[MITgcm-devel] some idea for pkg/shelfice

[Jean-Michel Campin]

Hi Martin,

Some here have been using pkg/shelfice and, in order to make
more progress, there are few things that could be changed:

A) add a new variable for the mass of ice (per unit area):
 I propose to do this, and to allow to specify (through a file)
 the ice-shelf mass. 
 To start with, fully backward compatible with no result changes,
 will allow either to specify:
   a) R_shelfice & SHELFICEloadAnomaly (same as now)
or b) R_shelfice & SHELFICE_Mass
 and with the relation between the 2:
  SHELFICE_Mass = SHELFICEloadAnomaly/gravity - Ro_surf*rhoConst
 (since Ro_surf is negative).
 At the end of shelfice_thermodynamics.F, will compute SHELFICEloadAnomaly,
 and in case of (a), SHELFICEloadAnomaly from the file will be converted 
 (in shelfice_init_varia.F) to SHELFICE_Mass using the same relation.
Advantages:
1) mass of ice is (much) more natural than SHELFICEloadAnomaly;
  + it does not dependent on hFacMin truncation.
2) should allow in the future to step forward SHELFICE_Mass.
3) they are few places where SHELFICE_Mass can be used
 (e.g., pressure @ the base of the ice-shelf = SHELFICE_Mass*gravity, 
  instead of R_shelfice as it is now, 
  also total thickness of the ice = SHELFICE_Mass/rho_ice ...)
 but will let you decide what should be modified or not.

B) implement real-fresh-water flux with pkg/shelfice:
 I think it would not be too hard to allow the surface forcing
 to be applied at the surface level (kSurf) even if kSurf <> 1
 This way, if we put the SHELFICE melting into PmE, we will get
 the real-fresh-water formulation with little more modif.

C) current isomip experiment: found a problem when turning on
 the non-lin free-surf (without z*), got some strange (and wrong)
 currents because of the CD-scheme. Would need confirmation,
 Likely due to a Pb in CD-scheme with non-lin free-surf and
 large variations of hFac (+ kSurf <> 1 ?), will need to be confirmed.
 Can you remember us why CD-scheme is used (since the resolution 
 is high enough, ~10.km, to run without) ?

Cheers,
Jean-Michel