Full report

Original page 1

NASA Technical Memorandum 104566, Vol. 23 SeaWiFS Technical Report Series Stanford B. Hooker, Editor NASA Goddard Space Flight Center Greenbelt, Maryland Elaine R. Firestone, Technical Editor General Sciences Corporation Laurel, Maryland James G. Acker, Technical Editor Hughes STX Lanham, Maryland Volume 23, SeaWiFS Prelaunch Radiometric Calibration and Spectral Robert A. Barnes Man Tech, Inc. Wallops Island, Virginia Alan W. Holmes Hughes Santa Barbara Research Center Santa Barbara, California National Aeronautics and Space Administration Goddard Space Flight Center Greenbelt, Maryland 20771 1994 Characterization William L. Barnes Wayne E. Esaias Charles R. McClain Goddard Space Flight Center Greenbelt, Maryland Tomas Svitek Orbital Sciences Corporation Dulles, Virginia

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.

Original page 3

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek ABSTRACT Based on the operating characteristics of the Sea-viewing Wide Field-of-view Sensor (SeaWiFS), calibration equations have been developed that allow conversion of the counts from the radiometer into Earth-exiting radiances. These radiances are the geophysical properties the instrument has been designed to measure. SeaWiFS uses bilinear gains to allow high sensitivity measurements of ocean-leaving radiances and low sensitivity measurements of radiances from clouds, which are much brighter than the ocean. The calculation of these bilinear gains is central to the calibration equations. Several other factors within these equations are also included. Among these are the spectral responses of the eight SeaWiFS bands. A band's spectral response includes the ability of the band to isolate a portion of the electromagnetic of that region. The latter is termed out-of-band response. the instrument are produced by radiance in the out-of-band spectrum and the amount of light that lies outside In the calibration procedure, some of the counts from region. The number of those counts for each band is a function of the spectral shape of the source. For the SeaWiFS calibration equations, the out-of-band responses are converted from those for the laboratory source into those for a source with the spectral shape of solar flux. The solar flux, unlike the laboratory calibration, approximates the spectral shape of the Earth-exiting radiance from the oceans. This conversion modifies the results from the laboratory radiometric calibration by 1-4%, depending on the band. These and other factors in the SeaWiFS calibration equations are presented here, both for users of the SeaWiFS data set and for researchers making ground-based radiance measurements in support of SeaWiFS. 1. INTRODUCTION not a specific instrument. NASA has entered a contrac- In addition to its role as an ocean color experiment, the tual agreement with Orbital Sciences Corporation (OSC) Sea-viewing Wide Field-of-view Sensor (SeaWiFS) serves to obtain, at a fixed price, an ocean color data set. Seaas a satellite procurement experiment for the National WiFS is a data buy. OSC has entered into an agreement Aeronautics and Space Administration (NASA). In a stanwith the Hughes/Santa Barbara Research Center (SBRC) dard procurement, NASA provides the instrument builder to build, as a subcontractor, the satellite sensor required with a detailed specification for the design of the sensor. to provide this data set. In this arrangement, SBRC has In this procedure, NASA also maintains detailed control the freedom to design an instrument to meet the predeterover the construction of that instrument. The builder promined set of performance specifications. The actual design vides NASA with the satellite sensor on a cost-plus basis. of the instrument has been left to SBRC. NASA eventually obtains the specified instrument, but at a price. The price includes a substantial supervisory over- For the SeaWiFS Project, NASA is procuring data, Based on SBRC's design of the SeaWiFS radiometer, head by the agency, a substantial documentation overhead it has been possible to develop a set of radiometric calion the builder showing compliance with the detailed NASA bration equations for the sensor. These equations, and the specification, and often a substantial increase in the cost philosophy behind them, are presented below. The equaof the instrument as the specifications from NASA change tions are presented in a form that allow for their update during the construction of the instrument. This process as relative changes in the instrument's radiometric sensiis the price that must be paid for the construction of new tivity are detected during on-orbit operation. In addition, satellite sensors which expand the definition of state of the the set of prelaunch radiometric calibration coefficients is art. presented. SeaWiFS does not break radically new ground in sen- This report also presents the spectral response of the sor design. The necessary improvements in the SeaWiFS eight SeaWiFS bands, using laboratory measurements from measurements, over those of its predecessor--the Coastal SBRC. On orbit, there is no method for determining spec- Zone Color Scanner (CZCS)--are straightforward techni- tral shifts in the instrument's response. The primary labocally (Barnes and Holmes 1993 and Hooker et al. 1993). ratory method of determining the spectral response of the From NASA's point of view, SeaWiFS is considered an SeaWiFS bands has been accomplished through measureimproved replacement instrument for its predecessor. Such ments of individual parts in the instrument's optical train. an instrument does not require the financial and supervi- In addition, system level response measurements have been sory overheads of a standard NASA procurement. These made using a monochromatic light source. The system overhead items have been replaced with a well defined and level measurements are used as a confirmation to the piece well scrubbed set of performance specifications. part measurements, since the system level apparatus does

Original page 4

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization respon-variations. The scrambler eliminates the need for individnothavethesensitivityto measuretheout-of-band small. ual compensators to remove residual polarization at each sesat wavelengthswherethe responseis extremely Thepiecepart resultsarepresentedhere. 2. INSTRUMENT DESCRIPTION SeaStar is being built by OSC and is currently scheduled to be placed into orbit by an OSC stretch-Pegasus booster in 1995. The SeaWiFS radiometer is shown mounted on the SeaStar satellite in Fig. 1. The instrument located on the top spacecraft shelf between the three antennas. SeaWiFS is the only research instrument that will be carried by SeaStar. SeaWiFS consists of a scanner and an electronics module. The scanner, which contains optics, detectors, preamplifiers, and scan mechanisms, elec- to improve signal-to-noise ratios (SNRs). The signal from located on the nadir face of the instrument shelf. The com- each detector is amplified, processed through a selectable tronics module, which contains signal conditioning, is lo- gain stage, and digitized with a 12 bit analog-to-digital mand and telemetry, and power supply components, cated directly opposite the scanner on the inside surface focal plane assembly. The scrambler consists of two optical wedges that act as a variable wave plate to convert incident polarized light into several cycles of circular, horizontal, and vertical polarized light across the instrument's aperture. The sensitivity of the output of SeaWiFS to polarized light is measured in the laboratory, using a source producing plane polarized light. The rotation of the pois larized light through 360 ° produced changes of less than 0.5%in the eight SeaWiFS bands. Two instrument bands, which have four detectors each, form a focal plane. The four detectors in each group are the added using a time delay and integration (TDI) technique is of converter (ADC). The four digital words from a band are instrument then: delayed, summed to obtain the TDI advantage, trunthe instrument shelf. The total weight of the is about 110 pounds. radi- bus through the electronics module. A solar calibrator is SeaWiFS is designed to measure Earth-exiting ances. The sensor's instantaneous field-of-view (IFOV) cated to 10 bits, and transmitted to the SeaStar satellite is mounted on the instrument so that, if desired, the optical 1.6 by 1.6 mrad per pixel, with one scan covering +58.3 ° system views a solar illuminated diffuser when passing over about nadir. The SeaWiFS scanner can be tilted to +20, the South Pole. The entire spacecraft can be rotated to al- 0, or -20 ° relative to nadir to minimize the number of low the instrument to view the nearly full moon, which is glint contaminated measurements in the data set. Each considered to be a stable calibration source for the purpose pixel value is digitized to 10 bit accuracy, with a typical of monitoring the long-term repeatability of the SeaWiFS scene producing about 600 counts with about one count of measurements (Woodward et al. 1993). noise. The SeaWiFS scanner is illustrated in Fig. 2. Light first strikes the primary mirror, an off-axis parabola, 3. BILINEAR GAINS and then is reflected from a second surface polarization scram- Tests of the instrument at SBRC in the spring of 1993 bler and the half angle mirror before reaching the field revealed that SeaWiFS measurements would be contamistop. The half angle mirror removes the rotation of the nated by stray light from clouds in adjacent pixels on orbit. image from the scan of the telescope. The mirror rotates These tests also showed the stray light contamination to at exactly half the rate of the telescope and polarization be roughly proportional to the brightness of the adjacent scrambler, and it uses alternating mirror sides on succes- cloud. As part of the set of instrument improvements, sive telescope scans. The field stop is actually 50% larger SBRC has changed the sensor's electronics to allow onthan the detectors, and it restricts stray light through system. After the field stop, the light is collimated the orbit measurements of the radiances from clouds. These by changes also maintain the sensitivity of the SeaWiFS meaanother off-axis paraboloid and directed to the aft optics surements of the ocean as specified in the initial requireassembly. Dichroic beam splitters divert the light into four ments for the instrument. focal plane assemblies, each containing two spectral bands delineated by narrowband filters in close proximity to the detector. The optical paths in the aft optics assembly shown in Fig. 3. Attention in the design of SeaWiFS has been given to minimizing the sensitivity of the instrument to polarized light. This consideration is the principal reason for splitting the telescope into two sections, with each rotating at a different speed. This design minimizes the incidence angle of light on the mirrors. There are other possible The new electronic configuration uses the four detector circuits in each SeaWiFS band to create bands with are bi-linear gains. The response for SeaWiFS band 1,412 nm, is shown in Fig. 4. The channel has a high sensitivity, i.e., considerably less than 1 mW per count, over threequarters of the band's dynamic range. Above this point, the channel's sensitivity is reduced, allowing the measurement of cloud radiances up to 60 mW (Fig. 4, top). For in- ocean measurements, the band will have the same response complexity, as before, except over a reduced number of counts (Fig. 4, strument designs which reduce this mechanical but they require large incidence angles on one or more rors, producing unacceptable polarization variations, a polar- can be illustrated using the radiance levels for band 1. ticularly in the blue. In addition, SeaWiFS uses these The values for this band are given in Tables 1-4. The ization scrambler (Fig. 2 and 3) to further reduce mir- bottom). par- The operation of the four channels in a SeaWiFS band

Original page 5

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek Fig. 1. The SeaStar spacecraft. The SeaWiFS instrument three antennas. DIRECTION OF FLIGHT HALF ANGLE MIRROR AFT OPTICS NADIR BENCH DIRECTION OF SCAN (WEST TO EAST) is mounted on the top payload shelf between the t Fig. 2. The SeaWiFS scanner assembly. The scanner mounts to the payload shelf using the four mounting points at the top of the figure. 3

Original page 6

SeaWiFS Prelaunch Radiometric Calibration and Spectral Characterization I I ! AFT OPTICS LENSES FILTER/FPA COLLIMATOR HALF-ANGLE MIRROR SCRAMBLE_MIRROR I PRIMARY Fig. 3. The SeaWiFS aft optics assembly. values are given for Science Gain 1, the standard gain using ence Gain 1. Details of the stepwise calculation algorithm SeaWiFS ocean measurements. The values are given is the are given with the tables. the output from all four channels for this band. This The The SeaWiFS channels are digitized at 10 bits. The standard detector configuration for the instrument. come output from each channel range from 0-1,023 counts. When input radiances, counts, and zero offsets (Table 1) from measurements of the radiometric calibration of for for SeaWiFS band 1. The calculations are made for Scithe the zero offsets are removed, the saturation counts for the instrument. With the zero offsets removed, the net counts, four channels in SeaWiFS band 1 range from 1,000-1,005 along with the sphere radiance, are used to calculate sensitivity for each channel (Table 1). Channel 1, with from saturation radiances for each of the channels. These values sensitivity, allows measurement of the high radiances allow are given in Table 2. For radiances greater than the satclouds. Channels 2, 3, and 4, with high sensitivities, the counts (Table 2). From these saturation counts and the low sensitivities for the channels, it is possible to calculate the measurements of the low radiances from oceans. These uration radiances, the output from the SeaWiFS channels fundamental will remain at their saturation count levels. sensitivities, in mW of radiance per count, are conversion The saturation radiances in Table 2 give the three knee to the calibration of SeaWiFS. They allow the of the counts from the instrument into radiances at the radiances and the maximum radiance for band 1 (gain 1). instrument's optical input. The minimum radiance is zero for zero counts, i.e., for zero Tables 1-4 describe, in stepwise fashion, the determi- counts after the removal of the offset. Using the saturation gains radiance levels and the sensitivities, it is possible to calcunation of knee and endpoint locations for the bilinear

Original page 7

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain, and T. Svitek 60- O O E - 40 q O- {D v E v 20 U C O O fE O O i 1 1 i i i i i ! i [ i i i i i i ! i i | ! i i l i ! i i i | i i i ! i ! i i i | i l l i i i i i i 200 400 600 800 1000 Band I Output (Counts) 15 o (9 E L- I0 E CT E v 5 (9 c- O "I3 O n 0 0 ........ 1'(3C)....... 2C)()....... 30() ....... 40() ....... 5(3C)....... i()C)........ 700'........ I_C)O Bond 1 Output (Counts) Fig. 4. Instrument response for SeaWiFS band 1 with a bilinear gain. The top panel shows the response from 0 to 1,000 counts. The slope from 0 to 800 counts shows greater sensitivity, i.e., fewer radiance units per count. The bottom panel shows the response from 0 to 800 counts. This is the working range for ocean measurements from band 1.

Original page 8

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization Table1. In thefollowingtables(1-4),theprocessof calculating the knee and endpoint locations for the bilinear gains of SeaWiFS band 1 is shown. These calculations are made for Science Gain 1. Table 1 gives the input values and calculated sensitivities for the four channels of band 1. The sensitivities are calculated from the sphere radiances and the net counts. Channel Radiance Meas urement [mW] [counts] 9.246 175 9.246 871 9.246 859 9.246 871 Offset Net Counts Sensitivity [counts] [mW/count] 21 154 0.060039 23 848 0.010903 18 841 0.010994 21 850 0.010878 Table 2. Saturation counts and saturation radiances for the four channels. The saturation radiances are alculated from the saturation counts and the sensitivities. The offset has been removed from both the zero and he saturation counts. Channel Zero Saturation Counts Saturation Radiance 1 0 2 0 3 0 4 0 [mW] 1,002 60.159 1,000 10.903 1,005 11.049 1,002 10.899 Table 3. Calculated instrument output at the saturation radiances for the four channels. The counts at the mees are calculated from the knee radiances and the sensitivities. The counts for each channel cannot exceed ;he saturation counts. Zero Knee 1 Knee 2 Knee 3 Saturation Counts--Channel 1 0.00 181.54 181.60 184.03 1,002.00 Counts--Channel 2 999.64 1,000.00 1,000.00 1,000.00 0.00 Counts--Channel 3 991.39 991.75 1,005.00 1,005.00 0.00 1,002.00 1,002.00 1,002.00 1,002.00 Counts--Channel 4 0.00 3,174.57 3,175.35 3,191.03 4,009.00 Sum of Counts 0.00 793.64 793.84 797. 76 1,002.25 [Sum of Counts]/4 0.00 Radiance 0.00 10.90 10.90 11.05 60.16 Table 4. Knees and endpoint locations for the bilinear gains. These are the values in the last two rows of Table 3. Location Radiance [m W] Counts Zero Knee 1 Knee 2 Knee 3 Saturation three The radiances and counts at the three knees and the late the number of counts from each channel at the knees and the two endpoints of the bilinear gains (Table The counts from the four channels are summed process operations, the zero offsets are removed at the start of divided by four in Table 3. This duplicates the channels, the calculations. This initial step opens up a direct rewithin SeaWiFS. On orbit, the output from the com- lationship between counts and radiances for ocean meaas selected by the instrument's electronics based on mands from the ground, will be summed. The result be sent from SeaWiFS to the SeaStar spacecraft. This which calculation of the sensitivity of band 1: (10.899/793.64), put will be sent minus its two least significant bits, means that the output will be sent from the instrument the spacecraft after division by four. 0.000 0.00 10.899 793.64 10.903 793.84 11.049 797.76 60.159 1,002.25 3). two endpoints are given in Table 4. The counts at zero and radiance are zero. For these calculations, and for on-orbit will surements. Below the first knee in the radiance region for out- ocean measurements, Table 3 gives the information for the to or 0.013773 mW per count. Above the third knee in the radiance region for cloud measurements, the sensitivity

Original page 9

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek is (60.159-11.049)/(1002.25-797.76),or 0.240158mW per differences in the instrument's output (Section 4.2). In adcount.Theoceanportionofthebilineargainis 17.5times dition, it will be possible to check scan-to-scan differences moresensitivethan the cloudportion. This differenceon orbit using Earth-exiting radiances from statistically in sensitivitiesbecomesgreaterin sequencefor bands2 uniform bodies of water, that is, from clear water regions. through8. These conditions lead to the following philosophy for Theeffectofslightlydifferentsensitivitiesin thethree monitoring the long-term repeatability of the radiometric highsensitivitychannelsis shownin Fig.5. In this case,sensitivity of SeaWiFS. On-orbit and ground-based meathekneeisnotsharp,but hastwointernalsegments,rather surements will be used to monitor the change, on average, thanjust onesegmentin andonesegmentout. This in- of the eight instrument bands. Thus, neither of the two creasestheregionof uncertaintyin thetransitionbetweenmirror sides is considered as prime, and neither as secthe oceanandcloudsensitivityregimes. ondary. Measurements of the average radiometric calibra- TherevisedelectronicconfigurationthatgivesSeaWiFStion coefficients combined with measurements of scan-tothe ability to detectclouds,alsodefinesthe methodin scan differences will be used to calculate the coefficients whichthe kneesfor the bilineargainsarecalculated(see for the two mirror sides, in a manner that the values for Tables1-4). If the detectorconfigurationfor a bandis the mirror sides are equally distant from the average. In addition to the measurements presented, ocean measurechanged,thenthe kneesforthebilineargainsmustberements on orbit can be used to check the magnitude of scan calculated.If thesensitivitiesofthehighsensitivitychanline-to-scan line differences. nelsfor a bandarechanged,i.e.,by changingthegainfor theband,thenthekneesforthebilineargainsmustberecalculated.However,the sensitivityofthe cloudchannel,4.2 Scan-to-Scan Difference Equation in this casechannel1 of band1, remainsfixed. It does Side-to-side differences in the half-angle mirror will be notchangewith gainchangesin theinstrument.Onlythe characterized by the manufacturer before launch. There sensitivitiesofthethreehighsensitivitychannelsforeach may be changes in the side-to-side characteristics of the bandcanbechanged.Forthis reason,theupperendpoint half-angle mirror during the course of the mission. The inthebilineargainsdoesnotchangewhentheband'sgains average reflectance for the two mirror sides is an inherent arechanged. part in the 20 scans that cover the surface of the moon in a lunar measurement and in the 480 scan lines of a diffuser 4. MIRROR SIDES measurement. However, the magnitude of the side-to-side differences in the half-angle mirror can be tracked during 4.1 Scan-to-Scan Differences standard measurements of the instrument's diffuser. During each diffuser measurement, there are 240 pairs of half- For the purpose of instrument calibration, SeaWiFS angle measurements between the two mirror sides to be is to be considered as two separate, but nearly identical used to calculate the differences. These differences transinstruments: mirror side 1 and mirror side 2. This separa- form into scan line-to-scan line differences in the ocean tion is necessary since the two sides of the half angle mir- measurements. The magnitude of the scan to scan difror (Fig. 2) have slightly different reflecting properties and ferences will be tracked as part of the onboard calibration slightly different alignments in the sensor's optical path. information from SeaWiFS and will be used in the determi- Mirror side differences translate into scan to scan differ- nation of the two sets of calibration coefficients. Strictly ences since the two mirror sides are used alternately, i.e., speaking, these differences are not part of the long-term from scan to scan in the SeaWiFS measurements. For this radiometric calibration for the instrument. report the authors have not considered the geometric scan- Mirror side differences for a given SeaWiFS band are to-scan differences generated by the small differences in the treated independently in the following equation: two mirror sides. The concern here is the radiometric effects of the differences. As shown below, these differences nl = (c, + c_)/(c_ + c_), (1) are a few tenths of a percent. where R1 is the radiance value measured with side 1 of the For SeaWiFS, there will be two sets of radiometric half-angle mirror, relative to the average radiance from the calibrations, identical in form but with slightly different two sides, with a value very close to unity; (C1 + C1)/(C1 + calibration coefficients. Prelaunch calibration coefficients, C2) is the calculation used to obtain R1, with a value very tailored for the two mirror sides, are provided by SBRC, close to unity; and C1 and C2 are measured values of the the instrument's manufacturer. After launch, however, tai- flight diffuser for sequential scan lines, as described above, lored information of this sort will not be available. For the in counts. R1 and R2 (defined immediately below) are most part, SeaWiFS measurements of the moon will check dimensionless quantities. only the average response of the two mirror sides, since the half-angle mirror, relative to the average radiance from the reflectivity of the lunar surface is not constant across its two sides, can also be calculated directly: face. However, with diffuser measurements, it will be possible to monitor scan-to-scan (mirror side to mirror side) The value for the radiance measured with side 2 of the t_ = (c2 + c2)/(c1 + c_), (2)

Original page 10

SeaWiFS Prelaunch Radiometric 12.00 1 1.75 o ._o 11.50 E 1 1.25 E 0 11.oo E --, 10.75 U C o 10.50 "ID o nr" 10.25 10.00776....... ....... ....... Bond 1 Output (Counts) Calibration and Spectral Characterization ....... ....... ........soo'.........805'........ Fig. 5. An expanded view of the change in slope of the bilinear gain for band 1. This shows the effect of differences in the sensitivities of the three high sensitivity channels. The two segments between 793 and 798 nm are created as the two higher sensitivity channels saturate in turn. For "real world" measurements, the change in slope does not occur at one radiance 2 of the launch. In addition, there will be no means of isolating where R2 is the radiance value measured with side half-angle mirror, relative to the average radiance from two sides, with a value very close to unity. level only. the transmission changes from other elements in the SeaWiFS optical train. On orbit, it will be possible to monitor only Both R1 and R2 are relative values, multipliers that relative changes in the output of the radiometer's bands. will be applied to the calibration constants, below, to give Information on the type of internal changes within those the values for each of the mirror sides. These two constants bands will not be available. Before launch, there are absoman- lute quantities in the prelaunch radiometric constants for are calculated from laboratory measurements by the ufacturer (Table 5). 5. CALIBRATION EQUATIONS As with the equations for mirror side differences, radiometric calibrations are identical in form for each of the instrument that come from laboratory measurements. There is a program in progress to check the transfer of the prelaunch radiometric calibration of SeaWiFS to orbit at the start of on-orbit operations (Biggar et al. 1993). After the that check is made, measurements on orbit will give only the changes in the instrument's sensitivity relative to that the SeaWiFS bands. Thus, the descriptions given here are at launch. representative of each band. The prelaunch radiometric calibrations for SeaWiFS are based on measurements the manufacturer. On orbit, there will be no mechanism within There are several factors that must be considered in for monitoring changes of individual components by 5.1 Calibration Equation Factors the radiometer. SeaWiFS carries no onboard calibrators. the determination of the radiometric sensitivity of the Sea- Among other things, there will be no means of determining WiFS radiometer. Laboratory measurements of the instruspectral shifts in the instrument's interference filters after ment's sensitivity have been made with a fixed set up of the 8

Original page 11

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek commandable variables for the radiometer. For example, sensitivity detectors will be treated comparatively, i.e., exthese measurements have been made at Science Gain 1, amining the change with time in the output of any one of and the instrument's sensitivity at the other three gains the high sensitivity detectors in comparison to the grouphas been determined relative to Science Gain 1. How- ing comprised of the other detectors. Sufficient anomalous ever, on orbit the long-term repeatability of the SeaWiFS behavior of a single detector will cause the removal of that measurements will be made, using scans of the moon at detector from the measurement set. Again, this process is the lunar gain. The principal gain for monitoring instru- discussed in more detail in Woodward et al. (1993). ment operation on orbit is different from the principal gain Other factors in the radiometric calibration equations for laboratory measurements. For practical considerations, are not expected to change over time. These include the the knees and endpoints for the eight SeaWiFS bands have dependence of the radiometric sensitivity on the temperabeen calculated in terms of the radiometric sensitivity of ture of the focal planes and the dependence of the radiothe four channels in each band. metric sensitivity on the scan angle of the measurement. Although there will be thermistors to measure the tem- Table 5. Prelaunch mirror side factors. R1 and R2 perature of each focal plane during the SeaWiFS mission, are dimensionless quantities. there will be no on board means of measuring changes in Band Center R1 R2 the temperature dependence of the output from the de- Wavelength [nm] tectors. In a like manner, no means has been found of 412 1.002 0.998 checking, after launch, changes in the instrument's radiometric sensitivity at the edge of the SeaWiFS scan relative 443 1.001 0.999 490 1.001 to the sensitivity at nadir. These factors are determined 0.999 510 1.001 0.999 solely from prelaunch measurements by the manufacturer, 555 1.002 0.998 as explained in a review of the CZCS calibration by Evans and Gordon (1994). 670 1.002 0.998 765 1.001 0.999 865 1.003 0.997 5.2 Long-Term Sensitivity Long-term changes in the radiometric sensitivity of the The term gain has been present in discussions of the op- eight SeaWiFS bands will be made using measurements of the reflected solar flux from the surface of the moon. These eration of the instrument since the inception of the Ocean measurements will be complimented by vicarious, ground- Color Specifications. It is preferable to refer to the gains as and ocean-based measurements. The sun is considered as sensitivity factors, with units of radiance per count. The a stable, non-changing light source, and the surface of the higher the gain for a measurement, i.e., the greater the moon is considered as unchanging over periods that are number of counts per unit radiance, the lower the saturashort on a geologic time scale. Variations in the solar flux tion radiance, and the higher the sensitivity per count. incident on the lunar surface due to changes in the Earth- This format for the calibration equations reflects the sun distance can be removed (Woodward et al. 1993). In on-orbit conditions for SeaWiFS. The SeaWiFS instrument a similar manner, variations in the reflectance of the moon has no onboard mechanism to monitor the absolute accudue to small differences in the lunar phase angle and due racy of the measured radiances. During the anticipated to small changes resulting from lunar libration can also be 5-10 year lifetime of the radiometer, the SeaWiFS Project removed. With these corrections the effective reflectance will be able to detect relative changes in the sensitivity from the surface of the moon can be made essentially conof the instrument only. After launch, lunar measurements stant. Changes in the SeaWiFS measured values for the will be used to monitor these changes as described below, lunar radiance will be used directly to detect changes in and onboard measurements will be used to detect relative the sensitivity of the instrument. changes among the four electronic gains for each SeaWiFS band and among the four photodiodes for each band. is not possible to separate changes in the instrument sen- The relationships between individual gains are measitivity from changes in the diffuser reflectance in such a For solar measurements with the SeaWiFS diffuser, it sured with a common electronic voltage, a calibration pulse, direct manner. All that can be derived from the diffuser added to the photodiode output while the instrument views measurements themselves is the product of the change in the inner surface of the back of the instrument. This pro- the instrument and of the change in the diffuser. There is cess is discussed in more detail in Woodward et al. (1993). an assumption, however, that will allow the diffuser and Relative changes among the individual detectors in a Seathe lunar measurements to be tied together. Basically, the WiFS band will be monitored by having the detectors change in the reflectance of the diffuser is assumed to be view the instrument's diffuser sequentially several times linear over time periods of approximately one month (Cewhile the diffuser is illuminated by the sun. This can be bula et ah 1988 and Herman et al. 1990). Experience with done, since each possible detector combination can be com- diffusers on previous satellite instruments has led to the exmanded from the ground. The relationship among the high planation that diffuser degradation has been caused by the

Original page 12

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization andsilicon- 1. Lnadir is the measured radiance at nadir, in mW coatingof the diffuserwith solarizedorganic fromthespacecraft. cm -2 sr- 1 #m- 1. basedmaterialsthat haveoutgassed The accumulationof thin layersof thesematerialsdoes 2. K1, which is a function of the time, is the primary not leadto sharpstepfunctionsin the reflectivityof the instrument sensitivity factor expressed as the indiffuserovertime. verse of the number of days post-launch, and there- Overlongertime periodsof oneto severalyears,the fore equal to unity at the start of the SeaWiFS mischangein the reflectanceof thediffuseris expectedto be- sion. This factor applies equally to each of the four comeanexponentialfunctionof time,asymptoticallyapproachingzerochangewith timeasthe SeaWiFSmission ends.However,this 5-10yearlongexponentialfunction canbetreatedasa seriesofmanylinearsegmentsthatare a monthin duration. measure- count. Usingnearlysimultaneouslunaranddiffuser gains in a SeaWiFS band. 3. t is the time, in days, after the start of on-orbit measurements by SeaWiFS. 4. K2 is the gain factor, in mWcm-2sr-l#m -1 per ments,it is possibleto separatechangesin thesensitivity 5. gs is the gain selected by command from the ground of the instrument,determinedfromlunarmeasurements, and present in a datum from the spacecraft engifromchangesin the reflectanceof the diffuser.Whenthe neering telemetry that precedes each scan line. time-seriesof diffuserreflectancesis normalizedby lunar measurements,it ispossibleto usetheassumptionofalinearchangein diffuserrefiectivitytoidentifystepchangesin the instrumentsensitivitythat mayoccurbetweenlunar measurements.Suchchangeswerefoundin CZCSmeasurements(EvansandGordon1994). 5.3 Long-Term Sensitivity Equation The long-termsensitivityequationforthe radiometer containsonlyfactorsthat canchangeduringthelifetimeof the SeaWiFSmission.Theseincludetwocommandvari- 6. ds is the detector configuration selected by command from the ground and present in a datum from the spacecraft engineering telemetry that precedes each scan line. 7. (1 + K3(T-Tref)) is the focal plane temperature factor, a dimensionless value very close to unity. 8. K3 is the temperature dependence of the output from the detector, in terms of inverse degrees Celsius, with a value very close to zero, as measured by the manufacturer. theselectionof 9. T is the measured temperature of the focal plane ablesthatcanbechangedfromtheground; They assembly, in degrees Celsius, calculated from a datheelectronicgains,andthedetectorconfiguration. canbe tum in the engineering telemetry stream associated alsoincludetime-dependentvariableswhosechange ofthefo- with each scan line. monitoredduringthemission,thetemperatures calplanesasmeasuredby thermistors,andthe long-term 10. Tref is the reference temperature for the temperaradiometricsensitivitiesoftheinstrumentbandsasdeterminedthroughlunaranddiffusermeasurements. Thelong-termsensitivityequationdoesnotincludethe ofthe 12. Cdark is the instrument dark restore value, in counts. scanangledependenceoftheradiometricsensitivity ture dependence, 20 ° C, set by the manufacturer. 11. Cout is the instrument output, in counts. radiometer(alsocalledscanmodulation),i.e.,the sensi- The calculation of the focal plane temperatures (T) tivity oftheinstrumentformeasurementsat thelimbrel- from the data in the spacecraft engineering telemetry reativeto thoseat nadir. Thesevaluesaredeterminedby quires several steps. As a result, the presentation of the the manufacturerandcannotbeindependentlyverifiedon temperature calculations is deferred to Section 8.0, where orbit. Theeffectofscanmodulationwill beaddedto the the eight values for/(3, one for each band, are also listed overallradiometricsensitivityequationin thenextsection.(Tables 10 and 11). In addition,the long-termsensitivityequationdoes The primary instrument sensitivity for each SeaWiFS not includechangesin the linearityin the detectorre- band, Kl(t), is calculated from a combined set of lunar sponsewith inputradiance.Again,thesevaluesaredeter- and diffuser measurements. There are also plans to derive minedsolelythroughmeasurementsby the manufacturerthis factor independently using ground based calibrations andcannotbecheckedafterlaunch.A discussionof the manufacturer'smeasurementsis givenbelow. The on-orbitequationfor the time dependentradiometricsensitivityofeachSeaWiFSband,at nadir,hasthe followingform: Lnadir ----- KI(t) × K2(gs, ds) x (1 + K3(T-Tref)) X (Cout - Cdark). The variables in (3) are defined as follows: 10 once the radiometer is collecting data on orbit. It is anticipated that the time series of K1 (t) values for periods through each lunar measurement will be updated and sent to the SeaWiFS stations approximately two weeks after that measurement. As the ensemble of lunar and diffuser measurements grows, improved time histories of the K1 (t) values will be supplied. In addition, as the rate of change of (3) the K1 (t) values becomes understood, it will be possible to give improved, educated guesses regarding future changes in the K1 (t) values. Until reasonable time series for K1 (t)

Original page 13

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek valuesareestablished,theseKl(t) values will be treated the eight values that are required for the calculation of the on orbit as constants during the time periods between the temperature correction. most recent and the next lunar measurement and then will Thermal models of the operation of the SeaWiFS inbe updated using the next lunar measurement. strument indicate that the anticipated day-night tempera- Since all four gains for each SeaWiFS band--the four ture differences for a single orbit are on the order of 1° C. values of K2(gs)--will be used frequently during the ra- However, as the instrument's thermal blankets age over diometer's mission, calibration coefficients for all of the the 5-10 year duration of the SeaWiFS mission, the av- SeaWiFS bands will be contained in a periodically updated erage temperatures of the focal planes are anticipated to list that will be provided to each of the ground stations. increase slowly at the rate of 1-2 ° Celsius per year. These values will be given in a table of the knees and endpoints for the bilinear gains, using the form of Table 4. 5.4 Off-Nadir Measurement Differences In addition, one word showing the command variable that determines the choice of the gain will be present in the enric response of the SeaWiFS instrument have been made gineering telemetry that precedes each scan line from the in the laboratory at nadir, i.e., pixel 643. Although not exsatellite. plicitly stated in the SeaWiFS performance specifications, The relative values for these gains will be monitored it is important to know how the measurement at pixel 643 regularly during the mission (Woodward et al. 1993). The can be transferred to the other pixels in the scan. SBRC value for the lunar gain for each SeaWiFS band, i.e., the ran such tests during its characterization of the instrument. radiance per count for the lunar gain, will not change with In the presentation of the test results, SBRC referred to The primary laboratory measurements of the radiomettime during the SeaWiFS mission. Since the change in the the effect as scan modulation. sensitivity of the instrument is determined through measurements of the moon, changes in the lunar gain will be at nadir and at 100 pixel intervals above and below nadir. found in Kl(t) for each band only. Changes in the values Data were taken for both mirror sides, and the off-nadir re- The data from these tests were taken approximately of the other gains relative to the lunar gain will be updated sponses for the two mirror sides were identical at the 0.2% as determined. Changes in the gain ratios are anticipated level, which is the resolution limit for the laboratory meato be small over time, if detectable at all. However, time surements. Overall the measurements can be summarized series for the gain ratios for the eight instrument bands as follows: will be provided by the Project. It is anticipated that the standard detector selection, ds, will not change during the SeaWiFS mission. For a given instrument band, that standard selection uses the sum of the output of the four detectors divided by four. Using the comparative method of the previous section, it is anticipated that none of the detectors will exhibit anomalous behavior and have to be removed from the measurement set. Should this occur, however, a new table of knees and endpoints for the bilinear gains will be supplied by the Project. In addition, a datum showing the command variable that determines the detector configuration will be present in the engineering telemetry that precedes each scan line from the instrument. The focal plane temperature factor, 1 + K3(T-Tref), contains two constants that are supplied from measurements by the radiometer's manufacturer: K3 and Tref. For 1. All bands show changes in their responses with pixel number. These changes are small, ranging from 1% to 2% over the 1,285 pixels of a scan line. 2. The scan modulations for odd bands (bands 1, 3, 5, and 7) show a marked similarity with each other, as do the scan modulations for the even bands. The odd and the even bands, however, show distinctly different responses from each other. 3. The scan modulations can be fitted to quadratic equations easily. It is possible to account for the scan modulation in all eight bands with two equations, one for the even bands and one for the odd ones. 4. The measurements by SBRC show no mirror side differences in the scan modulation. The data provided by SBRC did not include error bars all eight SeaWiFS bands, the constant Tre_ is set to 20 ° C. for the individual radiometric measurements. From the The calculation of the temperature of the two detectors results, however, it appears that the data have a scatter on each focal plane requires the conversion of engineering at one standard deviation of about ± 0.3%. This scatter telemetry from the SeaStar spacecraft, in counts, into an falls above and below the linear results versus wavelength analog signal from SeaWiFS, in volts, and then into tem- discussed above. The average values for the responses of perature, in degrees Celsius. The conversion from volts to the odd and even SeaWiFS bands are given in Fig. 6. The temperature requires a series of algebraic manipulations. quadratic curves in Fig. 6 show the off-nadir corrections These manipulations are not conceptually difficult, but from the manufacturer for the odd and even bands. they do require several steps of computation. The series The cause of the scan modulation has been explained of calculations required to obtain the focal plane temper- as vignetting (shadowing) of the detectors during portions atures is found in Section 8.0. This section also contains of the scan. The detectors can be considered as having 11

Original page 14

SeaWiFSPrelaunchRadiometricCalibration and Spectral Characterization I ,01 tgJ {:: .(2_1.00 ID e'. E C3 0.99 t- O {D I J, IX: 0.98 0 ...............6&400'.........600'.........800'........15'0a.......i'2'0'" Pixel Number 1.01 - I#} / L) f .9 1.00 E I) E 0.99 t- O P, I# n,," 0.98 0 ......... 200'......... 400'........ 606 ....... 806 ....... 1"000........ 1200'.... Pixet Number Fig. 6. Scan modulation measurements for the odd and even SeaWiFS bands. The results are given as relative differences from the nadir (pixel 643) value. The symbols show the average values from the laboratory measurements of the odd and even bands. The quadratic curves show the scan modulation corrections--see text for details. The top panel shows odd bands, and the bottom panel shows the even bands. 12

Original page 15

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain,andT. Svitek the imageof the fieldstopimposeduponthem.Anyim- of the variables have any mirror side dependence. Based on perfectionof thealignmentoftheinstrument'sopticswill considerations in the methods for determining long-term causethe imageof the fieldstopto shift relativeto the changes in the instrument's sensitivity, K1 has also been detectors.Tominimizestraylightenteringtheaft optics, defined without a mirror side effect. the imageof the fieldstopis 50%greaterthan the size of the detectorsfor eachband. In the directionnormal 5.5 Full-Up Equations to the scanplane,this givesonlya one-quarterpixelleealignment The use of the radiometric calibration equations on orwayon eithersideof the detectors.Imperfect oftheopticscausestheimageofthefieldstopto moverel- bit will require the consideration of the two sides of the ativeto thefocalplaneassemblies.Increasingthesizeof half-angle mirror. This factor has been described above, the fieldstop,or cockingit slightly,couldeliminatemuch including the factor R1 for mirror side 1 and R2 for mirror of this effect. Thesechangeswerenot part of the stray side 2. When these two factors are included, (5) can be light improvementsby SBRC.However,a portionof the modified to produce two equations with the form: straylightrepairsto SeaWiFSinvolvedworkonthe focal planeassemblies.Therepositioningofthefocalplanesduringtheirreinstallationdidaffectthefinalscanmodulation valuesslightly. The quadraticresponsecurves,calculatedby SBRC, arebasedonmeasurementsoftheoutputoftheSBRCin- and tegratingsphereovera periodof about30minutes.To viewthe sphere,the instrumentwasrotatedto sevenangularpositions,andthe radiancesfromthe spherewere recorded.The positionsfor the measurementswereap- LS1 = R1 x Kl(t) x K2(gs, ds) x (1 + g3(T-Tra)) (6) X g4(Pxl ) x (Cout--Cdark), LS2 = R2 x Kl(t) x K2(gs, ds) × (1 + K3 x (T-Tref)) (7) x Ka(Pxl) x (Cout--Cdark). proximately220pixelsapartandcoveredthe rangefrom The variables in (6) and (7) are defined as follows: 1-1,285pixels.Smallchangesin theoutputofthe sphere mayhaveaddeda scatterof a fewtenthsof a percentto the measurementsbythe instrument. The scanmodulationresponsecurvesequalunity at nadir.Forotherpixels,the curvesderivedby SBRCgive thedifferenceofthe instrumentmeasurements,in relative terms,fromthevalueat nadir.Thecorrectionsto thecalibrationequationsaretheinverseofthecalculatedinstrumentresponsesfromSBRC.The SBRCresponsecurves showsthe outputof oddbandsat pixel1 to be1%lower thanthoseat pixel 643.Tocorrectfor this,1%mustbe calibration equations for SeaWiFS. addedto thevaluesfromtheoddbandsat pixel1. Thisis thescanmodulationcorrection: 5.6 Instrument Response Linearity 1 K4(Pxl) = 1 + A0(Px1-643) + Bo(Px1-643) 2' (4), the measured radiance is a linear function of the number of 1. LS1 is the measured radiance for mirror side 1, in mW cm-2 sr-1 #m -1. 2. R1 is the dimensionless multiplier for mirror side 1 from (1). 3. LS2 is the measured radiance for mirror side 2, in mW cm-2 sr-1 pm-1. 4. R2 is the dimensionless multiplier for mirror side 2 from (2). Equations (6) and (7) are the two on-orbit radiometric Equations (6) and (7) are based on the assumption that where Pxl represents the pixel number, and where Ao = counts in the instrument's output. The SeaWiFS perfor- 3.115 x 10 -6 and B0 = -1.929 x 10 -s for odd bands, and mance specifications call for the output to be linear with A0 = 1.713 x 10 -5 and Bo = -1.456 x 10 -s for even bands. radiance at the 1% level or better. For the six shortest For all bands, K4(Pxl) is equal to unity at nadir. As a wavelength SeaWiFS bands (412, 443, 490, 510, 555, and result, this factor can be applied to (3) to cover all pixels 670 nm), the measurements at SBRC show the linearity to in each scan. For a SeaWiFS band, the values over a scan be at the 0.5% level. For bands 7 and 8 (765 and 865 nm), line are given as the laboratory measurements show linearities that are better than the specifications, but not at the levels of the other Lsca, = Kl(t) x K2(gs, ds) six bands. × (1 + K3(T-Tref)) (5) The measurements of the linearity of SeaWiFS were x K4(Pxl) × (Cout--Cdark), made using an integrating sphere. The radiances from the sphere were determined using a laboratory radiometer where Lscan is the measured radiance at any pixel in a scan, that compared the sphere's output to that from a standard in mWcm -2 sr -1 #m -1. lamp that has been calibrated by the National Institute of Equation (5) does not consider the mirror side differ- Standards and Technology (NIST). The radiances required ences in the instrument. Except for the variable, K1, none for measurements of SeaWiFS at 765 and 865 nm are very 13

Original page 16

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization to the outputfromthe standardlisted in Table 6 and are based on laboratory measurements smallwhencompared radiometer,by SBRC. The gain ratios are given with respect to Science lampat thesewavelengths.Forthe transfer measure-Gain 1. theselevelsweresolowthatnoisein thetransfer mentapproached1%. The gain ratios in Table 6 show little channel-to-channel isthattherelatively variability. The four channels in band 1 can be taken as Fromthisevidence,theconclusion poor linearity in the SeaWiFS laboratory measurements an example. For channel 1, the gain ratios for the four at 765 and 865 nm is due to limits in the quality of the gains in Table 6 are all equal to unity. Channel 1 is the characterization of the sphere itself. It is also concluded linear calibration data from SBRC, the measured gain ratios for that the readings for all eight SeaWiFS bands are with input radiance at the level of 0.5% (Barnes et 1994). There is no term in the instrument's calibration equations for non-linear response. 5.7 Polarization Sensitivity The degree of polarization of the eight SeaWiFS bands is measured in the laboratory using a polarizer that is rotated through 360 ° in 22.5 ° intervals. Over this angular range, the instrument bands should show two cycles of response. The response from SeaWiFS band 1 is shown in Fig. 7. There is little indication of the predicted twocycle response. It is more reasonable to conclude that the variations in the instrument output shown in Fig. 7 are indicative of changes in the transmission of the polarizer with angle. Barnes et al. (1994) used Fourier analysis expose the two-cycle signal. Their results indicate that polarization in the instrument is 0.25% or less. The measurements in Fig. 7 are taken at pixel 633, near nadir. The response in this figure is also representative of the instrument responses at 30 ° and 55 ° from nadir. From the polarization measurements at these three angles, the polarization of SeaWiFS remains at 0.25% or less over a scan. There is no term in the instrument's calibration equations for polarization sensitivity. 6. GAIN FACTORS The relative differences in the four sensitivity coefficients for each band result from different selectable gains in one of the intermediate amplifiers in the instrument's electronics. The SeaWiFS specifications give the gain requirements for gains 2, 3, and 4 relative to gain 1. Gain 1 is the principal science gain. Gain 2 is the secondary science gain with a radiometric sensitivity twice that for gain 1, that is, one count from Science Gain 2 represents half the radiance for one count from Science Gain 1. 6.1 Gain Ratios 6.1.1 Individual Channels been unit of radiance per count. The gains in the intermediate amplifiers have also lunar For gain 2 calculations, 1/1.988 is the fractional mulset up to give a three-quarters full scale output for cloud channel, and those values should all be unity. In the al. band 1-channel 1 are 1.016 (G2/G1), 1.005 (G3/G1), and 1.010 (G4/G1). These differences are assumed to come from noise in the measured counts for each gain. That noise amounts to 2 or 3 counts out of 220. The data for each gain have a single calibration datum and a single zero. On orbit, there will be the capacity to take 25 calibration points and 25 zeros from each calibration pulse scan. This will improve the on-orbit gain ratios significantly when compared with the prelaunch measurements. As a result, the gain ratios for the cloud channels have all been set to unity to eliminate the effects of noise in the measurements. In addition, there has been an averaging of gain ratios in the high sensitivity channels. For band 1-gain 2, the three high sensitivity bands all have the same gain ratio in Table 6 (1.988). In the calibration data to from SBRC, the measured gain ratios are 1.987 (channel 2), 1.982 (channel 3), and 1.995 (channel 4). As with the measurements of the cloud channel, there is noise in these gain ratios. In the prelaunch data, the gain ratios have been set to the average of the three channels to remove noise. On orbit, it will be possible to measure these ratios with greater precision. Until then, it is not possible to distinguish between noise in the gain measurements of the high sensitivity channels and small differences in their gains. 6.1.2 Combinations of Channels The gain ratios in Table 6 are given for individual channels. It is possible to calculate these ratios for combinations of channels. Such ratios can be calculated for the standard SeaWiFS detector configuration, 4:1 TDI, where the output is the sum of the four channels. The calculations are based on the procedure outlined in Tables 1-4. They can be explained by using the values from SeaWiFS band 1. The gain ratios for the high sensitivity channels for band 1-gain 2 are 1.988 (Table 6). This means that a channel will produce 1.988 counts per unit radiance at gain 2 if the channel gives one count per unit radiance at gain 1. Conversely, if gain 1 has a sensitivity of one unit of radiance per count, gain 2 will have a sensitivity of 1/1.988 measurements and for solar measurements, using the flight tiplier applied to the sensitivities for channels 2, 3, and 4 can from Table 1. For gain 2, the sensitivities of these channels diffuser. Details of the lunar and solar measurements be found in McClain et al. (1992) and in Woodward et al. are approximately 0.005 mW per count (Table 1). With (1993). The gain values for the 32 SeaWiFS channels 14 are these three sensitivities and the sensitivity for channel 1,

Original page 17

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain, and T. Svitek Table 6. Gain ratios for the SeaWiFS channels, relative for each band do not change with the gain selection. Band Channel Gain 1 1 1.000 2 1.000 3 1.000 4 1.000 1 1.000 2 1.000 3 1.000 4 1.000 3 1 1.000 2 1.000 3 1.000 4 1.000 4 1 1.000 2 1.000 3 1.000 4 1.000 1 1.000 2 1.000 3 1.000 4 1.000 1 1.000 2 1.000 3 1.000 4 1.000 1 1.000 2 1.000 3 1.000 4 1.000 1 1.000 2 1.000 3 1.000 4 1.000 to Gain 1. The gains for the cloud detection channels Gain 2 Gain 3 Gain 4 (G2/G,) (G3/GI) (G4/GI) 1.000 1.000 1.000 1.988 1.320 1.681 1.988 1.320 1.681 1.988 1.320 1.681 1.989 1.319 1.682 1.989 1.319 1.682 1.989 1.319 1.682 1.000 1.000 1.000 1.000 1.000 1.000 1.989 0.896 1.681 1.989 0.896 1.681 1.989 0.896 1.681 1.989 0.789 1.682 1.989 0.789 1.682 1.989 0.789 1.682 1.000 1.000 1.000 1.000 1.000 1.000 1.989 0.642 1.595 1.989 0.642 1.595 1.989 0.642 1.595 1.989 0.364 0.665 1.989 0.364 0.665 1.989 0.364 0.665 1.000 1.000 1.000 1.000 1.000 1.000 1.987 0.311 0.575 1.987 0.311 0.575 1.987 0.311 0.575 1.991 0.261 0.499 1.991 0.261 0.499 1.991 0.261 0.499 1.000 1.000 1.000 which has not changed, it is possible to work through the the bilinear gains. If the gain of a channel or combination calculations in the manner of those in Table 2 and 3. The of channels changes, then new knees must be calculated results give three new knees for the bilinear gains, both in using the procedure from Tables 1-4. radiance and in counts. The upper endpoint remains the same, since the sensitivity for channel 1 has not changed. 6.2 Lunar Measurements Using the zero values plus those from the first knee in the bilinear gain, it is possible to calculate the sensitivity Once on orbit, the monitoring of the long-term repeataof band 1 in the radiance region for ocean measurements. bility of SeaWiFS measurements will be done by viewing The ratio of the sensitivities for gain 1 and gain 2 give the moon. On orbit, the changes in the sensitivities of the the gain ratio for these gains with the standard detector science gains will be based on measurements of the moon configuration (Table 7). Table 7 also lists the gains that are at the lunar gains. The lunar gains will be treated as a used for lunar and solar measurements. These lunar and constant throughout the SeaWiFS mission. Changes in solar gains agree with the gains required by the SeaWiFS the measured lunar radiance, with the real lunar radiance performance specifications. remaining constant, will result in corresponding changes From this discussion, it follows that the calculation to the variable K1, also described in (3). The lunar gains method in Tables 1-4 is fundamental to the operation of in these interchannel measurements will be kept constant, 15

Original page 18

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 1.03 _ 1.02 ¢- 0 _ 1.ol E a _-I.00 * 0.99 o .>--0.98 0 n,- 0.97 iiillllllll|lllll|lllllJlllllllllllllllllJlllllJlllllJlllllJlllllllllll| 0 30 60 90 120 150 180 210 240 270 300 330 360 PolGdzer sensitivity of SeaWiFS band 1 (412 nm). Fig. 7. Polarization and the gain ratios for the other gains will be allowed instru- on orbit. This calibration plan has been developed to make vary. These are important considerations for an ment whose long-term repeatability will be monitored Angle (Degrees) to instruments, in conjunction with SeaWiFS measurements by both the vicarious and satellite measurements traceable to lunar, rather than terrestrial, measurements. Among other a single set of laboratory standards so that ground and on in viewpoint orbit measurements can be reconciled. For the vicarious things, these considerations require a change from terrestrial to lunar measurements by those tracking measurements, there is an ongoing program of round-robin changes in the instrument's sensitivity on orbit. Table 7. SeaWiFS gain values for the standard detector configuration. These values are given relative to gain 1. The nominal value for gain setting 2 is 2. The gains used for the lunar measurement and for the solar diffuser measurement are shown. These gains cover the measurement regions from zero radiance to the first knees in the bilinear gains. Band Gain 1 Gain 2 Gain 3 Gain 4 1 1.000 1.931§ 1.302 1.6421 2 1.000 1.940 1.303f 1.648§ 3 1.000§ 1.951 0.900t 1.655 4 1.000§ 1.955 0.7961 1.658 5 1.000§ 1.961 0.6521" 1.579 6 1.000 1.969 0.376i- 0.671{] 7 1.000 1.969 0.3231 0.583§ 8 1.000 1.975 0.2721 0.507§ Gain used for solar diffuser measurement. Gain used for lunar measurement. 1992) ocean, sites. The ground based measurements will com- The SeaWiFS calibration plan (McClain et al. pri- plement the set of satellite based, long-term repeatability includes a program of ground based measurements, field measurements described here. Ultimately, the combined marily measurements of water-leaving radiances with 16 intercalibrations of field instruments using NIST-traceable standards (Mueller 1993). The reconciliation of vicarious and on-orbit measurements also requires knowledge of the atmospheric transmission of light at the SeaWiFS wavelengths. For the SeaWiFS mission, this knowledge is derived from a set of models of the atmosphere that include light scattering by molecules of air, Rayleigh scattering, and by small particles, Mie scattering (H. Gordon, pets. comm.). These models will play a fundamental role in the reconciliation of the on-orbit radiance measurements from SeaWiFS and the water-leaving radiance measurements from field instruments. The on-orbit monitoring scheme for the long-term repeatability of SeaWiFS measurements as described here, and also found in Woodward et al. (1993), comprises a closed set of measurements made by the satellite instrument only. The set of ground based measurements proposed in the SeaWiFS calibration plan (McClain et al. 1992) will be performed in conjunction with SeaWiFS measurements during satellite overpasses of the ground, or

Original page 19

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek Table8. InputvaluesandcalculatedsensitivitiesfortheeightSeaWiFSbands.Theradiances,themeasurement counts,andtheoffsetcountscomefromlaboratorydata.Thesensitivitiesarecalculatedfromtheradiancesand the net counts. The values are given for Science Gain 1, the standard gain for SeaWiFS ocean measurements. Band Channel Radiance Measurement Offset Net Counts Sensitivity [mW] [counts] 1 9.246 175 2 9.246 871 3 9.246 859 4 9.246 871 1 9.122 883 2 9.122 887 3 9.122 878 4 9.122 153 1 7.216 127 2 7.216 899 3 7.216 905 4 7.216 903 1 5.970 856 2 5.970 855 3 5.970 856 4 5.970 111 1 4.692 98 2 4.692 840 3 4.692 837 4 4.692 828 1 1.682 540 2 1.682 538 3 1.682 544 4 8.058 168 1 9.885 253 2 2.057 915 3 2.057 913 4 2.057 922 1 1.063 671 2 1.063 670 3 1.063 671 4 10.283 320 set of satellite based and ground based measurements will form the basis for the understanding of the SeaWiFS data set. 7. CALIBRATION CONSTANTS 7.1 SeaWiFS Radiometric Calibration The laboratory results of the SeaWiFS radiometric calibration are given in Table 8. The 32 listings in the table show the input radiances and output counts for each channel of each band from the laboratory measurements. The [counts] [mW/count] 21 154 0.060039 23 848 0.010903 18 841 0.010994 21 850 0.010878 18 865 0.010546 21 866 0.010533 16 862 0.010582 18 135 0.067570 21 106 0.068075 22 877 0.008228 21 884 0.008163 19 884 0.008163 21 835 0.007150 20 835 0.007150 19 837 0.007133 21 90 0.066333 26 72 0.065167 22 818 0.005736 22 815 0.005757 17 811 0.005785 21 519 0.003241 17 521 0.003228 33 511 0.003292 21 147 0.054816 23 230 0.042978 20 895 0.002298 21 892 0.002306 27 895 0.002298 20 651 0.001633 24 646 0.001646 18 653 0.001628 20 300 0.034277 counts. The values from Table 8 are sufficient to calculate the knees and endpoints for each band for all detector combinations at Science Gain 1. The procedure outlined in Tables 1-4 can be applied to any combination of channels. For example, it is possible to combine channel 1 of band 1, taken once, with channel 2 of that band, taken three times, to create a truly bilinear response, a response with only one knee. As discussed in Section 6.1, the gain ratios can be applied to the sensitivities in Table 8 to calculate the knees for each band for gains 2, 3, and 4. The endpoints for these gains will not change. The zero remains zero, and the upvalues are given for Science Gain 1. The sensitivities for per endpoint is a function of the sensitivity of the cloud the channels are calculated from the radiances and the net channel, which also does not change. Both the radiance 17

Original page 20

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization ofthe for band 1 when it is exposed to a 5,900 K source, as mealevelsandcountvaluesforthe kneesarea function ofthreeof thechannelssured on orbit, than when it is exposed to a 2,850 K source, gainsettingsincethesensitivities change,whilethe cloudchannelremainsconstant. 7.2 Out-of-Band Correction measured in the laboratory. The counts per unit radiance of this band will be smaller on orbit than they are in the laboratory. It will take a 3.7% greater radiance to saturate the There is an artifact of the radiometric calibration that channels of band 1 on orbit than is required to saturate has not been discussed in the calibration data from SBRC. The radiometric calibration of the instrument was performed with an integrating sphere as the source of radithem in the laboratory. Thus, the radiances at the knees and the upper endpoint for band 1-gain 1 in Table 9 are 3.7% greater than those in Table 4. These are the radiances ance. The sphere has a wavelength dependent output that at which each channel of the band saturates. can be approximated by a 2,850 K blackbody. This source The fractional change from the out-of-band correction has a peak output near 1,000nm (Section 9.8, Fig. 17). On has only changed the radiances at the knees and the uporbit, SeaWiFS will measure input radiances with spectral shapes that are similar to that from a 5,900 K blackbody, which has a peak output near 490 nm. The bandwidths for the SeaWiFS bands are 20 nm wide for bands 1 through 6, and 40 nm wide for bands 7 and 8. For these bandwidths, the spectral shape of the source has per endpoint. The count values have remained the same. This results from the fact that the out-of-band correction affects the sensitivity of each channel in a band identically. For band 1, each channel changes by 3.7%. For the other bands, the fractional change is different (Section 9.10, Table 12). The out-of-band corrections have been applied to little effect (Section 11). The SeaWiFS bands, however, all bands and all channels in Table 9. have small out-of-band responses (Section 9). The photodiodes in each SeaWiFS band have a slightly different output when illuminated by a 2,850 K or 5,900K source. This output difference is of the order of a few percent 7.4 Out-of-Band Contributions or The out-of-band responses for the eight SeaWiFS bands less (Section 9.10). This output difference is also the same are parts of the instrument's radiometric calibration. In for each channel of a band. The cloud and high sensitivity channels for a band all have the same fractional change that calibration, the instrument views a broad area of known radiance, and records the output from the bands in in output when measuring these two sources. The out-of- counts. The counts from each band include the out-of-band band effect changes the sensitivities of all four channels a band identically. 7.3 Bilinear Gain Calibration Constants The calibration constants for the SeaWiFS bands are listed in Table 9. The table gives the endpoints and the knees for each band and gain. These values are given for of the The prelaunch calibration equations for SeaWiFS conthe standard detector combination, i.e., for the sum four channels in each band. Using the radiance levels and count values in Table 9, it is possible to calculate the radiance corresponding the band correction for a 5,900 K source factored into them. each count for each band. Table 9 is the basis for For The 5,900 K spectral shape closely duplicates the spectral calculation of the gain factors (K2 values) used in (3). computational purposes, it is also possible to prepare lookup arise from the use of the 5,900 K out-of-band corrections lookup table for each band and for each gain. The about for ocean measurements are estimated to be small, i.e., a table would have approximately 32,000 entries, with eight few tenths of a percent. If an alternate out-of-band cor- 1,000 entries for each of the four gains for each of the standard rection is to be used, then the 5,900 K correction must be bands. This lookup table applies only to the detector configuration. Any changes to that configuration also band correction inserted in its place. would require the recalculation of Table 9. It would require the creation of a new lookup table, if one was used in the on-orbit data reduction. The knees and endpoints for band 1-gain 1 in Table 9 are not the same as those in Table 4. The radiances channel 1-gain 1 in Table 9 are 3.7% greater than the corof the The information about the temperature sensors for the responding radiances in Table 4. This is the result Sec- focal plane assemblies includes the temperature dependenapplication of the out-of-band correction discussed in tion 9.10. There is a smaller output from the photodiodes 18 in contribution. Those out-of-band contributions are functions of the spectral shape of the source that is measured. The SeaWiFS laboratory calibration has the out-of-band correction for a 2,850 K source factored into its results. If the instrument measures a source with this spectral shape, these measurements automatically contain the appropriate out-of-band corrections. tain correction terms that convert the out-of-band responses from a 2,850 K source to a 5,900 K source. As a result, to the SeaWiFS calibration equations now have the out-ofa shape for SeaWiFS ocean measurements. The errors that removed from the measurement results, and a new out-of- 8. TEMPERATURE FACTORS for 8.1 Dependence Coefficients cies of the output of the eight SeaWiFS bands, the /{3

Original page 21

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek Table9. PrelaunchcalibrationconstantsfortheSeaWiFSbilineargains.Thesevalueshavebeencorrectedfor out-of-bandresponse(seeTable12forthe conversionfactor).Thesevaluesaregivenforthestandarddetector configuration(eachchannelusedonce).Forallbandsandatall gainsettings,the instrumentwascalibratedto givezerocountsat zeroradiance. Band Gain Knee 1 Knee 2 Knee 3 Saturation Radiance Counts Radiance Counts Radiance Counts Radiance Counts 1 1 793.64 11.313 793.84 11.317 797.76 11.469 1,002.25 62.445 2 771.09 5.691 771.27 3 782.64 8.571 782.83 4 775.26 6.731 775.45 5.693 774.89 5.769 1,002.25 62.445 8.547 786.60 8.688 1,002.25 62.445 6.734 779.12 6.824 1,002.25 62.445 2 1 789.10 10.734 791.34 10.778 792.93 10.837 1,004.75 69.062 2 769.69 5.397 771.85 3 779.67 8.140 781.86 4 773.27 6.382 775.44 3 1 779.55 8.343 780.61 2 764.62 4.194 765.64 3 783.05 9.315 784.11 4 767.39 4.963 768.42 4 1 778.79 7.175 779.00 2 765.37 3.607 765.57 3 786.02 9.098 786.23 4 767.85 4.267 768.05 5 1 769.72 5.794 771.63 2 758.77 2.913 760.64 3 782.02 9.029 783.97 4 761.50 3.631 763.38 6 1 763.41 3.211 763.48 2 756.05 1.615 756.12 3 789.32 8.831 789.40 4 770.87 4.829 770.95 7 1 759.48 2.300 763.07 2 752.87 1.158 756.41 3 788.94 7.390 792.73 4 769.31 3.999 772.97 8 1 762.22 1.618 762.77 5.420 773.33 5.449 1,004.75 69.062 8.174 783.40 8.218 1,004.75 69.062 6.408 776.94 6.444 1,004.75 69.062 8.360 782.00 8.401 1,002.25 69.576 4.202 766.96 4.224 1,002.25 69.576 9.333 785.52 9.379 1,002.25 69.576 4.973 769.74 4.998 1,002.25 69.576 7.178 779.28 7.185 1,002.75 66.599 3.609 765.84 3.612 1,002.75 66.599 9.101 786.52 9.110 1,002.75 66.599 4.268 768.32 4.272 1,002.75 66.599 5.815 774.33 5.872 1,001.25 65.556 2.923 763.23 2.952 1,001.25 65.556 9.062 786.80 9.152 1,001.25 65.556 3.645 766.00 3.681 1,001.25 65.556 3.212 764.36 3.223 1,000.00 54.322 1.615 756.97 1.620 1,000.00 54.322 8.832 790.36 8.861 1,000.00 54.322 4.830 771.85 4.846 1,000.00 54.322 2.317 763.69 2.323 1,000.25 43.193 1.166 757.01 1.169 1,000.25 43.193 7.442 793.43 7.460 1,000.25 43.193 4.027 773.61 4.036 1,000.25 43.193 1.620 763.74 1.626 1,002.50 34.001 2 756.28 0.813 756.82 0.813 757.77 0.817 1,002.50 34.001 3 796.06 6.206 796.64 4 774.20 3.243 774.76 6.213 797.74 6.237 1,002.50 34.001 3.246 775.78 3.258 1,002.50 34.001 values. The reference temperature for the temperature de- 3 and 4, and so forth. Thus, the coefficients for the calcupendence (Tref) has been set to 20 ° C by SBRC. The vari- lation of these pairs of temperatures should be identical in able Ka and the constant Tref are used in (3), (5), (6), and the following sections. (7). The K3 values for the eight SeaWiFS bands are listed in Tables 10 and 11. 8.2 Data Conversion The temperature corrections require knowledge of the temperatures of the focal plane assemblies. The convervoltages from SeaWiFS. sion of the engineering information in the scan lines from SeaStax into focal plane temperatures requires two steps (Sections 8.2 and 8.3). Since there is only one temperature sensor per focal plane assembly, the derived temperatures In this section, the term voltages is used to refer to 8.2.1 SeaStar Counts to Voltage The voltages for every analog telemetry output from for bands 1 and 2 will be identical, as will those for bands SeaWiFS range from 0-5.11 volts. These voltages are trans- 19

Original page 22

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization as8 bit binarywords. 1. VT is the output from the focal plane temperature mittedto the groundby SeaStar measurements sensor, in volts. SincethecountsfromtheSeaWiFSradiance aretransmittedin 10bit binarywords,theanalogteleme- 2. K5 is the spacecraft analog-to-digital (AD) convertry valuesarealsopackagedin 10bits. However,forthe sion factor, in volts per count. analogtelemetryvalues,onlyeightof those10bits per 3. is an 8-bit digital word in the SeaStar telemewordareused. Ctemp try, in counts. dependence 4. K6 is the offset in the spacecraft AD conversion, in Table 10. Valuesin thetemperature equations.Thistablecontainsthetemperaturedependencycoefficients(K3) andthe coefficientsto convertOSCtelemetrycountsto temperaturesensotoutputvolts(Ks andK6). Band K3 K5 K6 [(o C)-X] [volts/count] [volts] 1 0.000901 0.020 0.0 2 0.000585 0.020 0.0 3 0.000420 0.020 0.0 4 0.000390 0.020 0.0 5 0.000391 0.020 0.0 6 0.000151 0.020 0.0 7 0.000106 0.020 0.0 8 0.000078 0.020 0.0 Table 11. Current output for the current source diodes (KT). These data are given for a temperature of 20 ° C. A correction must be applied to these data at other temperatures. Values are given for the prime and backup temperature sensors for each focal plane assembly. The backup sensor for bands 3 and 4 is ino )erative. Band K7 (prime) K7 (backup) [mA] [mA] 1 0.493 0.484 2 0.493 0.484 3 0.492 Inoperative 4 0.492 Inoperative 5 0.491 0.497 6 0.491 0.497 7 0.486 0.492 8 0.486 0.492 The SeaStar spacecraft will convert the analog voltages from SeaWiFS using an ADC with a 5.12 volt reference voltage. This will eliminate the need for scaling amplifiers between SeaWiFS and the ADC. Since an 8 bit binary sigvolts. 8.2.2 Voltage to Focal Plane Temperature The temperature detection circuit for each focal plane assembly uses a precision thermistor in parallel with a 16.2 Kohm resistor. The current through the thermistorresistor pair is provided by a current source diode that has a nominal output of 0.48 mA. The actual output of the current sources at 20 ° C are given in Tables 10 and 11. The output from the temperature sensor is the voltage across the thermistor-resistor pair as caused by the current from the diode. With knowledge of this voltage drop and of the current from the source, it is possible to calculate the effective resistance for the thermistor-resistor pair. With knowledge of the value of the resistor, 16.2 Kohm, it is then possible to calculate the actual resistance of the thermistor. From the thermistor resistance, the temperature of the focal plane assembly is derived using a conversion equation that is specific for the type of thermistor in the temperature sensor. There is a small, additional complication to this calculation. The output from the current source diode has a small temperature dependence. The correction for current source diodes requires an approximate temperature (TC) for the focal plane assembly. TC is good to about 2°C over the range from 5°C to 45 ° C. The details of these calculations will be presented in a series of steps below. The calculated focal plane temperatures are accurate to 0.3 ° C. This includes the uncertainty in the digitization of the temperature sensor voltages by the SeaStar spacecraft. 1. Calculate the approximate temperature, with TC = (5--VT) × 40/3, (9) where TC is the approximate focal plane temperature, in degrees Celsius. 2. Calculate the current from the current source diode, including its temperature correction, with nal ranges from 0-255 counts, each count will have a value ICS = K7-(0.0013 x (TC-20)), (10) of 0.02 volts, with 0 counts equal to 0 volts. This corresponds to temperature sensor voltages that range from 0- where ICS is the current from the current source diode, 5.10 volts. This is the basis for the coefficients K5 and K6 in mA; /<7 is the current from the diode at 20 ° C, in mA in Tables 10 and 11. The conversion equation for SeaStar (Tables 10 and 11); and (0.0013 x (TC-20)) is the temvoltages perature correction for the diode source, in mA. telemetry counts to SeaWiFS temperature sensor is linear: (8) resistor pair, with V T = K 5 X Ctemp q-K 6. The variables in (8) are defined as: 20 3. Calculate the effective resistance for the thermistor- RE = VT/ICS, (11)

Original page 23

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek whereRE is the effective resistance for the thermistorresistor pair, in Kohms. 4. Calculate the resistance for the thermistor, with RT = (16.2 x RE)/(16.2--RE), (12) where RT is the resistance of the thermistor, in Kohms. 5. Calculate the focal plane temperature from the thermistor resistance using the following conversion equation. T = -341 + 5398.94/(ln(254898 x RT)), (13) where T is the focal plane temperature, in degrees Celsius, and the constants are given values for the type of thermistor used in the temperature sensors. The variable T, given for each SeaWiFS band, is used in equations (3), (5), (6), and (7). The working range for these temperature calculations is about +5 ° C to +45 ° C, which roughly corresponds to the range 1.7-5.0 volts, or about 85-250 counts. 9. SPECTRAL RESPONSE SBRC's primary method for determining the spectral response of SeaWiFS is based on the measurement of the individual piece parts in the instrument's optical train. The transmission and reflection characteristics of the piece parts and the responsivity of the detectors are multiplied together to give the optical throughput of the eight Sea- WiFS bands at 1 nm intervals. The results of the calculations are given as amperes of electrical current from the photodiode per watt of radiant flux at each 1 nm interval. This throughput must be combined with a radiance source having the spectral shape of the sun to give the total output of each band for the full range of wavelengths. SBRC's secondary method is based on a system level measurement using a monochromatic light source. This source employs a single monochromator to provide light input with a spectral width of about 1 nm and with a known radiant flux at each wavelength. In this method, the output of the photodiode is measured, in amperes, for a known radiant flux from the monochromator, in watts, at each 1 nm interval, allowing the calculation of the throughput of the instrument. This throughput is equivalent to that from the piece part measurements. For SBRC, the system level measurements give a double check of the piece part calculations. The piece part values give the prime measure of spectral response. The system level measurements work reasonably well in the wavelength range where there is at least a minimal output from the band's photodiode. There are, however, regions outside of the pass bands of the radiometer's optical components where the throughput of the system is sufficiently close to zero, such that essentially no measurable current comes from the photodiode. In these wavelength regions, noise in the system level measurements dominates the results. The piece part results give much better values, since they are based on measurements of individual components, each having a small throughput. At the system level, the measurements in these regions include the product of these small throughput contributions, which creates an overall throughput at the system level that is immeasurably small. The following sections contain discussions of the spectral responses of the individual parts for each SeaWiFS band. These individual parts are shown in the aft optics schematic (Fig. 3). It is important to note that the piece part measurements in the following sections have been made of interference filters taken from the production run for the SeaWiFS flight filters. The piece part measurements have not been made with the actual flight filters. 9.1 Mirrors There are five mirrors in the optical train for SeaWiFS. Four of these are shown in Fig. 3. They are the primary mirror, the polarization scrambler, the half-angle mirror, and the collimator. The fifth mirror, a folding flat mirror (not shown) is located after the collimating mirror (Fig. 17 of Woodward et al. 1993). Each of these mirrors uses silver as the reflecting surface, and four of these mirrors have essentially the same spectral response. The reflectivity of one of these four nearly identical mirrors has been measured at several wavelengths in the laboratory by SBRC. Values between these wavelengths have been calcuiated by linear interpolation. Figure 8 (top) shows the reflectivity of one of the silver mirrors. As part of the modifications to reduce stray light in Sea- WiFS, the polarization scrambler was reworked to move ghosts, in the direction normal to the scan plane, onto the primary image. In this modification, the front surface of the optical plate over the silver mirror in the scrambler was tilted slightly, relative to the mirror. This change has given the scrambler a slight wedge shape. A spectral scan was made by SBRC of the reflectivity of the reworked polarization scrambler. This spectral response is also shown in Fig. 8 (top). The model for the combined reflectances of the five mirrors, including that for the polarization scrambler, is shown in Fig. 8 (bottom). At each wavelength, the reflectance for the silver mirror in Fig. 8 (top) is multiplied by itself four times and then multiplied by the reflectance of the polarization scrambler. This calculation accounts for the net reflectance of these five mirrors in series in the SeaWiFS optical train. 9.2 Lenses There are four sets of lenses in the SeaWiFS aft optics (Fig. 3). These sapphire lenses are spectrally flat over wavelengths from 300-1,000 nm, covering the region of the 21

Original page 24

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 1.0 0.8 (D c- O •_ O.6 r-- Symbols - the polarization scrambler £ Curve - a standard silver mirror _0.4 ¢- *5 I1) 0.2 0.0 f00' " '466'"'''"6_6'''"'"86o500 700 900 ] 000 11 O0 1200 .... ' .... ' .... ' .... ' Wavelength (nm) 1.0 0.8 "2" I/1 D C 0 ' 0.6 C E C 0 Curve - combined mirrors and polarization scrambler rV 0.2 o.o300 .... 4-00''" _6o"' ;56o"' 760.... 800' .... 900' .... 1000' .... 11 b6"" 1200 Wavelength (rim) Fig. 8. SeaWiFS mirror reflectances. The top panel shows measured values for a standard silver mirror and for the polarization scrambler. The bottom panel shows the reflectance model for the four mirrors and the polarization scrambler in series in the SeaWiFS optical train. 22

Original page 25

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek SeaWiFSmeasurements.Thelenseshavetransmittancesin SeaWiFS are designed without light absorbing compoof99%.Theyarenotincludedin thespectralresponsecal- nents. In regions where the reflectance is near 100%, the lightthat has transmittance is near 0%. culations.The lensesfocusthe collimated passedthroughthe dichroicsontothe focalplaneassem- Figure 9 (bottom) gives the reflectance curve for diblies.Theanglewith whichthis lightconvergesmustbe chroic 2. Its cutoff wavelength is between 443 nm (band 2) responseof and 490nm (band 3). Dichroics 1 and 2 keep most of the consideredin themeasurementsofthespectral thenarrowbandinterferencefilters. 9.3 Dichroics The SeaWiFS radiometer uses three dichroics. In the longer wavelength radiation from the Earth from reaching bands 1 and 2, which limits the wavelength range over which the interference filters need to work. For wavelengths above 1,140nm, the SBRC measurements of dichroic 1 (Fig. 9, top) gave reflectances lower SeaWiFS application, these dichroics transmit light for than the resolution of the laboratory equipment. For these wavelengths above a reference wavelength in the dichroic's design and reflect light below the reference wavelength. The three SeaWiFS dichroics perform a prefiltering of the radiance from the Earth, limiting the range of wavelengths that reach the interference filters on each focal plane. The dichroics and focal plane assemblies in SeaWiFS are shown in schematic form in Fig. 3. In the schematic, wavelengths, the laboratory readings were zero. The values from the dichroics contribute to spectral responses presented in the following sections. These results are presented as a set of logarithmic plots. As a result, the zero values for dichroic 1 in this technical memorandum have been replaced with amounts that are equal to the lowest reflectance in the rest of the dichroic measurements. Simthe light to the focal plane assembly for bands 7 and 8 ilar measures have been taken at other places where the (765 and 865nm) passes in a straight line through two dichroics. These dichroics are numbered 1 (nearest the collimator) and 3. The light for bands 5 and 6 (555 and 670 nm) is transmitted by dichroic 1 and then reflected by dichroic 3. In Fig. 3, the focal plane assembly for bands 5 and 6 is located on the far right. The focal plane assembly for bands 1 and 2 (412 and 443nm) is located behind the BG39 filter in Fig. 3. The light for this focal plane is reflected off dichroics 1 and 2. Dichroic 2 is located to the left of dichroic 1 in Fig. 3. The remaining focal plane assembly, for bands 3 and 4 (490 and 510nm), receives light reflected from dichroic 1 and transmitted by dichroic 2. This focal plane assembly is located on the left most side of Fig. 3. For each focal plane assembly, and its pair of Sea- WiFS bands, light is either transmitted or reflected by two dichroics. These dichroics play a major role in forming the shapes of the spectral responses of the eight SeaWiFS bands. This is particularly true in establishing the out-oh band response of the instrument. The performance specifications sent from SBRC to its filter manufacturer call for the use of the dichroics to remove the transmission spikes in regions well away from the pass band of the interference filters. This has limited the number of filter elements required for the interference filters and has helped increase the transmission of the interference filters to almost 100% at their center wavelengths. Figure 9 gives the spectral responses for the two dichroics in the optical train for band 1. Figure 9 (top) gives the reflectance curve for dichroic 1. The cutoff for reflection by this dichroic falls between 510nm (band 4) and 555 nm (band 5), separating the four shortest wavelength bands from the four longest. When viewing Fig. 9 (top), it should be remembered that a dichroic either transmits light or reflects it. The dichroics and the interference filters SBRC data set has values of zero. 9.4 Broadband Filters There are two pieces of Schott color glass filters in the SeaWiFS aft optics assembly (Fig. 3). These broadband filters were not part of the original SeaWiFS optical design. The two filters (BG39 and BG26 in the figure) were added to the instrument after the SeaWiFS critical design review. Both broadband filters transmit in the blue-green region of the visible spectrum and reflect in the red. The transmission curve for the BG39 filter is shown in Fig. 10 (top). This filter was added to the focal plane assembly for bands 1 and 2 to provide additional suppression of transmission spikes from the interference filters in the red. These spikes do not present a significant problem for the 412 and 443 nm bands on orbit. The broadband filter was added to aid in the radiometric calibration of bands 1 and 2 in the laboratory. The light source used for the radiometric calibration of the instrument does not have the same spectral shape as the sun. The sun, with the approximate spectral shape of a 5,900K blackbody, has a peak flux output at a wavelength of 491nm. The integrating sphere in the SBRC laboratory has the spectral shape of a 2,850 K blackbody and a maximum light output at 1,017 nm. This is true of all integrating spheres since all use tungsten lamps as light sources. Blackbody radiation is described in Section 9.8. In relative terms, the laboratory light source produces considerably more light in the red portion of the spectrum than in the blue-green. In relative terms, this excess of red light from the integrating sphere accentuates the outof-band transmission spikes from the interference filters. When using the integrating sphere, the so-called red leak for band 1 (412 nm) is about 30% of the transmitted light through the filter's pass band. 23

Original page 26

SeaWiFSPrelaunch Radiometric Calibration and Spectral Characterization 10 0.8 {D C O _ 0.6 C E 123 _ 0.4 c- O I'Y 0.2 0.0 300 400 500 600 700 800 900 1000 1100 1200 Wavelength (nm) 1.0 f 0.8 .&-. (D C 0 '_ 0.6 C E _0.4 C 0 0.2 0.0 ' I I I I | I I I I | I I I I | I I I I | | I a I i I I I I i I I I I | I I I ' I I I I ! I 3OO 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 9. Reflectance curves for SeaWiFS dichroics 1 and 2. These dichroics are in the optical train for bands 1 and 2 (412 and 443 nm). The top panel shows the reflectance curve for dichroic 1. This element separates bands 1 through 4 from bands 5 through 8. The bottom panel shows the reflectance curve for dichroic 2. This element separates bands 1 and 2 from bands 3 and 4. 24

Original page 27

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek Thesun(ora 5,900K blackbody)producesmuchmore data from 406-490 nm. These fitted results replace the labradianceflux in theblue-greenregionthanin thered.For oratory measurements over wavelengths from 380-490 nm. sunlight,theredleak through the 412 nm interference filter is less than 1%. As a result, the BG39 filter has little use to SeaWiFS on orbit. The filter has been of great utility in the laboratory measurements which are required before the instrument is placed in orbit. The BG26 filter has been added to the instrument to reduce the amount of light in the two longest SeaWiFS bands (765 and 865 nm). This reduced optical signal has helped the stray light characteristics for these bands. Although the BG26 filter has not been added to change spectral shapes in the instrument, its spectral effects must still be considered. The transmission spectrum for the BG26 filter is given in Fig. 10 (bottom). Both colored glass filters have been measured in the laboratory at SBRC. The measurements have been made at several wavelengths. Values between these wavelengths have been calculated by linear interpolation. With its position at the entrance to the focal plane assembly, this filter affects the transmission of all eight Sea- WiFS bands. Figure 10 (bottom), however, shows that the The measured values and the fitted curve are shown in Fig. 11 (bottom). The extrapolation of the fitted curve to 380 nm removes unwarranted structure in the out-of-band response of the bands below 400 nm, although the extrapolation creates some additional error in the instrument level spectral response curves below 400nm. This error, however, is a small part of values that are already close to zero. The spectral response calculations in this technical memorandum use the fitted curve from 380-490 nm. Elements of the SeaWiFS optical train, other than the photodiodes, transmit or reflect a fraction of the input radiance. In the calculation of the optical throughput of the instrument, they are dimensionless. It is the photodiodes that convert mW of radiant flux into mA of electrical current. The diodes integrate the radiant flux over the wavelength range where the optics have a non-zero response. For each channel, the electronics transform this current into voltage, amplify the voltage, and change the voltage into counts with an ADC. This process gives the fundamental radiometric conversion factor for SeaWiFS, mW of curve has very little spectral shape over the pass bands of radiance flux per count. the eight SeaWiFS interference filters. The BG26 broadband filter is spectrally flat over SeaWiFS band 1 and adds next to nothing to its spectral shapes. 9.5 Photodiode Responsivity The responsivity of the photodiode used in SeaWiFS, expressed in terms of mA of current out per mW of radiant flux in, has been measured at SBRC. These measurements are made for one of the diodes in the production batch used for the instrument. SeaWiFS employs 32 photodiodes, and it is impractical to measure the spectral response of each. As a result, there has been only one measurement at SBRC. The diode's response curve shows little structure over the 10 and 20 nm half-widths of the eight SeaWiFS bands (Fig. 11, top). There is, however, structure in Fig. 11 (top) around the 412 and 443nm wavelengths. This results from noise in the measurement at SBRC and not from the wavelength dependence of the efficiency of the photodiode. The SBRC measurements of the spectral response curve were performed using the same equipment as the system level test of the wavelength dependence from SeaWiFS. Light from a monochromator was used to illuminate a NIST calibrated photodiode and then to illuminate the SeaWiFS diode. The response curve in Fig. 11 (top) was calculated using these data and the NIST calibration curve for the reference photodiode. The monochromator used a tungsten lamp for its source, and the measurements had low light levels in the blue wavelengths and increased noise at wavelengths below 500 nm. This artificial structure at the blue end of Fig. 11 (top) was removed using a second order polynomial fit of the 9.6 Detector-to-Detector Differences For the ocean color instrument, the performance specifications address spectral uniformity for an instrument with multiple detectors. SeaWiFS has four detectors per band which are cut from a single piece of silicon substrate and lie in a single line. The individual detectors are 0.025 cm square with a 0.005 cm saw cut in the substrate between them. The total length of the detector array is 0.117cm. All four detectors lie under the same narrowband interference filter. Assuming a reasonable uniformity in the silicon substrate and in the interference filters, the spectral responses for the four detector elements in each band are assumed to be essentially identical. The response curve for each band is dominated by the band's interference filter, with the photodiode adding little to the band's spectral response. The specifications call for the central wavelength of each channel to be within +0.5 nm of the central wavelength of all four channels. The prelaunch acceptance report (Barnes et al. 1994) shows the SeaWiFS instrument to meet this specification. 9.7 Narrowband Filters The spectral shape for each SeaWiFS band is dominated by the band's narrowband interference filter. The response of this filter gives the principal definition of the following for each SeaWiFS band: the center wavelength; the band edges, from the full width at half maximum; and the extended band edges, from the 1% power points. The transmission curve for the 412 nm narrowband interference filter from 380-580nm is shown in Fig. 12 (top). This 25

Original page 28

SeaWiFSPrelaunchRadiometricCalibration and Spectral Characterization 1.0 0.8 q_ O °_ _o.6 E (.1 '-0.4 E ffl C 0 _-0.2 0.0 3o'''"""'"osoo40066d"''"86d""700 900 ....'....'....'100011O01200 Wavelength (nm) 1.0 0.8 I) C .9 o.6 q) E _ 0.4 0 ° E Ul C 0 _-0.2 0.0 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 10. Transmission curves for the SeaWiFS color glass filters. The top panel shows the transmission curve for broadband filter BG39. This filter forms a part of the optical train for bands 1 and 2 (412 and 443 nm). The bottom panel shows the transmission curve for broadband filter BG26. This filter is part of the optical train for all eight SeaWiFS bands. 26

Original page 29

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain,andT. Svitek 0.8 0.6 o E e_ E O = 0.4 E e_ 0 )0.2 0 ° 17 0.0 300 .... 400' .... 500''" _6o.... 700''" 6o.... 900' .... 1000' .... 11 O0' .... 1200' Wavelength (nm) 0.3 Symbols = meosured volues Curve = " 0 "----0.2 E (/} 13. E 121 ° E _0.1 0 -10 0 0.0 36_llltlllllllllllllllllllllllllltlllllilllllllllllillllllllllllllltllll380 400 420 j- /- /f 440 460 480 500 Wovelength (nm) Fig. 11. Spectral response curve for the SeaWiFS photodiodes. The top panel shows the measured response curve for the diodes from 380 to 1150 nm. The bottom panel shows the measured values for the photodiodes and the fitted curve from 380 to 550 nm. The spectral response calculations in this technical memorandum use the fitted curve from 380 to 490 nm. 27

Original page 30

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 10 0.8 o] C O #o.6 E 123 Q) c0.4 0 E r'o _-02 0.0 v I I ! I I I I I I i i I ! f I J i ! I I I I I I I I I I t w I i I l I 1 w I I 1 I I I I r I i I 80 400 420 440 460 480 500 520 540 560 580 Wavelength (nm) 1.0 -- 0.8 - C 0 (f) c0.6 - E c5 '-0.4 • E ffl 0 -0.2 0.0 ..... J,,,, ,,,,,,, :.,, .... , .... , .... , .... , .... , 300 400 500 600 700 800 900 1000 11 O0 1200 Wovelength (nm) Fig. 12. Transmission curve for the 412nm narrowband interference filter. Vertical scale in linear units. The top panel shows the transmission curve from 380-580 nm covering the pass band of the filter and the out-of-band response near 500 nm. The bottom panel shows the transmission curve for the filter from 380 to 1150 nm showing the complete set of out-of-band responses. 28

Original page 31

R.Barnes,A. Holmes,W. Barnes,W. Esaias, C. McClain, and T. Svitek ,i01 -" t- O C E 10 -2 .-- _ t::) 2 E 10 -5_ t- O b-- 4 10 300 Fig. 13. Transmission curve for the 412nm filter. The vertical scale is given in logarithmic units. This accentuates the out-of-band response of the filter. curve, with its linear ordinate, shows the features and the shape of the pass band of the filter. Figure 12 (top) also shows the out-of-band transmission leak near 500 nm. Figure 12 (bottom) shows the response of the filter from The measurements of the SeaWiFS interference filters at SBRC have been made with laboratory equipment having a resolution of 0.0001 transmission units. There are data points in the test results where the transmission of 380-1,150 nm. The curve for the same filter is also given in the filters is less than that threshold. In the laboratory Fig. 13 with a logarithmic ordinate. Figure 13 shows the large and the small out-of-band components of the filter. Some of these peaks are substantial. The SeaWiFS optical results, these points have been given transmission values of zero. For this report, the zero values have been replaced with one-half of the resolution of the measurements, i.e., train for bands 1 and 2, however, contains dichroics and a at 0.00005 transmission units. This allows for plots of the BG39 broadband color glass filter to reduce these out-ofband components. The SeaWiFS radiometer uses a scrambler to remove any polarization that may be present in the input radiance to the instrument, although polarization is not a consideration in the measurement of the interference filters. The lenses between the dichroics and the focal plane assemblies (Fig. 3) convert the collimated light into converging light with a relative aperture (f-number) of 2. For this f-number, the distance between the lenses and the focal plane assemblies is twice the diameter of the opening in the lens assemblies. This converging angle has been included in the measurements of the narrowband interference filters in the laboratory. results on a logarithmic axis. For each SeaWiFS band, the dichroics substantially reduce the out-of-band transmission spikes. Figure 14 (top) shows the net result for the interference filter for band 1 (from Fig. 12, top) combined with the response curves for the band's two dichroics (from Figure 9, top and bottom). This figure uses a logarithmic ordinate to emphasize the characteristics outside of the filter's pass band. As shown in Fig. 14 (top), the out-of-band response is reduced significantly when compared with Fig. 13. A BG39 filter has been added to the instrument to further reduce the out-of-band response for the 412 nm channel. The inclusion of the BG26 and BG39 filter responses to Fig. 14 (top) give the result in Fig. 14 (bottom). The 29

Original page 32

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization out-of-bandresponseof band 1 above 600 nm is more than four orders of magnitude below the transmission peak at 412nm. Figure 15 shows the addition of the mirror and the photodiode responses to Fig. 14 (bottom). Figure 15 gives a shape of the sun better than tungsten lamps. These labthe spectral response of SeaWiFS band 1 (412nm) to spectrally fiat light source. At each wavelength from 380- 1 tegrating spheres using sets of tungsten lamps. In addi- 1,150nm, this theoretical source provides a radiance of mW cm -2 sr -1 #m -1. At each wavelength, the photodiode that the out-of-band response for the SeaWiFS bands be determined for a source with the spectral shape equivalent to the sun. There are xenon arcs and other laboratory sources that mimic the wavelength dependent shape of the blackbody oratory sources do not have the inherent stability of intion, these other sources cannot be easily adjusted over converts this radiance into picoamperes of current. The in- the range of intensities necessary for the calibration of tegral of this current over wavelength gives the total output of the photodiode, in nanoamperes, for such a source. the instrument. As a result, laboratory radiometric measurements of the SeaWiFS bands have been made using Figure 15 provides an intermediate step in the spectral re- sources with tungsten lamps and then calculated for a sponse calculations, since SeaWiFS views sources that have solar-equivalent spectral shape. specific spectral shapes, both in the laboratory and on or- A 5,900 K blackbody closely approximates the spectral bit. In addition, Fig. 15 marks the point in the spectral calculations where the measurements of the band edges, shape for the maximum anticipated radiances from the Earth found in the SeaWiFS performance specifications i.e., 50% power points, and extended band edges, i.e., 1% (Barnes et al. 1994). When the 5,900 K curve is normalized power points, are made to show compliance with the Sea- to the eight cloud radiances in the specifications (Fig. 16), WiFS specifications. The inclusion of the photodiode and the mirror responses to the data set in Fig. 15 has the effect of increasing the out-of-band response for band I relative to the response the relative standard deviation of the curve values from the cloud radiances is 5%, with the values from the curve falling low in bands 1, 2, and 6. A fine tuning of the temperature of the solar equivalent blackbody temperature to for the data sets in Fig. 14 (top and bottom). This can be other temperatures near 5,900K will have no significant seen in the relative size of the peak near 650 nm in Fig. 15 effect on the spectral response calculations for SeaWiFS. (bottom). The diode and mirror responses have enhanced this peak in Fig. 15 (bottom) between four and five times Figure 17 (top) shows the response of SeaWiFS band 1, 412 nm, to a source with the shape of a 5,900 K blackbody. relative to its size in Fig. 14 (top and bottom). The 650 nm The flux from the blackbody curve has been normalized at peak lies below the resolution of Fig. 14 (bottom), and is 412 nm to the typical saturation radiance for band 1 from four orders of magnitude below the transmission peak in the SeaWiFS performance specifications, 13.63 mWcm -2 Fig. 15 (bottom). 9.8 Blackbody Radiation For an idealized blackbody radiator, the radiance can be given as a function of temperature and wavelength. For the purposes of calculations, the equation can be used in the following form (Wyatt 1978): 1.191066 x l0 s (14) of SeaWiFS band 1 at the SeaWiFS saturation radiance L(A) = _5(el.43883x 104/)_T _ 1)' where A is the wavelength, in #m; T is the temperature, in kelvins; and L is the radiance, in Wm -2 sr -1 _m. The radiance values from (14) can be converted directly into units of mWcm-2 sr -1 #m -1, the units for SeaWiFS in- SeaWiFS radiometric calibration in the laboratory to the measurements. The radiance values from this equation crease at all wavelengths with increasing temperature, and the peak of the curve shifts toward shorter wavelengths signifi- spectrally fiat source with a radiance of 1 mWcm -2 sr -1 with increasing temperature. These two effects are are mum -1. The two blackbody radiance curves in Fig. 17 cant, since the laboratory measurements for SeaWiFS having use the same units. This allows a direct calculation of the made with light sources that use tungsten lamps ap- response of SeaWiFS band 1 to 2,850 K and 5,900 K sources blackbody temperatures near 2,850K, while the sun blue that have a radiance of 13.63mW at 412 nm. Figure 18 proximates a 5,900K source for wavelengths in the this shows the differences between the output of SeaWiFS band and the visible (Allen 1973 and Warneck 1988). For require 1 when exposed to 2,850 K and 5,900 K blackbody sources. reason, the SeaWiFS performance specifications 3O sr -1 #m -1 (Barnes et al. 1994). Figure 17 (bottom) shows the same response but to a 2,850 K blackbody normalized at 412 nm. In addition, the vertical scale for Fig. 17 (bottom) is 15 times greater than that for Fig. 17 (top). The radiation curve for the 2,850K blackbody peaks at 1,017nm, far to the red of the normalization wavelength, 412 nm. The normalized 5,900 K curve in Fig. 17 (top) would appear as a small hump in Fig. 17 (bottom) The illumination level with the SBRC integrating sphere also illuminates this band with considerable light in the red region. At 1,000nm, the illumination is about 16 times greater than the illumination at the band's center wavelength. This is a factor that must be considered in the transfer of the calibration on orbit. The system response curves in Fig. 15 are given for a

Original page 33

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek 1 - 10 °' C 0 c E o 10 -2. c 0 E _ 10 -_- 8 I-z 10 -'" I | I I' i I' I I ! I I I I I I i i i i I i r i i I i l _ [ 1 ' i i i 1 i i i w I 300 400 5OO 60O 700 800 900 1000 1100 1200 Wavelength (rim) 1q I I 10 -': c 0 C E _ 10 -2. g 10 -3 _ - Z 10 -'i I i I l I I I I I I I I I I I 1 I I I I I I I I I I I _ I I I I I v I I 300 ''' 4OO 500 600 700 800 900 1000 11 O0 1200 Wavelength (nrn) Fig. 14. Combined transmission curves for the interference filter, dichroics, and broadband filters in SeaWiFS band 1. The top panel shows the combined curve for the interference filter and dichroics. Compared with Fig. 10, the out-of-band response has been reduced by about two orders of magnitude. The bottom panel shows the combined curve for the interference filter, dichroics, and broadband filters. The out-of-band response above 600 nm has been further reduced by two orders of magnitude. 31

Original page 34

SeaWiFSPrelaunchRadiometricCalibration and Spectral Characterization 10- 8 2 A o C O "O O rv __ 6 o _ 4 E O n," E _ 2 IZ) i 0 l i i i 1 I l v , I i , i i I i i i w I _ I i i I i s i i I I l i , I i i i i I ' i a v I ' i i , I 380 400 420 440 460 480 500 520 540 560 580 Wavelength (nm) IO__ v I0-'- O 0 n- 10 -L t" 0 10 -" {D g lo -" C}. In I 0 -5. E & 10 -" _o-'3oo ", _6o'....._66 6oo'......_o' 8oo'.... 9oo'......._o'o6 i1_6 _2'oo Wavelength (rim) Fig. 15. System response curve for SeaWiFS band 1. This includes the optical components and the photodiode response to a spectrally flat light source with a radiance of 1 mWcm-2sr -1 #m -1. The top panel shows the system response on a linear scale. The out-of-band response near 500 nm is barely noticeable. The bottom panel shows the system response on a logarithmic scale. 32

Original page 35

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek 6O + + O Lo E 5O E u 4O _r vi E ----30 ID i.I C O "U o 20 he" Symbols = SeaWiFS cloud radiances N Curve = Normalized 5900K _10 Z blackbody 0 I I I I i I I ! I J I I I l J 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 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 16. Radiance curve for a 5,900K blackbody normalized to the cloud radiances from the SeaWiFS performance specifications. The spectral response characteristics for the other seven SeaWiFS bands are calculated in a manner that is the same as the method that has been described here for band 1. The spectral responses of all eight bands to a 5,900K source are shown below (Figs. 19-24). The integrated output of the instrument for the 2,850K and 5,900K blackbody temperatures must be used in ratios to convert the laboratory measured radiance calibration for SeaWiFS to the in-flight calibration. For SeaWiFS band 1, the laboratory measurements will give a slightly greater number of counts per unit radiance than will the measurements on orbit. The results of these calculations for all eight SeaWiFS bands are given below. The performance specifications for SeaWiFS allow outof-band responses that are less than 5% of those within the extended band edges, which are defined by the 1% power points. These out-of-band response specifications cover the response of the instrument to a radiance source that mimics the sun. Calculations for all eight bands, using the 5,900K blackbody, are also given in Section 12. In addition, the band edges and extended band edges for all of the SeaWiFS bands are tabulated in Section 11. These values are calculated for a spectrally flat light source, a 5,900 K blackbody, and for a 2,850 K source. 9.9 Absolute Throughput Using additional information, including the collection aperture for the instrument and the instrument's field of view, SBRC has calculated the absolute throughput for the eight SeaWiFS bands. Using these data and the output of the instrument when illuminated by an integrating sphere, SBRC has calculated the sphere radiance based ultimately on the piece part curves. The worst comparison between the calculated and measured sphere radiances was for band 1 (412 nm). For this band, the output of the instrument was about 50% greater than expected in the calculations. Absolute instrument throughput is not investigated here. 9.10 Spectral Response Curves In previous sections, the spectral responses of SeaWiFS band 1 for three light sources have been given: spectrally flat, 5,900K blackbody, and 2,850K blackbody. The radiances from the two blackbody light sources were normalized to the expected saturation radiance for band 1 at 33

Original page 36

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 15 2 O E13 E 10 Offl E _ 8 U C O o 5 112 "O N Symbol = SeaWiFS typicaltiradiance at 412 nm E O Z o300' """'g_o'"400 '0_o.... 700 900 1000 I I O0 1200 "" _6o.... ' .... ' .... ' .... ' Wavelength (nm) Symbol = SeoWiFS typical radiance at 412 nm 225 - "E" 0 b 200 E 175 E 150 125 E _oo C:: O o 75 o-i ,,,, 50 -5 E I,,,. o 25z 0 i ! i i | i i i i I ! i i i I i i I i I ! i i 1 I 1 i I i I i i I i I i i i i I i i i i I 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 17. Blackbody radiance curves normalized to the saturation radiance at 412nm from the SeaWiFS specifications. The top panel shows the curve for a 5,900 K blackbody that approximates the solar spectrum. The bottom panel shows the curve for a 2,850K blackbody that approximates the SBRC integrating sphere. 34

Original page 37

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek Symbols = Instrument response to 2850K blackbody 150 Curve = Instrument response 125 75 E 5o- 252 ' t to 5900K bleckbody 0 " ' I i I _ ' I ' ' I ' I ' I I I Ill ' Illl _ I I;:II;:I I I I I I _iI I _ IIII'I I ........... _ I I 380 400 420 440 460 480 500 520 540 560 580 Wavelength (nm) Symbols = Instrument response to 2850K blockbody I0 _ Curve = Instrument response 10 2 10 "E E O 0 1 (9 Q. 10 -' 13. I 0 -2 10 -" 2 10 -4 I 0 -5 10 -6 300 '''1 400 .... 500I .... 600I'll 17(_6 to 5900K blackbody I' 8(_0 .... 900, .... 1000I.... 11 O0I.... 1200I Wavelength (rim) Fig. 18. Instrument output for the two blackbodies. The differences in the instrument's responses to the laboratory and flight sources are a factor in the on-orbit radiometric calibration of the SeaWiFS. The top panel shows the instrument output using a linear scale. This demonstrates the in-band differences from the two light sources. The bottom panel shows the instrument out-of-band differences. output using a logarithmic scale. This shows the 35

Original page 38

SeaWiFSPrelaunchRadiometricCalibration and Spectral Characterization 2OO Q. v O 15O t9 O m ,.I 0 0 or) Ln IO0 o c- O (31. 5O E A ....,....,.... .... .... 320 340 360 380 400 Wavelength (nm) 250 v "_ 200 u 0 t _ 150 0 0 t- O e,- E , T I ' ' ' i I t i _ , I 420 440 460 480 500 520 I i l _ I l , i I "_ i I f l I i I i I I v I I , I i , t I I f ' 4OO 420 440 460 480 Wavelength (nm) Fig. 19. Instrument output for the SeaWiFS bands. 500 520 54O 56O 580 600 These are the responses of the instrument to a 5,900K blackbody. The linear scale shows the in-band responses. The top panel shows Bands 1 and 2 (412 and 443nm). The bottom panel shows Bands 3 and 4 (490 and 510nm). 36

Original page 39

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek / 0 520 540 560 580 600 620 640 660 680 700 720 Wavelength (nm) 80- O. >., 3 0 60 U 0 0 0 u 4O 0 m C 0 n a:20 E ID 0 i i ' ' I ' ' i i I ' i , i 1 _ r , , 1 1 i , I ' i , i I i 1 1 i I ' i , ' 1 720 740 760 780 800 820 840 860 880 900 920 Wavelength (nm) Fig. 20. Instrument output for the SeaWiFS bands. These are the responses of the instrument to a 5,900 K blackbody. The linear scale shows the in-band responses. The top panel shows Bands 5 and 6 (555 and 670 nm). The bottom panel shows Bands 7 and 8 (765 and 865 nm). 37

Original page 40

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 10 . 0 o. 10 2. "0 o 10 ,,,e o 0 o 10 ° O5 if') 2 10 -2. g 0 -' ne 10 -" E "_ -5 _10 • 10 • I I ! I l ! ! I ! I I ! ! ! | I I I ! I ! ! _ I I ! I I I I I I ! ! | ! I ! I | l ! I I ] 3OO 400 500 600 700 800 900 1000 I 1O0 1200 Wavelength (nm) 10 5. . 10 2. >, o 10 0 0 o 10 - if3 _ 10 -2. ffl g 10 -_o:: 10 -'" E - o 10 • I I l 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 l 1l I I I l I I | I | 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 21. Instrument output for the SeaWiFS bands. These are the responses of the instrument to a 5,900K blackbody. The logarithmic scale shows the out-of-band responses. The top panel shows Band 1 (412am). The bottom panel shows Band 2 (443 nm). 38

Original page 41

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain,andT. Svitek 10 3 a. 10 2. o 10 1.9 0 0 i o 10 -_ if) _ 10 -2. g 10 -_ 10 -' E10 -5 1 0 -6. I I I 1 I I I I I I I I I I I I I I 1 I I 1 I I ] I I I I I I I I I I I I I 1 I I I [ I I 300 400 500 600 700 800 900 1000 1100 1200 Wovelength (nm) 10 3. o. 10 2. o 10 (9 0 0 0 10 -" if) : o 10-2_ g 10 "_ ID a: I0 -" E "N 10 -s. 3_ 10 -6_ 3oo'" 500 700 .... 800''" 1000.... 1100' .... 1200' Wovelength (nm) Fig. 22. Instrument output for the SeaWiFS bands. These are the responses of the instrument to a 5,900 K blackbody. The logarithmic scale shows the out-of-band responses. The top panel shows Band 3 (490 nm). The bottom panel shows Band 4 (510 rim). 39

Original page 42

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 10 - O c 10 2. "O o 10 .Q O -z 0 o I0 -' O_ o 10 -2. {D g 10 -_ {D r_ 10 -_ E >,I0 4. or) 10 -6. ! 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 300 400 500 600 700 800 900 1000 I 100 1200 Wavelength (nm) I03 ._ 10 o 10 0 0 0 O 10 -I o 2 10 -2. ID g 10 -3. Q. n,- 10 -_ E 10 -5. O_ 10 1 _ I I 1 I I | I I I I | I I I I | I" I I I l I I I I j I I I ! J I I I I J I I' I I | I I I I I 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 23. Instrument output for the SeaWiFS bands. These are the responses of the instrument to a 5,900 K blackbody. The logarithmic scale shows the out-of-band responses. The top panel shows Band 5 (555 nm). The bottom panel shows Band 6 (670 am). 40

Original page 43

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain,andT. Svitek 10 3. a_ 10 - "0 o 10 O 0 o 10-' (y) if') o 10 -2 (1) g 10 - a:lO • E "_ 10 -' >.. I 0 -6 J r I ; I ; _ F r I _ r , , I _ I , I I I I I I l I I I I I I i i i I i i i i I _ r i , I 300 400 500 600 700 800 900 1000 11 O0 1200 Wovelength (nm) 10 . alO 2o 10 ..Q O _ 0 1 - 0 o 10 -. (Y) o 10 -2 g 10 -'c 10-' E _ 10 -s 1 0 -8 , 1 r i I i , i v I i r i I I I T r I I i , i ! I I _ _ I I i i r I I I I I i I I t T I I 300 400 500 600 700 800 900 1000 1100 1200 Wovelength (nrn) Fig. 24. Instrument output for the SeaWiFS bands. These are the responses of the instrument to a 5,900 K blackbody. The logarithmic scale shows the out-of-band responses. The top panel shows Band 7 (765 nm). The bottom panel shows Band 8 (865 nm). 41

Original page 44

SeaWiFS Prelaunch Radiometric Calibration and Spectral Characterization the band's nominal center wavelength, 412nm. For the The in-band responses for SeaWiFS bands 7 and 8 are overview of the responses of the eight SeaWiFS bands in shown in Fig. 20 (bottom), and the out-of-band responses this section, the response of the eight bands to a 5,900 K blackbody source, a source that approximates the solar spectrum will be shown. For each of the bands, the 5,900 K radiance has been normalized to the expected saturation radiance at that band's nominal central wavelength. Figure 19 (top) gives the in-band responses for the 412 and 443 nm channels, and Fig. 21 gives their out-of-band responses. The plots in Fig. 21 show that the interference filters for the two bands exhibit similarities, especially evident in their out-of-band transmittances near 500-560 nm. The similarities between the filters seem reasonable, since their peak transmittances are only 30 nm apart. For bands 1 and 2, the components in the optical train---other than are shown in Fig. 24. These bands have been designed with twice the spectral widths of the other SeaWiFS bands. The small scale structure, i.e., the lack of smoothness in the peak of the curve for band 7 (765 nm) is an artifact of the measurement of the band's interference filter by SBRC. Interference filters of the type used in SeaWiFS exhibit a much smoother spectral response. Band 8 (865 nm) shows the greatest out-of-band leakage for the SeaWiFS set, both in absolute and relative terms. However, its 3.7% out-ofband response is well within the 5% maximum value in the SeaWiFS performance specifications. Figures 21-24 shows the output of the photodiodes for each SeaWiFS band in picoamperes per nanometer (from the interference filters--are the same, so the similarities in 380-1,150nm). The actual current from the photodiode Fig. 21 come from the filters, themselves. For band 1, and to a lesser extent for band 2, there has been a reduction in the blue side of the in-band response of the filter. This can be seen most easily in a comparison of the size of the left side transmission peak in the band 1 interference filter near 406 nm in Fig. 12 (top) with the corresponding system response peak in Fig. 15 (top) Both the mirror reflectances (Fig. 8, bottom) and the photodiode output (Fig. 11, top) act to trim the blue side of the output from the interference filter for band 1. Combined with the shape of the 5,900K blackbody response near 412nm (Fig. 17, top), these elements are responsible for the reduction in the portion of the peak nearest the blue end of the spectrum of band 1 in Fig. 19 (top). To a lesser extent, these elements are also responsible for the reduced is the sum of the band output over the entire wavelength range. The photodiode serves to integrate the instrument's spectral response. The integrals, i.e., summations, of the spectral responses of the eight SeaWiFS bands are given in Table 12. These summations are given for the responses to a 5,900K blackbody (Figs. 21-24) and to a 2,850K blackbody (not shown). In the calculations for each band, both sources give the saturation radiance for the band at the band's nominal center wavelength (for band 1, 13.63 mW at 412nm; for band 2, 13.25 mW at 443nm; and so forth). Table 12 also provides factors which allow conversion between the 2,850K light source used in the laboratory and a 5,900K source that mimics the sun. For band 1 (Table 12), the output of the photodiode in response to blue side peak in the response of band 2. For both bands 1 a 5,900K source is almost 4% less than the output in reand 2, the system responses are about one-half nanometer more to the red side than the responses of the interference filters. The in-band responses for SeaWiFS bands 3 and 4 are shown in Fig. 19 (bottom), and the out-of-band responses are shown in Fig. 22. As was seen for bands 1 and 2, the out-of-band responses for bands 3 and 4 show marked similarities (Fig. 22). This is particularly true for the transmission curves for the 490 and 510nm interference filters (not shown) above 800 nm. In this region, the interference filters transmit 70% to 80% of the incident radiation. However, the two dichroics in the optical path for these bands effectively remove most of this flux before it reaches the filters. Figure 20 (top) gives the in-band responses for the 555 and 670 nm channels, and Fig. 23 gives their out-of-band responses. Figure 20 (top) shows an out-of-band leak between 690 and 700 nm. This peak and a second out-of-band peak around 450 nm are more clearly seen in Fig. 23. In relative terms, the out-of-band response of band 5 is just over twice that for band 6. The calculations of the out-of-band sponse to a 2,850 K source. Correction factors, similar to the ones in Table 12, must be used in the transfer of the laboratory radiometric calibrations to orbit. The out-of-band responses for the eight SeaWiFS bands are part of the instrument's radiometric calibration. In that calibration, the instrument views a broad area of known radiance, and records the output from the bands in counts. The counts from each band include the outof-band contributions which are functions of the spectral shape of the source that is measured. The SeaWiFS laboratory calibration has the out-of-band correction for a 2,850K source factored into its results. If the instrument measures a source with that spectral shape, those measurements automatically contain the appropriate out-of-band corrections. The prelaunch calibration equations for SeaWiFS contain correction terms that convert the out-of-band responses from those for a 2,850K source to those for a 5,900K source. As a result, the SeaWiFS calibration equations now have the out-of-band correction for a 5,900 K source factored into them. The 5,900K spectral shape dupliresponses for the eight SeaWiFS bands are summarized in cates the spectral shape for SeaWiFS ocean measurements Section 12.0. 42 closely. The errors that arise from the use of the 5,900 K

Original page 45

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek Table12. IntegratedinstrumentresponsesfortheeightSeaWiFSbands.Theresultsaregivenfor responsesin nanoamperesto 2,850K and5,900K blackbodyradiancesources.Theradiancesforthe blackbodysourcesare normalizedto theexpectedsaturationradianceforeachbandat the nominalcenterwavelengthforeachband. Theconversionfactorgivesa fractionalmultipliertoconvertthelaboratoryinstrumentresponseusinga 2,850K blackbodyto the responseon orbit usinga 5,900K source.Theseresultsarecalculatedoverthe wavelength rangefrom380-1,150nm. Band Response to Response to Conversion Normalization Normalization 2,850 K Source 5,900K Source Factor Wavelength Radiance [nA] [nA] [nm] [mWcm -2 sr-1 #m -1] 2.275 2.190 0.963 412 13.63 3.493 3.435 0.983 443 13.25 4.424 4.336 0.980 490 10.50 4.624 4.613 0.998 510 9.08 3.749 3.717 0.991 555 7.44 2.069 2.092 1.011 670 4.20 2.875 2.859 0.995 765 3.00 2.249 2.274 1.011 865 2.13 out-of-band corrections for ocean measurements are esti- Fig. 25 and normalizing the result to 100%. This gives the mated to be small, or a few tenths of a percent. If an alter- system level response to a light source that is spectrally nate out-of-band correction is to be used, then the 5,900 K flat over the wavelength range from 380-1,100 nm. For the correction must be removed from the measurement results piece part measurements, the responses to a spectrally flat and a new out-of-band correction inserted in its place. source have also been calculated (Fig. 15). Data from the SeaWiFS instrument in the system re- 10. SYSTEM LEVEL RESPONSE sponse tests has been taken from the output of the preamplifiers for each band. These voltages have been ampli- System level measurements of the spectral response of fied with a lock-in amplifier and digitized with a 12 bit SeaWiFS were made using a 0.5 m monochromator, illu- ADC. This modification to data acquisition gives the tests minated with a 100W halogen lamp, as a light source. a sensitivity to low light levels that is orders of magnitude The slits on the monochromator were adjusted to give a better than the resolution of the standard digital output spectral resolution for the source (full width at half maxfrom the sensor. Piece part calculations, however, still give imum) of 0.9 nm at 546 nm. The wavelength accuracy of results that are one to two orders of magnitude more sensithe monochromator was checked using five emission lines, tive than the minimum detectable limit of the system level in first and second order, from a mercury lamp. measurements. The relative output energy from the monochromator The piece part and the system level spectral response was measured using a photodiode with a known quantum curves for the eight SeaWiFS bands are shown in Figs. 26 efficiency, i.e., with a known number of mA of current out and 27. The system level measurements have been made per mW of radiant flux in. The output of this photodi- over 60 nm wavelength ranges, centered for each band's ode, after amplification and conversion to mW of radiant nominal pass band. In Figs. 26 and 27, the piece part energy, is shown in Fig. 25. No effort was made to deter- wavelength ranges have been trimmed to the same ranges mine the complete set of geometric (goniometric) factors as the system level results. For this reason, the spectral in the illumination of the calibrated photodiode, such as curves for bands 7 and 8 float above the horizontal axis the area of illumination of the diode and the solid angle in Fig. 27 (bottom). The system level measurements for of the illuminating light. Figure 25 gives relative values, bands 7 and 8 do not come down to the abscissa. only. Since the comparisons in this section are of spectral The band edge values from the piece part and system responses normalized to 100%, the calculation of absolute level measurements are given in the next section. All of the radiances is not necessary. band edges, which give the full width at half maximum for In previous sections, the calculated piece part results the spectral response, are within 1 nm for the two sets of for the SeaWiFS bands were convolved with the spectral measurements here--with the exception of two, the lower shapes of blackbody sources to give the responses of the band edges for band 3 and band 7 (Fig. 26, bottom, and instrument to laboratory and solar spectra. For the com- Fig. 27, bottom). In all cases, however, each band edge parisons in this section, the shape of the halogen lamp and measurement is within the performance specifications for monochromator light source must be removed from the the instrument. system level measurements of the SeaWiFS bands. This is The shapes of the peaks for the two measurement sets done by dividing the output of the bands by the values in show differences. Requirements for these shapes are not 43

Original page 46

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 2OOO C o 1500 c 0 0 I000 0 o E £ 500 0 c 0 0 300 400 500 600 700 800 900 1000 1100 1200 Wavelength (nm) Fig. 25. Spectral shape of the light source for the system level spectral response Measurements. The curve gives the output of a photodiode with known quantum the monochromator. Geometric (goniometric) factors, not included in the calculation of the curve, so absolute graph. part of the SeaWiFS specifications. Both the piece part efficiency which is illuminated by the output slit of such as the illuminated area of the photodiode, are radiance values are not given in the ordinate for the For band 1 (Fig. 26, top), the shape of the system level and the system level results are within the requirements of spectral response resembles the transmission curve for the the specifications, and the measurement sets are final and will not be repeated. The most reasonable explanation for the differences is that the system level measurements have been made with the flight filters, and the piece part measurements have been made with filters from flight spares that have come from the same lot as the flight parts. During the thermal vacuum testing of SeaWiFS, in the 412nm interference filter (Fig. 12, top). It seems likely that the model for the mirror reflectances (Fig. 8, bottom) shows too sharp a reduction in reflectance at the lowest wavelengths. Such an effect would remove too much of the blue side of the piece part results in Fig. 26 (top). There are differences between the piece part and the system level response curves. The piece part results are spring of 1993, the focal plane assembly for bands 5 and 6 considered prime, and the system level measurements are was replaced due to an electrical problem. When the interference filter for band 5 was removed from the old focal plane assembly, it was chipped. A replacement filter was placed in the new assembly. The system level measurements for bands 5 and 6 were not repeated after thermal vacuum testing. As a result, the system level measurements for band 5 no longer represent the actual flight unit but represent a band with an interference filter that has come from the same lot as the flight part, like the piece part measurements. 44 considered backup checks. The differences between the two sets of measurements are sufficiently small, and the system level results verify the piece part results. The comparison of the out-of-band measurements for the eight bands are given in Figs. 28-31. The comparisons are given for the responses to a spectrally flat radiance source. This is the arrangement used to calculate the band edges to check agreement with the SeaWiFS performance specifications. Figures 28-31 show the limited dynamic range for the system level measurements. To improve the sensitivity of the

Original page 47

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek Curve - System Level Measurements Curve with Symbols - Piece Part Calculations 100- D k N E O Z q c 0 5 60 O C D O 40 C O Q. r E ID t_ O 320 360 400 440 480 520 Wavelength (nm) Curve - System Level Measurements Curve with Symbols - Piece Part Calculations 1 O0 - N O o 80 Z c 0 :6 60 0 rY c o 40u} c- O (3. rr 20 , JJ 0 i i i i i i i i v I i i i i i i I i i i i v i r I i i 400 440 480 520 560 600 Wavelength (nm) Fig. 26. Spectral responses for the SeaWiFS bands. The curves show the response of the bands to a spectrally flat source as measured at the system level and as calculated from piece part measurements. The responses are normalized to 100%. The top panel shows Bands 1 and 2 (412 and 443 nm). The bottom panel shows Bands 3 and 4 (490 and 510rim). 45

Original page 48

SeaWiFS Prelaunch Radiometric Calibration and Spectral Characterization Curve - System Level Measurements Curve w;th Symbols - Piece Part Calculations 1O0 (D /,4 .m E O 80 Z v U c 0 60 0 O_ c 0 40 E 0 Q. o) q) O_ 20 E >, 0 520 560 600 640 680 720 Wavelength (nm) Curve - System Level Measurements Curve with Symbols - Piece Part Calculations 100 J N -6 • E J 0 80 Z E 0 5 60- 0 r E D 0 40. E O Q.. (/1 ,! 20 _ E N ! 720 760 800 840 880 920 Wovelength (nm) Fig. 27. Spectral responses for the SeaWiFS bands. The curves show the response of the bands to a spectrally flat source as measured at the system level and as calculated from piece part measurements. The responses are normalized to 100%. The top panel shows Bands 5 and 6 (555 and 670 nm). The bottom panel shows Bands 7 and 8 (765 and 865 nm). 46

Original page 49

R.Barnes,A. Holmes,W. Barnes,W. Esaias,C.McClain,andT. Svitek 10 _ Curve - System Level Me(]surements "t3 Curve with Symbols - Piece Part Calculations N 2 =10 0 E L_ o 10 z _ 1 c 0 1 5 o 10 - c D 10 -2. 0 LO C 0 10 -'- {# EY E 10 - _ 10 -6 3OO II' 400l .... 500I .... 600I .... 700l .... 800l .... 900I .... 1000I .... 1100I'''112_ 0 Wavelength (nm) 10 3. A Curve - System Level Measurements "1:9 Curve with Symbols - Piece Pert Calculations .N 10 2- -6 E O 10 Z O) (D 1 c 0 =6 I0 - c I0 -2 0 10 -_ m C O 1 0 -' cF E 10 -5 N 1 0 -6 ]00 I 1 l l I I I I I I I 1 1 I I I I 1 I I I I I 1 I I F I I I I I I I 1 1 1 1 1 I I I l 1 I 400 500 600 700 800 900 1000 1100 1200 Wavelength (nm) Fig. 28. Out-of-band measurements for the SeaWiFS bands. The curves show the response of the bands to a spectrally flat source as measured at the system level and as calculated from piece part measurements. The responses are normalized to 100%. The top panel shows Band 1 (412 nm). The bottom panel shows Band 2 (443 nm). 47

Original page 50

SeaWiFS Prelaunch Radiometric Calibration and Spectral Characterization 10 j- Curve - System Level Measurements Curve with Symbols - Piece Part Calculations .N TO 2- -5 O I0 Z U I C 0 =6 10 "' 10 -2 2 10 -- I g 10 - 10 "- t/) 1 0 -6__ 300 .... 400' .... 500' .... 600' .... 700' .... 800' .... 900' .... 1000' .... 1_'o6" '#oo Wavelength (nm) Curve - System Level Measurements Curve with Symbols - Piece Part Calculations .- 10 2 -6 E o 10 z O /f -6 a_ 10 -' D 10 -2 0 10 -5 (_ C O -4 _o cr -5 -6 mlo '1 i l i I i t , ''I l _ i , I i T I I I ' I I I I I I I I I I l I I I _ T f , I l F , T I 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 29. Out-of-band measurements for the SeaWiFS bands. The curves show the response of the bands to a spectrally fiat source as measured at the system level and as calculated from piece part measurements. The responses are normalized to 100%. The top panel shows Band 3 (490nm). The bottom panel shows Band 4 (510 nm). 48

Original page 51

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek 10 Curve - System Level Measurements CD Curve with Symbols - Piece Part Calculations .N 102 -6 E o 10 z u 1 c 0 c :D 10 -2 0 .)., ® 10 -_ c 0 10 -" 0,# c_ "E 10 -s @ >, c_ 10 -6 300 .... 400, .... 500, .... 600I .... 700_ .... 800, .... 900, .... 1000L .... 11 O0'''' 1'200 Wavelength (nm) 10' Curve - System Level Measurements "0 Curve with Symbols - Piece Part Calculations ¢) ._ 10 2 o o 10 z c 0 =6 a_ 10 -_ _ 10 -2 0 e 10 -_ C 0 _10 -' Q) E 10 -5 m 10 -6 i i , 1 I i i , , I T V i I I I 1 T i I r ) I V I I i ) e I i i I I I ' I I I ] , I i i l 300 400 500 600 700 800 900 1000 11 O0 1200 Wavelength (nm) Fig. 30. Out-of-band measurements for the SeaWiFS bands. The curves show the response of the bands to a spectrally flat source as measured at the system level and as calculated from piece part measurements. The responses are normalized to 100%. The top panel shows Band 5 (555 nm). The bottom panel shows Band 6 (670 nm). 49

Original page 52

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization 10 3. Curve - System Level Measurements Curve with Symbols - Piece Part Calculations D 10 2. -5 E O 10 Z 1 c 0 10 -I t- I 0 -2. o 10 - ffl C o 10 -' CK E 1 0 -5 IJ3 I0 -6 .300' ' ' :o.... 500' .... 600' .... 700 800 900 1000 1100 12 0 ' .... ' .... ' .... ' .... ' .... 'o Wavelength (nm) 10 3 Curve - System Level Measurements Curve with Symbols - Piece Part Calculations .N 10 2 -6 E o 10 Z c O 7O a 10 - c D 10 -2_ O 00 10 -_ c O _ 10 -'- C_ E 10 -" _ 10 -6. .500"'"' 400 .... .500' .... 600' .... 700' .... 800' .... 900' .... 1000' .... 11'o" 1'2b0 Wavelength (rim) Fig. 31. Out-of-band measurements for the SeaWiFS bands. The curves show the response of the bands to a spectrally flat source as measured at the system level and as calculated from piece part measurements. The responses are normalized to 100%. The top panel shows Band 7 (765 nm). The bottom panel shows Band 8 (865 nm). 50

Original page 53

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek system measurements, the lock-in amplifier was set to sat- the response between the extended band edges. The inurate at about 10% of the full-scale output for each band. strument is within these specifications for all eight bands. The lower limit for the system level measurements, for instance, the output from 600-1,100 nm in Fig. 28 (bottom), 13. SUMMARY has been set at one count in these figures. Where the actual output of the lock-in amplifier was zero or negative in that are specific to the spacecraft instrument. For example, the measurements, they have been set to one count in this there are small differences in the output of SeaWiFS from presentation. One count sets the resolution limit for the side to side of the half angle mirror. Most ground based system level measurements. instruments do not scan in the same manner as SeaWiFS. Above the resolution limit for the system level measure- In addition, SeaWiFS uses bilinear gains to allow high senments, there are structures in the piece part results that sitivity measurements of ocean-leaving radiances and low are not seen in the output from the lock-in amplifier. In sensitivity measurements of radiances from clouds, which general, these differences are at levels that are four to five are much brighter than the ocean (Section 3.0). orders of magnitude below the peak transmission of the bands. Again, the differences in the two sets of measurecommon to SeaWiFS and to other radiometers. The apments are sufficiently small that the system level results plication of these factors to the SeaWiFS calibration equaverify the piece part results. tions has been presented both for users of the data set from the satellite instrument and for researchers making 11. BAND EDGE WAVELENGTHS ground-based radiance measurements in support of Sea- WiFS. Ground based radiometric measurements must ac- The band edge measurements are given for the wave- There are aspects to the SeaWiFS calibration equations There are, however, many calibration factors that are lengths at which the instrument response equals half that count for many of these calibration factors. at the peak. The extended band edges give the points at spectral responses of the eight SeaWiFS bands. The diswhich the output is 1% of the peak. The results (Table 13) In particular, there has been a detailed discussion of the come from the data sets that make up the sets of figures in cussions have been developed to show the integrated specthis memorandum. The results from the piece part calcu- tral responses within SeaWiFS as functions of the shape of the radiant source measured by the instrument (Section lations are given for three source spectral shapes which are 9.10). These out-of-band responses of the bands contribute a spectrally flat source, a 5,900 K blackbody, and a 2,850 K to the laboratory calibration of SeaWiFS. For the instrublackbody. ment's calibration coefficients presented here, the out-of- Table 13 includes the band edge calculations for the band response for the SBRC integrating sphere has been piece part measurement of each interference filter, plus the replaced with the out-of-band response for a source with values from the system level measurements. For the system the spectral distribution identical to the solar flux (Seclevel measurements, the spectral shape of the monochrotions 7.2 and 7.3). Such spectral considerations may also mator light source has been removed. Thus, the system be important for experimenters making ground based ralevel measurements are equivalent to the measurement of diance measurements in support of SeaWiFS. piece part values using a spectrally flat light source. Since the data in the measurement sets are tabulated at one nanometer intervals, the values in Table 13 have been obtained by linear interpolation. The authors and editors wish to acknowledge the editorial assistance of Barbara A. Leroux, who worked on this document ACKNOWLEDGEMENTS 12. OUT-OF-BAND RESPONSE during her participation in the NASA/Goddard Prince George's County Teacher Intern Program. The values in Table 14 are based on the measurements shown in Figs. 21-24. The locations of the upper and lower extended band edges come from Table 13. All of the values in Table 14 come from the piece part calculations for the SeaWiFS bands as illuminated by a 5,900 K blackbody. This is the arrangement set forth in the SeaWiFS performance specifications (Barnes et al. 1994). Figures 21 24 give the response of the instrument in picoamperes per nanometer. Table 14 gives the following sums of these responses over three wavelength ranges: from 380 nm to the lower extended band edge; from the lower to the upper extended band edge; and from the upper ex- GLOSSARY AD Analog-to-Digital ADC Analog-to-Digital Converter CZCS Coastal Zone Color Scanner IFOV Instantaneous Field-Of-View NASA National Aeronautics and Space Administration NIST National Institute of Standards and Technology OSC Orbital Sciences Corporation SBRC Santa Barbara Research Center tended band edge to 1,150 nm. The SeaWiFS specifications SeaWiFS Sea-viewing Wide Field-of-view Sensor require that out-of-band responses be no more than 5% of SNR Signal-to-Noise Ratio 51

Original page 54

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization Table 13. Bandedges(halfmaximumwavelengths)andextendedbandedges(1%wavelengths)for SeaWiFSbands 1-4. The centerwavelengthis calculatedfromthe upperandlowerbandedges.Resultsaregivenfor threelight sources:spectrallyfiat, 5,900K blackbody,and2,850K blackbody.Resultsarealsogivenfor the interferencefilter only,andfor thesystemlevelmeasurementusingthe monochromatorasa lightsource. Band Nominal Lower Extended Lower Band Edges Band Edge Band Edge Into] [nm] [nm] 1 402-422 394.9 403.1 395.1 403.3 395.9 404.5 393.6 402.4 393.8 402.3 2 433-453 424.0 434.1 424.1 434.2 424.8 435.1 423.3 433.5 422.3 433.6 3 480-500 470.7 480.8 470.7 480.8 471.3 481.5 470.1 480.5 468.1 479.1 4 500-520 488.1 498.9 488.1 498.9 488.9 499.4 487.8 498.7 487.2 498.6 5 545-565 536.4 545.5 536.3 545.4 536.9 545.8 536.6 545.5 535.3 544.6 6 660-680 646.8 658.3 646.7 658.3 646.9 658.5 646.8 658.4 646.2 658.8 7 745-785 728.0 744.7 727.6 744.6 728.4 745.1 725.1 744.3 743.3 8 845-885 826.7 845.7 826.4 845.5 826.8 845.7 826.5 845.6 845.6 t Calculated from measurements of the narrowband interference Source Center Upper Upper Extended Wavelength Band Edge Band Edge Into] [nm] [nm] 413.2 423.3 433.4 Spectrally Flat 413.3 423.4 433.6 5,900 K 414.1 423.7 434.8 2,850 K 412.6 422.7 432.3 Filter Only t 413.0 423.7 433.8 System Level§ 443.9 453.7 463.7 Spectrally Flat 444.0 453.8 463.7 5,900K 444.6 454.1 464.3 2,850 K 443.5 453.5 464.6 Filter Onlyt 444.1 454.6 463.8 System Level§ 491.1 501.4 511.8 Spectrally Flat 491.1 501.4 511.8 5,900 K 491.6 501.6 512.3 2,850 K 490.8 501.2 511.3 Filter Onlyt 490.1 501.1 511.7 System Level§ 510.1 521.3 530.7 Spectrally Flat 510.1 521.2 530.7 5,900 K 510.5 521.5 531.1 2,850 K 509.9 521.0 532.9 Filter Onlyt 510.3 522.0 530.9 System Level§ 554.6 563.8 577.3 Spectrally Flat 554.6 563.8 577.2 5,900 K 554.9 563.9 577.9 2,850 K 554.6 563.8 577.1 Filter Onlyt 554.2 563.9 577.0 System Level§ 668.2 678.2 692.7 Spectrally Flat 668.2 678.1 692.5 5,900 K 668.4 678.3 692.9 2,850 K 668.3 678.2 692.8 Filter Onlyt 668.8 678.8 692.2 System Level§ 764.9 785.0 814.5 Spectrally Flat 764.6 784.6 812.9 5,900 K 765.1 785.1 815.6 2,850 K 765.0 785.7 814.2 Filter Onlyt 763.8 784.2 System Level§ 866.4 887.0 908.2 Spectrally Flat 866.1 886.7 907.5 5,900 K 866.5 887.2 908.4 2,850 K 866.2 886.9 908.0 Filter Onlyt 866.4 887.2 System Level§ filter only. § Calculated from system level measurements using a monochromator as the light source. Outside of the range of the system level measurements. 52

Original page 55

R. Barnes, A. Holmes, W. Barnes, W. Esaias, C. McClain, and T. Svitek Table 14. Calculated out-of-band responses for the eight SeaWiFS bands. The instrument responses are given as the output of the photodiode in picoamperes. The 5,900K radiances in the calculations are normalized to the expected saturation radiance for each band at the nominal center wavelength for each band. The upper and lower extended band edges come from Tables 7 and 8. These results axe calculated over the wavelength range from 380nm to ll50nm. Band Lower Lower In-Band Out-of-Band Extended Response Response [pA] Band Edge [nm] [pA] 1 3.38 395.2 2,175.35 2 9.59 424.1 3,418.80 3 6.48 470.7 4,301.14 4 17.32 488.1 4,586.23 5 39.14 536.6 3,631.84 6 12.66 646.7 2,071.19 7 10.17 727.3 2,818.97 8 66.36 826.4 2,191.97 TDI Time Delay and Integration SYMBOLS A0 Coefficient for linear term in scan modulation correction equation. B0 Coefficient for power term in scan modulation correction equation. C1 Measured value for the flight diffuser on a given scan line, in counts. C2 Measured value of the flight diffuser for the scan line immediately sequential to the first scan line used to measure the flight diffuser, i.e., $1, in counts. Upper Upper Out-of-Band Extended Out-of-Band Response Band Edge [nm] Response [pA] [%] 433.6 11.77 0.70 463.7 1.56 0.33 511.7 28.08 0.80 530.7 8.96 0.58 577.2 46.14 2.35 692.2 7.84 0.99 813.4 29.58 1.41 907.5 15.43 3.73 R1 Multiplier for mirror side 1. R2 Multiplier for mirror side 2. RE Effective resistance for the thermistor-resistor pair. RT Resistance of the thermistor. T Measured temperature of the focal plane assembly. TC Approximate focal plane temperature. Try; Reference temperature for the temperature dependence (20 ° C). VT Focal plane temperature sensor voltage output. REFERENCES Cdark Instrument dark restore value, in counts. Allen, C.W., 1973: Astrophysical Quantities, 3rd Edition. Atha- Cout Instrument output, in counts. Ctemp Temperature sensor output, in counts, represented lone Press London, 310pp. by an 8 bit digital word in the SeaStar telemetry. Barnes, R.A., and A.W. Holmes, 1993: Overview of the Seads Detector configuration datum. WiFS Ocean Sensor. SPIE, 1,939, 224-232. G1 Gain setting 1. Barnes, R.A., W.L. Barnes, W.E. Esaias, and C.L. McClain, G_ Gain setting 2. G3 Gain setting 3. G4 Gain setting 4. gs Gain selection datum. ICS Current from the current source diode. Biggar, S F., P.N. Slater, K.J. Thome, A.W. Holmes, and R.A. K1 Primary instrument sensitivity factor. K2 Gain factor. l': a Temperature dependence of detector output. Cebula, R.P., H. Park, and D.F. Heath, 1988: Characterization f<4 Scan modulation correction factor. K_ Spacecraft analog to digital conversion factor. K6 Analog-to-digital offset in spacecraft conversion. K_ Current from the diode at 20°C. 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. Barnes, 1993: Pre-flight solar-based calibration of Sea- WiFS. SPIE, 1,939, 233-242. of the Nimbus-7 SBUV radiometer for long-term monitoring of stratospheric ozone. J. Atmos. Ocean. Technol., 5, 215-227. L(A) Radiance. Evans, R.H., and H.R. Gordon, 1994: Coastal zone color scan- Lnadir Measured radiance at nadir. LS1 Measured radiance for mirror side 1. LS2 Measured radiance for mirror side 2. Herman, J.R., R.D. Hudson, and G.N. Serafino, 1990: An nscan Measured radiance at any pixel in a scan. Pxl Pixel number, i.e., the numerical designation of a pixel in a scan line. ner "system calibration": a retrospective examination. J. Geophys. Res., 99, 7,293-7,307. analysis of the 8 year trend in ozone depletion from alternate models of SBUV instrument degradation. J. Geophys. Res., 95, 7,403-7,416. 53

Original page 56

SeaWiFSPrelaunchRadiometricCalibrationandSpectralCharacterization Hooker,S.B.,C.R.McClain,andA.W.Holmes,1993:Ocean Vol. 5 ColorImaging:CZCStoSeaWiFS,Mar. Tech. Soc. J., 27, Mueller, J.L., and R.W. Austin, 1992: Ocean Optics Protocols 3-5. for SeaWiFS Validation. NASA Tech. Memo. 104566, Vol. 5, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard McClain, C.R., W.E. Esaias, W. Barnes, B. Guenther, D. En- Space Flight Center, Greenbelt, Maryland, 43 pp. dres, S.B. Hooker, B.G. Mitchell, and R. Barnes, 1992: Tech. Vol. 6 SeaWiFS Calibration and Validation Plan, NASA Firestone, Firestone, E.R., and S.B. Hooker, 1992: SeaWiFS Technical Memo. 104566, Vol. 3, S.B. Hooker and E.R. Eds., NASA Goddard Space Flight Center, Greenbelt, Report Series Summary Index: Volumes 1-5. NASA Tech. Maryland, 41 pp. Mueller, J.L., 1993: The First SeaWiFS Intercalibration Round- Robin Experiment, SIRREX-1, July 1992. NASA Tech. Memo. 104566, Vol. 14, S.B. Hooker and E.R. Firestone, Greenbelt, Darzi, M., 1992: Cloud Screening for Polar Orbiting Visible Eds., NASA Goddard Space Flight Center, Maryland, 60 pp. Warneck, P., 1988: Chemistry of the Natural Atmosphere. Academic Press, 757 pp. Memo. 104566, Vol. 6, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 9 pp. Vol. 7 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. 8 Woodward, R.H., R A. Barnes, C.R. McClain, W.E. Esaias, Hooker, S.B., W.E. Esaias, and L.A. Rexrode, 1993: Proceed- W.L. Barnes, and A.T. Mecherikunnel, 1993: Modelling the SeaWiFS Solar and Lunar Observations, NASA Tech. of ings of the First SeaWiFS Science Team Meeting. NASA Tech. Memo. 104566, Vol. 8, S.B. Hooker and E.R. Fire- Memo. I04566, Vol. 10, S.B. Hooker and E.R. Firestone, stone, Eels., NASA Goddard Space Flight Center, Green- Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 26 pp. belt, Maryland, 61 pp. Vol. 9 Wyatt, C.L., 1978: Radiometric Calibration: Theory and Meth- Gregg, W.W., F.C. Chen, A.L. Mezaache, J.D. Chen, J.A. ods. Academic Press, 200 pp. THE SEAWIFS TECHNICAL REPORT SERIES 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. 1, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 24 pp., plus color plates. Vol. 2 SeaWiFS: 1993: Analysis of Selected Orbit Propagation Models for Gregg, W.W., 1992: Analysis of Orbit Selection for Ascending vs. Descending Node. NASA Tech. Memo. Eds., II, S.B. Hooker, E.R. Firestone, and A.W. Indest, Eds., 104566, Vol. 2, S.B. Hooker and E.R. Firestone, Mary- NASA Goddard Space Flight Center, Greenbelt, Mary- NASA Goddard Space Flight Center, Greenbelt, land, 16pp. Vol. 3 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. Vol. 4 of GAC van, 1994: Case Studies for SeaWiFS Calibration and Val- McClain, C.R., E. Yeh, and G. Fu, 1992: An Analysis Sampling Algorithms: A Case Study. NASA Tech. Memo. 104566, Vol. 4, S.B. Hooker and E.R. Firestone, Mary- Space Flight Center, Greenbelt, Maryland, 52pp., plus NASA Goddard Space Flight Center, Greenbelt, land, 22 pp., plus color plates. 54 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, 17 pp. Vol. 10 Woodward, R.H., R.A. Barnes, C.R. McClain, W.E. Esaias, 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, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 26pp. Vol. 11 Patt, F.S., C.M. Hoisington, W.W. Gregg, and P.L. Coronado, the SeaWiFS Mission. NASA Tech. Memo. 104566, Vol. land, 16 pp. Vol. 12 Firestone, E.R., and S.B. Hooker, 1993: SeaWiFS Technical Report 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. Vol. 13 McClain, C.R., K.R. Arrigo, J. Comiso, R. Fraser, M. Darzi, J.K. Firestone, B. Schieber, E-n. Yeh, and C.W. Sulliidation, Part 1. NASA Tech. Memo. 104566, Vol. 13, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Eds., color plates.

Original page 57

R. Barnes,A. Holmes,W. Barnes,W. Esaias,C. McClain, and T. Svitek Vol. 14 Vol. 19 Mueller, J.L., 1993: The First SeaWiFS Intercalibration Round- McClain, C.R., R.S. Fraser, J.T. McLean, M. Darzi, J.K. Fire- 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, 60 pp. Vol. 15 Gregg, W.W., F.S. Patt, and R.H. Woodward, 1994: The Simulated SeaWiFS Data Set, Version 2. NASA Tech. Memo. Vol. 20 104566, Vol. 15, S.B. Hooker and E.R. Firestone, Eds., Hooker, S.B., C.R. McClain, J.K. Firestone, T.L. Westphal, NASA Goddard Space Flight Center, Greenbelt, Maryland, 42 pp., plus color plates. Vol. 16 Mueller, J.L., B.C. Johnson, C.L. Cromer, J.W. Cooper, J.T. McLean, S.B. Hooker, and T.L. Westphal, 1994: The Sec- Vol. 21 ond SeaWiFS Intercalibration Round-Robin Experiment, Acker, J.G., 1994: The Heritage of SeaWiFS: A Retrospec- 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. Vol. 17 Vol. 22 Abbott, M.R., O.B. Brown, H.R. Gordon, K.L. Carder, R.E. stone, F.S. Patt, B.D. Schieber, R.H. Woodward, E-n. Yeh, 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. E-n. Yeh, and Y. Ge, 1994: The SeaWiFS BiG-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. tive on the CZCS NIMBUS Experiment Team (NET) Program. NASA Tech. Memo. 104566, Vol. 2I, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 43 pp. Evans, F.E. Muller-Karger, and W.E. Esaias, 1994: Ocean Barnes, R.A., W.L. Barnes, W.E. Esaias, and C.R. McClain, 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. Vol. 23 Vol. 18 Barnes, R.A., A.W. Holmes, W.L. Barnes, W.E. Esaias, C.R. Firestone, E.R., and S.B. Hooker, 1994: SeaWiFS Technical Report Series Summary Index: Volumes 1-17. NASA Tech. Memo. 10.4566, Vol. 18, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, (in press). 1994: Prelaunch Acceptance Report for the SeaWiFS Radiometer. NASA Tech. Memo. 104566, Vol. 22, S.B. Hooker, E.R. Firestone, and J.O. Acker, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 32 pp. 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. 55

Original page 58

REPORT DOCUMENTATION Form Approved PAGE ova No.0Z04-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing 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 Reports, 1215 Jefferson Davis Highway, Suite 1204, Artin_ton, MA 22202-4302, and to the Office of Management and Bud_let, Paperwork Reduction Project 10704-0188/, Washington, DC 20503. 1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE October 1994 4. TITLE AND SUBTITLE SeaWiFS Technical Report Series Volume 23-SeaWiFS Prelaunch Radiometric Calibration Characterization 6. AUTHOR(S) 3. REPORT TYPE AND DATES COVERED Technical Memorandum 5. FUNDING NUMBERS and Spectral Code 970.2 Robert A. Barnes, Alan W. Holmes, William L. Barnes, Wayne E. Esaias, Charles R. McClain, and Tomas Svitek Series Editors: Stanford B. Hooker and Elaine R. Firestone Technical Editor: James 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 11. SUPPLEMENTARY NOTES 8. PERFORMING ORGANIZATION REPORT NUMBER 94B00144 10. SPONSORING/MONITORING AGENCY REPORT NUMBER TM-104566, Vol. 23 Robert A. Barnes: ManTech, Inc., Wallops Island, Virginia; Alan W. Holmes: Hughes Santa Barbara Research Center, Santa Barbara, California; Tomas Svitek: Orbital Sciences Corporation, Dulles, Virginia; Elaine R. Firestone: General Sciences Corporation, Laurel, Maryland; and James G. Acker: Hughes STX, Lanham, Maryland 12a. DISTRIBUTION/AVAILABlUTY STATEMENT Unclassified-Unlimited Subject Category 48 12b. DISTRIBUTION CODE Report is available from the NASA Center for AeroSpace Information, 800 Elkridge Landing Road, Linthicum Heights, MD 21090-2934, (301) 621-0390. 13. ABSTRACT (Maximum 200 words) Based on the operating characteristics of the Sea-viewing Wide Field-of-view Sensor (SeaWiFS), calibration equations have been developed that allow conversion of the counts from the radiometer into Earth-exiting radiances. These radiances are the geophysical properties the instrument has been designed to measure. SeaWiFS uses bilinear gains to allow high sensitivity measurements of ocean-leaving radiances and low sensitivity measurements of radiances from clouds, which are much brighter than the ocean. The calculation of these bilinear gains is central to the calibration equations. Several other factors within these equations are also included. Among these are the spectral responses of the eight SeaWiFS bands. A band's spectral response includes the ability of the band to isolate a portion of the electromagnetic spectrum and the amount of light that lies outside of that region. The latter is termed out-of-band response. In the calibration procedure, some of the counts from the instrument are produced by radiance in the out-of-band region. The number of those counts for each band is a function of the spectral shape of the source. For the SeaWiFS calibration equations, the out-of-band responses are converted from those for the laboratory source into those for a source with the spectral shape of solar flux. The solar flux, unlike the laboratory calibration, approximates the spectral shape of the Earth-exiting radiance from the oceans. This conversion modifies the results from the laboratory radiometric calibration by 1-4%, depending on the band. These and other factors in the SeaWiFS calibration equations are presented here, both for users of the SeaWiFS data set and for researchers making ground-based radiance measurements in support of SeaWiFS. 14. SUBJECT TERMS 15. NUMBER OF PAGES 55 SeaWiFS, Oceanography, Radiometer, Prelaunch Radiometric Calibration, Spectral Characterization, System Level Measurements 17. SECURITY CLASSIRCATION 18. SECURITY CLASSIRCATION OF REPORT OF THIS PAGE Unclassified Unclassified NSN 7540-01-280-5500 16. PRICE CODE 19. SECURITY CLASSIRCATION OF ABSTRACT Unclassified 20. LIMITATIONUnlimited OF ABSTRACT Standard Form 298 (Rev. 2-89)