Re: [ferret_users] FERRET erroneously reports '0' for values of field

[Keith Lindsay <klindsay@xxxxxxxx>]

Gary,

The MOC variable in CESM output is five dimensional. The MOC metadata 
from an ncdump on a file of mine is:

        float MOC(time, transport_reg, moc_comp, moc_z, lat_aux_grid) ;
                MOC:long_name = "Meridional Overturning Circulation" ;
                MOC:units = "Sverdrups" ;
                MOC:coordinates = "lat_aux_grid moc_z moc_components transport_region time" ;
                MOC:_FillValue = 9.96921e+36f ;
                MOC:missing_value = 9.96921e+36f ;

> From right to left, the dimensions are:

latitude
depth
component of velocity used in MOC computation
   Eulerian Mean
   Eddy-Induced (bolus)
   Submeso
spatial region used in MOC computation
   Global Ocean - Marginal Seas
   Atlantic Ocean + Mediterranean Sea + Labrador Sea + GIN Sea + Arctic Ocean + Hudson Bay 
time


I do not know what ferret does under the covers for a five dimensional 
variable, but I suspect that this is the root of the problem. Perhaps 
there is a nf_get_vara call inside ferret that isn't specifying what to 
do with the unexpected additional dimensions.

When I have used ferret to examine MOC in the past, I have first used 
NCO to extract along the component and region dimensions, as well as 
remove the resulting singleton dimensions.

For instance,

ncwa -a moc_comp,transport_reg \
     -d moc_comp,0 -d transport_reg,1 \
     -v MOC infile.nc outfile.nc

will extract the Eulerian Mean MOC for the Atlantic etc. region. Because 
MOC in the resulting file is now three dimensional, in COARDS order, 
ferret can handle it without any problem.

Using ncwa is rather heavy handed. I wouldn't be surprised if there is a 
different more elegant work around.

Keith

******************************************************************
Keith Lindsay                http://www.cgd.ucar.edu/oce/klindsay/
email: klindsay@xxxxxxxx   phone: 303-497-1722   fax: 303-497-1700