Hi Maghnus:
The point obs file format was designed for this problem:
http://www.unidata.ucar.edu/software/netcdf-java/formats/UnidataObsConvention.html
The idea is to use the record dimension to append data as it comes in, which is much more efficient.
However, different sampling rates are a problem not explicitly dealt with.
Magnus Hagdorn wrote:
Hello world,
we are currently considering to use the netCDF library to store data
coming from our instruments. The data represents multiple variables
together with a timestamp of the measurement. Some variables are updated
more often than others. Sample rates are of the order of 100Hz to
1000Hz. I was thinking of storing the data as a table with 3 columns,
timestamp, ID and value together with some appropriate header. Do you
think this is a reasonable idea, are there other more suitable standard
formats?
Regards
magnus
==============================================================================
To unsubscribe netcdfgroup, visit:
http://www.unidata.ucar.edu/mailing-list-delete-form.html
==============================================================================
==============================================================================
To unsubscribe netcdfgroup, visit:
http://www.unidata.ucar.edu/mailing-list-delete-form.html
==============================================================================
