Electromagnetic radiation -- sunlight and earthlight -- is the fundamental energy source for all atmospheric motions and so a particularly important parameterization in an earth system model. The radiation parameterization in version 0 and 1 of DOE's Energy Exascale Earth System Model (E3SM), known as RRTMG, has become increasingly outdated with respect to both the scientific content and the computational practices. Its successor, RTE+RRTMGP, is the default treatment of radiation for the Multiscale Modeling Framework (MMF) version of the E3SM and is fully integrated into the very high-resolution Simple Cloud-Resolving E3SM Atmosphere Model (SCREAM) and is being considered for inclusion in upcoming versions of the E3SM.
The team proposes to implement key improvements to the computational efficiency and capabilities of RTE+RRTMGP to ensure that the radiation code in the E3SM is efficient, accurate, and allows users to address key scientific priorities. As in PIs’ past DOE-funded work on RTE+RRTMGP, the proposed development will involve extensive interactions with the E3SM computational and scientific staff and will proceed collaboratively with the E3SM team. The effort will focus on the following tasks:
- Develop and/or collaborate on new computational kernels, using specialized programming languages and techniques, to improve efficiency with existing hardware and meet new computational platforms as they are acquired or considered by DOE;
- Extend and maintain a C++ front end to RTE and RRTMGP suitable for use in SCREAM, the ultra-high-resolution version of E3SM;
- Develop alternative sets of spectral data with varying spectral detail and sensitivity so that users of E3SM can balance computational cost against application-specific needs;
- Investigate alternative spectral decompositions for RRTMGP to better represent radiation within vegetation canopies and the ocean;
- Potentially regenerate data used by RRTMGP to be consistent with an updated scientific knowledge of the absorption of radiation by gases.
The accuracy of RTE+RRTMGP will be enhanced by 1) treating the largest effects of a spherical atmosphere, including refraction and extended path lengths, in calculations of the direct solar beam and 2) accounting for correlations between gases and clouds in the spectral dependence of scattering and absorption.