Abstract. Establishment of a uniform long-term record of "top-of-the atmosphere" (TOA) Earth radiation budget (ERB) components, on a scale appropriate to the study of cloud radiation interactions, requires that the data obtained from different observation missions satisfy two basic conditions: (1) the broadband shortwave (SW:0.2–4 µm) and longwave (LW: 4–50 µm) radiances must be demonstrably made on the same absolute scale; and (2) the methods used first to convert the instantaneous (filtered) radiances into (unfiltered) SW and LW radiant fluxes, and then to perform the space-time integrations to yield regional monthly means, must be consistent. Here we consider mainly the second point, with regard to the ScaRaB/Meteor mission in orbit since 25 January 1994 and observing the Earth since 24 February 1994. The objective of this mission is to determine the TOA ERB components and so to provide a continuation of the NASA ERBE scanner mission (November 1984–February 1990). We show how results compatible with ERBE can be obtained by taking into account the instrumental characteristics and the satellite orbit parameters: spectral response of the broadband channels, Earth local time of observation. Considering the spectral response of the ScaRaB broadband channels, we show that no spectral correction is required in the longwave domain, whereas a correction of +4.5% must be applied in the shortwave domain for clear and partly cloudy ocean, in order to compensate for underestimation at the shortest wavelengths. Despite possible differences between ERBE and ScaRaB procedures in values assumed for certain parameters of the scene/cloud identifications, application of these procedures to the same set of ERBE data (spectrally corrected, i.e. "unfiltered" radiances) shows that scene identification agreement is close to 90% and that, where there is disagreement, resulting differences in LW fluxes are negligible, those in SW fluxes small. We show that regional and global mean quantities are in excellent agreement, considering that differences between (ERBS+NOAA-9) and (NOAA-9 only) results may be taken as illustrating time-sampling effects. We find that biases may occur from the undersampling, specifically for the night-time clear-sky estimation over land and desert. Preliminary results using ScaaB data of March 1994 show that clear-sky regional estimates may be less numerous than in ERBE scanner products, due to either the larger pixel size or the auxiliary parameters used in the scene identification, and that expected uncertainties in the global monthly mean values depend mainly on the instrument radiometric calibration.
