The net radiative heating of the atmosphere for all sky conditions is investigated for both LW and SW fluxes at over 30 long-term sites for January-August 1998. Data used include: 1) high quality surface observations (includes data from ARM/SGP, BSRN, SURFRAD, CMDL and other sites worldwide), 2) upgraded and validated broadband TOA observations from the CERES scanner on board the TRMM satellite, and 3) auxilliary broadband radiative transfer calculations from the CERES operational processing Surface and Atmospheric Radiation Budget (SARB) product. The radiative transfer calculations use a modified Fu-Liou code, cloud properties inferred by the CERES Team from the VIRS narrowband imager, ECMWF soundings, and aerosols from the Collins-Rasch MATCH assimilation. For LW, calculations and observations are in excellent agreement (i.e. within 4W/m-2 at the surface) attesting to the quality of time mean retreivals of cloud fraction and cloud base height. This is not the case in the SW. Our previous analyses focused on the differences of net atmospheric heating for all-sky and clear-sky which used simpler ERBE-like CERES fluxes at the TOA and found no outstanding discrepancies between all-sky and clear-sky SW absorption of the atmospheric column. Now using more advanced CERES Single Satellite Flux (SSF) TOA data, we find discripancies between observed and computed SW heating of the atmosphere for both all sky and clear conditions. Diagnostics with AERONET sun photometers and cloud screening with temporally intensive surface radiometery can account for most of the clear sky discrepancy. In rigorously examined clear sky simulations, surface albedo inferred from large satellite footprints are systematically lower than those observed at a single point via surface radiometery. We report on the application of CERES/ARM Radiation Experiment (CARE) helicopter measurements (August 1998 at SGP) of near-surface albedo to resolve the clear-sky discrepancy.