Observational estimates of global top-of-atmosphere radiation on
monthly, seasonal, annual, and longer time-scales require estimates of
the diurnal variability in insolation and the asymmetry of surface and
atmospheric reflection. We compare EPIC and NISTAR observations from the
DSCOVR satellite with CERES hourly synoptic fluxes, which are filled
through geostationary observations, and find that a Fourier analysis of
these data substantially agree, showing strong relative power at
sub-diurnal, diurnal, seasonal, and annual time-scales, and power
growing from diurnal to seasonal time-scales. Frequency analysis of
fluxes from several models shows that they distribute too much power
over periods greater than 1 day but less than one year, indicating that
a closer look is needed into how models achieve longer-term stability in
reflected shortwave radiation. Model developers can consider using these
datasets for time-varying energetic constraints, since tuning parameter
choices will impact modeled planetary shortwave radiation across
timescales ranging from sub-diurnal to decadal.