The EOSCERES and NISTAR broadband radiometers do not directly measure the same quantity, first because of the full day required for CERES to see every pixel on Earth compared to NISTAR's instantaneous view, and second because the flux measured by NISTAR bears little relation to the "top-of-atmosphere flux" data product from CERES. They share only the use of the same broad shortwave and longwave wavebands (CERES has no NIR channel).
Nevertheless, the barriers to using NISTAR as an integral test for CERES data products are surmountable. We think the effort to do so is worthwhile not because NISTAR is better calibrated (it isn't), but because NISTAR tests exactly those approximations which CERES must use to move from a single-time, single-angle view to a time- and angle-averaged view. While the CERES approximations are highly evolved, they necessarily involve assumptions which make the final data products much less certain than the underlying calibrated radiation measurements.
Via modeling and interpolation using auxiliary data sets and assumptions about diurnal scene variations (Young et al., 1998), CERES is producing a synoptic data set of albedo and OLR (outgoing longwave radiation) on a 1° grid every 3 hours. This synoptic data set can be convolved with scene-appropriate angular-directional models (ADMs) and summed over all Triana-viewed Earth pixels to simulate the NISTAR measurement at each synoptic time. The EPIC will be required for scene identification.
Any one such comparison would of course be hard to interpret. As a large catalog of such comparisons is accumulated, however, sharper conclusions may be drawn by subsetting the catalog in various ways. For example, each synoptic hour will provide a comparison over a particular part of the globe so that problematic geographical regions may be discernible.
Differences between the NISTAR and convolved CERES radiances will constitute an upper bound on the total error in the CERES global albedo and OLR because the Triana viewing geometry is in one of the most sensitive portions of shortwave ADMs. Flux errors from CERES will likely be less than the Triana-based estimate because the multiple angle view from the CERES scanners will average out many of the ADM errors.
If the Europeans successfully launch GERB, the first Geostationary Earth Radiation Balance experiment, in 2001, it should also be possible to test GERB against NISTAR when GERB views only daylit parts of the Earth.