The large range in aerosol radiative forcing across CMIP5 GCMs may come from either a true range in aerosol climatology or radiative parameterization errors. One of the goals of the Radiative Forcing Model Intercomparison Project (RFMIP) is to quantify the contribution of radiative parameterization errors to the range in aerosol forcing on a global scale. This is achieved by running a benchmark-quality line-by-line code at the native resolution of each participating GCM using the atmospheric state from that GCM as input, including aerosol optical properties. Thus, the framework provides a benchmark estimate of flux for every model grid point. The resulting radiative fluxes can then be compared to those produced natively by the GCM radiative parameterization. In this way the radiative parameterization error is quantified globally and regionally while maintaining the variability in atmospheric state between models. This calculation is performed with and without aerosols to further provide an estimate of aerosol radiative forcing.

This intercomparison is unique in that it performs calculations over the entire globe on each model's native grid and with each model’s meteorological fields as input, thus elucidating the spatial distribution of errors and representing a wider range of atmospheric conditions in its characterizations. Previous radiative intercomparison efforts have relied on only a handful of atmospheric profiles, which are often idealized, to represent the variability of conditions in the atmosphere.

For this study we present a proof of concept of the RFMIP protocol by intercomparing the results for two GCMs, GFDL’s AM4 and NCAR’s CESM. We demonstrate the contribution RFMIP has on reducing the range in modeled aerosol radiative forcing and the potential it can have with the participation of additional modeling centers.