[MITgcm-devel] setting hFacW & hFacS with pkg/shelfice

[Jean-Michel Campin]

Hi Ichiro etal,

I am moving this discussion to mitgcm-devel since it's a broader issue.

In short, the problem is that when hFacW & hFacS are define as the
min of adjacent hFacC, the total column thickness at W & S edge (as computed 
from the sum of hFacW / hFacS * drF) does not always match the difference
 rSurfW - rLowW or rSurfS - rLowS (which is needed for rStar to work).

What has been done previously was to adjust rSurfW & rSurfS to match
 Sum hFacW/S * drF ; I ended up removing accidentally this part of code
in 2014 but just put it back last Friday.
So, this should fix the problem you guys bump into at JPL when using pkg/shelfice
with z-star.

But I think it would be better (at least as an option) to enable to
keep rSurfW & rSurfS (set according to ice-shelf geometry) - with some
minor adjustement due to hFacMin criteria - and to calculate hFacW & hFacS
in the same way as hFacC is computed. The advantage is that this way, dry/wet 
column at W or S edge are less dependent on vertical resolution.

I am trying to investigate how this works (getting machine truncation issues), 
but it seems that I will need to modify ini_masks_etc.F and also 
shelfice_update_masks.F, which are also modified and duplicated for 
shelfice-remeshing development. Will probably need to contact this group.

Cheers,
Jean-Michel