[ferret_users] Problems to find nearest neighbours in coordinate fields

[Martin Schmidt <martin.schmidt@xxxxxxxxxxxxxxxxx>]

Hi ferreters,

I am trying to extract station data from model output fields. The output 
of the same model run has been written to two separate files, one
covering the total model region with time averages over long periods, 
one for a subregion, but with higher output frequency. As a test, the 
time average of quantities in the high frequency regional output file 
should be exactly the same as in the output file of the full model domain.

Testing this for "stations" with given coordinates, this is really the 
case, if coordinates are specified "exactly" at the model grid. To be 
more specific:

Load the two output files, for example

use  ocean_day3d.nc
use rregionocean_stolpe.nc

show data

gives for example

TEMP     Conservative temperature         1:666     1:723     1:77      1:2
TEMP     Conservative temperature           1:61       1:49     
1:77      1:240

Indeed,

list/x=17.5/y=55.175/k=1/prec=6 temp[d=1,l=1],temp[d=2,l=1:120@ave]

gives the same results, y=55.175 specifies a grid point. But

list/x=17.5/y=55.2/k=1/prec=6 temp[d=1,l=1],temp[d=2,l=1:120@ave]

does not give the same result. Inspecting both files with ncdump, 
55.19167 and 55.20833 are found as the nearest neigbours for the 
y-coordinate. Surprisingly ferret gives also different results for 
coordinates in both the files:

list/x=17.5/y=55.2/k=1/prec=6 y[gy=temp[d=1]],y[gy=temp[d=2]]
Column  1: Y[Y=55.2N] is Y (axis YT_OCEAN)
Column  2: Y[Y=55.2N] is Y (axis YT_OCEAN_SUB1)
               Y      Y
I / *:     55.1917  55.2083

Hence,  for both files a different neighbour is found as the nearest 
one. Subtracting 55.2 reveals, that the distance to both the neighbours is
almost the same.
I / *:    -0.00833511  0.00833130

So there come up the question:
- why does ferret not read the coordinates from both files in the same 
manner? There is no attribute that defines eventually an equidistant 
grid spacing.     
float yt_ocean_sub1(yt_ocean_sub1) ;
               yt_ocean_sub1:long_name = "tcell latitude" ;
               yt_ocean_sub1:units = "degrees_N" ;
               yt_ocean_sub1:cartesian_axis = "Y" ;

Greetings,
Martin Schmidt