Re: X-Y Dimension resolution in Ferret?

[Mark A Ohrenschall <Mark.A.Ohrenschall@xxxxxxxx>]

Hi Joe,

Thanks for your help. I'm not running Ferret directly, only through LAS. However, I've created a small example that reproduces the problem. If you like, you can try viewing this sample yourself in Ferret. You can grab the netCDF file at http://panther.ngdc.noaa.gov/dods/test.nc (1 Kb).

Here is a description of the netCDF file:

[mao@panther dods]$ ncdump -c test.nc
netcdf test {
dimensions:
        lat = 11 ;
        lon = 21 ;
variables:
        float lat(lat) ;
                lat:units = "degrees_north" ;
        float lon(lon) ;
                lon:units = "degrees_east" ;
        short test(lat, lon) ;
                test:long_name = "Test Corner of Cell Registratin" ;
                test:units = "categories" ;

// global attributes:
                :title = "Test corner of cell registration" ;
                :Conventions = "COARDS" ;
data:

 lat = 90, 72, 54, 36, 18, 0, -18, -36, -54, -72, -90 ;

 lon = -180, -162, -144, -126, -108, -90, -72, -54, -36, -18, 0, 18, 36, 54,
    72, 90, 108, 126, 144, 162, 180 ;
}
[mao@panther dods]$
And from the LAS XML configuration file:
<datasets>
        <test name="Test Corner of Cell Registration">
                <variables>
                        <test name="Test Corner of Cell Registration" url=""file:test.nc">file:test.nc" units="categories">
                                <link match="/lasdata/grids/test_grid"/>
                        </test>
                </variables>
        </test>
</datasets>
<grids>
        <test_grid>
                <link match="/lasdata/axes/test_lon"/>
                <link match="/lasdata/axes/test_lat"/>
        </test_grid>
</grids>
<axes>
        <test_lon type="x" units="degrees_east">
                <arange start="-180" step="18" size="21"/>
        </test_lon>
        <test_lat type="y" units="degrees_north">
                <arange start="-90" step="18" size="11"/>
        </test_lat>
</axes>
It's my intent (in this example) to describe a grid (or perhaps a lattice)
in which each data point is centered on whole multiples of ten degrees
of latitude and longitude (as opposed to grid cell edges which align with
whole multiples of ten degrees of latitude and longitude). So for this
ten degree grid, for example, the first data point is at 180 W, 90 N (and
not 175 W, 85 N). To belabor the point, this means that all data values
in the first (and last) row of the grid are identical (since they're all
at 90 N -- or 90 S). It also means that the first and last data value in
each row are also identical (since their longitudes are 180 W and 180 E,
respectively).

Thanks,

Mark

Joseph Mclean wrote:

Hi Mark,

Are you using Ferret to change the axis definition?  If so, could you send me an example of the command you're using.  It seems like this can be fixed using the /EDGE and /MODULO qualifiers.  What are the exact coordinates of the cell corners and centers?

Joe
---------------------------------------------------------

Mark A Ohrenschall wrote:

I'm getting an internal server error when attempting to display a dataset with LAS and Ferret. GenericLAS_debug.txt reports:

 **ERROR: illegal limits: "LWSCP07" is not in the range X=164W:164E(524)
          Axis extremes are X=179.8E(-180.2):179.8W

I get this particular message when I shrink the window somewhat from the full region. This is the error message I get for the full region:

 **ERROR: dimensions improperly specified: must be a 2D region
          shade/levels=64/set lwscp07

What's unusual about this grid is that it has 361 rows and 721 columns, but the grid spacing is still half-degree (you would expect 360 rows by 720 columns). I get an image without error if I change the grid spacing so that it is 0.4993 (instead of 0.5), i.e., so that a ncdump -c shows:

 slon = -180, -179.5007, -179.0014, -178.5021, -178.0028, -177.5035,
    -177.0042, -176.5049, -176.0056, -175.5062, -175.0069, -174.5076,
    -174.0083, -173.509, -173.0097, -172.5104, -172.0111, -171.5118,

But what I'm trying to represent is a grid with corner of cell registration and not center of cell registration, hence the grid spacing of 0.5.

Incidentally, ncBrowse displays the image just fine with the 0.5 grid spacing.