Full report

Original page 1

NASA Technical Memorandum SeaWiFS Technical Report Stanford B. Hooker, Editor Goddard Space Flight Center Greenbelt, Maryland Elaine R. Firestone, Technical Editor General Sciences Corporation Laurel, Maryland 104566, Vol. 27 Series James G. Acker, Technical Editor Hughes STX Lanham, Maryland Volume 27, Case Studies for SeaWiFS Calibration and Validation, J. L. Mueller San Diego State University San Diego, California R. S. Fraser NASA Goddard Space Flight Center Greenbelt, Maryland S. F. Biggar, K. J. Thome, and P. N. Slater University of Arizona Tucson, Arizona A. W. Holmes Santa Barbara Research Center Goleta, California National Aeronautics and Space Administration Goddard Space Flight Center Greenbelt, Maryland 20771 1995 Part 3 R. A. Barnes ManTech Environmental Technology, Inc. Wallops Island, Virginia C. T. Weir, D. A. Seigel, and D. W. Menzies University of California at Santa Barbara Santa Barbara, California A. F. Michaels Bermuda Biological Station for Research Ferry Reach, Bermuda G. Podesta University of Miami Miami, Florida

Original page 2

This publication is available from the NASA Center for AeroSpace Information, 800 Elkridge Landing Road, Linthicum Heights, MD 21090-2934, (301) 621-0390. 1

Original page 3

Case Studies for SeaWiFS Calibration and Validation, Part 3 PREFACE he scope of the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Calibration and Validation Program encompasses a broad variety of topics, as evidenced by the contents of two previous case studies volumes in the SeaWiFS Technical Report Series--Volumes 13 and 19. Each case studies volume contains several chapters discussing topics germane to the Calibration and Validation Program. Volume 27, the third collection of case studies, further demonstrates both the breadth and complexity of the issues that the Program must address, and provides further justification for a comprehensive calibration and validation effort. The chapters in this volume present discussions of: a) Results on the measurement of immersion coefficients for submersible radiometers; b) The effect of oxygen absorption on the 765 nm SeaWiFS channel; c) The results of the second SeaWiFS ground-based solar calibration experiment, which was performed after the instrument was modified to reduce internal stray light; d) Ship shadow effects on subsurface radiance and irradiance measurements; and e) The definition of the SeaWiFS data day for level-3 data binning. GreenbeIt, Maryland -- C. R. McClain January 1995

Original page 4

Original page 5

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta Table of Contents Prologue ................................................................................................. 1 1. Comparison of Irradiance Immersion Coefficients for Several Marine Environmental Radiometers (MERs) ........................................................ 3 1.1 Introduction ....................................................................................... 3 1.2 Method and Results ................................................................................ 3 1.3 Discussion ......................................................................................... 4 2. The Effect of Oxygen Absorption on Band-7 Radiance ............................................. 16 2.1 Introduction ...................................................................................... 16 2.2 Theory ........................................................................................... 16 2.3 Conclusion ....................................................................................... 19 3. Second SeaWiFS Preflight Solar Radiation-Based Calibration Experiment ......................... 20 3.1 Introduction ...................................................................................... 20 3.2 Experimental Method ............................................................................. 20 3.3 Results ........................................................................................... 21 3.4 Discussion ........................................................................................ 24 4. In Situ Evaluation of a Ship's Shadow ............................................................ 25 4.1 Introduction ...................................................................................... 25 4.2 Experimental Methods ............................................................................ 2{} 4.3 Results ........................................................................................... 27 4.3.1 Simultaneous Comparison ........................................................................ 27 4.3.2 Multi-Distance Comparison ....................................................................... 28 4.4 Discussion ........................................................................................ 28 5. SeaWiFS Global Fields: What's In a Day? ........................................................ 34 5.1 Introduction ...................................................................................... 34 5.2 Temporal Definition .............................................................................. 34 5.3 Spatial Definition ................................................................................. 35 5.3.1 Beginning of the Data Day ....................................................................... 36 5.3.2 Advantages of the Spatial Definition .............................................................. 38 5.3.3 An Alternative Explanation ...................................................................... 40 5.4 Other Issues ...................................................................................... 42 GLOSSARY ............................................................................................... SYMBOLS ................................................................................................ REFERENCES ............................................................................................ 43 43 44 THE SEAWIFS TECHNICAL REPORT SERIES ............................................................ 45 o,. 111 PRECEDING PAGE BLANK NOT FILMED

Original page 6

Original page 7

Mueller, Fraser, Biggar, Thome, Slater, Holmes, Barnes, Weir, Siegel, Menzies, Michaels, and Podesta ABSTRACT This document provides brief reports, or case studies, on a number of investigations sponsored by the Calibration and Validation Team (CVT) within the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Project. Chapter 1 describes a comparison of the irradiance immersion coefficients determined for several different marine environmental radiometers (MERs). Chapter 2 presents an analysis of how light absorption by atmospheric oxygen will influence the radiance measurements in band 7 of the SeaWiFS instrument. Chapter 3 gives the results of the second ground-based solar calibration of the instrument, which was undertaken after the sensor was modified to reduce the effects of internal stray light. (The first ground-based solar calibration of SeaWiFS is described in Volume 19 in the SeaWiFS Technical Report Series.) Chapter 4 evaluates the effects of ship shadow on subsurface irradiance and radiance measurements deployed from the deck of the R/V Weatherbird H in the Atlantic Ocean near Bermuda. Chapter 5 illustrates the various ways in which a single data day of SeaWiFS observations can be defined, and why the spatial definition is superior to the temporal definition for operational usage. Prologue The purposes of the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Project is to obtain valid ocean color data of the world ocean for a five-year period, to process that data in conjunction with ancillary data to meaningful biological parameters, and to make that data readily available to researchers. The National Aeronautics and Space Administration (NASA) Goddard Space Flight Center (GSFC) will develop a data processing and archiving system in conjunction with the Earth Observing System Data and Information System (EOSDIS), which includes a ground receiving system; EOSDIS will oversee a calibration and validation effort which is designed to ensure the integrity of the final products. The Calibration and Validation Team (CVT) has three main tasks: 1) Calibration of the SeaWiFS instrument; 2) Development and validation of the operational atmospheric correction algorithm; and 3) Development and validation of the derived product algorithms, such as chlorophyll a concentration. Some of this work will be done internally at GSFC, while the remainder will be done externally at other institutions. NASA and the Project place the highest priority on assuring the accuracy of derived water-leaving radiances globally, and over the duration of the entire mission. If these criteria are met, the development of global and regional biogeochemical algorithms can proceed on many Sea WiFS Technical Report Series, the CVT has decided to publish volumes composed of brief, but topically specific, chapters. Volume 13 was the first volume, and consists primarily of contributions related to atmospheric correction methodologies, ancillary data sets required for level-2 processing of Coastal Zone Color Scanner (CZCS) and Sea- WiFS data, laboratory techniques for instrument calibration relevant to calibration round-robins, and field observations designed for transferring the prelaunch calibration to orbit, and in interpreting the on-orbit lunar calibration data. The second case studies volume, Volume 19, contains chapters on atmospheric and glint corrections, solar-, lunar-, and integrating sphere optical measurements, data format considerations, and the use of ancillary data (including surface wind velocities) in SeaWiFS processing. Volume 26 is the third in the set of such volumes. A short synopsis of each chapter in this volume is given below. 1. Comparison of Irradiance Immersion Coefficients for Several Marine Environmental Radiometers (MERs) This chapter describes how spectral immersion coefficients were measured experimentally for 12 irradiance collectors on underwater profiling radiometers. These coefficients are used to convert spectral radiance responsivity calibration factors, measured in air, for use underwater. At any given wavelength, the immersion coefficients typically had standard deviations between collectors ranging from 3-5%. The total variations at some wavelengths were as large as 10%. Repeated measurements on several of fronts. These various activities are discussed in detail in the collectors showed that experimental uncertainty is not The SeaWiFS Calibration and Validation Plan (McClain et al. 1992). Because many of the studies and other works undertaken with the Calibration and Validation Program are not extensive enough to require dedicated volumes of the greater than 1%. The primary conclusion of this study is that accurate underwater radiometry absolutely requires experimental characterization of each individual irradiance collector, rather than assuming a value based solely on its design and material specifications.

Original page 8

Case Studies for SeaWiFS Calibration and Validation, Part 3 2. The Effect of Oxygen Absorption on Band-7 Radiance Atmospheric oxygen absorbs about 13% of the available sunlight reflected from the Earth in band 7 of the Sea_,ViFS instrument. If a correction is made for the average amount 4. In Situ Evaluation of a Ship's Shadow In situ measurements of optical properties made from a ship can be biased by the ship's shadow. In an effort to evaluate the ship shadow perturbation created by the R/V Weatherbird II, profiles of downwelling irradiance, of absorption, then the measured radiance would vary by Ed(Z,)_); upwelling radiance, Lu(z,A); and derived ap- -t-0.004 for a two standard deviation (2a) variability in the amount of oxygen in a vertical column. For comparison, the instrumental noise is about one-half, or 0.002, of this parent optical properties (AOPs), were obtained at four distances--1 m, 3 m, 6 m, and 20 m or more--off the ship's stern. Two statistical analyses of these data are explored. variation. A correction based on the regional changes in The first analysis uses data from pairs of simultaneouslyabsorption as a function of season would not significantly reduce the statistical variation in absorption. 3. Second SeaWiFS Preflight Solar Radiation-Based Calibration Experiment This paper describes the second solar radiation-based 1 parisons of profiles obtained at least 3 m off the ship's stern. calibration of SeaWiFS. The experiment was done on November 1993 in the rock garden at the Santa Barbara Research Center (SBRC). The results of the calibration are presented, along with a comparison to the spherical integrating source (SIS) calibration done at SBRC. The estimated uncertainty of the SIS calibration is 2.8%, compared to the 4% estimated uncertainty for the solar-based calibration. There is also an uncertainty in the value of the exoatmospheric solar irradiance used to make this comparison, which is probably on the order of 1%. In addiobtained light profiles, one profile obtained at a distance greater than 20 m from the stern of the ship, and the other taken either 1 or 6 m off the stern. The second analysis compares the derived AOPs for each profiling distance from the ship, using data obtained throughout the length of the experiment. Significant differences are rare in com- At 1 m off the stern, however, significant discrepancies are intermittently observed. This work illustrates that the inherent sources of noise in determining radiative fluxes and AOPs in the upper ocean are generally greater than the effects incurred by the ship's own shadow under optimal conditions. 5. Sea WiFS Global Fields: _Vhat's In a Day? tion, the integrated out-of-band blocking for SeaWiFS is This chapter defines the procedure to be employed to in the 1-3% range, which can introduce significant differences between the lamp- and solar-based calibrations. This agreement is better than anticipated, and better than the agreement achieved in March 1993 from the first experiment. The better agreement is probably due to a sphere recalibration between the two experiments. delineate data corresponding to one day of SeaWiFS operation. The definition is required for data analysis with minimal temporal aliasing in tile same region of observation. The definition also allows proper assignment of data into daily fields that will be used for the generation of weekly and monthly average products.

Original page 9

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta Chapter 1 Comparison of Irradiance Immersion Coefficients for Several Marine Environmental Radiometers (MERs) JAMES L. MUELLER San Diego State University San Diego, California ABSTRACT Spectral immersion coefficients were measured experimentally for 12 irradiance collectors on underwater profiling radiometers. These coefficients are used to convert spectral irradiance responsivity calibration factors, measured in air, for use underwater. All of the irradiance collectors were the same design, however, 11 were made of Plexiglas® diffusing material and 1 was made of Teflon ®. At any given wavelength, the immersion coefficients typically had standard deviations between collectors ranging from 3-5%. The total variations at some wavelengths were as large as 10%. The coefficients of the Teflon diffuser were well within the bounds of one standard deviation from the sample mean. Repeated measurements on several of the collectors showed that experimental uncertainty is not greater than 1% (one standard deviation, or la). The primary conclusion of this study is that accurate underwater radiometry absolutely requires experimental characterization of each individual irradiance collector, rather than assuming a value based solely on its design and material specifications. 1.1 INTRODUCTION The spectral responsivities of underwater (irradiance) radiometers are calibrated in air using an FEL lamp, which has a spectral irradiance scale traceable to the National Institute of Standards and Technology (NIST). The spectral immersion coefficients for an underwater irradiance meter represent the differences between the instrument's spectral responsivities in air and in water. The responsivity will change due to the fact that the refractive index of the plastic (or Teflon ® ) diffuser is smaller, relative to the refractive index of water, than it is relative to the refractive index of air. Less incident light is reflected at the waterplastic interface, and therefore, more of the light reflected from the collector's inner surface escapes back into the water. The net result is that a smaller fraction of incident flux is transmitted through the irradiance collector in water and therefore, the instrument's irradiance responsivity is decreased. At present, the practice of the oceanographic community is to experimentally characterize the spectral immersion coefficients of only a small sample of irradiance collectors in a given class of collectors (with the same materials and design specifications). These coefficients have subsequently been associated with all collectors in that class, which assumes negligible variability between individual items. In a previous report (Mueller 1994), this assumption was tested by comparing irradiance immersion coefficients for several MER-series radiometers manufactured by Biospherical Instruments, Inc. (BSI) of San Diego, California. The results of that preliminary comparison between measured spectral immersion coefficients for six Plexiglas ® diffusers of the same material and design specifications show significant variations, with standard deviations (a) ranging from 3.2-3.5% and ranges (maximum minus minimum) as large as 9%. In this earlier work, however, the immersion tests on each instrument were not replicated, and thus, no estimates of the experimental uncertainty of the mehsurements could be made. The conclusions were, therefore, only tentative. This report presents results of an extended series of immersion characterization experiments on an expanded sample of irradiance collectors. The new series of characterization experiments were repeated two or more times, at different lamp-to-collector distances, to provide an estimate of the uncertainty in this laboratory's experimental determinations of immersion coefficients. The new results were then pooled with the earlier sample (Mueller 1994) to analyze overall variability between individual collectors. 1.2 METHOD AND RESULTS The laboratory procedure for determining an irradiance meter's spectral immersion coefficients is described in 3

Original page 10

CaseStudiesfor SeaWiFSCalibrationandValidation,Part3 taken the experimental mean values, for each individual collector MuellerandAustin(1992).Immersioncoefficients experi- (Tables 2-8) are illustrated in Figs. 2-8, respectively. fromMueller(1994)wereeachbasedon a single a newsetof experi- Linear regression analyses provide a reasonable fit for mentalmeasurement.Subsequently, of many of the instrumental immersion coefficients, Fi, to an mentalmeasurementswererepeated2-4timeson each es- equation of the form sevenirradiancecollectorsasa basisfor uncertainty timates.In eachcase,the immersiontest (Muellerand Austin1992)wasdonefirst at onelamp-to-collectordistance,andthenthe lampwasmoved12.5cmtowardthe collectorandthetestwasrepeated.In addition,theentire procedurewasrepeatedon differentdaysfor four of the collectors,andfor a fifth instrument[MER-1012f,Serial Number(S/N) 8107]the singledayresultsfromthis laboratorywerecombinedwith theresultsofanindependent immersioncharacterizationby BSI. Table 1. Immersioncoefficientsfor severalMERs characterizedat CHORS. The column headings denote the MER model number and irradiance type. The data is extracted from Mueller (1994). Wavelength MER-lO48t MER-2040_ [nm] E_l E,, Ed E,, 408 1.3686 410 1.4315 1.3882 1.3019 439 1.4289 440 1.4456 441 1.3846 1.3212 465 1.3751 1.3158 488 1.4138 1.3591 1.3012 489 1.4169 518 1.3841 519 1.3952 520 1.3470 1.2865 548 1.3822 550 1.3704 560 1.3693 1.3265 1.2695 589 1.3654 1.3699 1.3088 1.2557 632 1.3339 655 1.3425 1.2804 1.2332 671 1.3236 1.3344 683 1.3997 693 1.3724 1.3233 709 1.3389 MER S/N 8302. $ MER S/N 8716. Table 1 lists spectral irradiance immersion coefficients for two MER instruments (i.e., four irradiance collectors) characterized at the San Diego State University (SDSU) Center for Hydro-Optics and Remote Sensing (CHORS) bA F_ = a -- 10---_ (1) where A is the wavelength in nanometers. Regression coefficients a and b; residual standard deviations, s,y; and squared linear correlation coefficients, R2; are compared for these instruments in Table 9, and the regression lines are illustrated in Figs. 2 and 4-8. Mean, range, and a of immersion coefficients for subsamples of different collectors were computed at selected wavelengths and are presented in Table 10. 1.3 DISCUSSION This report compares the experimentally determined immersion coefficients for the irradiance collectors on nine MER-series underwater radiometers manufactured by BSI. There are 12 irradiance collectors involved in these experiments, all having the same basic design--ll have Plexiglas diffusers and 1 (MER-2040 S/N 8738) has a Teflon diffusers (Table 8 and Fig. 8). The replicated experiments summarized in Tables 2- 8 show that, for the majority of channels tested, the la uncertainties in experimentally determined immersion coefficients are approximately 1% or less. The notable exceptions are the Ed(A) channels at wavelengths greater than 550nm of the MER-1012f S/N 8107 (Fig. 2 and Table 2) and MER-1015 S/N 8205 (Fig. 3 and Table 3). It is suspected that these larger uncertainties may indicate nonlinearities in the responsivities of these channels. Nonlinearity may be due to a voltage discontinuity across a gain change in replicated experiments at different lampto-collector distances. The total range between the immersion coefficients of the 12 collectors is as large as 15% at some wavelengths (Fig. 1 and Table 10). The standard deviation of dispersion in immersion coefficients is generally between 3.5-5%, at least at those wavelengths for which collector sample sizes were large enough to estimate a reasonable standard deviation (Table 10). Immersion coefficients vary linearly with wavelength for many of the diffusers (Table 9, and Figs. 2 and 4-8), with residual standard deviation s,y x 100% <_1% (except for during 1993 (Mueller 1994). Immersion coefficients for the MER-2040 S/N 8725, where s, x 100% = 1.7%). The seven additional collectors are listed in Tables 2-8, together with mean, standard deviation, and range for replicated tests at each wavelength. Spectral immersion coefficients from all collectors are illustrated in Figl 1 (which includes all data from Table 1, and mean coefficients from Tables 2-8). Coefficients from replicate experiments, and 4 immersion coefficients for the MER-1015 S/N 8205 are not a well-behaved linear function of wavelength. The dispersion in immersion coefficients between these irradiance collectors (Fig. 1 and Table 10) is too large to neglect, and the results of the replicated experiments indicate that only a small fraction of this variation can be

Original page 11

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodest_ Table 2. Immersioncoefficientsmeasuredexperimentallyfor the MER-1012f(S/N 8107).The dataare )resentedherein orderof decreasinglamp-to-collectordistance. Wavelength 24 June 1994 24 June 1994 [nm] 144.5 cm 130.0 cm 406.9 1.3510 1.3492 442.5 1.3673 1.3643 487.1 1.3499 1.3454 517.8 1.3383 1.3344 564.9 1.3178 1.3388 632.3 1.3130 1.2961 681.3 1.3044 1.2636 17 March 1994t Immersion Coefficient 106.3 cm p a Range 1.3512 1.3505 0.0009 0.0020 1.3685 1.3667 0.0018 0.0042 1.3501 1.3485 0.0022 0.0048 1.3391 1.3373 0.0020 0.0047 1.3195 1.3254 0.0095 0.0210 1.3246 1.3112 0.0117 0.0286 1.2807 1.2829 0.0167 0.0408 t Immersion coefficients calculated from experimental measurements by BSI. Table 3. Immersion coefficients measured ex _erimentally for the MER-1015 (S/N 8205). Wavelength 1 September 1994 2 September 1994 Immersion Coefficient [nm] 152.2 cm 137.8 cm 152.2 cm 137.8 cm p a Range 406.0 1.4159 1.4217 1.4233 1.4104 1.4178 0.0051 0.0130 438.5 1.4335 1.4473 1.4351 1.4398 1.4389 0.0054 0.0138 462.4 1.4253 1.4371 1.4309 1.4291 1.4306 0.0043 0.0118 485.5 1.4210 1.4248 1.4227 1.4177 1.4215 0.0026 0.0071 517.9 1.4031 1.4083 1.4071 1.4022 1.4052 0.0026 0.0060 536.8 1.3929 1.4004 1.3965 1.3930 1.3957 0.0031 0.0075 558.5 1.3843 1.3196 1.3879 1.3183 1.3525 0.0336 0.0696 588.4 1.3104 1.4189 1.3169 1.4171 1.3658 0.0522 0.1086 624.2 1.4494 1.4065 1.4448 1.3955 1.4240 0.0235 0.0539 673.3 1.4325 1.3993 1.4482 1.3861 1.4165 0.0249 0.0621 696.2 1.3846 1.3507 1.3730 1.3418 1.3625 0.0171 0.0428 762.4 1.3120 1.3131 1.3151 1.3109 1.3128 0.0016 0.0042 Table 4. Immersion coefficients measured experimentally for the MER-1032 (S/N 8301). The data are _resented here to allow comparison for similar lamp-to-collector distances. Wavelength 30 December 1993 22 July 1992 [nm] 120.0 cm 138.0cm 123.5cm 411.1 1.4053 1.4172 1.4136 441.9 1.3946 1.4302 1.4016 452.9 1.3897 1.4091 1.4129 489.6 1.3754 1.3887 1.3918 508.9 1.3795 1.3877 1.3854 528.6 1.3529 1.3790 1.3753 555.3 1.3451 1.3664 1.3618 588.9 1.3329 1.3540 1.3450 632.1 1.3071 1.3337 1.3271 654.8 1.3013 1.3222 1.3153 670.7 1.3032 1.3285 1.3232 26 July 1992 Immersion Coefficient 137.9 cm 123.5 cm Iz a Range 1.4112 1.4037 1.4102 0.0057 0.0135 1.4146 1.4005 1.4083 0.0143 0.0356 1.4089 1.4096 1.4060 0.0093 0.0232 1.3945 1.3925 1.3886 0.0077 0.0191 1.3889 1.3819 1.3847 0.0039 0.0094 1.3795 1.3735 1.3720 0.0110 0.0266 1.3679 1.3622 1.3607 0.0091 0.0229 1.3577 1.3490 1.3477 0.0096 0.0248 1.3348 1.3280 1.3261 0.0112 0.0276 1.3184 1.3151 1.3145 0.0079 0.0209 1.3279 1.3219 1.3209 0.0103 0.0253

Original page 12

CaseStudiesfor SeaWiFSCalibrationand Validation, Part 3 ['able 5. Immersion coefficients measured ex )erimentally for the MER-1032 (S/N 8301). Wavelength 22 July 1994 26 July 1994 Immersion Coefficient [nm] 131.6 cm 123.5 cm 131.6 cm 117.2 cm # a Range 411.3 1.4532 1.4508 1.4458 1.4427 1.4481 0.0041 0.0105 1.4165 1.4156 1.4188 0.0027 0.0060 442.0 1.4216 1.4214 1.3987 1.3961 1.3991 0.0020 0.0056 489.9 1.3999 1.4017 509.1 1.3859 1.3877 1.3845 1.3819 1.3850 0.0021 0.0058 555.4 1.3664 1.3658 1.3664 1.3595 1.3645 0.0029 0.0069 529.1 1.3788 1.3783 1.3773 1.3735 1.3770 0.0021 0.0053 633.0 1.3108 1.3271 1.3103 1.3204 1.3172 0.0070 0.0168 670.9 1.3249 1.3068 1.3167 0.0078 0.0181 1.3237 1.3113 Table 6. Immersion coefficients measured ex )erimentally for the MER-2040 (S/N 8724). Wavelength 22 July 1994 26 July 1994 Immersion Coefficient [nm] 121.6 cm 107.2 cm 121.6 cm 107.2 cm # a Range 453.2 1.4086 1.4077 1.4116 1.4117 1.4099 0.0018 0.0040 440.3 1.4144 1.4143 1.4186 1.4187 1.4165 0.0021 0.0044 486.7 1.3942 1.3931 1.3987 1.3980 1.3960 0.0024 0.0056 516.9 1.3783 1.3825 1.3828 1.3870 1.3827 0.0031 0.0087 530.2 1.3770 1.3807 1.3766 0.0029 0.0083 1.3724 1.3762 565.1 1.3560 1.3554 1.3604 1.3602 1.3580 0.0023 0.0050 664.0 1.3175 1.3167 1.3218 1.3213 1.3193 0.0022 0.0050 Table 7. Immersion coefficients measured experimentally for the MER-2040 (S/N 8725). Wavelength 2 September 1994 [nm] 125.4 cm lll.0cm 408.6 1.3150 1.3177 438.9 1.3451 1.3380 484.9 1.3449 1.3394 617.6 1.3366 1.3398 564.0 1.3223 1.3206 662.6 1.3011 1.2997 Immersion Coefficient # a Range 1.3164 0.0014 0.0028 1.3415 0.0035 0.0071 1.3421 0.0028 0.0056 1.3382 0.0016 0.0032 1.3215 0.0008 0.0016 1.3004 0.0007 0.0014 Table 8. Immersion coefficients measured experimentally for the MER-2040 (S/N 8738). Wavelength 24 June I994 29 June 1994 Immersion Coefficient [nm] 120.7 cm 106.2 cm # a Range 340.0 1.4281 1.4170 1.4225 0.0055 0.0111 380.0 1.4127 1.4033 1.4080 0.0047 0.0094 1.3917 1.3956 0.0039 0.0079 412.0 1.3996 443.0 1.3870 1.3804 1.3837 0.0033 0.0066 1.3977 1.4015 0.0038 0.0076 395.0 1.4053 665.0 1.3126 1.3078 1.3102 0.0024 0.0047 1.3772 1.3803 0.0032 0.0064 455.0 1.3835 1.3652 1.3683 0.0031 0.0062 490.0 1.3714 510.0 1.3638 1.3580 1.3609 0.0029 0.0058 532.0 1.3570 1.3467 1.3518 0.0052 0.0103 555.0 1.3486 1.3380 1.3433 0.0053 0.0106 570.0 1.3447 1.3325 1.3386 0.0061 0.0122 6

Original page 13

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 1.5 + o 1.4 8÷ # x 8 + #0# $ # × % 1.3 ® E q [J_ t.2 1.1 1.0 400 450 500 550 MER-I012 SN 8107 Ed: MER-I015 SN 8205 Ed: MER-I032 SN 8301 Ed: MER-IO48 SN 8302 Ed: MER-2040 SN 8716 Ed: MER-2040 SN 8724 Ed: MER-2040 SN 8725 Ed: MER-2040 SN 8738 Ed: f o _ # & _ o 600 650 700 7 O, Wavelength {nm) x * o Eu + # Eu: @ & Eu: @ 0 % F Fig. 1. Immersion coefficients (F,) for several MER-series radiometers. 7

Original page 14

CaseStudiesfor SeaWiFSCalibrationandValidation,Part3 1.5 1.4 1.3 E E h 1.2 1.1 ;.o ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I ' ' ' ' I 550 600 650 700 750, 400 450 500 BSI Test 6/24/94, 6/24/94. MEAN: Wave]ength (nm) (June 94): x R - 130.0 cm: R= 144.5 cm: o + Fig. 2. Immersion coefficients from replicate immersion tests on the MER-1012f (S/N 8107) Ed channels. The legend in the figure provides experiment dates and lamp-to-collector distances. The solid line is the least-squares regression fit to the mean coefficients.

Original page 15

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 1.5 1.4 ! t 1.3 E "j tJ i.2 i.i i.0 400 " 450 500 5 0 # + " ÷ 600 6 0 700 Wave]ength (nm) 9/1/94, R = 152.2 ¢m: x 9/1/94, R = 137.8 cm: 9/2/94, R = 152.2 cm: o 9/2/94, A = 137.8 cm: ÷ NEAN: # Fig. 3. Immersion coefficients from replicate immersion tests on the MER-1015 (S/N 8205) Ed channels. The legend in the figure provides experiment dates and lamp-to-collector distances. 9

Original page 16

Case Studies for SeaWiFS Calibration and Validation, Part 3 1.5 w 1.4 x 1.3 "'-_1 u. 1.2 1.1 1.0 ''''l''''l'''' I I I I I I I I I I I I I I I I I I I I I 400 45O 5OO 550 12/30/93, 7/22/94, 7/22/94, 7/26/94, 7/26/94, MEAN: 600 650 700 750 Wavelength (nm) R = 120.0 cm: x R = 137.9 cm: w R = 123.5 cm: o R = 137.9 cm: + R = 123.5 cm: # Fig. 4. Immersion coefficients from replicate immersion tests on the MER-1032 (S/N 8301) Ed channels. The legend in the figure provides experiment dates and lamp-to-collector distances. The solid line is the least-squares regression fit to the mean coefficients. 10 =

Original page 17

Mueller, Fraser, Biggar, Thome, Slater, Holmes, Barnes, Weir, Siegel, Menzies, Michaels, and Podesta 1.5 1.4 "-4-.. 1.3 E E LI_ 1.2 1.1 i.0 400 450 5o0 550 GO0 G50 Wavelength (nm} 7/22194, R = 131.5 cm: x 7/22194, R = 123.5 cm: 7/26194, g - 131.6 cm: o 7/26194, R = 123.5 cm: + MEAN: I Fig. 5. Immersion coefficients from replicate immersion tests on the MER-1032 (S/N 8301) Eu channels. The legend in the figure provides experiment dates and lamp-to-collector distances. The solid line is the least-squares regression fit to the mean coefficients.

Original page 18

Case Studies for SeaWiFS Calibration and Validation, Part 3 1.5 1.4 1.3 E LL 1.2 1.1 ;-T_W I If If I,_, I''' 'I'''' I''''I''''I 450 500 550 600 650 700 750. 7120/g4, 7/20/g4, 7/21/g4, 7/21/94, MEAN: Wavelength (nm) R = 121.6 cm: x R = 107.2 cm: R = 121.5 cm: o R = 107.2 cm: ÷ # Fig. 6. Immersion coefficients from replicate immersion tests on the MER-2040 (S/N 8724) Ed channels. The legend in the figure provides experiment dates and lamp-to-collector distances. The solid line is the least-squares regression fit to the mean coefficients. 12

Original page 19

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 1.5 1.4 1.3 '4 I h 1.2 l.! 1.0 4O0 I I I I450J I I I I500J I I I I550J I I I I 600J I I I I 650J I I I I 700I I I I 17_ 0 I_ave]ength {nm) 9/2/94, R = 125.4 cm: x 9/2/94. R = 111.0 cm: MEAN: o Fig. 7. Immersion coefficients from replicate immersion tests on the MER-2040 (S/N 8725) Ed channels. The legend in the figure provides experiment dates and lamp-to-collector distances. The solid line is the least-squares regression fit to the mean coefficients. 13

Original page 20

Case Studies for SeaWiFS Calibration and Validation, Part 3 1.5 tL 1.2 1.1 450 500 550 600 650 700 7 0 Wavelength (nm) 5/24/94. R = 120.7 cm: x 5/29194. R = 106.2 cm: HEAN: 0 Fig. 8. Immersion coefficients from replicate immersion tests on the MER-2040 (S/N 8738) Ed channels. The legend in the figure provides experimental dates and lamp-to-collector distances. The solid line is the least-squares regression fit to the mean coefficients 14

Original page 21

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta Table9. Linearregressionfitsto immersioncoefficients,asexpressedin (1),forirradiancecollectorsonseveral MERunderwaterradiometersmanufacturedby BSI.In addition,thesquaredcorrelationcoefficient,R2, and the residual standard deviation, sxu, are listed. MER Number of Measurement Regression Coemcient Model S/N Channels Type 1012 8107 7 E d 1032 8301 11 Ea 1032 8301 8 E_ 2040 8724 7 Ed 2040 8725 6 Ed 2040 8738t 12 Ed a b R 2 sxu 1.4745 2.6759 0.900 0.010 1.5833 4.0050 0.980 0.005 1.6496 5.1186 0.980 0.007 1.6084 4.3755 0.998 0.002 1.3826 1.0907 0.355 0.015 1.5405 3.5163 0.998 0.002 t MER-2040 S/N 8738 is equipped with a Teflon diffuser. All other instruments tested have Plexiglas diffusers. Table 10. Statistics of variability between immersion wavelengths. Wavelength Number of d:2 nm Collectors 406 3 410 7 442 8 489 8 555 3 664 3 670 5 explained by experimental uncertainty in the characterization procedure. The observed scatter far exceeds the allowable uncertainty implied by the radiometric calibration goals (that is, 1% uncertainty) of the SeaWiFS Callcoefficients for different irradiance collectors at selected Immersion Coefficient # a Range 1.3791 0.0346 0.0668 1.3966 0.0520 0.1513 1.3920 0.0377 0.1244 1.3744 0.0387 0.1157 1.3562 0.0113 0.0212 1.3194 0.0191 0.0382 1.3424 0.0419 0.0998 bration and Validation Program. It is essential, therefore, to experimentally characterize immersion factors for every profiling irradiance sensor to be used as part of SeaWiFS validation. 15

Original page 22

CaseStudiesforSeaWiFSCalibration and Validation, Part 3 Chapter 2 The Effect of Oxygen Absorption on Band-7 Radiance ROBERT S. FRASER NASA Goddard Space Flight Center Greenbelt, ABSTRACT Maryland Atmospheric oxygen absorbs about 13% of the available sunlight reflected from the Earth in band 7 of the SeaWiFS instrument, which has a bandwidth spanning 745-785 nm. If a correction is made for the average amount of absorption, then the measured radiance would vary by +0.004 for a two standard deviation (2a) variability in the amount of oxygen in a vertical column. For comparison, the instrumental noise is about one-half, or 0.002, of this variation. A correction based on the regional changes in absorption, as a function of season, would not significantly reduce the statistical variation in absorption, but a correction based on surface pressure would be accurate. 2.1 INTRODUCTION Atmospheric oxygen absorbs approximately 13% of the radiant energy in SeaWiFS band 7, which has a width In this simulation, sunlight is considered to reflect from both the sea surface and a concentrated layer of air just above the sea surface. The combined reflectance (for the sea surface and the layer of air) is equal to p. Light would spanning 745-785 nm. Since the total amount of oxygen is be absorbed along a two-way path through the entire at- : proportional to the surface pressure, the amount of oxygen varies directly with the variation in surface pressure. The standard deviation (la) of the surface pressure over the entire ocean is only about 1%. It is expected, therefore, that oxygen will cause a variability, equivalent to 2a, in the radiance in band 7, i.e., about 0.02 x 0.13 =0.0026. This variability is about 1.5 times the instrument radiance noise (Hooker et al. 1992). The detailed analysis presented here supports this conclusion. 2.2 THEORY The radiance (Lt), measured at a satellite, can be expressed as Lt = Latm -t- Lsfc, (2) where Latin is the radiance of light reflected from the atmosphere, and Lsfc is the radiance of light leaving an ocean surface and passing through the atmosphere. Although Lat m receives contributions from light scattered throughout the atmosphere, a slightly more conservative approach to estimating the effect of oxygen absorption is to assume that the scattering occurs in a thin layer near the surface, and that the absorbing oxygen lies entirely above the surface layer. 16 mosphere. Then the total radiance measured by the Sea- WiFS instrument (in orbit) would be Lt = /p(A)f(A)T())S(A)d), (3) where f isthe instrument spectralresponse function,S is the solarspectralirradiance,T isthe two-way transmission through the volume of oxygen T(A,O,00) = e -m(°'°°)r°x(:q (4) where fox is the oxygen absorption optical thickness, and m represents the air mass: 1 1 m =cosOo + cos 0" (5) In (5), 00 is the solar zenith angle, and 0 is the zenith angle of the line-of-sight in a plane-parallel atmosphere. In (4), the oxygen absorption optical thickness, %x(A), is defined as Vo×(A) = k(k)N, (6) where k is the molecular absorption cross-section area, and N is the total number of oxygen molecules per unit area in a vertical column of the atmosphere.

Original page 23

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 0 0.6 - ........... 0 ? ,.Q 0.4 ............ < .2 -- ........... 0.0 758 760 762 764 766 768 770 772 Wavelength [nm] Fig. 9. Oxygen absorption band (A-band), the integrand of (13), for an air mass m = 2. The curve is adapted from data given by Wu (1985). Two models, one with oxygen and another without oxygen, will be examined and compared with regard to the radiance SeaWiFS will measure in orbit. In the model without absorbing oxygen, L0 represents the radiance. The number of absorbing molecules above the reflecting layer, N, is then equal to 0, the optical thickness r=0, the transmission T=I, and from (3) the radiance is L0 = pS(Ax)./f(A)dA (7) = pS(A1)B. In this expression, the reflectance, p, can be considered constant without loss of generality, and S(765) has a value of 122.5mWcm-2#m -1 when A1 is equal to 765nm. The width of band 7, B, is approximately 40.5 nm, as defined by the integrated SeaWiFS spectral response function, f(A), supplied by Barnes (1994). The wavelength A1 = 765 nm is selected so that (3) and (7) are equal. If oxygen absorption is accounted for, the radiance, L, of light transmitted through the absorbing oxygen along a two-way path is, from (3) and (4), L = /p(A)f(A)e-ra(O'O°)r(1)S(A)dA. (8) J With T = 1 in (3), the absorbed radiation is then found as the difference between (3) and (8): AL = Lo - L, (9) = /p(A)f(A)[1-e-m(O'O°)r(X)]S(A)dA, (10) = ill) = pf(A2)S(,k2)W, (12) w = f[1-e-m(°'°°)r()]d_, (13) with the following values in effect: A2 = 764nm, f(_2) = 0.94, and S(A2) = 124mWcm-2#m -1. Note that A2 is a wavelength within the absorbing band (758 < As < 771 nm) and not within the total band (735-800 nm). The integral in (13) represents an equivalent bandwidth, W, of complete absorption and depends on the amount of oxygen. The integrand of (13) is given in Fig. 9. The absorption band is restricted to a width of 13nm (758 < A < 771 nm). 17

Original page 24

CaseStudiesfor SeaWiFSCalibrationandValidation,Part3 0 -a ' ' ' I .... I ' ' " ..... •,,i .: -¢'- ] 200 %!", "L 400 2 3 \ • ,, 600 .......................*.........................::..........................i............................................................... Ni • _ 800 ......................., .........................• ..........................":..................................-...................................... i ! \ ",i Sea! L_v..l i ........................................ .-,,,...... i..................... 1000 ............................... ""': ""'='"'*"'"""':"'-'":'""'"" "; 1200 ........ ., t .... 0.0 1.0 2.0 ""'i ""'"':" " " : ...... "m : • i,,,, t .... t .... 3.0 4.0 5.0 6.0 Equivalent Width [nm] Fig. 10. The equivalent width, W, of the oxygen A-band for sunlight reflected from surfaces at increasing pressure levels in the atmosphere. Curves are given for air masses 2 and 3. This figure is adapted from computations made by Curran (pers. comm.). Table 11. Equivalent width, W, of oxygen A-band for sunlight passing to the surface (1,013 mb) and reflected into space. The data presented are taken from Fig. 10. The symbol p represents surface pressure. Air W A W / Ap b-actional Mass [nm] [nm/mb] Absorption W/B 2 4.6 0.0031 0.1] 3 5.2 0.0035 0.13 The change in W with respect to the total surface pressure of all atmospheric gases (N2, 02, etc.) is shown in Fig. 10. Table 11 gives W for band 7, calculated for sunlight reaching sea level and then reflected to space. The change in W with respect to pressure appears in the third column, and the fraction of energy absorbed in band 7, from (15), appears in the last column. The relative loss of radiance caused by oxygen absorption is found by dividing (12) by (7): AL pf(A)S(A2)W Lo pS(A1)B (14) W =-E' 18 where for air mass m = 2, f(=)s() s() 0.94 x 124 (15) 122.5 = 0.95. The relative amount of absorbed energy taken out of the band is approximately that given by letting fl = 1 in (14): AL W (16) Lo B The relative loss in radiance equals the ratio of the equivalent width W of the oxygen band, to band 7 width B. The change in the equivalent bandwidth caused by a change in the amount of oxygen can be calculated from (14): AL AW Lo B (17) AW Ap Ap B'

Original page 25

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel, Menzies, Michaels, and Podesta where p is the total sea level pressure of all atmospheric gases, i.e., surface pressure. Statistical data on the atmospheric surface pressure are not available with high spatial and temporal resolution. The median value of pressure over the oceans is 1,012.5 mb, and the extreme monthly values are 969 and 1,043 mb (Mc- Clain et al. 1994). The extremes and reference pressure (P_ef) data are given in Table 12. The deviations of -44 and +30 mb are the changes of the monthly extreme pressures from 1,013 mb (United States Navy 1978). The maximum presssure variations occur in the middle latitudes during the winter. The deviations of -30 and +34 mb, appearing in the last column of the second row, are based on the maximum data given by Cantor and Cole (1985); the deviations are computed as the local average (+ 2a) minus the worldwide average. For example, the All ocean deviations=l,015 =t=32 - 1,013 = -30, +34 mb. The tropical tions in the tropics would have a small effect on absorption changes in band 7. Table 13. Absorption changes for band 7. Air AW/Ap Pdev W Absorption Mass [nm/mb] [mb] Change 3 0.0035 -44 -0.15 -0.004 3 O.0035 +34 +0.12 +0.003 4 0.0041 -44 -0.18 -0.004 4 0.0041 +34 +0.14 -0.003 2.3 CONCLUSION The simplest correction for oxygen absorption is based on a constant amount of atmospheric oxygen in a vertical direction. In this case, oxygen absorbs about 13% of the deviation for a 2a variability in pressure (-6, +2 mb) is radiant energy available for remote sensing by SeaWiFS much weaker. Because extreme low-pressure cyclones are associated with strong winds, rain, and overcast clouds, satellite observations of the surface would not be possible under such conditions. Table 12. Sea level pressure data for oxygen absorption estimation. The deviation column values (Pdev) are the differences between the minimum and maximum surface pressures compared to 1,013 mb. StatisticaI Pref Pdev Basis [mb] [mbl All-ocean averaget 1,015 -30, +34 Monthly extreme low 969 -44 Monthly extreme high 1,043 +30 Tropicst 1,011 -6, +2 Cantor and Cole (1985); all others from McClain et al. (1994). Absorption changes caused by extreme variations in band 7. The absorption depends on the length of the path through the atmosphere: from the sun [i.e., total solar irradiance at the top of the atmosphere (TOA)] to the surface, and back to SeaWiFS. Even if the amount of oxygen in a vertical direction is assumed to be constant, oxygen corrections have to be adjusted for the geometry. Statistical data are not available for making a precise estimate of the variation in absorption caused by the change in the amount of oxygen that occurs when the atmospheric pressure varies from an average value of 1,013mb. For a 2a variation in the amount of oxygen, however, the variable absorption in band 7 is not more than twice the instrumental noise. The operational procedure for making an atmospheric correction for molecular scattering will use the surface pressure in the correction algorithm. This pressure data can also be utilized to make a correction for oxygen absorption. In this case, the statistical variations discussed here would not occur. In the above discussions, the vertical gradient of atoxygen absorption are given in Table 13 for air masses 3 mospheric optical properties has been neglected, but Ding and 4. The minimum (-44 rob) and maximum (+34 mb) and Gordon (1994) have shown that the vertical profile pressure changes are taken from Table 12. The changes in must be included for the derivation of the water-leaving equivalent bandwidth (AW/Ap) are taken from Table 11 for air mass 3 and extrapolated for air mass 4. The magnitude of the variations is 0.003-0.004, which can be compared with the instrumental noise, 0.002, estimated from the signal-to-noise ratio (SNR) for band 7 (Hooker et al. 1992). As seen from Table 12, the small pressure variaradiance to be sufficiently accarate. Corrections based on a single reference profile are adequate except for the following three cases: large amounts of stratospheric aerosol, such as occurred after the volcanic eruptions of El Chich6n and Mount Pinatubo; thin cirrus clouds; and aeolian dust, such as that from the Sahara or Gobi Deserts. 19

Original page 26

Case Studies for SeaWiFS Calibration and Validation, Part 3 Chapter Second SeaWiFS Preflight 3 Solar Radiation-Based Calibration Experiment STUART F. BIGGAR KURTIS J. THOME PHILIP N. SLATER Remote Sensing Group, University of Arizona, Optical Sciences Center Tucson, Arizona ALAN W. HOLMES Santa Barbara Research Center Goleta, California ROBERT A. BARNES ManTech, Inc. Wallops Island, Virginia ABSTRACT This paper describes the second solar radiation-based calibration of SeaWiFS. The experiment was done on 1 November 1993 in the rock garden at SBRC. The results of the calibration are presented, along with a comparison to the SIS calibration done at SBRC. The estimated uncertainty of the SIS calibration is 2.8%, compared to the 4% estimated uncertainty for the solar-based calibration. There is also an uncertainty in the value of the exoatmospheric solar irradiance used to make this comparison, which is probably on the order of 1%. In addition, the integrated out-of-band blocking for SeaWiFS is in the 1-3% range, which can introduce significant differences between the lamp- and solar-based calibrations. This agreement is better than anticipated, and better than the agreement achieved in March 1993 from the first experiment. The better agreement is probably due to a sphere recalibration between the two experiments. 3.1 INTRODUCTION The basic concept for a solar radiation-based calibration of a satellite sensor is to attempt to simulate the solar irradiance incident on the diffuser in space while doing the experiment on the ground. A thorough discussion of the concept was presented in April 1993 at the Society of Photo-optical Instrumentation Engineers (SHE) meeting in Orlando, Florida (Biggar et al. 1993). This presentation included results from the first calibration performed on SeaWiFS. After this calibration, a stray light, or transient response, problem was discovered in SeaWiFS. The sensor was subsequently modified to reduce the response to outhaze and smoke from fires in the Southern California area, prevented the calibration from taking place on 29-31 October. On 1 November, sky conditions were good enough for calibration purposes. The sensor was taken outside at about 1115 Pacific Standard Time (PST), and measurements were taken at about 1215, 1255, and 1400 PST. The instrument was covered immediately after the last data set was collected. 3.2 EXPERIMENTAL METHOD The transmittance along the path to the sun was measured with a solar radiometer possessing 10 bands coverof-field radiation, such as that caused by clouds, within 10 ing the spectral range of about 370 1,040nm. Table 14 pixels of the SeaWiFS instantaneous field-of-view (IFOV). After the modifications, the sensor was recalibrated in the laboratory using the 100em SIS at SBRC. Another solar radiation-based calibration was scheduled for late October 1993, to coincide with a look at the full moon on 29 October. Cloud and visibility conditions, caused in part from 2O shows representative radiometer data for the nine bands that were not affected by water vapor absorption. The sky conditions on 1 November 1993 were not sufficiently favorable to allow a Langley plot determination of the optical depth; therefore, instantaneous measurements of the transmittance were made by using previous calibrations

Original page 27

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta Table 14. Solarradiometeropticaldepthmeasurements. Band Wavelength Optical Number [nm] 368.9 1.0109 0.0061 399.1 0.8279 0.0070 440.0 0.6623 0.0061 518.6 0.4633 0.0050 Depth ¢r 5 608.5 0.3499 0.0045 6 669.0 0.2756 0.0039 7 779.7 0.2008 0.0033 8 869.8 0.1643 0.0028 10 1,027.2 0.1284 0.0020 Table 15. SeaWiFS measurements on 1 November pixel 0 (zero offset) were used for the determinations 1993 at 1400 PST. Pixel 312 (the center pixel) and of the total and diffuse only signals. Band Shaded Shaded Unshaded Unshaded Diffuse/ A Unshaded Number (zero offset) (312) (zero offset) (312) Global vs. Shaded 20 60 17 58 20 52 20 55 20 254 0.17094 194 17 294 0.14801 236 20 282 0.12261 229 20 332 0.11218 277 22 60 22 421 0.09520 361 24 60 24 508 0.07438 448 24 55 24 508 0.06405 453 21 58 21 591 0.06491 533 of the instrument zero-airmass intercept. Measurements by the same solar radiometer were also made, after the calibration described here, during subsequent satellite calamount, and the barometric pressure, the optical depth components can be computed for each of the SeaWiFS bands. MODTRANwas used to compute the effects of gaseous ibration campaigns at White Sands National Monument in absorption for each band. The only band that exhibits any New Mexico. For the intercepts done before and after the solar radiation-based calibration, the standard deviation (a) is about 1% of the intercept value. It is expected that the error in the transmittance measurement when using these intercepts will be less than approximately 3% at the measurement wavelengths. The optical depths and the barometric pressure are used to separate the optical depth components due to Rayleigh scattering, aerosol scattering and absorption, and absorption due to ozone (Biggar et al. 1990). The procedure employed here assumes a Junge power law distribution for aerosol particle size. The results on 1 November give a Junge parameter of 3.50-3.43, and a derived columnar ozone amount of 0.246-0.260 cm-atmt for the three measurement times. Using the Junge parameter, ozone The centimeter-atmosphere (cm-atm) is a measure of trace gas columnar amount. It can be envisioned as if the entire trace gas content within a 1 cm _ column of the Earth's atmosphere was accumulated at the base of this column under standard temperature and pressure conditions. The cm-atm would give the length of this volume of gas in centimeters. significant absorption is band 7, and the oxygen slant path transmittance for this band is computed to be 0.927. The transmittance measurements described above were performed when the sensor was actually taking data from the illuminated solar diffuser. Measurements of the diffuser were made with the diffuser illuminated by the sun and the sky (unshaded), and by only the sky (direct beam blocked, or shaded). Data from pixel 312, the center pixel from the diffuser, along with that from pixel 0 (the zero offset) were used to determine the total and diffuse only signals. The choice of pixel 312 is not critical, as the variation across the diffuser is no more than one digital count (DC) for any band. 3.3 RESULTS The measurements from the diffuser are presented in Table 15. These data are from the 1400 PST measurement sequence. The data in the table include the zero offset for each band for both the shaded and unshaded measurements, the actual measurements, and two computed 21

Original page 28

Case Studies for SeaWiFS Calibration and Validation, Part 3 Table 16. Transmittance and predicted DCs in orbit. Band Vertical Number Path 1 0.4651 2 0.5274 3 0.6004 4 0.6238 5 0.6654 6 0.7575 7 0.7507 8 0.8437 Table 17. Sphere calibration computations. Band Integrated Calibration Number Irradiance ( E) 1 170.827 0.00704 2 188.992 0.00810 3 193.383 0.01045 4 189.022 0.00923 5 187.431 0.00744 6 151.546 0.00629 7 121.715 0.00512 8 98.168 0.00422 Bidirectional Reflectance Distribution Function quantities--the ratio of the diffuse signal to the global (or total) signal, and the difference between unshaded and shaded measurements. This difference is the direct solar beam signal without the diffuse sky contribution. A forward-scatter correction must be made to account for the small amount of forward-scattered diffuse light that is blocked by the disk. This corrected measurement of the direct solar beam is then further corrected for the transmittance in order to compute an expected solar diffuser measurement at TOA, i.e., in orbit. The forward-scatter correction is wavelength dependent, but in all cases it is very small, i.e., less than 1 DC. This correction is subtracted from the difference in measurements. The DCs for each SeaWiFS band are then divided by the slant path transmittance, which has been computed for each band. This transmittance is computed with Beer's Law, using the optical depth components in each band, and further multiplied by the oxygen transmittance in band 7. The transmittance values computed for each band are given in Table 16. These values correspond to a time of 14:00:30 and an effective airmass of 1.6354. The uncertainty in airmass for a five-minute period around this time is 0.0079, which corresponds to an uncertainty in transmittance of less than 0.5%. The predicted DCs have some associated uncertainties, which can be difficult to quantify. The major source of uncertainty is the transmittance measurement. Other 22 slant Predicted Path TOA DCs 0.2860 677.5 0.3512 671.1 0.4342 526.8 0.4622 598.7 0.5136 702.2 0.6349 705.1 0.6565 689.7 0.7573 703.5 BRDFt Spectral Offset Predicted Correction DCs 0.02682 0.963 20 695.8 0.02786 0.983 17 678.3 0.02751 0.980 20 539.5 0.02806 0.998 20 595.8 0.02755 0.991 22 722.4 0.02800 1.011 24 691.3 0.02854 0.995 24 705.9 0.03011 1.011 20 712.8 sources of uncertainty are interpolation from the radiometer wavelengths to the SeaWiFS wavelengths, the forwardscatter correction, the oxygen transmittance computation for band 7, and atmospheric variability. The transmittance measurement uncertainty is probably less than 3%, as this uncertainty is dependent on the radiometer calibration. The interpolation uncertainty can be estimated by using the calculated optical-depth components to compute the expected transmittance, which was also measured in the solar radiometer bands. The largest difference is for band 4 of the radiometer, and it corresponds to an error in transmittance of about 0.9%. The second error term in Equation 3 of Biggar et al. (1993), representing the airmass uncertainty, is less than 0.2% for all bands. The atmospheric variability can be estimated by comparing the standard deviation of the transmittance measurements to the average. The variability in transmittance in the radiometer bands was about 1%. The average transmittance was used so that this variability should not cause a significant uncertainty. A total uncertainty on the order of 4% is expected, with the radiometer calibration being the dominant term. The solar radiometer gives a much more repeatable measurement, so it would be better if the sky conditions were stable enough for a Langley plot determination of the transmittance. Conditions in Santa Barbara were not good enough, however, for such a determination.

Original page 29

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 1 0 D Mar. 8, 1993 Results 09 * Nov. 1, 1993 Results (/1 a9 -EO.8 0 go.7 E Eo.6 oJ 0 go.5 ,-_" t- O _0.3 e.n0.2 E 0 _0.1 0.0 -------El i I ; a I i i i i I i i i i I i i i i I i i i ! I ! i i i I I i I i I i i i i I I I ! i I i i , v t 400 450 500 550 600 Wavelength (nm) Fig. 1 1. Measured atmospheric transmittances during 650 700 750 800 850 900 the two solar calibration experiments. The transmittances are given at the wavelengths of the Arizona solar radiometer. iillr199ests -11 "-" - 312-I.... , .... , .... _ .... , .... 400 450 500 550 600 Wovelength (nm) , .... _ .... , .... , .... , .... 650 700 750 800 850 900 Fig. 12. Comparison of the laboratory- and solar-based calibrations of SeaWiFS. The laboratory measurements were made by SBRC with a 100 cm SIS. For the 1 November 1993 solar calibration, the two techniques agreed to better than 3%. 23

Original page 30

CaseStudiesforSeaWiFSCalibration and Validation, Part 3 3.4 DISCUSSION As a check on the results, and in an effort to relate these results to the International System of Units (Sit), these results were compared to the calibration of the Sea- WiFS system with the SBRC 100 cm SIS. The SIS results were taken from the SeaWiFS Calibration and Acceptance Data Package, which is described in Barnes et al. (1994). Exoatmospheric solar irradiance data, based on Neckel and Labs (1984), were used; these data were integrated over the SeaWiFS band and then divided by the radiance-to- DC calibration. This value was then multiplied by the diffuser BRDF and divided by the correction for spectral shape. The zero signal offset was then added to give the experiment. The better agreement is probably due to a sphere recalibration between the two experiments. It is also instructive to compare the SeaWiFS band transmittances from the two dates (Fig. 11). The atmosphere was clearer in March compared to late October, when smoke from fires in the Los Angeles area affected conditions in Santa Barbara. The measurements in March were also at a lower solar zenith angle than in November. Both of these factors could possibly make the November data more uncertain; however, the results still compare well with laboratory measurements. This favorable comparison leads to the conclusion that the transmittance measurements are fairly accurate even though a Langley plot was not possible. Figure 12 shows the perpredicted TOA DC value. The results are summarized in centage difference between the calibration methods for the Table 17. The results in Table 17 can be easily compared to those in Table 16. The summary of this comparison is shown in Table 18. For all bands, the differences between the solar-based and sphere-based methods are less than 3%. The estimated uncertainty of the SIS calibration is 2.8%, compared to the 4% estimated uncertainty for the solar-based calibration. There is also an uncertainty in the value of the exoatmospheric solar irradiance used to make this comparison, which is probably on the order of 1%. In addition, the integrated out-of-band blocking for SeaWiFS is in the 1-3% range, which can introduce significant differences between the lamp- and solar-based calibrations. This agreement is better than anticipated, and better than the agreement achieved in March 1993 for the first The SI acronym is derived from the original French title, Syst_me International d' Unitds. 24 two dates. Table 18. Comparison between solar radiationbased and laboratory (SIS based) calibrations. Band Solar SIS Difference Number Based Based [%] 677.5 695.8 2.6 671.1 678.3 1.1 526.8 539.5 2.3 598.7 595.8 -0.5 702.2 722.4 2.8 705.1 691.3 -2.0 689.7 705.9 2.3 703.5 712.8 1.3

Original page 31

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta Chapter 4 In Situ Evaluation of a Ship's Shadow CHRISTIAN T. WEIR DAVID A. SIEGEL DAVID W. MENZIES University of California at Santa Barbara Santa Barbara, California ANTHONY F. MICHAELS Bermuda Biological Station for Research Ferry Reach, St. George's, Bermuda ABSTRACT In situ measurements of optical properties made from a ship can be biased by the ship's shadow. In an effort to evaluate the ship shadow perturbation created by the R/V Weatherbird II, profiles of downwelling irradiance, Ed(Z, A); upwelling radiance, Lu(z, A); as well as derived AOPs were obtained at four distances--1 m, 3 m, 6 m, and 20 m or more--off the ship's stern. Two statistical analysis uses data from pairs of simultaneously-obtained analyses of these data are explored here. The first light profiles, one profile obtained at a distance greater than 20 m from the stern of the ship, and the other taken either 1 or 6 m off the stern. The second analysis compares the derived AOPs for each profiling distance from the ship, using data obtained throughout the length of the experiment. Significant differences are rare in comparisons of profiles obtained at least 3 m off the ship's stern. At 1 m off the stern, however, significant discrepancies are intermittently observed. This work illustrates that the inherent sources of noise in determining radiative fluxes and AOPs in the upper ocean are generally greater than the effects incurred by the ship's own shadow under optimal conditions. 4.1 INTRODUCTION Accurate measurements of AOPs are required to develop a detailed understanding of the processes regulating bio-optical property distributions and their relationship to remotely sensed signals. Instrumentation designed to measure properties of the underwater radiation field, when deployed at relatively close proximity to a ship, may encounter perturbations caused by the ship's shadow (e.g., Poole 1936, Strickland 1958, Gordon 1985, Voss et al. 1986, Waters et al. 1990, and Helliwell et al. 1990). This source of error is of obvious importance and must be accurately assessed. Poole (1936) estimated that the ship shadow erin downwelling irradiance rarely exceeds 2% as long as skies are clear and the sun is within 45 ° of the stern. At low solar elevations, however, these errors can increase to about 10%. Gordon (1985) also shows that the errors are reduced as the instrument is moved horizontally away from the ship, although errors during diffuse light conditions may remain as high as 30%. Voss et al. (1986) conducted an experiment with an extendable sea-going crane that showed values of upwelling radiance, L_ (z, A), decrease by 10-20% unless the instrument is deployed more than 5 m from the ship. Ship shadow perturbations are likely to be the greatest ror under diffuse skylight is about 10% at a depth of 5 m near the sea surface. This factor is critical for the develwhen the radiometer is deployed approximately 2 m off the opment of ocean color algorithms, as maximum accuracy stern, and also noted that this source of error decreases in must be sought for the determination of calculated pasignificance with increasing depth. The Monte Carlo simulations performed by Gordon (1985) indicate that the error Editors' Note: This chapter originally appeared as an article in Ocean Optics XII, published by SPIE (Weir et al. 1994), and is being included in this volume with the permission of the rameters such as the remote sensing reflectance, Rcs(A). Several studies have attempted to completely avoid the ship's shadow by floating optical instrumentation a considerable distance from a ship (Gordon and Clark 1980, Clark 1981, and Waters et al. 1990). These deployment strateauthors and SPIE. Minor editorial changes have been made to gies are difficult to conduct operationally, particularly in reflect the style of The Sea WiFS Technical Report Series. rough seas. Such strategies also place severe limits on the 25

Original page 32

Case Studies for SeaWiFS Calibration and Validation, Part 3 amount of data that can be collected, restrict linkages to other oceanographic observations, and limit time-scale resolution, e.g., Dickey and Siegel (1993). In order to completely avoid the ship's shadow when measurements of downwelling irradiance, Ed(z, A), are being made, Mueller and Austin (1992 and 1995) suggest that the deployment of optical instrumentation should be at a distance, d- Mueller and Austin (1992 and 1995) define d using sin(48"4°) (18) 44 = Kd(A) ' where Kd(A) represents the vertical attenuation coefficient for downwelling irradiance. Typical values for Kd(A) off Bermuda range from 0.02-0.08 m -1 (Siegel et al. 1994), which result in recommended deployment distances of 9- 4.2 EXPERIMENTAL METHODS The observations presented here were made from the Weatherbird II off Bermuda on 7 July and 9-10 July 1992. Two underwater spectroradiometers--the BBOP and the OFFI--were lowered simultaneously from the Weatherbird H to about 50 m (Fig. 13). The BBOP package consisted of a BSI MER-2040 underwater spectroradiometer, interfaced with a SeaTech transmissometer, chlorophyll fluorometer, and SeaBird conductivity, temperature, and pressure sensors (Siegel et al. 1994). The OFFI is a modified BSI MER-2020 underwater unit with a case outfitted with buoyant fins to provide stability and control in its descent rate (Waters et al. 1990). The BBOP was lowered at three distances off the stern (1, 3, and 6 m) using the extendable boom. The OFFI was fished out, i.e., deployed astern of the ship, at least 20m 40 m from a ship's stern. Similarly, recommendations for before descent. Profiles were made simultaneously so that deployment distances for E(z, ),) and L(z, A) (_ and _L, respectively) are given by Mueller and Austin (1992 and 1995) as 3 _'- gu(A)' (19) and 1.5 L- KL(A)' (20) where K,,(A) and KL(A) are the vertical attenuation coefficients for upwelled irradiance and radiance, respectively. Values of K,(A) and KL()) are roughly equal to values of Kd(A). Typical values of _ and L are usually greater than 30 m, which is nearly the length of the ship used in this study. The distances recommended by (18)-(20) are based only on geometric relations, and are independent of ship size, sky conditions, sea state, deployment method, and the orientation of the ship with respect to the solar beam. In this study, the effects of ship shadows upon data collected from the R/V Weatherbird//--length 35.05 m, beam 8.53 m, and draft 2.60 m--are examined. The Weatherbird H is used by the Bermuda Bio-Optics Project (BBOP) to make routine spectroradiometer casts in conjunction with the Joint Global Ocean Flux Study (JGOFS) Bermuda Atlantic Time-Series Study (BATS). Spectroradiometer profiles were made using an extendable boom to deploy the BBOP package, which allowed profiles to be made up to 6 m off the stern of the Weatherbird II. These data are compared with data collected using the optical free-falling instrument (OFFI) described by Waters et al. (1990). The OFFI data provide a control which can be used to search statistically for the effects of the ship's shadow. The measurements used were made under optimal conditions (i.e., clear skies, stern-to-ship solar orientation, near-constant illumination, etc.), and thus, provide the basis for evaluating the role of the ship shadows in developing ocean color remote sensing algorithms using the BBOP data set. 26 instantaneous fluxes from the two instruments could be compared. Both instruments were deployed with the sun off the stern so that the ship's shadow trailed away and behind, which is part of the normal BBOP sampling procedure. Both instruments sampled Eu(z,A) and Lu(z,A) in spectral wavebands centered at 410, 441,488, 520, and 565 nm. Using laboratory facilities at the University of California at Santa Barbara (UCSB), radiometric calibrations were performed on both instruments two weeks prior to, and two months after, this cruise. The same calibration lamp (UCSB lamp F-303) was used for both calibrations. The calibration coefficients for the BBOP instrument varied by less than 0.5% for irradiance, and less than 3% for radiance, between both calibration dates. The OFFI calibration coefficients differed by 1-4% for both the irradiance and radiance channels. Calibration coefficients, which were obtained from the precruise determinations, were used for the analysis presented here. The two individual bio-optical data sets were processed using the BBOP data processing system (Sorensen et al. 1994). The BBOP data processing system is used to: 1) Eliminate radiation values that are below a specified threshold; 2) Identify time segments where cloud perturbations are minimal; 3) Smooth specified channels of data and remove spikes, although not for Ed(Z, A) or L(z, A); and 4) Bin the data into 1 m vertical depth bins. The derived AOPs, such as Kd(z, A) and the remote sensing reflectance R_s(z, A), are also calculated. In addition, the downwelling irradiance and upwelIing radiance spectra just beneath the sea surface, Ed(O-, A) and Lu(0-, A), are determined by fitting profile data from the upper 20 m to the Beer-Lambert relation. Sorensen et al. (1994) gives a complete description of the BBOP data processing system and data handling procedures used by BBOP.

Original page 33

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta om [ >20m Fig. 13. Diagram of the BBOP ship shadow evaluation experimental design. 4.3 RESULTS Two distinct statistical analyses are performed to evaluate the effects of the shadow cast by the R/V Weatherbird II. The first analysis compares the statistical differences between simultaneously sampled BBOP and OFFI data as the BBOP radiometer is deployed at various distances off the ship's stern. Differences between the two data sets are interpreted here to indicate the effects of the ship shadow, after accounting for a constant calibration error and the occurrence of random errors, i.e., noise. This analysis will be referred to as the simultaneous comparison. The second analysis, referred to as the muIti-distance comparison, uses data obtained throughout the experiment to compare mean derived AOP values at each of the four distances (1, 3, 6, and greater than 20 m). The object of this comparison is to address whether any significant differences can be found among the AOP determinations. 4.3.1 Simultaneous Comparison The simultaneous comparison evaluates the statistical difference between the fluxes and AOPs measured by the BBOP profiler at two distances off the ship's stern, and identical parameters determined using the OFFI profiler. The BBOP casts with _ = 1 m are referred to as the BI two instruments are estimated to be collected simultaneously to within 5 sec. The statistical differences between OFFI (O20) and BBOP (B1 or B6) measurements of downwelling irradiance, upwelling radiance, and derived AOPs are compared. A positive difference means that the BBOP measurements underestimate the 020 values, which may indicate a ship shadow influence. The error bars shown correspond to 90% confidence intervals (c.i.) for the mean estimates throughout the analysis. For the experimental measurement of downwelling irradiance at 441 nm, there are no statistically significant differences (at the 90% confidence level) between the O20 casts and either the B1 or the B6 casts (Fig. 14, top). However, the B1 mean differences are consistently positive, suggesting that the B1 casts may be affected by the ship's shadow, although not in a statistically significant manner. The B1 - 020 difference increases as the sea surface is approached, which also suggests the signature of a ship shadow. The vertical profile of the B6 - 020 differences does not give any indication of a ship shadow influence. The other matching wavelengths (410, 488, 520, and 565 nm) for Ed(z, A) gave similar results, which are not shown here. The differences between the B1 and O20 upwelled radiance data at 441 nm, Lu(z,441), are significantly different from zero and show a ship shadow pattern casts, and the BBOP casts with _ = 6 m are denoted as with depth, as the mean difference increases significantly B6. A total of 15 paired OFFI-BBOP casts, 7 B6 and 8 towards the sea surface (Fig. 14, bottom). This divergence B1 casts, are used in this analysis. In the following discussion, the OFFI casts are designated 020. Data from the of the measurements is particularly apparent over the top 20 m, and the differences become smaller with increasing 27

Original page 34

Case Studies for SeaWiFS Calibration and Validation, Part 3 depth. The B6 - 020 mean difference for L,(z, 441) shows no statistically significant pattern with depth. This observed reduction in ship shadow effects with distance from the ship is consistent with previous studies. In terms of the derived AOPs Kd(z, ) and/8(z, ), no significant differences are found between either the B1 or B6 BBOP casts and the O20 data (Fig. 15). In particular, there are no consistent variations in these differences with depth that may be simply attributed to a ship shadow. This is true even for the mean Rrs(z, )) differences (Fig. 15, bottom) where the 020 -B1 Lu(z, ) observations showed some deviations attributed to ship shadow. This lack of a signature in Rrs(z, )) may be due to the fact that both the 020 - B1 Lu(z, )) and Ed(z, )) determinations axe affected by the ship shadow. The decrease in both L,(z, )) and Ed(z, )) due to the ship shadow, may actually cancel the effects of the shadow on values of Rr_(z, )). Accurate measurement of both the upwelling and downwelling light streams just beneath the sea surface is critical to the development of algorithms for estimating bio-optical properties from satellite sensors. The differences between the 020 and B1 or Be estimates of Ed(O-, )) and L_(0-, ), for both the B1 and Be distances off the stern, are shown in Fig. 16. The mean Ed(O-, ) differences show no significant differences from zero, or between the two deployment distances (Fig. 16, top). Significant divergence from zero is found, however, for the mean L(0-, ) differences for all wavelengths except 565 nm. Significant differences are also found for some of the wavelengths in the B6 - O20 comparison, although it is unclear how large of a calibration difference remains between the two instruments. In particular, the size of the disparity in the mean differences between the two BBOP deployment distances increases as ) is decreased. These spectral observations are consistent with numerical results that indicate that the ship shadow effects scale as c(A), where c(A) is the beam attenuation coefficient and _ is the distance from the ship (Gordon 1985). The value of c(A) at 565 nm is likely to be larger than its value at 441 nm. These results further show that the influence of the ship's shadow will be more critical for the upwelling light stream rather than for downwelling light, as is expected. These results can be compared to Mueller and Austin (1992 and 1995). 4.3.2 Multi-Distance Comparison The second analysis compares the mean values of derived AOPs using data obtained for each of the four distances (1, 3, 6, and greater than 20 m) throughout the experiment. All available casts are used for this analysis and the AOP determinations are classified by their distance from the stern of the Weatherbird II. The variations in Kd and R_s at 441 nm with depth and deployment distance are shown in Fig. 17. Only rarely are there statistically significant differences (i.e., non-overlapping error bars), for the 28 four deployment distances, though trends with distance are apparent. Similar results, not shown here, are found with the other wavelengths sampled by the BBOP. Spectral differences in the remote sensing reflectance just beneath the sea surface, R_ (0-,)'1, can be used to surmise the spectral structure of the ship's shadow (Fig. 18). Again, no significant differences are found for any of the wavelengths. This analysis again suggests that the effects of the ship shadow on the upwelled light field may be effectively canceled out when normalized by the downwelling irradiance. 4.4 DISCUSSION In order to correctly interpret the present results, it must be recognized that the observed mean differences are a composite of one or more signals: the actual ship shadow perturbation, a constant calibration difference between the OFFI and BBOP, and random errors due to the poor sampling of short time-scale noise (i.e., wave glint, small clouds, ship roll, and other effects). The influence of random noise can be reduced by averaging over many individual casts; however, the sample size for the simultaneous comparison is relatively small (N = 7 or 8). Comparisons between the pre- and post-cruise calibrations indicate the occurrence of only small calibration differences. The mean difference observed is, therefore, primarily composed of the ship shadow perturbation as modulated by an incompletely sampled random noise field. Elucidation of the ship shadow perturbation above this random noise element is the present goal. The results of the simultaneous comparison show that there are no significant differences between the comparison of 020 Ed(z, ) values and either B6 or B1 irradiance determinations (Fig. 14, top). This result is consistent with the irradiance direct beam cone remaining off the stern (Gordon 1985 and Voss et al. 1986). The determination of upwelled radiance, however, clearly shows the effects of the ship shadow for the B1 comparison, but not for the Be comparison (Fig. 14, bottom). A significant difference is also observed between the B1 and Be mean L (0-, A) differences (Fig. 16, bottom). At all wavelengths, B1 mean differences were consistently larger than those calculated from B6 data, with the 020 value being predictably and significantly greater than its simultaneous BBOP measurement. These results clearly show that BBOP data must be taken more than 1 m off the stern of the Weatherbird II, but does not have to be taken beyond 6 m, in order to avoid the ship's shadow perturbation. The results of the multidistance comparison provide additional information for fine tuning of the distance criteria for the Weatherbird II. The derived AOP profiles showed little variability among the four different deployment distances. The multidistance comparisons support the notion that the effects of the ship's shadow may be effectively masked by random errors associated with the many sources

Original page 35

Mueller,Fraser,Biggar,Tbome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta SIMULTANEOUS COMPARISON OF ED_441 -5 i 1 I : ._. : i : _ ,. I -2O "1- I-a. -25 i 0 LLI Q -3O i 0 I X- i -35 i 0 I X i -4O I 0 -45 i i _: -10 -5 0 i i O I 0 i i x ==> OFFI - 6m BBOP (N=7) o ==> OFFI - lm BBOP (N=8) i _ __ 5 10 15 IRRADIANCE [uW cm-2 nm-1] SIMULTANEOUS COMPARISON OF LU_441 -5 i i i i -1( I i X i -1.' I -2o I "1" I ),( i a. -25 iii tm -30 i O -35 i 0 X I -4O , , , , i i i i i 0 I 0 i 0 ' I x ==> OFFI - 6m BBOP (N=7) i o ==> OFFI - lm BBOP (N=8) , , , , , -d.o6 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 RADIANCE [uWcm-2 nm-1 sr-1] Fig. 14. Mean differences (with 90% c.i.) of simultaneous determinations of downwelling irradiance at 441 nm, shown in the top panel, and upwelling radiance at 441 nm, shown in the bottom panel. The differences are between 020 casts and B1 (o; I m off the stern) or B6 (x; 6 m off the stern) BBOP casts. 29

Original page 36

Case Studies for SeaWiFS Calibration and Validation, Part 3 SIMULTANEOUS COMPARISON OF K_ED441 -5 -10 i -15 P O I - X -2O l :E I- i X a. -25 LIJ c_ , X -3O I -35 •, X -4O I I 01 -0.005 0 )( i I 0 I I x ==> OFFI - 6m BBOP o ==> OFFI -lm BBOP , )< J I I 0.005 0.01 0.015 0.02 DIFFUSE AI-I'ENUATION COEFFICIENT [m-l] SIMULTANEOUS COMPARISON OF Rrs_441 -5 -10 I O -15 _' C -2O , :X: -25 UJ a -30 *, -35 i , .4O I I -45 -0.5 0 l .'- X l r >( (> I )(" , )( i 0 i X , 0 J 1 ! 0.5 1 1.5 REMOTELY SENSED REFLECTANCE [st- 1] x 10 .9 Fig. 15. Mean differences (with 90% c.i.) of simultaneous determinations of the diffuse attenuation coefficient for downwelling irradiance, Kd(z, 441), reflectance, Rrs(z, 441), shown in the bottom panel. shown in the top panel, and the remotely sensed The differences are between 020 casts and B1 (o; 1 m off the stern) or B6 (× ; 6 m off the stern) BBOP casts. 30

Original page 37

Mueller, Fraser, Biggar, Thome, Slater, Holmes, Barnes, Weir, Siegel, Menzies, Michaels, and Podesta ED410 @ z=0 58O i _ i I i , i I IO x I I 560 540 520 I I o x "1- I- _z 500 w .,,J I 1 w > 480 < 460 440 I x 0 I 420 I t )( IO I I I 4OO -15 -10 -5 0 II 0 x I I x ==> OFFI - 6m BBOP (N= I o ==> OFFI - lm BBOP (N=] 1 ] I I 5 10 15 20 25 IRRADIANCE [uW cm-2 nm-1] LU410 @ z=0 58O i 560 540 520 I-_1 o 2= I-- 500 z w -- I x II ,o w > 480 460 44( t x I I 42( I X ,.o8 0.05' 0'.1 - ,05 RADIANCE i i x ==> OFFI - 6m BBOP (N=7) o ==> OFFI - lm BBOP (N=8) I I o I I I 0 I 0.15' 0 2i 0.25' 0 3'. 0.35 [uW cm-2 nm-1 sr-1] Fig. 16. Mean differences (with 90% c.i.) of simultaneous determinations of the downwelling irradiance spectrum just beneath the sea surface, Ed(0-, A), shown in the top panel, and the upwelling radiance spectrum just beneath the sea surface, Lu (0-, A), shown in the bottom panel. The differences are between O20 casts and B1 (o; 1 m off the stern) or B6 (x; 6m off the stern) BBOP casts. 31

Original page 38

CaseStudiesforSeaWiFSCalibrationandValidation,Part3 NEAR-TIME COMPARISON OF Ked_441 -5 t i i X ! -10 J I -15 X -20 o. -25 LU i 0 -3O I _ -35 -4O I --i _ J x ==> ONE METER CAST (N=18) o ==> THREE METER CAST (N=50) • ==> SIX METER CAST (N=17) + ==> >20 METER CAST (N=23) t : _ o , I ! o.ols o.o2 0.025 0.03 0.035 0.04 0.042 0.05 0.055 Kd [rn-1] NEAR-TIME COMPARISON OF Rrs_441 -5 -10 -15 -2O I "1- I-- 0- -25 UJ a • C : -3O I I I -35 : 0 -40 , I -45 i i 0.008 0.01 0.012 , X , _ 0 : l I i c , I I x ==> ONE METER CAST (N=t2) o ==> THREE METER CAST (N=34) "==> SIX METER CAST (N=I 1) + ==> >20 METER CAST (N=19) i J I 0.014 0.016 0.018 0.02 REMOTELY SENSED REFLECTANCE [sr-1] Fig. 17. Mean (with 90% c.i.) determinations of the diffuse attenuation coefficient for downwelling irradiance, Kd(z, 441), shown in the top panel, and the remotely sensed reflectance, P_s(z, 441), shown in the bottom panel, as a function of depth for four different deployment distances (1, 3, 6, and > 20 m). 32

Original page 39

Mueller, Fraser, Biggar, Thome, Slater, Holmes, Barnes, Weir, Siegel, Menzies, Michaels, and Podesta NEAR-TIME COMPARISON OF EXTRAPOLATED Rrs at SEA SURFACE 58O g , 560 54_ 5201 :E k-- 500 Z LU ..J UJ > 480 460 440 420 t_t | { 4% ooo5 ool Rrs_441 [sr-1] ! l x ==> ONE METER CAST (N=14) o ==> THREE METER CAST (N=4; ° ==-> SIX METER CAST (N=14) + ==> >20 METER CAST (N=22) | . ,, n J ' i oo15 002 0 o25 Fig. 18. Spectral structure of the remote sensing reflectance just beneath the sea surface, Rrs(0-, )_), evaluated at four different deployment distances (1, 3, 6, and greater than 20 m). of geophysical and sampling noise. Voss et al. (1986) found long as the deployment distance off the stern is 3m or very little differences (less than 6%) in values of Kd(Z, A) greater. It is stressed that this analysis holds only for and Rrs(z, _) profiles taken at 0 and 9 m from the ship. The differences were 0-3% and 1-6%, respectively, in the upper 20 m. Below 20m, moreover, they found that the effects of the ship shadow have altogether disappeared, as is shown in this study. In conclusion, little variability in downwelling irradiance is observed for any of the deployment distances off the Weatherbird H. Significant variations are found, however, for upwelling radiance when the deployment distance is less than 3 m. These findings suggest that the ship's shadow exerts its greatest influence in the upper 20 m, with the largest perturbation at the sea surface. Measurements of optical properties from the Weatherbird H during clear skies with the stern pointed into the sun can be made as deployments made off the stern of the R/V Weatherbird H under clear sky conditions. The effects of variable sea state, sun glint, diffuse sky, and different hull shapes and sizes have not been evaluated, and additional experiments are necessary. ACKNOWLEDGMENTS This work was supported by the National Science Foundation (OCE 91-16372 and OCE 90-16990), NASA (NAGW-3145), and the SeaWiFS Project Office. Computational support from Digital Equipment Corporation's flagship research project, Sequoia 2000, is also gratefully acknowledged. 33

Original page 40

Case Studies for Sea'WiFS Calibration and Validation, Part 3 Chapter SeaWiFS Global Fields: 5 What's In a Day? GUILLERMO PODESTA Rosenstiel School ]or Marine and Atmospheric Sciences University of Miami Miami, Florida ABSTRACT This chapter defines the procedure to be employed to delineate data corresponding to one day of SeaWiFS operation. The definition is required for data analysis with minimal temporal aliasing in the same region of observation. The definition also allows proper assignment generation of weekly and monthly average products. of data into daily fields which will be used for the 5.1 INTRODUCTION the data collected between 00:00:00 Coordinated Universal Time (UTC) (or any other arbitrary start of the day) The basic products to be generated by the SeaWiFS and 23:59:59UTC. This definition is simple, intuitive, and Project are global daily fields of geophysical quantities, extremely easy to implement. Its negative aspects, howsuch as phytoplankton pigment concentration. The daily ever, will become apparent when one considers the orbital fields will be the basis of subsequent temporal compositing characteristics of the SeaStar spacecraft on which SeaWiFS into weekly and monthly products. One basic question, will be flown. however, is: what constitutes a day's worth of data? This question is the subject of this chapter. sented for the SeaStar spacecraft (Fig. 19). To simplify The need for a consistent definition of a data day is the visualization, only the descending tracks are displayed, only truly relevant to the production or analysis of global i.e., the spacecraft is flying from north to south. The Seadata fields. If one is dealing with a limited area (although Star descending tracks correspond to daytime data, which in this case, limited means anything less than global, and is the only data archived for SeaWiFS other than special can encompass entire ocean basins), one takes advantage calibration measurements. The nadir tracks were generof the fact that satellite sensors usually sample a region ated using the program SeaTrack, made available by the at approximately the same time, or times, every day. In SeaWiFS Project. A dummy set of orbital elements for the this way, data separated by approximately 24-hour periods SeaStar spacecraft [in North American Air Defense (NOcan be assigned to different data days. Analyses of the RAD) Command two-line format] was also obtained from resulting daily data fields will introduce a minimal amount the Project. of temporal aliasing, as the difference in sampling times will be on the order of a couple hours over an approximate begin the hypothetical 24-hour data day on 2 April 1994 repeat cycle of a few days. at 00:00:00 UTC, when the nadir track intersects the 180 ° In contrast, when daily global satellite data fields are meridian (marked _eg on Fig. 19). The descending orbit to be constructed, a consistent definition of a data day immediately after the beginning of the data day is labeled needs to be adopted. This definition should be easy to im- N. Subsequent descending tracks pass to the west, and are plement in practice and should minimize temporal aliasing offset by a distance of about 25 ° of longitude at the equaand discontinuities in the resulting products. In the foltor. The swaths viewed by SeaWiFS in consecutive orbits lowing sections, some of the various alternatives will be have an increasingly larger overlap with latitude. This explored. means that areas at high latitudes (greater than about 50 °) may be sampled twice or more during a data day. 5.2 TEMPORAL DEFINITION When an area is sampled in two consecutive descending To illustrate the problem, a plot of nadir tracks is pre- For comparison with subsequent cases, the choice is to The most obvious definition of a data day is a 24-hour orbits, measurements will be separated by about an hour period. For instance, a daily field would encompass all and a half. Unless one is concerned with features having 34

Original page 41

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 60"N 30"N O 30°S 60"S 0 ° 60°E 120°E 180"E 120°W 60"W Fig. 19. Descending SeaStar tracks for a 24-hour data day beginning on 2 April 1994 00:00:00UTC. The data day begins at the point labeled Beg. The day ends during an ascending orbit (not shown). The first orbit after the beginning of the data day is labeled N, and subsequent orbits are labeled N+I...N+14. very small scales, or with calculation of rates, it is prob- addition to a wide gap, large temporal discontinuities beably safe to assume that the ocean fields will not change tween data swaths from tracks N+14 and N+I. If there is significantly between consecutive passes; thus, temporal overlap between the two swaths, data collected far apart aliasing should be negligible. At the same time, at low in time may be averaged, once again introducing potential and intermediate latitudes (approximately between 50°S aliasing. Similar problems occur in the area south of track and 50 ° N), the swaths do not overlap, and there will be N (south of New Zealand), which is sampled by tracks N+14 gaps in the daily coverage (Hooker and Esaias 1993). and N+13 much later in the day. The SeaStar polar platform, which will carry SeaWiFS, The large gaps in coverage, as well as potential aliasing is planned to have an orbital period of approximately 99 and temporal discontinuity effects associated with the 24minutes. The actual period will depend on the spacecraft hour definition, are further complicated by the fact that altitude, and therefore, may vary with time as the altitude the locations where gaps occur change in time. Figure 20 shows the locations, along the SeaStar nadir tracks, of the of the satellite changes. Given an orbital period of about 99 minutes, the number of revolutions that the SeaStar boundaries between 24-hour data days for a 10-day period beginning on 2 April 1994. The dot labeled 1 corspacecraft will complete in a 24-hour period is approximately 14.55. The last descending orbit of the 24-hour responds to the beginning of the period on 2 April 1994 at 00:00:00UTC. The dot labeled 2 indicates the begindata day is labeled N+14. It is apparent from Fig. 19 that ning of the second 24-hour data day, and so forth. The the 24-hour day leaves a large gap in coverage north of the dot marked 11 corresponds to the end of the period on 12 beginning of the day, between orbits N+I and N+14. April 1994 at 00:00:00UTC. The shift in the location of A second problem inherent in the temporal definition the daily boundaries is a direct result of the difference beof a data day is the existence of areas on the global fields tween the 24-hour data day and the longer time it would with large temporal discontinuities in sampling times, even take the spacecraft to complete a number of revolutions though these areas may be spatially contiguous. For inwhich would ensure global coverage. stance, consider descending track N+14 in Fig. 19, the last track of the data day. To the north of that track, i.e., over 5.3 SPATIAL DEFINITION the Arctic Ocean north of Alaska, data are contributed by track N+I and, possibly N+2, although this orbit appears Because of the problems associated with a temporal to be too far north. These two tracks, however, were sam- data day definition, the implications of adopting a spatial pled near the beginning of the data day, more than 20 hours definition have been explored. In this case, the boundbefore track N+14. The daily fields will then contain, in ary between data days is not defined by time, but by a 35

Original page 42

Case Studies for SeaWiFS Calibration and Validation, Part 3 60°N 30=N O 30"S 7 60"S 0 ° 60"E 120°E 180"E 120°W 60°W Fig. 20. Locations of the boundaries of 24-hour data days for a 10-day period beginning on 2 April 1994 00:00:00 UTC, at the dot labeled 1. fixed geographic reference. A similar criterion is commonly used for designating orbit numbers in several spacecraft-the orbit number usually is incremented upon crossing the equator. For initial investigations, the 1800 meridian was selected as the boundary between data days. Figure 21 shows SeaStar nadir tracks for a spatiallydefined data day. Because the nadir tracks cross the reference line seven or eight times during a day, one of the crossings must be selected to be the beginning of a data day. An operational definition for the selection of the crossing, which initiates the data day, is presented in Fig. 21. For this discussion, the day is defined to begin on 2 April is applied on a pixel-by-pixel basis, that is, pixels along the same scan line on a given orbit can be assigned to different days, depending on whether they are on one side or the other of the 180 ° meridian. Figure 22 illustrates the pixel-by-pixel assignment of data to a given day. The figure shows a schematic description of the sampling pattern of the SeaWiFS instrument as it flies over the 180 ° meridian. Because there is not yet a scanner model for SeaWiFS, nadir tracks and scan lines are shown for the Advanced Very High Resolution Radiometer (AVHRR), which has slightly wider scans than SeaWiFS. The figure shows about 20 minutes of nadir track, i.e., 1994 at 00:00:00 UTC, when the spacecraft crosses the 180 ° ±10 minutes from the 180 ° meridian crossing. The scan meridian flying from north to south. Notice that this is the same time at which the 24-hour data day shown on Fig. 19 started, but it is entirely fortuitous that the 180 ° crossing took place at 00:00:00. The first descending track of the day is labeled N. In this case, the end of the data day is defined as the moment when the nadir track crosses the 180 ° meridian during revolution N+I5. This happens, for the example given, approximately on 3 April 1994 at 00:30:00UTC. The observation most readily apparent is that a spatial definition will result in a data day that does not necessarily correspond to a 24-hour day; in this case, the data day is approximately 24 hours and 30 minutes long. Note that Fig. 21 is approximate for two reasons. First, sometimes one less revolution is required to ensure almost complete global coverage, that is, the last orbit of the day would be N+I4. The data day would be about 23 hours and 22 minutes long in this case. Second, the spatial definition 36 lines shown on Fig. 22 are separated by one minute. Pixels along a given scan line that are located east of 180 ° are assigned to day K (K is an arbitrary designation for a given day). If pixels along the same scan line are west of 180 °, those pixels are assigned to the following day (K+I). It is appareat from Fig. 22 that even before the nadir track crosses the 180 ° meridian, pixels are already being assigned to day K+l. Conversely, after the nadir track has crossed the reference meridian, pixels east of the meridian are still being allocated to day K. It is this allocation mechanism that makes it difficult to precisely define the duration of a data day. 5,3.1 Beginning of the Data Day How is the spatial definition of a data day implemented in routine processing of global satellite data fields? The first step is to define a meridian, which will serve as the

Original page 43

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 60"N 30"N o 30"S 60°S 0" 60°E 120"E 180°E 120"W 60°W Fig. 21. SeaStar descending orbits for a spatially-defined data day beginning on 2 April 1994 00:00:00 UTC. At this time, the nadir track crosses the 180 ° meridian. The day ends when the nadir track crosses the 180 ° meridian (square labeled End) on 3 April 1994 00:30:00UTC. 60"N 30"N Day K+I 30"S Day K 120"E 180°E 120"W Fig. 22. AVHRR nadir track and scan lines for a 20 minute period centered about the 180 ° meridian crossing. Pixels to the east of the 180 ° meridian (marked in a thicker line) get assigned to a given data day K, whereas the pixels to the west of the meridian correspond to data day K+I. 37

Original page 44

CaseStudiesforSeaWiFSCalibrationandValidation,Part3 referenceforthespatialdefinition.The180° meridian used in the previous examples is a good choice, as it minimizes differences between actual dates and the dates assigned to the data days. As the spatial data days are not 24 hours long, a suitable naming convention will have to be established. A second step in defining a data day is to decide which of the descending crossings of the reference meridian will mark the beginning of each data day. As mentioned above, there are either seven or eight descending crossings of the reference meridian in a day. This pattern is illustrated in Fig. 23, which shows the latitude of descending crossings of the 180 ° meridian as a function of time for the SeaStar spacecraft, beginning on 2 April 1992 at 00:00:00UTC. A period of about 10 days duration, ending on 12 April 1994 at 00:37:00 UTC, is shown in the figure. Most of the crossings (shown as dots) take place at high latitudes, and one or two crossings per day occur at tropical-to-intermediate latitudes. For a given day, any of the crossings of the 180 ° meridian shown on Fig. 23 can be potentially selected as the one marking the beginning of a data day for descending orbits. For operational purposes, the following definition is proposed: A data day for descending orbits is defined to begin at the descending crossing of the 180 ° meridian that is closest to the equator. Crossings that satisfy this definition are shown as large squares in Fig. 23. Such a actually begins on 3 April 23:52 UTC and ends on 5 April 00:23 UTC. Table 19. The beginning times of 15 data days for descending orbits of the SeaStar spacecraft. The latitude of the 180 ° meridian crossing is also shown. Date Beginning Time Latitude of [April 1994] [UTC] 180 ° Crossing 2 00:00:00 -16.1 3 00:30:00 35.7 3 23:52:00 -22.9 5 00:23:00 27.4 5 23:46:00 -31.8 7 00:17:00 18.0 7 23:39:00 -39.5 9 00:11:00 7.8 9 23:32:00 -46.0 11 00:05:00 - 2.7 12 00:37:00 42.9 12 23:59:00 -13.0 13 23:52:00 -22.9 15 00:24:00 27.4 15 23:46:00 -31.7 5.3.2 Advantages of the Spatial Definition In the previous sections, a spatial definition was prodefinition will be the easiest to implement because there is posed for a data day, together with an objective definition always only one crossing in a day that fulfills the condition. Consecutive crossings may, however, in certain instances for the temporal beginning and end of such a data day. So far, however, the advantages or disadvantages of the have very similar absolute latitudes of intersection, one in proposed definitions have not been discussed. the Southern Hemisphere, and the other in the Northern Hemisphere. The alternating solid and dashed lines in Fig. 23 indicate consecutive data days. Initially, the latitude of data day initiation seems to follow a regular progression to the south, alternating between the Northern and Southern Hemispheres. Note, however, that the progression is interrupted near the end of the period illustrated. In this case, the next to last crossing would continue the progression, but the following crossing (the last square in the sequence) is actually closer to the equator. Following the proposed definition, the data day is extended until the next crossing, which is located in the Northern Hemisphere, i.e., the data day is slightly longer_ne more revolution in this case. The southward progression of the crossings subsequently resumes. Table 19 lists the start times of descending data days for a 15-day period beginning on 2 April 1994, as well as the latitude where the crossing of the 180 ° meridian occurs. It must be stressed that, because of the pixelby-pixel allocation described above, parts of the field will include data collected both before and after the times listed in Table 19. In addition, as stated above, an appropriate naming convention will have to be worked out for the data 4 time will once again be averaged. Elsewhere on the global days. For instance, the data day considered as April - 38 Problems associated with the temporal definition of the data day were: 1) The potential presence of gaps, 2) Aliasing and large temporal discontinuities, and 3) The changing locations of the 24-hour data day boundaries. The spatial definition avoids temporal changes in the location of boundaries, as the boundary is fixed, e.g., the 180 ° meridian. Furthermore, the spatial definition, to some extent, reduces gaps in the coverage. The presence of large temporal discontinuities among adjacent areas is still present, however. The large temporal discontinuities identified on Fig. 19, north of Alaska and south of New Zealand, are still present in Fig. 21. It is clear that the large temporal discontinuities occur in two places near the meridian that define the separation between data days. The first place is the area south of the first track of the data day and west of the reference line. The second area with discontinuities occurs north of the last track of the data day, east of the reference line. As a result of the large temporal discontinuities that occur between adjacent swaths when the swaths overlap at higher latitudes, data that were sampled far apart in

Original page 45

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta 9O -80/1 II II i I -go II II ' I I I I I I I I I 0 1 2 3 4 5 6 7 8 9 10 Days Fig. 23. Latitude of crossing of the 180 ° meridian for a period approximately 10 days long, beginning for SeaStar descending orbits. The data shown are on 2 April 1994 00:00:00UTC and ending on 12 April 1994 00:37:00 UTC. Crossings are indicated by small dots. Large squares indicate crossings that begin data days. The alternating solid and dashed lines indicate consecutive data days. fields, any given track is surrounded by tracks sampled one orbital period (about 99 minutes) earlier or later. The presence of temporal discontinuities, or the averaging of data collected at very different times, may not be too important for some applications, although users should certainly be made aware of the occurrence of these events. In other situations, however, such temporal discontinuities may cause significant problems. Examples of such applications may be the estimation of the translation speed of larger. certain features, or the computation of fluxes. In order to limit the large meridional temporal discontinuities near the data day boundary, the short track segments north and south of the first and last tracks of the data day could simply be eliminated (e.g., parts of N+I, N+2, N+3, N+13, and N+14). This approach is illustrated in Fig. 24, which shows descending tracks between 2 April 1994 00:00:00UTC and 3 April 1994 00:30:00UTC, i.e., the data day shown on Fig. 21. The map is now centered at 0 °, rather than at 180 °, as in Fig. 21. Note that the nadir tracks, for which segments were eliminated, seem to end a bit before or after the 180 ° line. This break occurs because positions were predicted at one-minute increments by the orbital model used. The elimination of segments may result in areas not being sampled, e.g., upper left and lower right corners of the map. These gaps might possibly be filled by the swath of the first and last tracks of the data day (tracks N and N+15 in the south and north, respectively). The size of the gaps is, however, a function of the latitude of the reference line crossing which defines the beginning of the data day. As shown in Fig. 23, this latitude changes with time, moving north and south approximately between 50°N and 50 ° S. When the crossing is farther north, the gap to the south of the first track will be larger. Conversely, when the crossing is further south, the gap north of the last track will get It is proposed that one additional swath be aclded at each end of the data day in order to replace the eliminated segments. Plots of nadir tracks for days in which the crossings are farthest nJth or south (not shown here) have shown that one additional swath is enough to fill each of the gaps, and a second swath would not make a significant contribution. The added swaths would be temporally continuous with the first and last tracks of each data day, thus eliminating the problems of temporal discontinuities. An operational scheme would involve the following steps: 1. The times corresponding to the beginning and end of a spatially-defined data day are found following the definition suggested above. These times will be referred to as the beginning and end of the data day. 2. Data east of the 180 ° meridian, collected up to 12 hours after the beginning of the data day, 39

Original page 46

CaseStudiesforSeaWiFSCalibrationandValidation,Part3 60°N 30"N @ 30"S 60°S 180"W 120°W 60°W 0 ° 60°E 120°E 180"E Fig. 24. SeaStar descending orbits for a spatially-defined data day beginning on 2 April 1994 00:00:00 UTC. Segments that introduce large north-south temporal discontinuities (see text) are excluded. will be excluded. Data west of the 180 ° meridian, sampled up to 12 hours before the end of the data day, will be similarly excluded. The net result of these actions is similar to the elimination of segments shown in Fig. 24. 3. To ensure full coverage, data collected up to 99 minutes before the beginning of the data day, and covering the area west of the 180 ° meridian, will be added to the beginning of the data day. This addition fills the gap to the south of the first track of the day. Data collected up to 99 minutes after the end of the data day, sampling the area east of 180 ° , are also added. These data fill the gap north of the last track of the data day. The end result is illustrated in Fig. 25. Figure 25 shows the descending orbits for the data day beginning approximately on 2 April 1994 00:00:00UTC. The gaps shown in Fig. 24 have been filled by the addition of two short segments, indicated by arrows and dotted lines, on Fig. 25. Note that these segments have been sampled before (N-l) and after (N+16)--the times estimated for the beginning and end of this data day (see Table 19). However, because the added segments are close in time to orbits N and N+15, the large temporal discontinuities have been eliminated. The segments excluded from this data day are the first portion of tracks N+I, N+2, and N+3 east of 180 °, and the last portion of tracks N+13 and N+14 west of 180 °. 4O 5.3.3 An Alternative Explanation To facilitate comprehension of the methodology, a simple analogy may be helpful. Envision a continuous strip chart on which the continents are drawn. Above the chart recorder there is a clock showing UTC time and date. As the chart moves from left to right, a pen draws descending tracks one at a time. The speed of the chart movement is appropriate to ensure that the nadir track's latitude and longitude, corresponding to any given UTC time, are correct, i.e., the nadir tracks should look similar to those on Figs. 24 and 25. Suppose the chart is positioned so that the pen is just crossing the 180 ° meridian, near the equator, on 2 April 1994. The clock time should be about 00:00:00 UTC. The chart recorder is then allowed to run for almost 24 hours, until a track crosses the 180 ° meridian again at about 36 ° N. The time should be about 00:30:00 UTC on 3 April 1994. If the cha:t is cut along the two 180 ° meridians drawn (left and r,ght), the tracks on the chart should look exactly like Fig. 24. As in Fig. 24, there will be some gaps in the coverage. On the right side of the chart, there is a gap south of the first track (N) of the day. This gap should have been filled by the last portion of tracks N+13 and N+14, which have been drawn to the left of the 180 ° meridian on the left side of the chart. These lines, however, were eliminated when the chart was cut along the left 180 ° line. Similarly, the gap north of the last track of the day should have been filled by the initial portions of tracks

Original page 47

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta N+16 60°N 30°N O 30"S 60°S I 180°W 120°W 60°W 0 ° 60'E 120"E 180°E Fig. 25. The data day beginning on 2 April 1994 00:00:00UTC, showing the addition of two segments (indicated by arrows and dotted lines) in order to minimize temporal discontinuities. The first track sampled after the estimated beginning time of the day (Beg) is track N. The added segment south of this track corresponds to the previous orbit (N-l). The last track before the estimated end time of the data day (End) is track N+15. The added segment to the north corresponds to the next orbit (N+16). N+I, N+2 and N+3. These segments were drawn east of the 180 ° meridian on the right side of the plot. As the 180 ° line ians. By doing this, the spatial pixel-by-pixel assignment of data is applied to a given data day. The end result was cut along on the right, however, these segments were should look exactly like Fig. 25. Finally, envision running excluded. It is apparent that the chart recorder analogy reproduces the action of eliminating tracks which cause the recorder for long periods and repeatedly cutting the long chart along the 180 ° meridians. Each of the maps large temporal discontinuities, as the end result looks ex- would correspond to one data day. actly like Fig. 24. The gaps can be filled in the global fields When discussing an elimination of orbital segments that using the same chart recorder analogy. would result in large temporal discontinuities, the presen- Now envision the case in which the chart recorder does tation could have given the impression that data in these not start at 00:00:00 UTC on 2 April 1994, but rather, the segments would be unused, and therefore wasted. If the chart is moved backwards and is started about 99 minanalogy presented above is followed, however, it is easy to utes earlier. If the recorder starts then, an additional see that the data will not be deleted, but rather the data track (N-l) will he drawn before the nadir track of orbit N will be assigned to the previous, or the following, data crosses the 180 ° meridian at 00:00:00UTC, which defines days. For instance, the northern portions of tracks N+I, the temporal beginning of the data day. The southern por- N÷2. and N+3 (not labeled) in Fig. 24 would be plotted to tion of track N-1 will fall west of the 180 ° meridian, filling the east of the right 180 ° meridian on the chart. When the gap previously existing in the south. Then the recorder the chart is cut, these portions get assigned to the previis allowed to run up to 99 minutes past the time originally ous data day, which begins on 1 April 1994 UTC. In the defined as the end of the day (3 April 1994, 00:30:00 UTC), same way, the southernmost portions of tracks N+13 and and again, an additional track will be drawn. If the last N+14 are plotted to the west of the left 180 ° meridian; track of the day is N+15, the northern portion of track thus, being assigned to the next data day after the chart N+16 will fill the northern gap. Once the recorder has is cut along the meridian. The end result of the scheme been allowed to run for the estimated duration of the data proposed is a daily global field where all parts of a field day, plus the additional 99 minutes on either end, a pair are temporally separated from adjacent areas by, at most, of scissors is used to cut the chart along both 180 ° merid- one orbital period. 41

Original page 48

CaseStudiesforSeaWiFSCalibrationandValidation,Part3 5.4 OTHER ISSUES An aspect that has not been discussed so far is that at both the extreme north and extreme south of the fields, data from several tracks might be averaged within a data day. At high latitudes, the spacecraft is flying in nearly an east-west direction and, thus, the scan lines have a northsouth orientation. For instance, there are five or six passes a day at high latitudes (Fig. 23) near the 180 ° meridian. Some of these passes are excluded at high latitudes, as described above. In other high latitude regions, however, the fields will contain the average of several passes. This overlap should not have too many consequences on SeaWiFS in continuous across the 180 ° meridian. These daily scenes products, as the areas affected will be mostly on land the Southern Hemisphere and under permanent ice cover in the Northern Hemisphere. Furthermore, in these regions the sensor may encounter limitations in available sunlight, which may preclude sampling. One final issue requiring discussion is that the spatial scheme proposed above will result in temporal discontinuities in areas that straddle the reference line. Suppose that a study is made of an area of the North Pacific Ocean, encompassed between 150 ° W and 150 ° E, and straddling the 180 ° line. If this study obtains a global field for a given data day, it must be realized that the portion of the study area west of 180 ° has been sampled much earlier than the 42 portion to the east. Again, this may not be relevant for some research, but it could be in some cases. A solution would be to place the reference line elsewhere, e.g., along 0 °, but there will always be some location where areas on either side of the line will be sampled far apart in time. Alternatively, a user might obtain product fields for two consecutive data days and paste the appropriate portions. In the Pacific example presented above, the eastern part of the study area would be extracted from data day K and the western part from day K+I. To study the daily data of a region that includes 180 ° longitude, two consecutive daily products should be joined at the seam. This procedure will produce data that are can then be averaged over time using a time binning algorithm to construct weekly or longer period composites of an area straddling the 180 ° meridian. This method would produce the most accurate long-term composites. One could, however, use the standard global products by joining areas east and west of 180 ° from the same weekly, monthly, or annual product. The latter method will result in a slight temporal discontinuity across the 180 ° meridian, since the earliest data contributing to the composite west of 180 ° is not matched by continuous data east of 180 ° . Similarly, data east of 180 ° is not matched by data west of 180 ° on the final day of the composite.

Original page 49

Mueller, Fraser, Biggar, Thome, Slater, Holmes, Barnes, Weir, Siegel, Menzies, Michaels, and Podesta GLOSSARY A-band Absorption Band AOP Apparent Optical Properties AVHRR Advanced Very High Resolution Radiometer BATS Bermuda Atlantic Time-Series Study BBOP Bermuda Bio-Optics Project BRDF Bidirectional Reflectance Distribution Function BSI Biospherical Instruments, Inc. CHORS Center for Hydro-Optics and Remote Sensing c.i. confidence interval CVT Calibration and Validation Team CZCS Coastal Zone Color Scanner DC Digital Count EOSDIS Earth Observing System Data Information System FEL Not an acronym; designates a type of irradiance lamp. GSFC Goddard Space Flight Center IFOV Instantaneous Field-of-View JGOFS Joint Global Ocean Flux Study MER Marine Environmental Radiometer NASA National Aeronautics and Space Administration NIST National Institute of Standards and Technology NORAD North American Air Defense (Command) OFFI Optical Free-Falling Instrument PST Pacific Standard Time R/V Research Vessel SBRC Santa Barbara Research Center SDSU San Diego State University k Molecular absorption cross-section area. Vertical attenuation coefficient for downwelling it- Kd(z, A) radiance. KL(Z, A) Vertical attenuation coefficient for upwelled radiance. Ku(z, h) Vertical attenuation coefficient for upwelled irradiance. L Radiance of light transmitted through absorbing oxygen. Lt Radiance measured at a satellite, i.e., orbiting sensor. L0 Model radiance without absorbing oxygen. L_tm Radiance of light reflected from the atmosphere. Lsfc Radiance of light leaving an ocean surface and passing through the atmosphere. L_ (z, A) Upwelled spectral radiance. Air mass. m N Total number of oxygen molecules per unit area in a vertical column of the atnmsphere. OFFI casts 20 m from the ship's stern. O20 P Surface pressure. Pdev Pressure deviation between the minimum and maximum surface pressures compared to 1,013 mb. Reference pressure. Pref R_ The square of the linear correlation coefficient. R_ (z, X) Remote sensing reflectance. S(A) Solar spectral irradiance. Residual standard deviation. sxu T(A) Two-way transmission through oxygen in the model •layer. SeaWiFS Sea-viewing Wide Field-of-view Sensor T(A, 9, 0) SI International System of Units (Systdme International d' Unitds) SIS Spherical Integrating Source S/N Serial Number SNR Signal-to-Noise Ratio SPIE Society of Photo-Optical Instrumentation Engineers TOA Top of the Atmosphere UCSB University of California at Santa Barbara UTC Coordinated Universal Time SYMBOLS a Regression coefficient. b Regression coefficient. B Band 7 width. B1 BBOP casts 1 m from the ship's stern. Bs BBOP casts 6 m from the ship's stern. ¢() Spectral beam attenuation coefficient. Ed Incident downwelling irradiance. Ed(z, ;q Downwelled spectral irradiance. E_ Incident upwelling irradiance. E_(z, ,_) Upwelled spectral irradiance. f(A) Instrument spectral response function. Fi Immersion coefficient. Two-way transmission through oxygen in the model layer in terms of zenith angle (0), and solar angle (00). W Equivalent bandwidth. AL The difference between L and Lo. Ap The difference in atmospheric pressure. 0 Zenith angle of the line-of-sight in a plane-parallel atmosphere. Oo Solar zenith angle. X Wavelength. P Reflectance. (7 Standard deviation. ox() Optical thickness due to oxygen absorption. Actual deployment distance. _d Calculated deployment distance for downweUing irradiance measurements. Calculated deployment distance for upweUing irradiance measurements. _L Calculated deployment distance for upwelling radiance measurements. X Proportionality constant. 43

Original page 50

CaseStudiesfor SeaWiFSCalibrationandValidation,Part3 REFERENCES Barnes, R.A., 1994: SeaWiFS Data: Actual and Simulated. [World Wide Web page.] From URLs: http://seawifs •gsfc. nasa. gov/SEAWIFS/IMAGES/spectral, dat and , J.C. Comiso, R.S. Fraser, J.K. Firestone, B.D. Schieber, E. Yeh, K.R. Arrigo, and C.W. Sullivan, 1994: Case Studies for SeaWiFS Calibration and Validation, Part 1. NASA Tech. Memo. 104566, VoL 13, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Green- Center, belt, Maryland, 52 pp., plus color plates. /spectra2.dat NASA Goddard Space Flight Greenbelt, Maryland. --, A.W. Holmes, W.L. Barnes, W.E. Esaias, C.R. McClain, and T. Svitek, 1994: SeaWiFS Prelaunch Radiometric Calibration and Spectral Characterization. NASA Tech. Memo. 104566, Vol. 23, S.B. Hooker, E.R. Firestone, and J.G. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 55 pp. Biggar, S.F., D.I. Gellman, and P.N. Slater, 1990: Improved optical depth components from Langley plot data. Remote Sens. Environ., 32, 91-101. , P.N. Slater, K.J. Thome, A.W. Holmes, and R.A. Barnes, 1993: Preflight solar-based calibration of SeaWiFS. SPIE, 1,939, 233-242. Cantor, A.J., and A.E. Cole, 1985: Handbook of Geophysics and the Space Environment, A.S. Jursa, Ed., Air Force Geophysics Laboratory, Air Force Systems Command, USAF, 15-48. Clark, D.K., 1981: Phytoplankton pigment algorithms for the Nimbus-7 CZCS. In: Oceanography from Space, J.F.R. Gower, Ed., Plenum Press, New York, 227-237. Dickey, T.D., and D.A. Siegel, 1993: Bio-Optics in U.S. JGOFS. U.S. JGOFS Planning Report Number 18, U.S. JGOFS Planning and Coordination Office, Woods Hole, Massachusetts, 180 pp. Ding, K., and H.R. Gordon, 1994: Analysis of the influence Mueller, J.L., 1994: Preliminary Comparison of Irradiance Immersion Coefficients for Several Marine Environmental Radiometers (MERs). CHORS Tech. Memo. 004-94, Center for Hydro-Optics and Remote Sensing, San Diego State University, San Diego, California, 4 pp. --, and R.W. Austin, 1992: Ocean Optics Protocols for Sea- WiFS Validation. NASA Tech. Memo. 104566, Vol. 5, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 45 pp. --, and R.W. Austin, 1995: Ocean Optics Protocols for Sea- WiFS Validation, Revision 1. NASA Tech. Memo. 104566, Vol. 25, S.B. Hooker, E.R. Firestone, and J.G. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 69 pp. Neckel, H., and D. Labs, 1984: The solar radiation between 3,300 and 12,500/. Solar Phys., 90, 205-258. Poole, H.H., 1936: The photo-electric measurement of submarine illumination in offshore waters. Rapp. Proc- Verb. Consell Expl. Mar., 101, 9. Siegel, D.A., A.F. Miehaels, J. Sorensen, M. Hammer, and M.C. O'Brien, 1994: Seasonal variability of light availability and its utilization in the Sargasso Sea. J. Geophys. Res., (submitted). of Ou "A" band absorption on atmospheric correction of Sorensen, J.C., M. O'Brien, D. Konoff, and D.A. Siegel, 1994: ocean color imagery. AppL Opt., (submitted). Gordon, H.R., 1985: Ship perturbation of irradiance measurements at sea. h Monte Carlo simulations. Appl. Opt., 24, 4,172-4,182. , and D.K. Clark, 1980: Atmospheric effects in the remote sensing of phytoplankton pigments. Bound.-Layer Meterol., 18, 299-313. Heltiwell, W.S., G.N. Sullivan, B. Macdonald, and K.J. Voss, 1990: Ship shadowing: model and data comparisons. Ocean Optics X, R.W. Spinrad, Ed., SPIE, 1,302, 55 71. Hooker, S.B., W.E. Esaias, G.C. Feldman, W.W. Gregg, and C.R. McClain, 1992: An Overview of SeaWiFS and Ocean Color. NASA Tech. Memo. 104566, Vol. 1, NASA Goddard Space Flight Center, Greenbelt, Maryland, 24 pp., plus color plates. --, and --, 1993: An overview of the SeaWiFS Project. EOS, Trans. AGU, 74, 241. McClain, C.R., W.E. Esaias, W. Barnes, B. Guenther, D. Endres, S. Hooker, G. Mitchell, and R. Barnes, 1992: Calibration and Validation Plan for SeaWiFS. NASA Tech. Memo. 104566, Vol. 3, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 41 pp. 44 The BBOP data processing system. Ocean Optics XII, J.S. Jaffe, Ed., SPIE, 2,258, 539-546. Strickland, J.D.H., 1958: Solar radiation penetrating the ocean: A review of requirements, data, and methods of measurement, with particular reference to photosynthetic productivity. J. Fish. Res. Bd. Canada, 15, 453-493. Voss, K.J., J.W. Nolten, and G.D. Edwards, 1986: Ship shadow effects on apparent optical properties. Ocean Optics VIII, P.N. Slater, Ed., SPIE, 637, 186-190. United States Navy, 1978: Marine Climatic Atlas of the World; South Atlantic Ocean, Vol. 4, NAVAIR 50-1C-531, US Government Printing Office, Washington, DC, 325 pp. Waters, K.J., R.C. Smith, and M.R. Lewis, 1990: Avoiding ship induced light-field perturbation in the determination of oceanic optical properties. Oceanogr., 3, 18-21. Weir, C.T., D.A. Siegel, D.W. Menzies, and A.F. Michaels, 1994: In situ evaluation of a ship's shadow. Ocean Optics XII, J.S. Jaffe, Ed., SPIE, 2,258, 815-821. Wu, M.C., 1985: Remote sensing of cloud-top pressure using reflected solar radiation in the oxygen A-band. J. Climate Appl. Meteorol., 24, 540-546.

Original page 51

Mueller,Fraser,Biggar,Thome,Slater,Holmes,Barnes,Weir,Siegel,Menzies,Michaels,andPodesta THE SEAWIFS TECHNICAL REPORT SERIES VoI. 10 Woodward, R.H., R.A. Barnes, C.R. McClain, W.E. Esaias, Vol. 1 Hooker, S.B., W.E. Esaias, G.C. Feldman, W.W. Gregg, and C.R. McClain, 1992: An Overview of SeaWiFS and Ocean Color. NASA Tech. Memo. 104566, Vol. I, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 24pp., plus color plates. Vol. 11 W.L. Barnes, and A.T. Mecherikunnel, 1993: Modeling of the SeaWiFS Solar and Lunar Observations. NASA Tech. Memo. 104566, Vol. 10, S.B. Hooker and E.R. Firestone, Eels., NASA Goddard Space Flight Center, Greenbelt, Maryland, 26 pp. Vol. 2 Patt, F.S., C.M. Hoisington, W.W. Gregg, and P.L. Coronado, Gregg, W.W., 1992: Analysis of Orbit Selection for SeaWiFS: Ascending vs. Descending Node. NASA Tech. Memo. 104566, Vol. _, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 16 pp. Vol. 12 Vol. 3 Firestone, E.R., and S.B. Hooker, 1993: SeaWiFS Technical Re- McClain, C.R., W.E. Esaias, W. Barnes, B. Guenther, D. Endres, S. Hooker, G. Mitchell, and R. Barnes, 1992: Calibration and Validation Plan for SeaWiFS. NASA Tech. Memo. 104566, Vol. 3, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Vol. 13 Maryland, 41 pp. McClain, C.R., K.R. Arrigo, J. Comiso, R. Fraser, M. Darzi, Vol.__..A McClain, C.R., E. Yeh, and G. Fu, 1992: An Analysis of GAC Sampling Algorithms: A Case Study. NASA Tech. Memo. 104566, Vol. 4, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 22 pp., plus color plates. Vol. 14 Mueller, J.L., 1993: The First SeaWiFS Intercalibration Round- Mueller, J.L., and R.W. Austin, 1992: Ocean Optics Protocols for SeaWiFS Validation. NASA Tech. Memo. 104566, Vol. 5, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 43 pp. Vol. 15 Vol. 6 Gregg, W.W., F.S. Patt, and R.H. Woodward, 1994: The Sim- Firestone, E.R., and S.B. Hooker, 1992: SeaWiFS Technical Report Series Summary Index: Volumes 1-5. NASA Tech. Memo. I04566, Vol. 6, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 9 pp. Vol. 16 1993: Analysis of Selected Orbit Propagation Models for the SeaWiFS Mission. NASA Tech. Memo. 104566, Vol. 11, S.B. Hooker, E.R. Firestone, and A.W. Indest, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 16 pp. port Series Summary Index: Volumes 1-11. NASA Tech. Memo. 104566, Vol. 12, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 28 pp. J.K. Firestone, B. Schieber, E-n. Yeh, and C.W. Sullivan, 1994: Case Studies for SeaWiFS Calibration and Validation, Part 1. NASA Tech. Memo. 104566, Vol. 13, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 52pp., plus color plates. Robin Experiment, SIRREX-1, July 1992. NASA Tech. Memo. 104566, Vol. 14, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 60pp. ulated SeaWiFS Data Set, Version 2. NASA Tech. Memo. 104566, Vol. 15, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 42 pp., plus color plates. Vol. 7 Mueller, J.L., B.C. Johnson, C.L. Cromer, J.W. Cooper, J.T. Darzi, M., 1992: Cloud Screening for Polar Orbiting Visible and IR Satellite Sensors. NASA Tech. Memo. 104566, Vol. 7, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 7pp. vol. s McLean, S.B. Hooker, and T.L. Westphal, 1994: The Second SeaWiFS Intercalibration Round-Robin Experiment, SIRREX-2, June 1993. NASA Tech. Memo. 104566, Vol. 16, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 121 pp. Hooker, S.B., W.E. Esaias, and L.A. Rexrode, 1993: Proceed- Abbott, M.R., O.B. Brown, H.R. Gordon, K.L. Carder, R.E. ings of the First SeaWiFS Science Team Meeting. NASA Tech. Memo. 104566, Vol. 8, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 61 pp. Vo/.9 Vol. 18 Gregg, W.W., F.C. Chen, A.L. Mezaache, J.D. Chen, J.A. Firestone, E.R., and S.B. Hooker, 1994: SeaWiFS Technical Re- Whiting, 1993: The Simulated SeaWiFS Data Set, Version 1. NASA Tech. Memo. 104566, Vol. 9, S.B. Hooker, E.R. Firestone, and A.W. Indest, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 17pp. Evans, F.E. Muller-Karger, and W.E. Esaias, 1994: Ocean Color in the 21st Century: A Strategy for a 20-Year Time Series. NASA Tech. Memo. 104566, Vol. 17, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 20pp. port Series Summary Index: Volumes 1-17. NASA Tech. Memo. 104566, Vol. 18, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 47pp. 45

Original page 52

CaseStudiesforSeaWiFSCalibrationandValidation,Part3 Vol. 19 McClain, C.R., R.S. Fraser, J.T. McLean, M. Darzi, J.K. Fire- Vol. 24 Firestone, E.R., and S.B. Hooker, 1995: SeaWiFS Technical Restone, F.S. Patt, B.D. Schieber, R.H. Woodward, E-n. Yeh, port Series Summary Index: Volumes 1-23. NASA Tech. S. Mattoo, S.F. Biggar, P.N. Slater, K.J. Thome, A.W. Holmes, R.A. Barnes, and K.J. Voss, 1994: Case Studies for SeaWiFS Calibration and Validation, Part 2. NASA Tech. Memo. 104566, Vol. 19, S.B. Hooker, E.R. Firestone, and J.G. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 73 pp. Vol. 20 Hooker, S.B., C.R. McClain, J.K. Firestone, T.L. Westphal, E-n. Yeh, and Y. Ge, 1994: The SeaWiFS Bio-Optical Archive and Storage System (SeaBASS), Part 1. NASA Tech. Memo. 104566, Vol. 20, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 40 pp. Vol. 21 Acker, J.G., 1994: The Heritage of SeaWiFS: A Retrospective on the CZCS NIMBUS Experiment Team (NET) Program. NASA Tech. Memo. 104566, Vol. 21, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 43 pp. Vol. 22 Barnes, R.A., W.L. Barnes, W.E. Esaias, and C.R. McClain, 1994: Prelaunch Acceptance Report for the SeaWiFS Radiometer. NASA Tech. Memo. 104566, Vol. 22, S.B. Hooker, E.R. Firestone, and J.G. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 32 pp. VoI. 23 Barnes, R.A., A.W. Holmes, W.L. Barnes, W.E. Esaias, C.R. McClain, and T. Svitek, 1994: SeaWiFS Prelaunch Radiometric Calibration and Spectral Characterization. NASA Tech. Memo. 104566 , Vol. 23, S.B. Hooker, E.R. Firestone, and J.G. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 55 pp. 46 Memo. 104566, Vol. 24, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, (in press). Vol. 25 Mueller, J.L., and R.W. Austin, 1995: Ocean Optics Protocols for SeaWiFS Validation, Revision 1. NASA Tech. Memo. I04566, Vol. 25, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 67 pp. Vol. 26 Siegel, D.A., M.C. O'Brien, J.C. Sorensen, D.A. Konnoff, E.A. Brody, J.L. Mueller, C.O. Davis, W.J. Rhea, and S.B. Hooker, 1995: Results of the SeaWiFS Data Analysis Round-Robin (DARR), July 1994. NASA Tech. Memo. 104566, Vol. 26, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 58 pp. Vol. 27 J.L. Mueller, R.S. Fraser, S.F. Biggar, K.J. Thome, P.N. Slater, A.W. Holmes, R.A. Barnes, C.T. Weir, D.A. Siegel, D.W. Menzies, A.F. Michaels, and G. Podesta 1995: Case Studies for SeaWiFS Calibration and Validation, Part 3. NASA Tech. Memo. 104566, Vol. 27, S.B. Hooker, E.R. Firestone, and J.G. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 46pp.

Original page 53

REPORT DOCUMENTATION Form Approved PAGE OMBNo.0704-0188 ii Public reporting burden for this collection of information is estimated to average 1 hour per response, Including the time for revlewlng instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Repods, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Mana_lement and Budget, Paperwork Reduction Project 10704-0188), Washin_lton, DC 20503. 1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE May 1995 4. TITLE AND SUBTITLE SeaWiFS Technical Report Series 3. REPORT TYPE AND DATES COVERED Technical Memorandum 5. FUNDING NUMBERS Volume 27-Case Studies for SeaWiFS Calibration and Validation, Part 3 Code 970.2 6. AUTHOR(S) James L. Mueller, Robert S. Fraser, Stuart F. Biggar, Kurtis J. Thome, Philip N. Slater, Alan W. Holmes, Robert A. Barnes, Christian T. Weir, David A. Siegel, David W. Menzies, Anthony F. Michaels, and Guillermo Podesta Series Editors: Stanford B. Hooker and Elaine R. Firestone Technical Editor: .lame._ G. Acker 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Laboratory for Hydrospheric Processes Goddard Space Flight Center Greenbelt, Maryland 20771 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) National Aeronautics and Space Administration Washington, D.C. 20546-0001 8, PERFORMING ORGANIZATION REPORT NUMBER 95B00084 10. SPONSORING/MONITORING AGENCY REPORT NUMBER TM-104566, Vol. 27 11. SUPPLEMENTARY NOTES James L. Mueller: San Diego State University, San Diego, California; Stuart F. Biggar, Kurtis J. Thome, and Philip N. Slater: University of Arizona, Tucson, Arizona; Alan W. Holmes: Santa Barbara Research Center, Santa Barbara, California; Robert A. Barnes: ManTech Environmental Technology, Inc., Wallops Island, Virginia; Christian T. Weir, David A. Siegel, and David W. Menzies: University of California at Santa Barbara, Santa Barbara, California; Anthony F. Michaels: Bermuda Biological Station for Research, Ferry Reach, Bermuda; Guillermo Podesta: University nf Miami Miami Flnrida: F.IAine R Fire_tnne! eneral ,qci_nce Cnrnc_ratinn 12a. DISTRIBUTION/AVAILABILITYSTATEMENT Unclassified-Unlimited Subject Category 48 I_autz_.l. MAry. land: and Jzme_ _ Acker: Hu_h_ .qTX I anham M_rvland 12b. Di,IFIUqiON CODE Report is available from the Center for AeroSpace Information (CASI), 800 Elkridge Landing Road, Linthicum Heights, MD 21090; (301) 621-0390. 13. ABSTRACT (MaximLcn 200 words) This document provides brief reports, or case studies, on a number of investigations sponsored by the Calibration and Validation Team (CVT) within the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Project. Chapter 1 describes a comparison of the irradiance immersion coefficients determined for several different marine environmental radiometers (MERs). Chapter 2 presents an analysis of how light absorption by atmospheric oxygen will influence the radiance measurements in band 7 of the SeaWiFS instrument. Chapter 3 gives the results of the second ground-based solar calibration of the instrument, which was undertaken after the sensor was modified to reduce the effects of internal stray light. (The first ground-based solar calibration of SeaWiFS is described in Volume i9 in the SeaWiFS Technical Report Series.) Chapter 4 evaluates the effects of ship shadow on subsurface irradiance Weatherbird H in the Atlantic Ocean near Bermuda. Chapter and radiance measurements deployed from the deck of the R/V 5 illustrates the various ways in which a single data day of SeaWiFS observations can be defined, and why the spatial definition is superior to the temporal definition for operational usage. 14. SUBJECT TERMS 15. NUMBER OF PAGES 46 SeaWiFS, Oceanography, Immersion Coefficients, MER, Band 7 Radiance, Solar Radiation, Ship Shadow, Data Day, Global Fields, Case Studies 16. PRICE CODE 18. SECURITY CLASSIRCATION 19. SECURITY CLASSlRCATION 17. SECURITY CLASSIRCATION OF REPORT OF THIS PAGE Unclassified Unclassified NSN 754001-280°5500 OF ABSTRACT Unclassified 20, UMITATIONUnlimited OF AB I HACT Standard Form 298 (Rev. 2-89)

Original page 54