DIURNAL BIAS IN CALIBRATION OF BROAD-BAND RADIANCE MEASUREMENTS FROM SPACE

We examine the problem of determining the separate shortwave (SW) and longwave (LW) components of the Earth radiation budget from space, Bec ause true broad-band longwave filters do not exist, daytime LW radianc e determinations can depend entirely or in part on subtraction of the measured SW radiance from the ''Total'' (TW) radiance involving integr ation over the entire electromagnetic pectrum, Examining radiances mea sured in the three channels (SW, imperfectly filtered broadband LW, TW ) of the Earth Radiation Budget Experiment (ERBE) scanners on board th e NOAA-9, ERBS and NOAA-10 satellites, we find small discrepancies in the daytime estimates of broad-band (''unfiltered'') LW radiances usin g the ERBE ''spectral correction'' procedure, We show that these resul t from errors (of order 2.5%) in the calibration of the SW channel and possibly in the spectral characterization of the SW and/or TW channel of the ERBE scanners on NOAA-9 and NOAA-10. Nighttime estimates show no such bias, and there appears to be no such error in the data from E RBS. Considering the LW radiant exitances determined from ERBE scanner data from the three satellites, we find systematic differences in ind ividual satellite estimates of simultaneous instantaneous regional mea ns and of regional monthly means, consistent with the radiance discrep ancies, instantaneous daytime LW estimates can be in error by 20% in t he extreme case of very bright cold cloud, and LW cloud radiative forc ing may be significantly biased, We consider the implications of these small SW-dependent errors on the determination of diurnal variation a nd of cloud radiative forcing in the longwave domain, We show how the ERBE estimates can be corrected; and consider how our procedures can b e used to validate results of future experiments (ScaRaB and CERES).