Many global climate models currently assume that longwave photons are non-scattering in clouds, and also have overly simplistic treatments of surface emissivity. Multiple scattering of longwave radiation and non-unit emissivity could lead to substantial discrepancies between the actual Earth’s radiation budget and its parameterized representation in the infrared, especially at wavelengths longer than 15 µm.

The evaluation of the parameterization of longwave spectral multiple scattering in radiative transfer codes for global climate models is critical and will require benchmarking across a wide range atmospheric conditions with more accurate, though computationally more expensive, multiple scattering models. We therefore present a line-by-line radiative transfer solver that includes scattering, run on a supercomputer from the National Energy Research Scientific Computing that exploits the embarrassingly parallel nature of 1-D radiative transfer solutions with high effective throughput. When paired with an advanced ice-particle optical property database with spectral values ranging from the 0.2 to 100 µm, a particle size and habit distribution derived from MODIS Collection 6, and a database for surface emissivity which extends to 100 µm, this benchmarking result can densely sample the thermodynamic and condensate parameter-space, and therefore accelerate the development of an advanced infrared radiative parameterization for climate models, which could help disentangle forcings and feedbacks in CMIP6.