Re: Netcdf Observation Dataset Conventions

Janet Scannell wrote:
> John,
>
> Janine Goldstein told me that she had been working with you about 6
> months ago to develop a standard for some of EOL's data.  I am currently
> working on converting some data to netCDF and I have some questions
> about the document that you have written called Unidata Observation
> Dataset Conventions.  The data in one netcdf file are observations at
> one particular station with a specific latitude, longitude and altitude
> for the station.  There are observations at multiple times in each
> file.  At each time, there are multiple heights at which the
> observations occur.  This sounds similar to your described Profile
> Observation dataset, except this data is not "Profiler" data and this is
> the cdm_datatype that is required in the global attributes.  Since we
> didn't want to say this is Profiler data, we were just going to use the
> Point Observation dataset type.  Could the cdm_datatype be changed to
> call it "Profile" instead of "Profiler"?

yes, it should be "profile", I think we intended to correct that.

>
> The other question I have is about the "Identifying the Coordinate
> Variables" section.  Since I have observations at multiple heights, does
> height need to be included in the coordinate variables or is just
> latitude, longitude, altitude, time enough?

im not sure...

>
> I would like the netcdf file I am creating to be IDV compliant, also.
> What is the best way to accomplish this?

we are proposing a new convention to CF, but in the meantime the nidata
Observation convention is
the best to work with. If you are looking for a long-term format, you should
consider switching to
the CF convention when its ready.

>
> I have also been reading the "Writing NetCDF Files: Best Practices" and
> "NetCDF Attribute Conventions for Dataset Discovery" documents.  I have
> a netcdf file created with this data if that is any help.

yes, send me the example....

>
> Thanks,
> Janet