[geotk] NetCDF Coverage Reading issue

Aaron Braeckel braeckel at ucar.edu
Fri Feb 18 13:49:47 EST 2011


On 02/18/2011 03:50 AM, Martin Desruisseaux wrote:
>> Note that this file is not CF-compliant and it is NetCDF 4 (HDF 5)
>> format.  I think it is the data model causing problems rather than
>> the format, though.
>
> Being not CF-compliant may be the cause - it depends if the UCAR
> library recognized the convention since Geotk relies it. However there
> is also unfinished work on the Geotk size. This issue is actually
> closely linked to the issue posted in the previous email.

To clarify my comments from a month or two ago, the NetCDF Java API can
interpret many data file formats (CF compliant NC4, non-compliant NC3,
GRIB 2, GRIB 1, MacIdas, etc. etc. etc.) into grids.  If you are
operating at the right level of the library stack (GridDatatype) then
you can leave it almost completely to the NetCDF API to care about the
actual format differences and turning those into geo-rectified grids.  I
believe this is one of the primary strengths of the NetCDF API.  With
this particular file, I am positive that the NetCDF API can parse it
into a geo-rectified grid because we use this dataset with the NetCDF
API directly to parse out grid objects.  My earlier code example was
intended to illustrate this level of use of the NetCDF API.  The NetCDF
API can also be used at a lower level (variables, dimensions,
attributes, etc.) in which case you need to deal with a lot of
lower-level details of how the data formats/conventions are represented
in the NetCDF data model.

> I filled an other JIRA task, which contains also a link to a proposed
> workaround:
>
> http://jira.geotoolkit.org/browse/GEOTK-155
>
> Could this workaround be applicable in the main time?

Since my main goal right now is to test out the Geotk resampling
capabilities, I can do so with other datasets that are working as
expected.  So I can make progress without this particular dataset, this
is merely a useful bugfix. 

Thanks,
Aaron


More information about the Geotoolkit mailing list