In order to obtain a top of the atmosphere radiation budget from scanning sensors onboard satellites, observed radiances must be converted to fluxes. Early satellite missions simply assumed that all surfaces were isotropic, so radiances in the visible wavelengths could be converted directly into albedos. Later, angular directional models (ADMs) were constructed empirically from satellite observations. These have been used almost exclusively since then to convert satellite radiances to fluxes. However, significant questions have been raised about the applicability of ADMs to satellite retrievals in general. In particular, biases have been shown in the retrieved albedos due to the way the ADMs were constructed, especially at large view angles.

At present, the best knowledge available concerning the earth's radiation budget comes from low Earth orbiting satellite scanners, which have used ADMs, such as those from Nimbus 7, to invert the observed radiances. This casts some doubt at least on the shortwave portion of the radiation budget for a couple of different reasons. First, these satellites have employed a cutoff in view angle for their observations due to the growth in the size of the observed region at large view angles, which makes scene identification difficult. Second, it is unclear whether or not polar orbiting satellites are able to provide adequate angular sampling at large view angles to allow for unbaised construction of ADMs.

Geostationary satellites, such as the Geostationary Operational Environmental Satellites (GOES) offer another perspective on the problem. These satellites have a spatial resolution higher than that used previously to construct ADMs - one square kilometer at nadir - and the geostationary orbit provides good angular resolution to view angles out to 89 degrees. Using an archive of three years of full resolution, full disk data from GOES-10, we have constructed new ADMs. Significantly, the GOES data allow us to directly study the assumptions used at large view angles in earlier ADMs. We will present the results of our study and show how the our ADMs compare to older ADMs, particularly at large view angles, and discuss the implications this has for albedo retrievals.