DECEMBER 1993 S T A Y L O RStability of the Earth Radiation Budget Experiment Scanner Results for the First Two Years of Multiple-Satellite Operation W. FRANK STAYLORAtmospheric Sciences Division, NASA/Langley Research Center, Hampton. Virginia(Manuscript received 6 October 1992, in final form 15 April 1993)ABSTRACT Clear-sky albedos and outgoing longwave radiation (OLR) determined from Earth Radiation Budget Experiment (ERBE) scanners on board the earth radiation budget satellite and NOAA-9 spacecraft were analyzed forthree target sites for the months February 1985-January 1987. The targets were oceans, deserts, and a multiscenesite covering half the earth's surfhce. Year-to-year ratios of the monthly albedos and OLR were within the0.98 -1.02 range with a standard error of about 1%. The data indicate that ERBE scanner measurements werestable to within a few tenths of a percent tbr the two-year period.I. Introduction The goal of the Earth Radiation Budget Experiment(ERBE) is to provide a dataset to study the regional,zonal, and global radiation properties of the earth atdaily, monthly, seasonal, and annual time scales(Barkstrom et al. 1989). This radiation dataset mustbe highly accurate in order to detect and understandthe often small, yet important global climate changes.To meet the stringent accuracy goals set for ERBE, thesatellite orbits had to permit frequent coverage of theentire globe and the instruments had to provide highradiometric accuracy. Coverage errors are basicallyfixed by the number of satellites and their orbits (Harrison et al. 1983), while radiometric errors are determined by the design of the sensors and the degree towhich their calibrations are known in flight. All of theERBE instruments were extensively calibrated andcharacterized in a ground test facility; nonetheless, onboard calibration systems were provided to recalibratethe sensors and to validate their long-term stability inflight. A description of the ERBE sensors and theirground and onboard calibration systems is given byBarkstrom (1984). In the early 1960s, radiometers exposed to the harshspace environment often degraded at alarming ratesand the need for in-flight calibration became apparent.Consequently, onboard calibration systems weresometimes added that generally took the form of solardlffuser plates, incandescent lamps, or both, for shortwave channels and temperature-controlled blackbodiesfor longwave channels (Williamson 1977). Unfortunately, in-flight calibration systems can also fail, de Corresponding author address: W. F. Staylor, NASA / Langley Research Center, Mail Stop 420, Hampton, VA 23665-5225.grade, or become unusable for numerous reasons. Furthermore, their applicability is often questionable asthey are usually activated in special calibration sequences (as were the ERBE scanners) that are substantially different from their earth viewing modes.Consequently, Staylor (1986) selected several desertsites that were to be used as natural earth targets toprovide an additional check on the stability of theERBE scanners and to allow a comparison of the Nimbus-7 and ERBE scanner results. Earth targets havealso been used by Jacobowitz et al. (1984), Whitlocket al. (1990), Brest and Rossow (1990), and Staylor(1990) to evaluate the stability of several scanners. The use of earth targets inherently assumes that thetargets will reflect (shortwave) or emit (longwave) predictable quantities of radiation. Stable surfaces such asdeserts and oceans observed under similar solar andmeteorological conditions at some later time shouldproduce similar radiations. The present paper examinesthe stability of the albedo (ratio of reflected to incidentshortwave radiation) and OLR (outgoing longwave radiation) results for several earth targets during the twoyear period from February 1985 through January 1987,a period when both the earth radiation budget satellite(ERBS) and the NOAA-9 ERBE scanners were operational. ERBE shortwave and longwave scanner stabilities will be assessed from comparisons of monthlyaveraged, clear-sky target albedos and OLR with thosefor the same month the following year when similarsolar and meteorological conditions exist.2. Satellite measurementsa. Instruments Identical ERBE scanner instrument packages wereflown on the ERBS, NOAA-9, and NOAA-IO spacecraft.828 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 10Each of the ERBE scanners had three broadbandchannels with thermistor bolometer detectors thatmeasured reflected shortwave (0.2-5 ~m), emittedlongwave (5-50 #m), and total (0.2->50 ~tm) radiances with a nadir ground resolution of about 40 km.The instruments scanned from horizon-to-horizon in4 s using a view of space as a zero-radiance referenceclamp. Normally, the instruments were scanned in across-track mode for global coverage, but infrequentlywere slewed 90- to scan along track to obtain data formodeling purposes. Each of the ERBE scanners hadthree onboard calibration systems--the internal calibration module ( ICM ), shortwave internal calibrationsource (SWICS), and the mirror attenuator mosaic(MAM). The ICM provided an electrically heatedblackbody for calibration of the longwave and totalchannels. SWICS had a three-level tungsten lamp usedfor calibrating the shortwave channel, and the MAMwas basically a solar reflector device that was used forshortwave and total channel calibrations. A morecomplete description of the ERBE scanners and theironboard calibration systems is given by Kopia (1986). The ERBS scanner obtained useful radiance datafrom November 1984 until its failure in February 1990,the NOAA-9 scanner from January 1985 to January1987, and the NOAA-IO scanner from November 1986to May 1989. The present paper is based on scannerdata taken from the ERBS and NOAA-9 spacecraft forthe two-year period from February 1985 through January 1987. January 1985 ERBS and NOAA-9 data wereeliminated because both spacecraft scanners were inalong-track scanning modes for a substantial numberof days during the month. The NOAA-IO data wereeliminated altogether because they provided only threemonths of overlap data with ERBS and NOAA-9 duringthe two-year period of interest.b. Orbits The ERBS spacecraft was launched 5 October 1984into a 600-km altitude orbit with an inclination of 57-.ERBS is in a precessing orbit with a repeat cycle of72.8 days, meaning that it observes a given locationan average of 20 min earlier each day. The NOAA-9spacecraft was launched 12 December 1984, into an850-km orbit with an inclination of 99-. This was basically a sun-synchronous orbit with an initial ascending equatorial crossing time of 1430; however, theNOAA-9 spacecraft precessed an average of 20 minlater per year during the two-year period of interesthere. From the equator to about 30- latitude, both theERBS and NOAA-9 scanners observe all regions twiceper day, once during daylight and once during nighttime hours. Above 30-, ERBS daily observations of aregion increase substantially up to about 70- latitude,where useful ERBS measurements end. NOAA-9 alsoobtains increased daily observations beginning at about50- latitude, reaching a maximum of 14 times per dayat the poles (i.e., once per orbit).c. Albedo and OLR estimates Six sequential steps are involved in the conversionof the basic scanner shortwave measurements into albedos: 1 ) Earth-reflected shortwave energy is absorbedby the thermistor bolometer chip producing a voltagethat is converted to counts and transmitted to a groundcenter. There, the zero-radiance counts are subtractedand the difference is multiplied by a gain producing afiltered radiance, so denoted because the optical transmissions of the scanner telescopes are not unity norspectrally flat. 2) Filtered radiances are converted tounfiltered radiances (i.e., unity shortwave spectraltransmission) using the modeling work of Avis et al.(1984), which is a function of solar and viewing zenithangles, azimuth angle, and the viewed scene type (clear,overcast; ocean, land; etc.). 3 ) Unfiltered radiances areconverted into albedos using angular direction models(Suttles et al. 1988 ) that are also functions of the sameparameters given in (2). 4) ERBE divides the earthinto 10 368 regions, each covering a 2.5- latitudex 2.5- longitude area. The 10-120 scanner measurements obtained from a region during each satellite overpass are first converted to albedos (steps i-3 ) and thenaveraged to produce a regional albedo for the meantime of the overpass. 5) The 2-14 regional albedosobtained each day are converted into a daily albedousing diurnal models developed by Brooks et al.(1986). 6) Daily albedos for each day of a month areaveraged to produce a monthly albedo for whateversky conditions existed for that region. Steps 1-6 arealso performed for the edited cloud-free scenes providing a clear-sky monthly albedo. Steps involved in the conversion of the basic scannerlongwave measurements into OLR are very similar tothose for the shortwave. The major exceptions are thatthe unfiltering model for longwave radiances (Avis etal. 1984) is only a function of viewing zenith angle andscene type, and unfiltered radiances are converted intoOLR using limb-darkening models (Suttles et al. 1989).All of the procedures (including the detector gains)involved in the estimation of the albedos and OLR asdetailed above remained fixed during the two-year period of interest here. Clear-sky albedos increase and OLR decrease withsolar zenith angle, and, therefore, one should expectthat monthly values would have annual cycles that respond to the varying solar conditions. However, barringsubstantial changes in the surface, atmosphere, or detector gain, one should expect a good comparison between site albedos or OLR with those for the samemonth in previous or following years (hereafter referredto as year-to-year comparisons) when solar conditionsare identical. An examination of the estimation procedures suggests that only step 5 might produce differDECEMBER 1993 S T A Y L O R 829cnccs that do not actually exist for the year-to-yearcomparisons. This could occur if the sampling timesof the instantaneous measurements were substantiallydifferent from year to year and were input into diurnalmodels that were not error-free.d. Sampling time The ERBS spacecraft precesses about 20 min perday or 10 h per month. Averaged with the relativelyconstant NOAA-9 sampling time, the mean samplingtime for the ERBS-NOAA-9 combination changesabout 5 h per month. However, ERBS precessed fivecomplete cycles in 364.1 days, and during the remaining 0.9 day of the year, precessed to about an 18-minearlier sampling time. As stated previously, NOAA-9precessed an average of 20 min later per year. Hence,the mean ERBS-NOAA-9 sampling time for a givenday or month in t 985 was only I min different for thesame day or month in 1986. This fortunate orbitaltiming virtually eliminates diurnal modeling as a sourceof any differences that might occur for year-to-yearcomparisons.3. Sites The present search for dcsirablc target sites reliedon previous works by Staylor (1986, 1990). Staylorfound that desirable properties of a site included: 1 )long-term stability, 2) high albedos and OLR combinedwith low solar zenith angles, 3) uniformity, 4) lowcloudiness, and 5) sufficient size. One of the "sites"chosen by Staylor (1990) included all latitudinal zonesfrom 30-N to 30-S, which covered halfthc earth's surface (3456 ERBE regions) and will hereafter be referredto as thc 30-N-30-S site. Shown in Fig. 1, this sitecontains large areas of all major scene types (oceans,vegetated land, deserts, tropical forest, etc.) exceptsnow, which is restricted mostly to a tkw high-elevationregions in thc Himalayan and Andes mountains. The30- latitudinal boundaries for this site were specificallyset to reduce the effects that interannual differences ofsnow coverage might have on albedo and OLR comparisons. There was concern that if shortwave sensor degradations were detected from the 30-N-30-S site comparisons, it would not be possible to determine whetherthe sensor had degraded uniformly across its broadbandor had degraded only in a portion of its response spectrum. This is because the 30-N-30-S site is "multicolored" containing "blue" ocean, "green" vegetation,and "red" desert areas. The selection of additional target sites that emphasize the shorter- or longer-wavelength extrema of the scanner shortwave responsespectrum seemed appropriate. Top-of-the-atmosphere (TOA), clear-sky spectralalbedos for ocean and desert surfaces are given in Fig.2 for typical surface, atmospheric, and solar conditions(Suttles 1981 ). The spectral albedo for oceans is higherat the shorter wavelengths and lower at the longerwavelengths. Deserts have the reverse spectral characteristics, meaning that oceans and deserts might provide the best spectral extrema sites. Locations of the ocean and desert spectral sites areshown in Fig. 1. The ocean site is located in the PacificOcean and is bounded by 45-N-45-S and 180-230-E. This site covers almost 10% of the earth's surface (720 ERBE regions), and its surface is about 99.9%water. The desert site is a composite of a mostly desertarea located in the vast Sahara Desert (20--30-N, 0-30-E) and a desert-arid land area located in Australia(20--30-S, 120--150-E), and together they cover1.3% of the earth's surface (96 ERBE regions). Thesetwo desert areas are equal in size ( 10- x 30-) and indistance from the equator, meaning that the compositedesert site is "equatorially balanced" as are the 30-N30-S and ocean sites. By compositing the albedos fromthe two desert areas, better temporal and solar samplingare acquired for the larger single site than would havebeen obtained for the two smaller sites separately.4. Site al~bed~s and cLR Monthly, clear-sky, regional albedos and OLR wereobtained from ERBE S-4 data tapes, which are availablefrom the archive at the National Space Science DataCenter. Albedos and OLR for each of the three sitest 30-N/30osFiG. 1. Locations and boundaries of the 30-N-30-S, ocean, and desert sites.830 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 10.5 - nl < .3 - o I I I I .3 .5 .7 1.0 3.0 WAVELENGTH, gmFIG. 2. TOA spectral albedos for oceans and deserts. Abscissa scale linear with cumulative spectral solar flux.considered here are computed from the individual (963456) regional values that comprise the sites. Regionalalbedos (12-144) for each of the latitudinal zones (436) within a site were averaged and then weighted bythe zonal areas and TOA solar fluxes. This procedureproduced a true site aibedo in that it is the ratio of thetotal shortwave energy that is reflected from the site tothe total solar energy that falls on the site. MonthlyOLR fluxes were obtained in a similar manner exceptthey were not weighted by the TOA solar fluxes. Monthly, clear-sky albedos for the 30-N-30-S,ocean, and desert sites for the 24 months from February1985 through January 1987 are given in Table 1. A1bedos for the first data year (February 1985-January1986) are listed in the first column, the second datayear (February 1986-January 1987 ) in the second column, and the ratio of the second-year albedos to thefirst-year albedos are listed in the third column for eachof the three sites. All of the monthly, year-to-year albedo ratios are very near unity and range only from0.98 to 1.02. The monthly site albedos are plotted inFig. 3, and the 30-N-30-S and the ocean site are seento exhibit dual annual cycles that peak at the solsticesand valley at the equinoxes. Peaks occur when the sitemean solar zenith angles are at maximum values, andvalleys occur at minimum angles. The composite desertsite exhibits only a single annual cycle primarily because the higher reflected fluxes from the brighter Sahara Desert overwhelm those from the darker Australian Desert in the composite albedo calculations. Monthly, clear-sky OLR for the three sites are listedin Table 2 in the same format used for the albedos inTable 1. Similarly, all of the monthly, year-to-year OLRratios are very near unity and range only from 0.99 to1.01.5. Albedo comparisons between satellites Two desert sites were chosen by Staylor (1986) toserve as in-flight validation targets. The sites were 2.5-x 2.5- ERBE regions located in the bright Saudi Desert(20.0--22.5-N, 50.0--52.5-E; ERBE region number3909) and in the dark Gibson Desert (25.0--27.5-S,120.0--122.5-E; ERBE region number 6673) as shownin Fig. 1. Nimbus-7 scanner data (taken in 1978 and1979) were used to produce clear-sky, bidirectional reflectance models for each of the sites. The models express reflectance in terms of the solar and viewing zenith angles and were used to compute monthly albedoscompatible with those from ERBE. Monthly albedosfor the Saudi and Gibson deserts from ERBE (ERBSand NOAA-9 combined) are compared to those fromNimbus-7 in Fig. 4. The comparison is quite good asthe mean values are virtually identical (ERBE 0.1%greater) and the standard deviation is only 3% of themean value.TABLE 1. Clear-sky site albedos, ERBS and NOAA-9 combined. FirstMonth year30-N-30-S Ocean DesertSecond First Secondyear Second/first year year Second/firstFirstyearSecondyear Second/firstFebruary 0.1279 0.1259March 0.1270 0.1271April 0.1296 0.1282May 0.1329 0.1311June 0.1342 0.1348July 0.1328 0.1330August 0.1317 0.1308September 0.1285 0.1270October 0.1283 0.1267November 0.1278 0.1294December 0.1300 0.1308January 0.1320 0.1316MeanStandard deviation0.9844 0.1093 0.1076 0.9844 0.2619 0.2611 0.99701.0008 0.1093 0.1092 0.9991 0.2641 0.2661 1.00760.9892 0.1079 0.1083 1.0037 0.2813 0.2785 0.99010.9865 0.1107 0.1092 0.9864 0.2858 0.2828 0.98951.0045 0.1107 0.1127 1.0181 0.2897 0.2898 1.00031.0015 0.1131 0.1126 0.9956 0.2856 0.2841 0.99470.9932 0.1106 0.1097 0.9919 0.2768 0.2777 1.00330.9883 '0.1099 0.1077 0.9800 0.2679 0.2654 0.99070.9875 0.1094 0.1092 0.9982 0.2595 0.2571 0.99081.0125 0.1120 0.1130 1.0089 0.2520 0.2512 0.99681.0062 0.1127 0.1143 1.0142 0.2515 0.2510 0.99800.9970 0.1152 0.1140 0.9896 0.2561 0.2562 1.00040.9960 0.9975 0.99660.0091 0.0120 0.0057DECEMBER 1993 S T A Y L O R 8 31.30.28 .26ALBEDO .24 .14.12.10 30oN~OOS_ OCEAN F M A M J J A S O N D MONTH ~O.3. Monthly, clear-skysitealbedos..4.1.2 .3 .4 NIMBUS 7 ALBEDOFIG. 4. Comparisons of the ERBE and Nimbus-7 monthly,clear-sky albedos for the Gibson and Saudi deserts. All of the ERBE albedos presented thus far wereobtained from combined ERBS-NOAA-9 measurements (see step 5 ). The ERBE S-4 data tapes also contain atbedos computed separately for both the ERBSand NOAA-9 spacecraft scanners. Monthly albedos forthe Saudi and Gibson deserts from NOAA-9 are compared to those obtained from ERBS in Fig. 5. Again,thc comparison is quite good as the mean values arenearly identical (NOAA-9 0.4% greater) and the standard deviation is only 2% of the mean value.6. Discussion of results A basic assumption of the present work was that anychanges in the monthly, year-to-year, clear-sky sitebedos or OLR would reflect changes in the scannersensor gains. This assumption was based on the factthat solar conditions, viewing angles, sampling times,computational procedures, etc. were identical for thesame month the following year. At the bottoms of Tables I and 2, the mean and standard deviation of themonthly albedo and OLR ratios are given for each ofthe three sites, and the means ranged only from 0.994to 0.999 and the deviations from 0.3% to 1.2%. Themean and standard deviation of the ratios for all 36site-months combined is 0.9967 and 0.90%, respectively, for the albedo and 0.9968 and 0.47%, respectively, for the OLR. Lee and Barkstrom (1991) statethat based on the in-flight ICS, SWICS, and MAM calibrations, the ERBE sensor gains were stable to aboutthe 1% level for all channels.TABLE 2. Clear-sky site OLR (W m-2), ERBS and NOAA-9 combined. FirstMonth year30-N-30-S OceanSecond First Second Firstyear Second/first year year Second/first yearDesertSecond year Second/firstFebruary 286.3March 287.4April 288.2May 288.6June 287.6July 286.2August 288.0September 287.7October 288.0November 287.0December 286.8January 288.1MeanStandard deviation285.6 0.9975 280.0 281.5 1.0054 287.5287.9 1.0018 280.2 282.0 1.0065 287.4286.3 0.9935 279.9 279.3 0.9976 289.8286.7 0.9934 281.5 279.4 0.9923 288.2287.0 0.9981 281.3 281.9 1.0020 289.9286.3 1.0002 282.4 281.8 0.9980 292.2287.0 0.9966 285.1 282.9 0.9921 295.8286.6 0.9962 283.7 282.0 0.9941 293.9287.3 0.9975 283.0 282.9 0.9994 285.4287.1 1.0004 281.7 281.3 0.9987 284.8286.5 0.9991 281.8 282.4 1.0023 285.8287.4 0.9977 284.2 282.3 0.9933 285.1 0.9977 0.9985 0.0026 0.0049285.3 0.9923286.8 0.9981286.7 0.9893286.1 0.9928287.5 0.9915290.5 0.9942292.4 0.9885290.3 0.9879286.5 1.0040285.7 1.0032283.3 0.9914284.2 0.9970 0.9942 0.0054832 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 10 0 .1 .2 .3 .4 ERBS ALBEDOFIG. 5. Comparisons of NOAA-9 and ERBS monthly,clear-sky albedos for the Gibson and Saudi deserts. The author had no difficulty in detecting a 6% peryear degradation for the visible channel of the Advanced Very High Resolution Radiometer that was alsoon board the NOAA-9 spacecraft (Staylor 1990).However, the author now finds it difficult to concludefrom the data here that any consistent degradation ofthe ERBS-NOAA-9 scanner sensors occurred duringthe two-year period from February 1985 through January 1987. For instance, if only the last three comparison months were considered (November-January),one might conclude that the shortwave sensor gainshad slightly increased (0.3%). Essentially the same results were obtained for all three ERBS-NOAA-9 targetsites suggesting that r~o spectral response shifts occurredduring the period. The excellent comparisons of theERBE and Nimbus-7 and the NOAA-9 and ERBS albedo results for bright and dark deserts provide positiveties between these datasets. The albedo of the earth is about 0.3, and therefore,its global reflected flux averages about 100 W m-2. Iffor the sake of argument it were assumed that the ERBEsensors degraded 0.3%, this would amount to a shortwave flux error of 0.3 W m-2 and a longwave flux errorof 0.7 W m-2. The accuracy goal set for ERBE priorto launch was 1 W m-2 (Barkstrom 1984).7. Summary Monthly, clear-sky albedos and OLR determinedfrom ERBE scanners on board the ERBS and NOAA-9spacecraft were compared for three target sites for twoyears, February 1985-January 1987. The targets included a multicolored site (30-N-30-S) covering halfthe earth's surface, a blue ocean site (Pacific), and acomposite red desert site (Sahara/Australia). Monthly, year-to-year comparisons for all sites fellwithin the 0.98 to 1.02 ratio range for the albedos andwithin the 0.99 to 1.01 ratio range for the OLR witha standard deviation of about 1% for both. The dataindicate that the ERBE albedos and OLR were stableto within a few tenths of a percent for the two-yearperiod. For the Saudi and Gibson desert sites, plots ofthe ERBS-NOAA-9 (combined) versus Nimbus-7 albedos and the NOAA-9 versus ERBS albedos showedagreement to within a few tenths of a percent.REFERENCESAvis, L. M., R. N. Green, J. T. Suttles, and S. K. Gupta, 1984: A robust pseudo-inverse spectral filter applied to the Earth Radia tion Budget Experiment (ERBE) scanning channels. NASA TM 85781, 32 pp.Barkstrom, B. R., 1984: The Earth Radiation Budget Experiment (ERBE). Bull. Amer. Meteor. Soc., 65, 1170-1185. - E. Harrison, G. Smith, R. Green, J. Kibler, R. Cess, and the ERBE Science Team, 1989: Earth Radiation Budget Experiment (ERBE) archival and April 1985 results. Bull. Amer. Meteor. Soc., 70, 1254-1262.Brest, C. L., and W. B. Rossow, 1990: Radiometric calibration and monitoring of NOAA AVHRR data for ISCCP. Int. J. Remote Sens., 13, 235-274.Brooks, D. R., E. F. Harrison, P. Minnis, J. T. Suttles, and R. S. Kandel, 1986: Development of algorithms for understanding the temporal and spatial variability of the earth's radiation bal ance. Rev. Geophys., 24, 422-438.Harrison, E. F., P. Minnis, and G. G. Gibson, 1983: Orbital and cloud cover sampling analyses for multisatellite earth radiation budget experiments. J. Spacecr. and Rockets, 20, 491-495.Jacobowitz, H., H. V. Soule, H. L. Kyle, F. B. House, and the Nimbus 7 ERB Experiment Team, 1984: The Earth Radiation Budget (ERB) Experiment: An overview. J. Geophys. Res., 89, 5021 5038.Kopia, L. P., 1986: Earth Radiation Budget Expei'iment scanner in strument. Rev. Geophys., 24, 400-406.Lee, R. B., Ill, and B. R. Barkstrom, 1991: Characterization of the Earth Radiation Budget Experiment radiometers. Metrologia, 28, 183-187.Staylor, W. F., 1986: Site selection and directional models of desertsused for ERBE validation targets. NASA TP 2540, 12 pp.--, 1990: Degradation rates of the AVHRR visible channel for the NOAA 6, 7, and 9 spacecraft. J. Atmos. Oceanic. Technol., 7, 411-423.Suttles, J. T., 1981: Anisotropy of solar radiation leaving the earth atmosphere system. Ph.D. dissertation, Department of Me chanical Engineering, Old Dominion University, 180 pp. , R. N. Green, P. Minnis, G. L. Smith, W. F. Staylor, B. A. Wielicki, I. J. Walker, D. F. Young, V. R. Taylor, and L. L. Stowe, 1988: Angular radiation models for the earth-atmosphere system. Volume l--Shortwave radiation. NASA RP 1184, 144 PP. --, G. L. Smith, B. A. Wielicki, I. J. Walker, V. R. Taylor, 'and L. L. Stowe, 1989: Angular radiation models for earth atmosphere system. Volume II--Longwave radiation. NASA RP 1184, 84 pp.Whitlock, C. H., W. F. Staylor, J. T. Suttles, G. Smith, R. Levin, R. Frouin, C. Gautier, P. M. Teillet, P. N. Slater, Y. J. Kaufman, B. N. Holben, W. B. Rossow, C. L. Brest, and S. R. LeCroy, 1990: AVHRR and VISSIR instrument calibration results for both cirrus and marine stratocumulus IFO periods. NASA CP 3083, 141-144.Williamson, L. E., Ed., 1977: Calibration Technology for Meteoro logical Satellites. First ed. Atmos. Sci. Lab. Monogr., Ser. 3, U.S. Army, 139 pp. [Available from DTIC as AD A041 662.]