[MITgcm-devel] exf interp_method

[Jean-Michel Campin]

Hi,

I've just checked-in these changes (adding run-time param
*_interpMethod, to select the interpolation method
with a value of zero switching off the interpolation)

The defaults are the same as before.
Regarding USE_NO_INTERP_RUNOFF options, I added a stop
if it's defined, to use instead "runoff_interpMethod=0" in data.exf
but the code itself is still there (will be removed soon).

I did not really checked that it works as it use to
for all possible fields (but Jeff just checked with Gael's set-up), 
so if there is something strange, let me know.

Notice few problems (fixed or to fix):
a) in case one of the 2 uwindFile/vwindFile is empty and in
   the case (usingCurvilinearGrid .OR. rotateGrid), no wind
   was loaded in exf_set_uv.F ; this should be fixed now.
b) ALLOW_CLIMSTRESS_RELAXATION code was not compiling without
   USE_EXF_INTERPOLATION (now fixed), but there are still issues
   regarding location of climustr,climvstr (A-grid/C-grid)
   and since it's not very clear what they are used for,
   will need to be fixed later.
c) default setting of interpolation parameters in exf_readparms.F
   (not fixed, but I am going to fix this soon):
   lon0 & lat0 for uwind,vwind,wspeed, and ustess,vstress should
   be also at center (since they will be interpolated at grid-cell
   center).
   _lat_inc is set in a funny way for most of the field:
   since "*_lat_inc(i) = (delY(i) + delY(i))*exf_half" is not so
   different from delY(i); should be:
   "*_lat_inc(i) = (delY(i) + delY(i+1))*exf_half"
d) some check in exf_check.F (e.g. usingCartesianGrid) should test
   if interpolation is really used (and not only if USE_EXF_INTERPOLATION
   is defined), and all the *_nlat should be checked (instead of
   just climsst_nlat and climsss_nlat).

Cheers,
Jean-Michel

On Fri, Dec 16, 2011 at 09:48:15AM -0800, Menemenlis, Dimitris (3248) wrote:
> same comment as Patrick
> 
> Dimitris Menemenlis
> 
> On Dec 16, 2011, at 8:48 AM, Patrick Heimbach wrote:
> 
> > Good idea and fine with me.
> > p.
> > 
> > On Dec 16, 2011, at 11:29 AM, Jean-Michel Campin wrote:
> > 
> >> Hi,
> >> 
> >> I propose to add a set of run-time param ({inputfield}_interpMethod),
> >> one for each input field, to select the interpolation method
> >> with a value of zero switching off the interpolation.
> >> 
> >> Motivations: There are fields other than runoff we may want to provide
> >> on the model grid (SST, SSS for relaxation); 
> >> and allowing to select the interpolation method (bilinear/bicubic)
> >> (presently hard-coded) could be useful too.
> >> 
> >> Any comments ?
> >> 
> >> Cheers,
> >> Jean-Michel
> 
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel