Your right, the String valued "station" is a extension to CF-1, and not
obvious, though it is not against any CF-1 rule and it should work in a
general CF-1 reader.
Looking at your UnidataObsConvention link, I must admit that this is a
very useful extension to CF-1, and I would like to see it included in
the successor to CF-1. I had some problems adding extensions to CF-1.0,
because in 2.6.1 it says that the Conventions string must be 'CF-1.0',
while the netcdf documentation allows for a comma-separated list of
conventions. I tried some programs (visad, cfchecker.py) and they had
problems when I created files with Conventions = "CF-1.0,GDV".
You leave me puzzled with your comment about "NATTOT_year_cf.nc". This
is a CF-1.0 grid file in polar_stereographic projection I have some
problems with reading into IDV. CF section 5.4 is about Timeseries of
station data. I thought I send you links to two files:
StationObs in CF-1.0: ftp://emepftp@xxxxxxxxxx/heikok/pointData.nc
polar_stereo grid in CF-1.0:
ftp://emepftp@xxxxxxxxxx/heikok/NATTOT_year_cf.nc
Best regards,
Heiko
John Caron wrote:
We dont yet have a Point-Data reader for CF-1, partly because I havent
seen a real file yet (until now), and partly because Im not sure the CF
spec is complete enough. Theres been some discussion of this on the CF
list that I need to review. From my POV, its not obvious what the
String valued "station" coordinate means. The CF spec doesnt mention
such a beast explicitly.
Anyway, looking at your file "NATTOT_year_cf.nc" , I would expect an
attribute coordinate = "lat lon" on each data variable; see CF section 5.4.
Ill look at the feasibility of adding a CF station reader when i get
back next week and see about reading your files into it ( I probably
need you to add the coordinate attribute). You might want to add
altitude info also if you can.
You might find this interesting:
http://www.unidata.ucar.edu/software/netcdf-java/formats/UnidataObsConvention.html
