Hi,
we are currently trying to put our forecast data on our thredds server.
We have already some oceanografic data out on
http://thredds.met.no/thredds/myocean/ARC-MFC/myoceanv1-class1-arctic.html
There, we gather the model-runs into 'Bulletins' and create a 'Best
estimate'-dataset by creating a very long netcdf-aggregation, joining
all the data we assume best line by line in the catalog.xml file. We
don't use the forecast_reference_time yet.
I've been looking at the aggregation as explained in
http://www.unidata.ucar.edu/software/netcdf/ncml/v2.2/FmrcAggregation.html,
which looks quite promising though I don't understand everything yet.
Are the netcdf-internal variables (_CoordinateAxisType) required, or
will it work with pure CF? Will there be many data consisting of
_FillValue since data is usually only defined close to the diagonal in a
time/forecast_reference_time matrix?
And, since examples often say more than explanations:
Does somebody have a FmrcAggregation Dataset on a publicly accessible
server? Eventually with access to the single model-runs as well?
And maybe, does somebody have grib-data on a publicly accessible server?
With aggregation?
Are there other examples of aggregations used for forecasts?
Best regards,
Heiko