The Earth Radiation Budget Experiment included a scanning and nonscanning radiometer package on the NOAA-9 and NOAA-10 spacecraft and on the dedicated Earth Radiation Budget Satellite (ERBS). The last spacecraft flew in an orbit of 57 degree inclination so as to precess through all local times with a 72-day period in order to gather information about the diurnal cycle of Earth radiation. Monthly mean maps of albedo and outgoing longwave radiation (OLR) are major data products from these radiometer instruments. The nonscanner package included wide field-of-view (WFOV) radiometers, which have operated
aboard the ERBS since November 1984, providing a data record of over 13 years to date. The OLR and albedo vary randomly due to synoptic effects and have cyclical variations which are denoted as diurnal cycles. The precession of the ERBS spacecraft causes sampling patterns which result in some regions having poor temporal sampling for computing the monthly-mean reflected solar radiation and OLR. Analyses have been developed for computing these temporal sampling errors based on a priori statistical descriptions of the synoptic and diurnal
variations. In this paper, the analyses are applied to compute temporal sampling errors of monthly-mean maps generated from ERBS WFOV measurements.
The analyses are first validated by use of monthly mean OLR and albedo data products based on combined NOAA-9 and ERBS WFOV data. Using sampling patterns from the two satellites is adequate (usually) for their combined product to be considered a "truth set". Zonal
discrepancies between NOAA-9 and ERBS data products compare well with the analytic error. Examples are shown for both good and poor temporal sampling cases, as determined by the orbital positions for the month and solar declination. With good sampling, the standard deviation of
OLR temporal sampling error is typically 3 to 4 W/m2, but with poor sampling, the errors grow to more than 12 W/m2. Time sampling errors tend to be somewhat higher for monthly mean albedo or shortwave flux. The analyses are now used for quality control, so that for a region which has a large temporal sampling error in monthly mean OLR or albedo, that data will be rejected