Re: Reference time in aggregated files

Hi Roy:

There is a problem when the units of the aggregation coordinate change from 
file to file.

There is a way around this, by generating the needed coordinate values and 
placing in the NcML, eg:

<netcdf xmlns="http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2";>

 <variable name="time">
    <attribute name="units" value="months since 2000-01-01 00:00"/>
    <values>0, 1, 2</values>
  </variable>

 <aggregation dimName="time" type="joinNew">
   <variableAgg name="T"/>
   <netcdf location="file:test/data/ncml/nc/time0.nc" />
   <netcdf location="file:test/data/ncml/nc/time1.nc" />
   <netcdf location="file:test/data/ncml/nc/time2.nc" />
  </aggregation>

</netcdf

I think this also works for joinExisting, in that it would override the 
existing values inside the file(s) (I need to double check that). I dont think 
there will be a performance penalty.

If the values are regularly spaced, you can use:
 <values start="0" increment = "1"/>

I think that this case of regular spacing may work for scanned datasets, in 
that the number of points will be taken from the files scanned (again i need to 
double check that).

Roy Mendelssohn wrote:
> Hi:
>
> This question has arisen working with a partner - and while I am pretty 
> certain I know the answer to this question, I want to make sure what is 
> the correct answer.
>
> The netcdf files that these people are creating have a changing 
> reference time fin the files.  So, for example, say they are creating 
> monthly files, then the 12 files for any year all have the same base 
> reference data  (the start of that year), and this changes with each year.
>
> My guess is that this will most likely break the aggregation - and if 
> not at least slow it down considerably.  Is my guess correct?
>
> TIA,
>
> -Roy M.

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