An updated evaluation of the surface radiation budget in climate model
s (1994-96 versions; seven datasets available, with and without aeroso
ls) and in two new satellite-based global datasets (with aerosols) is
presented. All nine datasets capture the broad mean monthly zonal vari
ations in the flux components and in the net radiation, with maximum d
ifferences of some 100 W m(-2) occurring in the downwelling fluxes at
specific latitudes. Using long-term surface observations, both from la
nd stations and the Pacific warm pool (with typical uncertainties in t
he annual values varying between +/- 5 and 20 W m(-2)), excess net rad
iation (R-N) and downwelling shortwave flux density (S(o)down arrow) a
re found in all datasets, consistent with results from earlier studies
[for global land, excesses of 15%-20% (12 W m(-2)) in R-N and about 1
2% (20 W m(-2)) in S(o)down arrow]. For the nine datasets combined, th
e spread in annual fluxes is significant: for R-N, it is 15 (50) W m(-
2) over global land (Pacific warm pool) in an observed annual mean of
65 (135) W m(-2); for S(o)down arrow, it is 25 (60) W m(-2) over land
(warm pool) in an annual mean of 176 (197)W m(-2).The effects of aeros
ols are included in three of the authors' datasets, based on simple ae
rosol climatologies and assumptions regarding aerosol optical properti
es. They offer guidance on the broad impact of aerosols on climate, su
ggesting that the inclusion of aerosols in models would reduce the ann
ual S(o)down arrow by 15-20 W m(-2) over land and 5-10 W m(-2) over th
e oceans. Model differences in cloud cover contribute to differences i
n S(o)down arrow between datasets; for global land, this is most clear
ly demonstrated through the effects of cloud cover on the surface shor
twave cloud forcing. The tendency for most datasets to underestimate c
loudiness, particularly over global land, and possibly to underestimat
e atmospheric water vapor absorption, probably contributes to the exce
ss downwelling shortwave flux at the surface.