The NASA/GEWEX Surface Radiation Budget project: Analysis of Release 3.0 shortwave dataset

The SRB project has completed a 24.5 year (July 1983-December 2007) dataset of satellite-derived surface longwave and shortwave radiative fluxes at a 3-hourly, 1°x1° resolution. Here we describe and examine the shortwave dataset.

Several improvements from the previous Release 2.81 have been incorporated into the new release. The primary inputs for SRB shortwave algorithm are the ISCCP DX radiances and cloud properties. Instances occasionally occur within a one degree grid cell where observed TOA radiances in pixels marked cloudy by the ISCCP cloud-screening algorithm are actually lower (darker) than nearby clear pixels. This was leading to occurrences of spurious negative cloud radiative forcing. Release 3.0 now declares all cloudy pixels darker than any clear pixel within the same grid cell to be clear.

The treatment of aerosol has also been improved. The algorithm uses a climatological first guess aerosol optical depth to determine surface albedo based on ISCCP clear sky composite radiance, then picks a retrieved AOD to match the TOA upward flux implied by the instantaneous observed clear sky radiance. Release 2.81 featured a crude initial guess based solely on surface type. Release 3.0 incorporates maps of monthly aerosols derived from output from the MATCH transport model. This has led to a substantial improvement in SRB flux products. In addition, retrieved aerosol optical depth is artificially capped at 0.05 over snow and ice surfaces, to reduce spuriously high AODs which were generated in previous versions, likely due to cloud/ice detection issues in ISCCP.

We compare Release 3.0 fluxes to ISCCP-FD, the QC version of SRB, various CERES products, and to surface data from BSRN. Surface validation efforts are discussed in more detail by Zhang et al., also in this conference.