The calculation of clearsky fluxes is done solely to provide diagnostic output (i.e., to calculate the "cloud radiative effect" by difference clear-sky with all-sky fluxes), and do not affect the model simulation. Furthermore, clear-sky fluxes should probably not change much from column to column, and certainly not enough to change the domain average (which is all that is output). Because clearsky fluxes are calculated by calling the radiative transfer routines an additional time, this is costing quite a bit of computation time in the radiative transfer. With RRTMGP, it should be straightforward to separate this out, and to only compute clearsky fluxes ONCE from the domain average gas and thermodynamic profiles. We might also be able to get away with only calculating the gas optics once per timestep as well, but we need to evaluate the reasonableness of this.
The calculation of clearsky fluxes is done solely to provide diagnostic output (i.e., to calculate the "cloud radiative effect" by difference clear-sky with all-sky fluxes), and do not affect the model simulation. Furthermore, clear-sky fluxes should probably not change much from column to column, and certainly not enough to change the domain average (which is all that is output). Because clearsky fluxes are calculated by calling the radiative transfer routines an additional time, this is costing quite a bit of computation time in the radiative transfer. With RRTMGP, it should be straightforward to separate this out, and to only compute clearsky fluxes ONCE from the domain average gas and thermodynamic profiles. We might also be able to get away with only calculating the gas optics once per timestep as well, but we need to evaluate the reasonableness of this.