The missing data all occur for the Absorbed Solar Radiation (ASR) and thus net radiation near the delimiter of the incoming radiation where it is difficult to obtain an accurate albedo. To fill the missing data, we firstly took advantage of the fact that the areas missing varied from year to year and we formed a climatology of the available data on albedo. A least squares fitted first harmonic was then derived for each point, and the missing climatological mean values determined and these were used to fill in the missing albedo. Finally, a 9 point smoother was applied to the replacement data and to data within 2 grids of a replacement value. The ASR and net radiation were then derived. The mean annual cycle probably involves more than a single harmonic, but use of the least squares fitting approach with more than one harmonic could occasionally result in albedos for missing points that exceeded unity. Therefore the approach used is conservative, but it produces quite reasonable numbers. Because the missing points are weighted by small incoming radiation, the impact on the ASR is not very great, but it is desirable to do this step rather than treat the data as missing.
For the first two years of ERBE data, the global mean outgoing longwave radiation (OLR) mean was 233.9 W/m^2 but it jumped to 236.5 W/m^2 after the loss of NOAA 9. Presuming that the values with 3 satellites are superior, we have adjusted the OLR everywhere uniformly downward from February 1987 on, justifying this as a bias most likely arising from the diurnal cycle. The net imbalance in annual mean net radiation was initially 4.2 to 6.0 W/m^2, and after the first adjustment to OLR, this was corrected for by applying a decrease in the albedo uniformly such that the twelve month running mean radiation budget balances. At the beginning and end of the ERBE record, the twelve-month mean is one-sided rather than centered. Note that each month is adjusted separately, not the twelve-month mean. Consequently, the adjustment in the ASR is non-uniform, and instead is greatest where the radiation is largest, consistent with the view that the imbalance is most likely associated with sampling, especially of the diurnal cycle. Therefore, following these two fairly minor adjustments, the 12 month running mean net radiation for most months is close to, but not exactly zero. It ranges from -0.14 W/m^2 to +0.14 W/m^2. Real interannual variability in the net radiation is not accounted for.