Background:
E3SM has developed the idiosyncratic but convenient habit of referring to specific humidity (=mass of water species divided by mass of total air) as wet mixing ratio and the actual mixing ratio (=mass of water species divided by mass of dry air) as dry mixing ratio.
Unlike EAMf90, the EAMxx atmospheric driver(AD) requires all variables passed to it to be wet mixing ratios. Each process called by AD can choose to use wet or dry mixing ratio, but must convert its variables (and - if required - the time tendencies it produces) back to wet mixing ratio. For processes which exclusively use wet mixing ratio, there is no need for any conversion. SCREAM provide following methods for converting dry mass mixing ratio (mmr) to wet mmr and vice versa via functions:
calculate_drymmr_from_wetmmr
calculate_wetmmr_from_drymmr
The same rules hold for volume mixing ratio (vmr). For processes requiring vmr, SCREAM provide the following functions:
calculate_mmr_from_vmr
calculate_vmr_from_mmr
vmr can be dry or wet. Appropriate conversion should be made based on parameterization requirements. Tendencies computed by the parameterization should be converted back to wet mmr before passing these back to AD.
Explanation for why each process is wet or dry:
Dycore:wet mmr - The dycore needs wet mixing ratios because ???.
Atmosphere Driver:wet mmr - Because many variables passed to the AD are advected by the dycore, we decided to insist that all variables handled by the AD are wet.
P3:dry mmr - Microphysics contains a lot of empirical parameterizations. Empirical parameterizations tend to assume variables are couched in dry mixing ratio since that’s what’s usually measured observationally.
SHOC:wet mmr - I should re-read the SHOC overview paper before explaining this, but basically SHOC is derived from the Navier-Stokes equations which should be able to be adapted equally well for wet or dry variables.
RRTMGP:wet vmr - Ben Hillman ?