[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Fwd: Re: bug in CF conventions for "Hurrell SST climatology" files]



Hi Brian:

We think maybe you are the person that should know about a problem with a file we are using as an example of CF Conventions.. If not, can you forward to correct person? If you have a file thats been fixed, let me know, and i'll swap it for the one on our examples page. We also have an experimental validator at

http://motherlode.ucar.edu:9080/thredds/cdmValidate.html

that might be helpful.



-------- Original Message --------
Subject:        Re: bug in CF conventions for "Hurrell SST climatology" files
Date:   Fri, 10 Feb 2006 19:09:45 -0700
From:   Brian Eaton <address@hidden>
Organization:   UCAR/Unidata
To:     John Caron <address@hidden>
CC:     Russ Rew <address@hidden>, Michael Burek <address@hidden>
References:     <address@hidden>



Hi John,

My guess from looking at the global attributes is that Brian Kauffman was
the person responsible for this file.  So it would be worth pointing this
out to him.  My guess is that later versions of this file may not contain
that error (it is definitely and error -- the units are not legal unless
they're contained in the udunits.dat file).  I think it would be good to
replace this file with one that conforms to conventions.  Either Brian
could provide a more recent version, or the NCO attribute editor (ncatted)
could be used to fix the attributes in the file you have.  Note that in
addition to the "degrees north" error, you should replace "degrees east" by
"degrees_east" and "degrees C" by "degree_C".  As far as I can tell there
are no plural forms of "degree" when it refers to a temperature unit.

Brian


On Fri, Feb 10, 2006 at 01:55:34PM -0700, John Caron wrote:
Hi Brian:

We just happened to be checking some example CF files with our CF validator, and noticed that

http://www.unidata.ucar.edu/software/netcdf/examples/sst_clim_hurrell_1x1_020827.nc

has an error in it that prevents recognizing the grids. (uses "degrees north" instead of "degrees_north")

Im not sure if this is systematic, and if we should notify Jim Hurrell or Brian Kaufman or T Craig or ? I wonder if CGD wants to systematically check their data?? Could the CDP check and send error messages? Should we change CF to allow this ??

any advice?


heres the global attributes:

:title = "Hurrell SST climatology (Reynolds SST + Hadley anomolies)";
:history = "Tue Aug 27 16:17:06 2002: ncatted -O -a source,global,a,c,\n created Aug 27 by T Craig from sst_clim_hurrell_1x1_02074.nc, set ice_cov < 1% (0.01) to zero sst_clim_hurrell_1x1_020827.nc Tue Aug 27 16:15:13 2002: ncks -A -v ifrac a1x1.nc sst_clim_hurrell_1x1_020827.nc
Tue Aug 27 15:51:00 2002: ncrename -v nifrac,ifrac a1x1.nc";
:Conventions = "CF-1.0";
:source = "/fs/cgd/data0/shea/jhurrell/MODEL.had+oiv2.sst.mnly.49-01.unf.nc /fs/cgd/data0/shea/jhurrell/MODEL.had+oiv2.ice.mnly.49-01.unf.nc created Aug 27 by T Craig from sst_clim_hurrell_1x1_02074.nc, set ice_cov < 1% (0.01) to zero"; :comments = "Jim Hurrell (NCAR) combined the Hadley Center anomalies with the Reynolds SST climatology to produce SSTs over the ocean only (2001?). Several tests were run to determine how values over land were to be created. /fs/cgd/home0/shea/ncld/jhurrell/sst.ncl Tests indicated that direct interpolation of the full SST grids would be satisfactory. Note: values over land are entirely bogus. In particular, values over Siberia are particularly unusual. However, values near coasts are close to those at nearby oceans. The grids produced for this file were created using the NCL function cssgrid which interpolates randomly spaced data on a sphere to a lat/lon grid. The original grid values were then used to overwrite the interpolated values over the oceans. The land values are the ONLY interpolated values on these grids. The NCL script is: /fs/cgd/home0/shea/ncld/jhurrell/sstDirect.ncl /fs/cgd/data0/shea/jhurrell/had+oiv2.sst.mnly.49-01.unf.nc This nc file was created from the binary source file using: /fs/cgd/home0/shea/ncld/jhurrell/sstBinary2nc.ncl ------------------------- Jim Hurrell provided the sea-ice percentages. Basically, linear interpolation [East-west] was used. /fs/cgd/home0/shea/ncld/jhurrell/iceDirect.ncl /fs/cgd/data0/shea/jhurrell/had+oiv2.ice.mnly.49-01.unf.nc This nc file was created from the binary source file using: /fs/cgd/home0/shea/ncld/jhurrell/iceBinary2nc.ncl -------------------------
Brian Kauffman (NCAR) formed a 12 month climatology and combined
the SST and ice fraction data (described above) into one file appropriate for used by the docn5 & dice5 CCSM2 data models.
The data were padded with -0.5 & 360.5 longitude bands
plus -90 and +90 latitude bands to facilitate interpolation
onto other grids (eg. pop shifted pole, etc.).";