Re: GRIB1 to GRIB2 transition on CONDUIT

Jerry,

NCO has a GRIB1->GRIB2 transition web page available here:
http://www.nco.ncep.noaa.gov/pmb/docs/GRIB1_to_GRIB2.shtml

The document does not as of yet have a final date when dual production
of
GRIB1 and GRIB2 products will cease. However, at this time, most of the
products on CONDUIT that are still being sent in GRIB1 format already
have
equivalent GRIB2 format replacements on the ftp server, so it is
possible to begin transitioning these products on CONDUIT as soon as
practical. As you noted in
your message, users have been informed of the GRIB2 transition for quite
some time, so they should be ensuring their applications are ready with
the available GRIB2 products available on the FTP servers. Several
CONDUIT users have already asked
for expedited transition of CONDUIT in order to better use available
bandwith and processing capabilities (and make room for other data
sets). 

We will provide timetables for transition of CONDUIT products when
available.

Steve Chiswell
Unidata User Support



On Wed, 2007-04-25 at 12:36 -0500, Jerrold Robaidek wrote:
> Steve,
> 
> When does the 180 day transition period start?   Is this the same transition 
> period that has been 
> talked about for the last 1 1/2 years?  Will the official time coincide with 
> a similar transition 
> period on the CONDUIT/ftp site? We have a lot of users needing to know the 
> schedule.  Thanks.
> 
> Jerry
> 
> 
> Steve Chiswell wrote:
> > David,
> > 
> > The difference is the parsing and creation of the GRIB2 metadata vs the
> > GRIB1 metadata where the routines used to create the metadata between
> > the "!" characters
> > is different. All of the "!grib2/" data will have the time as
> > yyyymmddhhFfff. And
> > the good news is, that is the direction the data is moving.
> > 
> > Since you are on the subject of GRIB2, we will need to begin the
> > transitioning of
> > GRIB1 products in the data stream to the GRIB2 versions as NCEP is now
> > creating both versions for most of the products. The 180 day period of
> > overlap allows
> > for a time of transition, and all users should be preparing their
> > applications that
> > rely on these products for GRIB2. We will provide notice for transition
> > of
> > products as changes can be scheduled with NCEP at the insertion end.
> > 
> > You can find the grib2 data on the ftpprd.ncep.noaa.gov server under the
> > pub/data
> > directory following the directory structure as in the CONDUIT products.
> > 
> > The CONDUIT GRIB1 data that are now available on NCEP's FTP server in
> > GRIB2 are:
> > 
> > NAM: awip3d, grbgrd, awipak
> > RUC2: [sbp]grb20 (grib2 is available on CONDUIT)
> > GFS: 1 degree pgrb (grib2 0.5 degree is available on CONDUIT)
> > 
> > Steve Chiswell
> > Unidata User Support
> > 
> > 
> > 
> > 
> > 
> > On Wed, 2007-04-25 at 09:54 -0500, David P Wojtowicz, Jr wrote:
> >> Hi,
> >>
> >>  
> >>
> >>  I noticed the following inconsistency in the CONDUIT data stream for
> >> the ensstat parts of the SPEC62MRF product.
> >>
> >>  
> >>
> >> data/nccf/com/gens/prod/gefs.20070425/00/ensstat/ensstat.t00z.pqpf_24hbc 
> >> !grib/ncep/SPEC62MRF/#002/200704250000/F024/PROB/sfc! 000000
> >>
> >>  
> >>
> >> Note the  / between the date and the forecast time
> >> (“200704250000/F024”) in the !! section.   This is different from
> >> pretty much all other products in the stream that combine the two
> >> without the slash separator (“20070425000F024”) such as:
> >>
> >>  
> >>
> >> data/nccf/com/gens/prod/gefs.20070425/00/pgrb2a/gep03.t00z.pgrb2af24 !
> >> grib2/ncep/SPEC62MRF/#000/200704250000F024/HGHT/925 Pa PRES! 000001
> >>
> >>  
> >>
> >> This throws off generic patterns in pqact.    One could make a special
> >> case pattern for it, but it seems like it is an inconsistency that
> >> should be corrected.
> >>
> >>  
> >>
> >>  
> >>
> >> --------------
> >>
> >> David Wojtowicz, Sr. Research Programmer
> >>
> >> IT Coordinator, SESE
> >>
> >> University of Illinois at Urbana-Champaign
> >>
> >> address@hidden   (217) 333-8390
> >>
> >>  
> >>
> >>
> 
> 
-- 
Steve Chiswell <address@hidden>
Unidata