Hi John,
My experience has been that it's difficult to force a set of in-situ
observations into a netCDF file, no matter what conventions are used.
It's difficult to read and difficult to write and whenever I have to
deal with this type of netCDF file I always end up yelling at my computer.
An alternative approach is to package a set of individual netCDF files
(each representing one profile or time series, using the existing CF
convention) into an uncompressed jar or zip file. It's a lot easier to
deal with and no new CF attributes or names are required. It works well
with large datasets (I've used this convention successfully to package
several million observations from the World Ocean Database 2005 for use
in our Dapper/DChart server).
Perhaps this could be considered by CF as an alternative to packaging
the observations into one netCDF file.
Joe
cf-metadata-request at cgd.ucar.edu wrote:
>
> Hi Don:
>
> My own opinion is that CF is not currently adequate for writing observational data to NetCDF. The basic limitation in section 5.4 is that
>
> float humidity(time,pressure,station)
> float pressure(pressure);
> double time(time);
>
> requires the same number and values of the time and pressure coordinates at each station.
>
> We've made up some conventions at Unidata, and have been testing them. You can see it at:
>
> http://www.unidata.ucar.edu/software/netcdf-java/formats/UnidataObsConvention.html
>
> I plan to submit to CF as a proposal, but I havent had time to revise it, or write it up as a proposal. Comments by anyone would be welcome. If there were a few others wanting to see this through, perhaps its time to do so.
>
> John
>
> Don Morton wrote:
>
>> Hello,
>>
>> I apologize if this question is too elementary, or not "directly" applicable
>> to this list, but I've Googled and searched around a bit, and this seems
>> like my best avenue.
>>
>> I am interested in taking lots of surface obs (currently in flat ASCII files -
>> one file per station, records sorted by obs time) and putting them in a
>> netCDF format. I'd also like to do this only once, and it appears that maybe
>> CF is the convention I should adhere to.
>>
>> I'm wondering if there are example scripts or programs out there that
>> demonstrate
>> the writing of time series of surface obs (e.g. METARs) into a CF-compliant
>> netCDF file. I don't need any hand-holding (I hope); I've used netCDF API in
>> Fortran before (for simple stuff), and am just hoping I can find some example
>> that shows how the variables are written into the file, what
>> atrributes are attached,
>> etc. I've seen the CF 1.0 document, but just feel like an example would be so
>> much more helpful.
>>
>> Thanks for any help,
>>
>> Don Morton
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20070815/001ba27a/attachment-0002.html>
Received on Wed Aug 15 2007 - 11:42:29 BST
