ZM convective mass flux (atmosphere_convective_mass_flux_due_to_deep_convection) units inconsistent with snapshot

[jimmielin]

Currently, the units for the deep convective mass flux in ZM is hPa s-1 as output in zm_convr CCPPized scheme:

[ mcon ]
  standard_name = atmosphere_convective_mass_flux_due_to_deep_convection
  units = hPa s-1
  type = real | kind = kind_phys
  dimensions = (horizontal_loop_extent,vertical_interface_dimension)
  intent = out

In current CAM, the deep convective mass flux (saved in the snapshot as CMFMC) is in the units of kg m-2 s-1. This is because:

1. The atmosphere_convective_mass_flux_due_to_deep_convection, or mcon or cmfmc is first computed in zm_convr_run with the units of hPa s-1.
2. In zm_conv_intr.F90, it is converted to kg m-2 s-1:

!
! Convert mass flux from reported mb/s to kg/m^2/s
!
   mcon(:ncol,:pverp) = mcon(:ncol,:pverp) * 100._r8/gravit
   mconzm(:ncol,:pverp) = mcon(:ncol,:pverp)

this is the output argument for zm_conv_tend

subroutine zm_conv_tend(pblh    ,mcon    ,cme     , &

3. mcon is the output argument used in convect_deep_tend in convect_deep.F90

  case('ZM') !    1 ==> Zhang-McFarlane (default)
     call pbuf_get_field(pbuf, pblh_idx,  pblh)
     call pbuf_get_field(pbuf, tpert_idx, tpert)

     call zm_conv_tend( pblh    ,mcon    ,cme     , &

which is the output argument for convect_deep_tend

subroutine convect_deep_tend( &
     mcon    ,cme     ,          &

4. This output argument is saved in the snapshot as cmfmc

    if (trim(cam_take_snapshot_before) == "convect_deep_tend") then
       call cam_snapshot_all_outfld_tphysbc(cam_snapshot_before_num, state, tend, cam_in, cam_out, pbuf, &
           flx_heat, cmfmc, cmfcme, zdu, rliq, rice, dlf, dlf2, rliq2, det_s, det_ice, net_flx)
    end if

    call convect_deep_tend(  &
         cmfmc,      cmfcme,             &

and thus the units for cmfmc in the snapshot are kg m-2 s-1.

In CCPPized ZM, the unit conversion is only done at the diagnostic output time in the zm_diagnostics scheme.

However, matching the output with the snapshots or for deep convection to work together with shallow convection requires that atmosphere_convective_mass_flux_due_to_deep_convection is in consistent units with the snapshot (kg m-2 s-1) and the shallow convective mass flux (also in kg m-2 s-1).

Thus, the unit conversion for atmosphere_convective_mass_flux_due_to_deep_convection should be moved from zm_diagnostics.F90 to zm_convr.F90 so that the units are passed consistently across schemes (CCPP cannot convert from hPa s-1 to kg m-2 s-1 on its own)