Tracer indices in GFS PBL schemes

[egrell]

I believe there is an error (with impact only on diagnostic output) in both the satmedmfvdifq and hedmf PBL schemes, in the indexing of the tracers.
In satmedmfvdifq.meta, the tracer indices are defined as:
[ntrac] = standard_name = number_of_vertical_diffusion_tracers
[ntoz] : standard_name = index_for_ozone
[ntcw] : standard_name = index_for_liquid_cloud_condensate
[ntiw] : standard_name = index_for_ice_cloud_condensate_vertical_diffusion_tracer
[ntke] : standard_name = index_for_turbulent_kinetic_energy_vertical_diffusion_tracer

I think all the nt* variables that represent index values should use the “index_for_*_vertical_diffusion_tracer” values, so ntoz and ntcw should be changed to
[ntoz] : standard_name = index_for_ozone_vertical_diffusion_tracer
[ntcw] : standard_name = index_for_liquid_cloud_condensate_vertical_diffusion_tracer

This is because prior to calling the PBL scheme, the tracers that need to be vertically diffused get copied into a vertically_diffused_tracer_concentration array in routine GFS_PBL_generic_pre, where the tracer indices are changed. For the GFSv15 and v16 physics suites, in which nvdiff /= ntrac and GFDL microphysics are used:
vdftra(i,k,1) = qgrs(i,k,ntqv)
vdftra(i,k,2) = qgrs(i,k,ntcw)
vdftra(i,k,3) = qgrs(i,k,ntiw)
vdftra(i,k,4) = qgrs(i,k,ntrw)
vdftra(i,k,5) = qgrs(i,k,ntsw)
vdftra(i,k,6) = qgrs(i,k,ntgl)
vdftra(i,k,7) = qgrs(i,k,ntoz)

For cloud water, ntcw = 2 anyway, so there is no problem.
For ozone, the original index (ntoz) is 4, but in the vertically_diffused_tracer array it becomes 7.
It seems this would result in do3dt containing rain water tendencies instead of ozone tendencies.
As do3dt is only diagnostic output, it should not change the results but would be good to correct.
The same issue exists in the hedmf scheme.

[climbfuji]

Thanks, Evelyn, definitely we need to look into this.

[climbfuji]

@SMoorthi-emc FYI

[SMoorthi-emc]

ntoz was 4 only for Zhao-Carr microphysics. as far as I know.