[MITgcm-devel] some thoughts on Pressure-Coordinate development

[Jean-Michel Campin]

Hi,

Just to have a record somewhere to list what I have in mind regarding future developments
related to pressure-coordinate and to get some feedbacks. This is also separated from
PR #334, branch "pcoord_fixes".

1) Implement Mass-coordinate
 -> needed to allow gravity to vary with height
 -> also more natural to deal with atmospheric loading in Ocean using Mass-coord:
    since Mass-Coord can be defined as Mass of water (liq + solid ?) above.
 -> more natural for non-boussinesq Non-Hydrostatic formulation
Suggestion:
  a) add Atmos & Ocean in Mass-coord formulations
  b) keep option of Atmos in P-coord ; remove option for Ocean in P-coord ?
Comment:
  would be easier to use conversion factor (such as: mass2rUnit/rUnit2mass or wUnit2rVel/rVel2wUnit)
  everywhere in the code instead of having explicit conversion factor with rhoConst and gravity
  so that it would work for both P-coord and Mass-coord 

2) Estimating the depth in Ocean in P-coord or Mass-coord:
 -> currently, use rhoConst*gravity: this is not very accurate (less accurate than anelastic 
    formulation) and not very satisfying to rely on a boussinesq reference density in a 
    non-boussinesq formulation.
 -> most accurate way would be to use PhiHyd, as done, e.g., in pkg/atm_phys ; however this
    might be tricky when using some approximation (e.g., Lin Free-Surf) and with bathymetry.
 -> compromise solution: use a conversion factor that only depends on vertical index k:
    this is robust and not less accurate than anelastic formulation.

3) Allow to provide bathymetry (in meter) as input file:
 -> this option exist for Atmos in P-coord, with some code converting orography ("topoFile",
    in meter) to reference surface pressure for model partial-cell settings.
 -> would be nice to have this option available also for Ocean in P-coord.
 -> this might change if we update the way the solve_for_pressure works, solving
    for increment of bottom pressure instead of the full bottom pressure anomaly,
    since the need to use a reference bottom pressure might not be necessary anymore.

Cheers,
Jean-Michel