Open white-alistair opened 2 years ago
With #123 this should become much easier as we can define tempC
and tempK
for a given column and use both within the parametrizations. Given that the tendencies in ˚C and Kelvin are the same, we then are left with the question whether we can carry temperature in ˚C around in the dynamics...
Most parts in the dynamical core now use the temperature anomaly anyway, so maybe less of an issue. Hoskins and Burridge 1975 normalise temperature by $a^2\Omega^2/R$ ($a$ radius, $R$ gas constant) which might be another option
Btw, IFS uses virtual temperature as prognostic variable, maybe that's a good idea too
Copying @milankl's comment from PR #82: