Glenn.Rutledge wrote:
Hello John,
Thanks for your note and offer to help and provide an alpha. Any
time-scale you have for delivery is good with me! Dan Swank can
provide details of our current configuration- Dan?
btw- Steven Anthony has just been hired to assist with our TDS and
he's cc:ed.
Do you guys convert the NCEP grib to NetCDF or do you have the OPeDNAP
portion of the server encapsulate?
The netcdf-java library converts GRIB-1 and GRIB-2 into the "Common
Data Model" on the fly, which we can then serve through OpeNDAP, WCS, or
write a NetCDF/CF file.
Finally, for a fist cut- we'll need basic (non-aggregation)
capabilities for grib1-2 and that will also help get SteveA up to
speed on TDS in general. Thanks so much- Glenn
Sounds good. Ethan will be the main support contact for NOMADS, but send
messages to support-thredds@xxxxxxxxxxxxxxxx, and whoever's available
will help.
John Caron wrote the following on 8/15/2006 5:03 PM:
Glenn.Rutledge wrote:
Hi John and Ethan
SteveH passed your note to me. I think you're aware NOMADS / NCDC
would be very interested in the new aggregation capabilities in the
TDS. Not sure if Steve was implying we install the code or hit
your new server as a beta. My preference is to adopt your latest
here for testing and ensure NOMADS can provide archived NWP.
Yes, we'd like to get you up and running.
I should be able to give you an alpha release in a week or so. There
are some performance issues for a site as large as yours. I assume
that will take another month or so to get a production-worthy version
done. Does that time frame work for you?
On our current install, we are having problems understanding the xml
to the point where we can point to a file (grid), and wonder if you
or Ethan can assist (is there a TDS helpdesk?). We need this TDS
of course since it's advertised to handle grib2- and not sure how
far along Jennifer and Brian are for that at COLA. For our GDS,
we've developed a grib2-->1 dumming down, for subsetting etc. But
the long term solution is the TDS I feel.
Especially for GRIB files, you likely need this new "Forecast Model
Run Collection Aggregation" capability. We have been working with our
IDD NCEP GRIB files, and there are some complications, especially
non-homogenaity due to missing records and variable time and vertical
dimensions, that cant really be solved by the current (index based)
aggregation.
Ethan and I will work closely with you guys to get this working. I'd
like to understand what you have in more detail, number and types of
files, how they are stored, etc. Can you or someone summarize?
John
