Re: [netcdf-java] aggregation of files with inconsistent content

On 9/10/2012 3:14 PM, Ellyn Montgomery wrote:
Aggregation gurus-

Please help with how to aggregate files that may contain an inconsistent list of variables. The files I want to use all contain temperature and salinity time series, but some also have conductivity, attenuation or other data. I'd like to include all the possible variables in a single aggregation. Seems like if one file of 50 is missing attenuation, that I have to exclude attenuation from the aggregation, and just operate on the variables that are present in all 50.

Have any of you found a way to also include the other "possibly present variables", or whether it's better to just make 1 joinExisting time aggregation per variable. I heard that in FMRC aggregations there was a way to specify a "protoDataset" (maybe not the right term) and it fills in appropriately for any files with variables missing from the prototype. I couldn't find a reference to this in the documentation, but if it's possible, would this approach be appropriate for the inconsistently present variables I'm trying to aggregate?

Or maybe someone has a good reason to supply several files (one file per variable aggregated) instead of one file with all the possible variables aggregated.

I'm interested in the pros and cons of each approach. Thanks for your help!


What is the format of the data files you are trying to aggregate? Grib versus NetCDF will make a difference, particularly if you have transitioned to TDS4.3. In the meantime, here is a link to documentation about protoDatasets and the aggregation of non-homogeneous files:

or, in 4.3:


  • 2012 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the netcdf-java archives: