Problem aggregating along time dimension

[Jon Blower]

Hi John,

Thanks for this - I could generate the coordinates in NcML but the
problem is that every day a new file gets added in to the timeseries
and I don't want to have to edit the NcML every day.  This is why I
wanted to do it with a <scan>.  Furthermore, missing data files are
possible so we can't assume that the files we have are equally spaced
in time.

I think this would be a good enhancement.  It might be hard to send
you the data files because they are (a) big and (b) commercial in
nature, but I might be able to sort something out.  It's probably not
too hard to create some dummy datasets though?

Cheers, Jon

On 2/9/07, John Caron <caron at unidata.ucar.edu> wrote:
> Hi Jon:
>
> At the moment, all you can do is generate the coordinates yourself in the NcML, following this section of the docs:
>
>   http://www.unidata.ucar.edu/software/netcdf/ncml/v2.2/Aggregation.html#JoinExistingTypes
>
> Otherwise we'll have to add a "request for enhancement" on dealing with join coordinates whose units change. It would be helpful to have some datasets with this problem for testing.
>
> Regards,
>
> John
>
> Jon Blower wrote:
> > Dear all,
> >
> > I've come across a problem when aggregating a dataset across the time
> > dimension using NcML and the latest (2.2.18) nj22 library.  I would
> > like to aggregate a number of files, which represent timesteps of an
> > ocean forecast model (one timestep per file).  I am aggregating using
> > an NcML file like this:
> >
> > <netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2">
> >   <aggregation dimName="t" type="joinExisting">
> >     <scan location="/path/to/files" suffix=".nc"/>
> >   </aggregation>
> > </netcdf>
> >
> > When reading this NcML file using NetcdfDataset.openDataset() I get an
> > aggregated dataset with the correct number of timesteps BUT all the
> > values along the time axis are the same.
> >
> > After much head-scratching I think I've found the source of the
> > problem, but I don't have a solution.  In each individual file, the
> > time dimension is specified with different units but the values remain
> > the same:
> >
> > File 1: time units: seconds since 2007-02-06 00:00:00, value: 43200
> > File 2: time units: seconds since 2007-02-07 00:00:00, value: 43200
> >
> > This seems to confuse the aggregation code, which I assume expects the
> > units to remain constant between files and the values to change.
> >
> > Is there a solution or workaround for this?  I don't control how the
> > netCDF files are made so I don't think I can change them.  Any help
> > will be much appreciated.
> >
> > Regards, Jon
> >
>


-- 
--------------------------------------------------------------
Dr Jon Blower              Tel: +44 118 378 5213 (direct line)
Technical Director         Tel: +44 118 378 8741 (ESSC)
Reading e-Science Centre   Fax: +44 118 378 6413
ESSC                       Email: jdb at mail.nerc-essc.ac.uk
University of Reading
3 Earley Gate
Reading RG6 6AL, UK
--------------------------------------------------------------

==============================================================================
To unsubscribe netcdf-java, visit:
http://www.unidata.ucar.edu/mailing-list-delete-form.html
==============================================================================