--James J. Butler (butler@ltpmail.gsfc.nasa.gov),
NASA Goddard Space Flight Center, Code 920.1, Greenbelt, MD 20771
--B. Carol Johnson (cjohnson@nist.gov), National Institute
of Standards and Technology, Optical Technology Division, Gaithersburg,
MD 20899
The second National Aeronautics and Space Administration (NASA)/National Institute of Standards and Technology (NIST) Earth Observing System (EOS)-sponsored spectral radiometric measurement comparison was conducted at NEC Corporation in Yokohama, Japan, and at the Mitsubishi Electric Corporation in Kamakura, Japan, November 5-15, 1996.
Radiance measurements were made by several participants on two integrating sphere sources. Participating institutions were Goddard Space Flight Center, NIST, the National Research Laboratory of Metrology (NRLM, Japan), and the University of Arizona (U of A). The two integrating sphere sources were used in the pre-flight radiance calibration of the Advanced Spaceborne Thermal Emission and Reflection radiometer (ASTER) Visible and Near Infrared Radiometer (VNIR) and the ASTER ShortWave Infrared Radiometer (SWIR). These instruments, along with the ASTER Thermal Infrared Radiometer (TIR), will be assembled onto the first EOS spacecraft, EOS AM-1, at Lockheed-Martin in Valley Forge, PA, in early 1997.
The measurement plan was coordinated by NRLM and the Japan Resources Observation System Organization (JAROS) with input from NIST and the EOS Project Science Office. Key issues to be addressed were similar to the first radiometric comparison experiment in August 1996 (Butler and Johnson 1996): measurement repeatability, evaluation of unknown systematic effects, and stability. Again, time constraints required that the number of sphere levels measured be a subset of those used during the calibration of the ASTER VNIR and SWIR.
The overall goals remained the same as at the first intercomparison: 1) compare the spectral radiance of the sphere sources as calibrated by the EOS instrument providers (i.e., NEC Corporation and Mitsubishi Electric Corporation) with that determined by NRLM and NIST using NRLM- and NIST-calibrated radiometers; 2) compare the spectral radiance determined by the participants from the other laboratories using the sphere sources as common targets; and 3) evaluate the findings in terms of measurement procedure and basic metrology.
Because the measurements at NEC were preceded by measurements by the same participants in February 1995 of the sphere source used to calibrate the ASTER VNIR (Sakuma et al. 1996), the results of the new measurements can be compared to the earlier work. Because NIST has not completed the EOS-sponsored portable radiometer for the spectral region required for the ASTER SWIR, the spectral radiance scale of the sphere source at Mitsubishi could not be compared to that determined by NIST.
At NEC, the 1-m diameter integrating sphere used in the radiometric calibration of the ASTER VNIR was measured at three different levels by four teams of researchers over a six-day interval. One level was measured on two different days. At Mitsubishi, the 1-m diameter integrating sphere used in the radiometric calibration of the ASTER SWIR was measured at four different levels by three teams of researchers over a four-day interval, and two levels were re-measured on different days.
At NEC, the participants included John Cooper from GSFC and Hughes STX Corporation, Carol Johnson from NIST, Stuart Biggar from the U of A, and Fumihiro Sakuma and Juntarou Ishii from NRLM. Participating instrumentation included the GSFC-EOS scanning single-grating monochromator measuring from 400 nm to 1100 nm, the NIST EOS Visible Transfer Radiometer (VXR), the U of A visible/near-infrared transfer radiometer, and three NRLM ASTER visible/near-infrared transfer radiometers.
With respect to the participating filter radiometers, the VXR has six image locations with separate interference filter/detectors at each location. The interference filters are narrow band (~10 nm). The U of A visible/near-infrared transfer radiometer uses a rotating filter wheel to alternately measure at selected wavelengths that correspond to those in the EOS Moderate-Resolution Imaging Spectroradiometer (MODIS), which is also scheduled for flight on the EOS AM-1 platform. The solid angle in the U of A transfer radiometer is determined by a pair of precision apertures that are separated by a fixed distance and located between the filter wheel and the detector system. The NRLM radiometers are separate units, each making measurements at or near particular ASTER bands. NEC calibrated the ASTER VNIR integrating sphere source from 400 nm to 1100 nm on October 28 and 29, 1996, using a variable temperature blackbody and a double grating monochromator. The radiance temperature of the blackbody was determined by comparison to a standard blackbody that is at the temperature of freezing copper.
Daily, a series of measurements was followed by the reporting of preliminary results. The typical measurement procedure was to turn on the ASTER VNIR sphere source to a given radiant level determined by the lamp voltages and the specific lamps illuminated, measure using the VXR, then measure using the other participants' radiometers, and then repeat the VXR measurement. In this manner, over the complete course of the comparison, the sphere was measured at the same level at least twice by NIST, and twice by all the participants for one sphere setting.
Koichi Suzuki of NEC adjusted the sphere lamp voltages according to NEC procedures, and an automated data acquisition system was used to record the voltage of the sphere lamps and the output of two monitor detectors. One monitor detector consisted of a small silicon photodiode mounted on the edge of the exit aperture of the sphere and the other monitor detector consisted of a radiation thermometer with a center wavelength of about 650 nm. The radiation thermometer measured the radiance at the center of the exit aperture along an optical axis that was about 45deg. from normal incidence.
One level was measured each day with all measurements taking place in the same clean room in Building 25 at NEC that had been used for the 1995 radiometric measurement comparison. However, for this comparison, the area around the integrating sphere had been enclosed with black curtains to shield the measurement area from sources of ambient light. The automation feature, the use of the monitor detectors, the construction of the dark area, and the calibration of the sphere over the broad spectral interval were recommended by NASA and NIST following the 1995 radiometric measurement comparison.
Preliminary results from the 1996 measurement comparison are very encouraging and indicate a scatter of 1% to 2% among the participating institutions and NEC.
The comparison participants packed their equipment during the afternoon of 11 November and traveled to Mitsubishi on 12 November. The equipment was unpacked and moved into the clean room before lunch, and measurements began in the afternoon. At Mitsubishi, the participating institutions, individuals, and instrumentation included GSFC (John Cooper), with a scanning single grating monochromator operated from 1200 nm to 2400 nm; the U of A (Paul Spyak and Stuart Biggar), with the U of A shortwave infrared transfer radiometer; and NRLM (Fumihiro Sakuma and Juntarou Ishi), with two ASTER shortwave infrared transfer radiometers. NIST (Carol Johnson) observed the measurements and documented the intercomparison. The design of the U of A shortwave infrared transfer radiometer is similar to the visible near-infrared instrument, except that a liquid-nitrogen-cooled indium antimonide (InSb) detector is used.
Radiance measurements were made on the ASTER SWIR sphere source on 12 to 15 November, with sessions for discussion of preliminary results interspersed among the measurement intervals. Mitsubishi attempted to calibrate the ASTER SWIR sphere source prior to the intercomparison, but problems with a germanium sphere monitor detector made these results suspect. The data supplied to the participants correspond to a previous calibration of the sphere performed in September 1995.
Instead of calibrating the sphere at each level to be used to calibrate the ASTER SWIR, Mitsubishi selects a single level for calibration and uses a filter/germanium photodiode that is fixed to the sphere wall to set the radiance to levels required for the various ASTER SWIR bands. The sphere radiance is adjusted by changing the number of lamps that are illuminated and/or the position of a variable aperture between the 1-m sphere and one of two small satellite spheres that are mounted on the main sphere. Unlike the sphere operation at NEC, the lamps are always operated at the same current and voltage. The output of the sphere monitor is recorded by a computer and is displayed on a computer monitor, but no data are recorded in electronic format for future reference. The Mitsubishi calibration method requires that the sphere monitor be stable and linear with radiant flux and that the spectral shape of the sphere be independent of the lamp configuration or the position of the shutter between the satellite sphere and the main sphere.
Each day consisted of a series of measurements at one or two radiance levels. The typical measurement procedure was to turn on the ASTER SWIR sphere source to a given radiance level, determined by which lamps were on and the position of the shutter on the satellite sphere, and measure using the NRLM radiometers, the U of A shortwave infrared radiometer, and the GSFC monochromator. For some sphere levels, NRLM made measurements before and after GSFC and U of A. Shigeki Akagi of Mitsubishi adjusted the satellite sphere shutter or the illuminated lamps until the germanium monitor detector gave the correct reading. Since the sphere parameters were not recorded automatically, Carol Johnson recorded the output of the monitor detector during the measurements by the participants, and periodically recorded the currents, voltages, and operating hours on the lamps.
In this fashion, the sphere configuration for the ASTER SWIR Band 4 (at 1650 nm) was measured on the afternoon of 12 November and the morning of 13 November. ASTER SWIR Band 5 (at 2165 nm) was measured on the afternoon of 13 November, Band 6 (at 2205 nm) on the morning of 14 November, and Band 9 (at 2395 nm) on the afternoon of 14 November. Regardless of the sphere configuration, the U of A made measurements at all of the channels in the shortwave transfer radiometer (from 1244 nm to 2463 nm); GSFC made measurements from 1200 nm to 2400 nm; and NRLM used both transfer radiometers, which had center wavelengths at about 1600 nm and 2200 nm.
On the last measurement day, 15 November, the measurement procedure was varied so that two levels could be measured by all participants in the morning followed by equipment packing and transport in the afternoon. Because the only difference between the sphere configuration for Band 4 and Band 5 is the position of the shutter between the two spheres, the configuration was changed in the middle of the measurement procedure for each of the participants. As a result, Band 4 was measured by all participants at least three times, Band 5 was measured at least twice, and Bands 6 and 9 were measured at least once.
Finally, a special test was devised to examine the change of spectral shape of the sphere radiance as a function of the position of the shutter between the two spheres. The preliminary results from 12 and 13 November indicated that this test would be useful, and on the afternoon of 14 November, the U of A and NRLM measured the sphere for a particular lamp configuration and for three settings of the shutter (i.e., open, half-open, and 90% closed).
The preliminary results for the entire experiment at Mitsubishi indicate that the stability as measured at the beginning and end of a measurement sequence using the NRLM radiometers was about 0.5%. The reproducibility for turning the sphere off and back on to the same level was about 1%, as measured using the NRLM radiometers. When the participants' results are compared to the Mitsubishi calibration values, the scatter in the results is up to 10%.
NIST is coordinating the data analysis from this radiometric measurement comparison through its Statistical Engineering Division. The participants have submitted all raw data files, copies of log sheets, and descriptions of radiometers. It is expected that examination of the raw data will lead to uniform procedures for comparison of results acquired with instruments with different spectral, temporal, and spatial resolutions. Recommendations in calibration metrology will also be made where appropriate.
Butler, J. J. and B. Carol Johnson, 1996: EOS Radiometric Measurement Comparisons at Hughes Santa Barbara Remote Sensing and NASA's Jet Propulsion Laboratory. The Earth Observer, 8 (5), NASA Goddard Space Flight Center, 17 - 19.
Sakuma, F., B. Carol Johnson, S. F. Biggar, J. J. Butler, J. W. Cooper, M. Hiramatsu, and K. Suzuki, 1996: EOS AM-1 Pre-flight Radiometric Measurement Comparison using the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) visible/near-infrared integrating sphere. SPIE-The International Society for Optical Engineering, Earth Observing System, W.L. Barnes, Editor, 2820, 184-196.