraft photopolarimetric observa- tions of the ocean, ice/snow, and clouds in coastal regions of the Antarctic Peninsula. Antarctic Journal of the U.S., this issue. Koepke, P. 1984. Effective reflectance of oceanic whitecaps. Applied Optics, 20:1,816-24. Holm-Hansen, 0. and M. Vernet. 1992. Distribution, abundance, and productivity of phytoplankton in Gerlache Strait during austral spring. Antarctic Journal of the U.S., this issue. Morel, A. 1988. Optical modeling of the upper ocean in relation to its biogeneous matter content (Case I waters). Journal of Geophysical Research, 93:10,749-68. Tanre, D., M. Herman, P. Y. Deschamps, A. Deleffe. 1979. Atmospheric modeling for measurements of ground reflectances, including bidirec- tional properties. Applied Optics, 18:3,587-96. Tanré, D., C. Deroo, P. Duhaut, M. Herman, J. J. Morcrette,J. Perbos, and P. Y. Deschamps. 1986. Simulation of the satellite signal in the solar spectrum. Laboratoired'optiqueatmosphérique technical report, Universite des Sciences et Techniques at Lille, France, 267 p. Sobolev, V. V. 1963. A treatise on radiative transfer. Princeton, New Jersey: D. Van Nostrand Company. Near-surface phytoplankton pigment concentration in the Gerlache Strait derived from aircraft-polarization- and-directionality-earth- reflectance data (POLDER) ROBERT FROuIN Scripps Institution of Oceanography La Jolla, California 92093-0221 Several aircraft missions were flown over the Gerlache Strait during the 1991-1992 Research on Antarctic Coastal Ecosystem Rates (RACER) campaign for the purpose of mapping near- surface phytoplankton pigment concentration and primary pro- duction, and, hence, to extend spatially the local observations made aboard R/V Polar Duke. The aircraft, a Twin Otter operated by the British Antarctic Survey, was equipped with an ocean color imager, the polarization and directionality of the earth reflectance (POLDER) instrument, which measured the specific intensity of sunlight reflected by the atmosphere and ocean in spectral bands centered at 450, 500, 570, 670, and 850 nanometers, as well as the polarization characteristics of the reflected light. Details about the instrument's concept, imaging principle, and characteristics can be found in Frown et al. During each mission, the aircraft flew one low-altitude leg at 61 meters with passage over R/V Polar Duke and several high- altitude legs at 3,962 meters or 4,572 meters. The objective of flying the high-altitude legs was to map the experimental site. Because the swath at 3,962 meters was 4.9 x 6.5 kilometers and the pixel resolution 17x 17 meters, it would have required flying six parallel legs to map the Gerlache Strait completely. This was generally not possible, however, because of fuel requirements. The objective of flying the low-altitude leg was to check atmospheric correction schemes and, when passing over the ship, to compare the aircraft measurements with in situ optical data. We focus on the low-altitude leg flown across the Gerlache Strait on 29 December 1991, on an exceptionally clear day (no clouds, low aerosol loading at the surface). During the few days preceding and following that date, 25 December through 30 De- cember 1991, R/VPolar Duke surveyed the Gerlache Strait (RACER fast grid C), and the data revealed a strong southwest-northeast gradient of surface phytoplankton pigment concentration (Chloro- phyll a + phaeophytin), with values reaching 17 milligrams per cubic meter in the southwest and as low as 2 milligrams per cubic meter in the northeast (Holm-Hansen and Vernet this issue). These conditions provided the opportunity to check, over algal biomass levels spanning almost an order of magnitude, the ability of the POLDER instrument to remotely sense ocean color accurately and, thus, to provide quantitative estimates of near-surface phytoplank- ton pigment concentration. First, we describe the FOLDER data and detail the procedure to correct the data for atmospheric effects. Then, we present the results, namely POLDER estimates of pig- ment concentration along the aircraft subtrack, and we compare the estimates with values measured during fast grid C. Finally, we discuss the accuracy of the estimates in terms of potential sources of errors, in particular, the specific optical properties of the phy- toplankton in the Gerlache Strait and the anisotropy of the water body reflectance. Figure 1 shows the aircraft subtrack across the Gerlache Strait on 29 December 1991. The atmospheric conditions were clear sky, with a horizontal visibility better than 70 kilometers at the surface. Sunphotometer measurements made aboard R/V Polar Duke, on the other hand, indicated a rather high aerosol-optical thickness (e.g., 0.2 at 450 nanometers), seemingly in contradiction with the high visibility reported by the ship. The high aerosol-optical thickness, however, was explained by the presence of stratospheric aerosols following the eruption on 15 June 1991, of Mount Pinatubo in the Philippines (Frouin, Panouse, and Devaux this issue). The sea was calm south of Brabant Island, but light foam was observed east of 62 15' W. Small floes were sometimes within the field of view of the POLDER instrument, but they minimally affected the measurements. Because the leg was flown in the middle of the Gerlache Strait, the effect of sunlight reflection by surrounding ice (adjacency effect) was negligible. At 161 meters, the pixel size at the ground is about 25 centime- ters. Owing to the speed of the aircraft (about 120 knots) and the 1992 REVIEW 205

Transcript of Near-surface phytoplankton pigment concentration in the ... · surface phytoplankton pigment...

Page 1: Near-surface phytoplankton pigment concentration in the ... · surface phytoplankton pigment concentration and primary pro-duction, and, hence, to extend spatially the local observations

McPherson and Rony Hermanto for helping with the data pro-cessing and model simulations.

References

]Brody, E., 0. Holm-Hansen, G. B. Mitchell, and M. Vernet. 1992. Species-dependent variations of the absorption coefficient in the GerlacheStrait. Antarctic Journal of the U.S., this issue.

Cox, C. S. and W. H. Munk. 1954. Measurements of the roughness ofthe sea from photographs of the sun's glitter. Journal of the OpticalSociety of America, 44:838-50.

Frouin, R. and R. Hermanto. 1992. Analytical modeling of the specificintensity of sunlight backscattered by the ocean. Antarctic Journal ofthe U.S., this issue.

Frouin, R., J. Y. Balois, P. Y. Deschamps, C. Verwaerde, M. Herman, M.Panouse, and J. Priddle. 1992. Aircraft photopolarimetric observa-tions of the ocean, ice/snow, and clouds in coastal regions of theAntarctic Peninsula. Antarctic Journal of the U.S., this issue.

Koepke, P. 1984. Effective reflectance of oceanic whitecaps. AppliedOptics, 20:1,816-24.

Holm-Hansen, 0. and M. Vernet. 1992. Distribution, abundance, andproductivity of phytoplankton in Gerlache Strait during australspring. Antarctic Journal of the U.S., this issue.

Morel, A. 1988. Optical modeling of the upper ocean in relation to itsbiogeneous matter content (Case I waters). Journal of GeophysicalResearch, 93:10,749-68.

Tanre, D., M. Herman, P. Y. Deschamps, A. Deleffe. 1979. Atmosphericmodeling for measurements of ground reflectances, including bidirec-tional properties. Applied Optics, 18:3,587-96.

Tanré, D., C. Deroo, P. Duhaut, M. Herman, J. J. Morcrette,J. Perbos, andP. Y. Deschamps. 1986. Simulation of the satellite signal in the solarspectrum. Laboratoired'optiqueatmosphérique technical report, Universitedes Sciences et Techniques at Lille, France, 267 p.

Sobolev, V. V. 1963. A treatise on radiative transfer. Princeton, New Jersey:D. Van Nostrand Company.

Near-surface phytoplankton pigmentconcentration in the Gerlache Straitderived from aircraft-polarization-

and-directionality-earth-reflectance data (POLDER)

ROBERT FROuIN

Scripps Institution of OceanographyLa Jolla, California 92093-0221

Several aircraft missions were flown over the Gerlache Straitduring the 1991-1992 Research on Antarctic Coastal EcosystemRates (RACER) campaign for the purpose of mapping near-surface phytoplankton pigment concentration and primary pro-duction, and, hence, to extend spatially the local observationsmade aboard R/V Polar Duke. The aircraft, a Twin Otter operatedby the British Antarctic Survey, was equipped with an ocean colorimager, the polarization and directionality of the earth reflectance(POLDER) instrument, which measured the specific intensity ofsunlight reflected by the atmosphere and ocean in spectral bandscentered at 450, 500, 570, 670, and 850 nanometers, as well as thepolarization characteristics of the reflected light. Details about theinstrument's concept, imaging principle, and characteristics canbe found in Frown et al.

During each mission, the aircraft flew one low-altitude leg at 61meters with passage over R/V Polar Duke and several high-altitude legs at 3,962 meters or 4,572 meters. The objective of flyingthe high-altitude legs was to map the experimental site. Becausethe swath at 3,962 meters was 4.9 x 6.5 kilometers and the pixelresolution 17x 17 meters, it would have required flying six parallellegs to map the Gerlache Strait completely. This was generally notpossible, however, because of fuel requirements. The objective offlying the low-altitude leg was to check atmospheric correctionschemes and, when passing over the ship, to compare the aircraftmeasurements with in situ optical data.

We focus on the low-altitude leg flown across the GerlacheStrait on 29 December 1991, on an exceptionally clear day (noclouds, low aerosol loading at the surface). During the few dayspreceding and following that date, 25 December through 30 De-cember 1991, R/VPolar Duke surveyed the Gerlache Strait (RACERfast grid C), and the data revealed a strong southwest-northeastgradient of surface phytoplankton pigment concentration (Chloro-phyll a + phaeophytin), with values reaching 17 milligrams percubic meter in the southwest and as low as 2 milligrams per cubicmeter in the northeast (Holm-Hansen and Vernet this issue). Theseconditions provided the opportunity to check, over algal biomasslevels spanning almost an order of magnitude, the ability of thePOLDER instrument to remotely sense ocean color accurately and,thus, to provide quantitative estimates of near-surface phytoplank-ton pigment concentration. First, we describe the FOLDER dataand detail the procedure to correct the data for atmospheric effects.Then, we present the results, namely POLDER estimates of pig-ment concentration along the aircraft subtrack, and we comparethe estimates with values measured during fast grid C. Finally, wediscuss the accuracy of the estimates in terms of potential sourcesof errors, in particular, the specific optical properties of the phy-toplankton in the Gerlache Strait and the anisotropy of the waterbody reflectance.

Figure 1 shows the aircraft subtrack across the Gerlache Straiton 29 December 1991. The atmospheric conditions were clear sky,with a horizontal visibility better than 70 kilometers at the surface.Sunphotometer measurements made aboard R/V Polar Duke, onthe other hand, indicated a rather high aerosol-optical thickness(e.g., 0.2 at 450 nanometers), seemingly in contradiction with thehigh visibility reported by the ship. The high aerosol-opticalthickness, however, was explained by the presence of stratosphericaerosols following the eruption on 15 June 1991, of Mount Pinatuboin the Philippines (Frouin, Panouse, and Devaux this issue). Thesea was calm south of Brabant Island, but light foam was observedeast of 62 15' W. Small floes were sometimes within the field ofview of the POLDER instrument, but they minimally affected themeasurements. Because the leg was flown in the middle of theGerlache Strait, the effect of sunlight reflection by surrounding ice(adjacency effect) was negligible.

At 161 meters, the pixel size at the ground is about 25 centime-ters. Owing to the speed of the aircraft (about 120 knots) and the

1992 REVIEW 205

Page 2: Near-surface phytoplankton pigment concentration in the ... · surface phytoplankton pigment concentration and primary pro-duction, and, hence, to extend spatially the local observations

633W 623W 6200W 6110 C 6IOO'W

643O S

MOW

Figure 1. Map of the experimental site showing the sub-aircraft trackat the surface during the low-altitude (61 meter) leg across theGeriache Strait on 29 December 1991.

time necessary to acquire the data (a few seconds for 9 filters andthe optical zero), the same point at the surface could not beobserved in all the spectral bands. We therefore assumed that theocean was sufficiently homogeneous spatially to consider simul-taneity of the measurements in the various spectral bands. Toreduce the errors, we only used the spectral bands centered at 500and 570 nanometers, because the corresponding interferencefilters were positioned next to each other on the rotating wheel.To avoid glitter, we selected pixels corresponding to backscatter-ing conditions, namely a 150 scattering angle, and we averagedthe aircraft reflectance over 5 x 5 pixels.

In a preprocessing stage, we using the sunphotometer data tocompute the atmospheric transmittance (direct plus diffuse) ataircraft altitude and used the values to convert the POLDER datainto reflectance. Figure 2 shows the resulting raw (not correctedfor atmospheric effects below the aircraft) POLDER reflectance at500 and 570 nanometers (upper curves). Reflectance generallyincreases with decreasing longitude at 500 nanometers, whereasit remains more or less constant with longitude at 570 nanom-eters. Because the atmospheric properties and the solar andviewing geometries remained practically unchanged along theaircraft path, the variations in reflectance already suggest that thewaters corresponding to the southwest part of the axis are richerin phytoplankton, corroborating the in situ observations.

To interpret quantitatively the POLDER data in terms of near-surface phytoplankton pigment concentration, it is necessary tocorrect the raw reflectance for atmospheric effects, namely, re-sidual absorption and scattering below the aircraft, as well asglitter contamination. This was accomplished by first examiningthe data in the 850-nanometer band, for which the ocean can beconsidered black. A glitter reflectance was deduced from the dataand (since Fresnel reflection does not depend on wavelength)was used to correct the data in the 500-and 570-nanometer bands.Aerosol amount was assumed negligible below the aircraft; con-sequently, the correction for the intrinsic atmospheric reflectanceonly included scattering by molecules.

Figure 2 shows the atmospherically corrected reflectance—that is, the water body reflectance just above the surface at 500 and570 nanometers (lower curves)—and how it compares with theuncorrected one (upper curves). The effect of the atmosphere is toincrease the water body reflectance by about 0.05, or 50 percent,at both wavelengths. A slightly larger difference between uncor-

rected and corrected values is noticeable in the middle andeastern portion of the axis (except near the eastern end), probablydue to increasing wind speed.

Figure 3 (top) shows the ratio of the atmospherically correctedreflectance at 500 and 570 nanometers. At 500 nanometers, phy-toplankton pigments absorb strongly (the maximum of absorptionis at 440 nanometers), whereas they do not at 570 nanometers. Theratio of the water body reflectance at the two wavelengths, there-fore, is a measure of the amount of biomass existing below (Morel1980). A sharp change in reflectance ratio is observed around 62 W,where the value increases from 1 t 1.6 over a 20-kilometer distance.This sharp change was noted visually during the flights, whenpredominantly dark green waters at the beginning of the flight(southwest part of the axis) became more blue-green.

Using the reflectance model of Morel (1988), near-surface phy-toplankton concentration was deduced from the reflectance ratio at500 and 570 nanometers. The formula applied, obtained by fittingempirically reflectance-pigment concentration pairs, reads

Log (C) = 1.5429 + -3.3788 log (R I R570),

where C is pigment concentration in milligrams per cubic meter,and R and R575 are water body reflectances above the surface at500 and 570 nanometers, respectively.

Figure 3 (bottom) displays the pigment concentrations ob-tained as a function of longitude. High values (7 to 8 milligramsper cubic meter), in some instances reaching 14 milligrams per,cubic meter, are estimated west of 62*6 W, whereas low values(about 1 milligram per cubic meter) prevail east of 61 50' W.Between these two longitudes, pigment concentration changesfrom about 4 milligrams per cubic meter to 0.8 milligrams per

C

a-E00LO

—63 —62 —61Longitude (Deg)

raw datacorrected data

a-

-63 —62 —61Longitude (Deg)

Figure 2. Raw and atmospherically corrected aircraft reflectance at500 nanometers (top) and 570 nanometers (bottom) across theGerlache Strait on 29 December 1991. The corrected data correspondto water body reflectance just above the surface.

206

ANTARCTIC JOURNAIr

Page 3: Near-surface phytoplankton pigment concentration in the ... · surface phytoplankton pigment concentration and primary pro-duction, and, hence, to extend spatially the local observations

cubic meter. Spatial variability is substantial on scales of 10 to 20kilometers in both the high and low biomass regions.

Compared to the pigment concentrations measured in situ (at a5-meter depth) during fast grid C (Holm-Hansen and Vemet thisissue), the POLDER-derived values are generally lower. In thenorthwest part of the Gerlache Strait, 2 to 3 milligrams per cubicmeter are measured instead of 1 milligram per cubic meter. Around623O' W, the in situ values are well above 10 milligrams per cubicmeter, whereas the POLDER-derived ones are consistently below10 cubic milligrams. The strong gradient around 62'W is well-retrieved, but the aircraft values suggest that it is actually sharperthan indicated by Holm-Hansen and Vemet (this issue), with themaximum lying slightly farther west (between stations 3 and 44,the change in pigment concentration is not linear).

The generally too-high POLDER-derived values should bediscussed in view of optical properties of the phytoplanktonpresent in the Gerlache Strait waters, which differ significantlyfrom those used in Morel (1988). According to Panouse, however,the diffuse attenuation coefficient, measured in situ (in the firstten meters), when corrected for its pure water and soluble mate-rial components (terrigenous materials were assumed to be non-existent) was, on average, close to Morel's (1988) chlorophyll-specific diffuse attenuation coefficient, k", at 500 nanometers.This disagrees with previous measurements by Mitchell andHolm-Hansen (1991). Spatial variability in the phytoplanktonpopulations (diatoms vs. cryptomonads), hence in the opticalproperties of their varied mixtures (see Brody et al.), may stillcontribute to the discrepancies.

The situation is more complicated, however, because thereflectance above water, as measured by the POLDER instru-

2.

2.1

0

0

1.c

0.

—01Longitude (Deg)

10.1

0.1—01

Longitude (Deg)

ure 3. Ratio of water body reflectance at 500 and 570 nanometersoss the Gertache Strait on 29 December 1991 (top). Near-surface'toplankton pigment concentration (bottom) deduced from theo of reflectances at 500 and 570 nanometers using the model ofrd (1988).

ment, is higher than the reflectance predicted by Morel (1988),after correction for refraction at the air-sea interface. For apigment abundance of 10 milligrams per cubic meter, for in-stance, Morel's (1988) model gives above-surface reflectances of0.006 and 0.008 at 500 and 570 nanometers, respectively, whichcompares to measured values of about 0.008 and 0.01. Yet,terrigenous materials could not have affected the reflectance,because they were quasi-nonexistent (Vernet personal communi-cation). Moreover, in situ measurements of flux reflectance justbelow the surface made at the passage of the aircraft, which oc-curred at the longitude of approximately 6r 45'W, revealed val-ues of 0.009 and 0.008 at 500 and 570 nanometers, respectively,when transformed into reflectance just above the surface, whereasthe corresponding POLDER-derived values read 0.013 and 0.008.Thus the agreement is not good at 500 nanometers. Convergingevidence therefore exists that the retrieved reflectances are toohigh, especially at 500 nanometers. One possible explanation isthat, unlike Morel's (1988) model and the in situ optical profiler,the POLDER instrument gives access to bidirectional reflectanceand not to flux reflectance. In backscattering conditions (recallthat the scattering angle is about 150 in our case), the backwardpeak of the phytoplankton phase function may act to increase thebidirectional reflectance when compared with that averaged overviewing angles (Frouin and Hermanto this issue).

In summary, our analysis of the POLDER data acquired duringthe low-altitude flight of 29 December 1991 in the Gerlache Straitindicates that the POLDER instrument meets the basic require-ments for ocean color remote sensing in the Antarctic. Non-negligible discrepancies between POLDER-derived near-surfacepigment concentrations and in situ measurements were found,however, and they could not be explained satisfactorily. A definiteassessment of the instrument capability will therefore requireexamining more closely the specific optical properties of the phy-toplankton in the area, including spatial variability, the bidirec-tional properties of the water body reflectance, as well as othersources of errors (e.g., nonsimultaneity of the measurements in thevarious spectral bands, atmospheric effects). Nevertheless, thepreliminary results described above are encouraging; they consti-tute a step toward achieving one of the major objectives of theaircraft missions, namely, mapping near-surface phytoplanktonpigment concentration over the RACER study areas.

This research was supported in part by the National Aeronau-tics and Space Administration under grant NAG W-2774 to Rob-ert Frouin, the National Science Foundation, the British AntarcticSurvey, the French Space Agency, and the Centre National de laRecherche Scientifique. We thank the British Antarctic Surveypersonnel at Rothera for their help with the aircraft missions;Jean-Yves Balois from the University of Lille, France, for operat-ing the POLDER instrument during the experiment and perform-ing the necessary calibrations; Michel Panouse from theObservatoire Oceanologique de Banyuls, France, for collectingthe in situ optical data; the captain and crew members of R/VPolar Duke for their assistance during the cruise, John McPhersonfor processing the POLDER data, and Maria Vernet, OsmundHolm-Hansen, and Eric Brody for helpful discussions.

References

Brody, E., 0. Holm-Hansen, G. B. Mitchell, and M. Vernet. 1992. Species-dependent variations of the absorption coefficient in the GerlacheStrait. Antarctic Journal of the U.S., this issue.

992 REVIEW 207

Page 4: Near-surface phytoplankton pigment concentration in the ... · surface phytoplankton pigment concentration and primary pro-duction, and, hence, to extend spatially the local observations

Frouin, R. and R. Hermanto. 1992. Analytical modeling of the specificintensity of sunlight backscattered by the ocean. Antarctic Journal oftheU.S., this issue.

Frouin, R., J . Y. Balois, P. Y. Deschamps, C. Verwaerde, M. Herman, M.Panouse, and J. Priddle. 1992. Aircraft photopolarimetric observa-tions of the ocean, ice/snow, and clouds in coastal regions of theAntarctic Peninsula. Antarctic Journal of the U.S., this issue.

Frouin, R., M. Panouse, and C. Devaux. 1992. Sunphotometer measure-ments of aerosol optical thickness in the Gerlache Strait and Marguer-ite Bay, Antarctica. Antarctic Journal of the U.S., this issue.

Holm-Hansen, 0. and M. Vernet. 1992. Distribution, abundance, and

productivity of phytoplankton in the Gerlache Strait during australspring. Antarctic Journal of the U.S., this issue.

Mitchell, B. G. and 0. Holm-Hansen. 1991. Bio-optical properties ofAntarctic Peninsula waters: Differentiation from temperate oceanmodels. Deep-Sea Research, 38(8-9):1,009-28.

Morel, A. 1980. In-water and remote measurements of ocean color.Boundary Layer Meteorology, 18:177-201.

Morel, A. 1988. Optical modeling of the upper ocean in relation to itsbiogeneous matter content (Case I waters). Journal of GeophysicalResearch, 93:10,749-68.

208

ANTARCTIC