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

Re: 20001031: WMO header now includes model type?? (fwd)




===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================

---------- Forwarded message ----------
Date: Tue, 31 Oct 2000 10:58:22 -0700
From: Steve Chiswell <address@hidden>
To: Anne Wilson <address@hidden>
Subject: Re: 20001031: WMO header now includes model type??



Anne,

This list of IDs has been part of the grib388 document for a long time 
(Table A-5), they've taken a long time transitioning everything from
KWBC to KWBx.... and still not complete. They started with NGM about 4 years
ago...this is not new.

Below is the output of a grep for grib products from nwstg, with the
resultant model id (PDS 6) and the 4 letter distribution CCCC.

AVN/            KWBC
MRF/            KWBC
NOW/            KWBC
NWS_0/          KWBC
RUC/            KWBC
SPEC62MRF/      KWBC
SSIAVN/         KWBC
SST/            KWBC
ETA_89/         KWBD
ETA_84/         KWBE
NGM/            KWBF
RUC/            KWBG
SPEC62MRF/      KWBH
SSIFNL/         KWBH
MRF/            KWBH
NOW/            KWBJ
NWW_121/        KWBJ
NWW_122/        KWBJ
NWS_120/        KWBM

NWS_185/        KMSR

ECMWF_197/      ECMF
UKM_15/         EGRR
UKM_45/         EGRR


You'll notice that they still do not have all RUC ID's as KWBG. For instance
this hour, we still get KWBC like:

Oct 31 16:44:05
grib/ncep/RUC/#211/200010311600/F000/UGRD/10 m above gnd/ YUQA98 KWBC 311600 

Oct 31 16:44:05
grib/ncep/RUC/#211/200010311600/F000/VGRD/10 m above gnd/ YVQA98 KWBC 311600


So, as of this hour, you can't omit that /m tag for RUC.

As for the model grid, that is given by the A(1) in the T(1)T(2)A(1)A(2) CCCC,
for instance ETA on grid 211 would have A(1) = Q, while ETA on 212 A(1) = R.
They still maintain a difference in KWBD and KWBE for off hour ETA.

The biggest confusion is probably the AVN/MRF nomenclature. You see multiple
entries for those KWBC and KWBH. This is because they have 3 different
model IDs (for the SSI, FNL and the forcast hours).


The lines I show above are from the ingestor I am running on jackie. The
TTAAII CCCC DDHHHH is prepended with more verbose information from the GRIB
PDS itself:
data type = grib
model center = ncep
model = RUC
grid = 211
model initial time = 200010311600
forecast time = F000
parameter = VGRD
level = 10 m above gnd

As Don mentions, the toughest thing is maintaining translation tables.

The primary question I would raise is, would users find it useful
to add the supplemental information, or do we force them to
know that KWBG is RUC? This seems to be something that time after time
they have problems with.


Chiz





On Tue, 31 Oct 2000, Anne Wilson wrote:

> Don Murray wrote:
> > 
> > >From: Anne Wilson <address@hidden>
> > >Organization: UCAR
> > >Keywords: 200010311604.e9VG4I402762
> > 
> > Anne-
> > 
> > >The following message suggests that we may not need the /m meta tag...
> > 
> > It's much easier to use a /m tag in a regular expression than the
> > last letter.  Also, I don't know whether the last letter is explicit
> > enough (Chiz probably does).  We get the ETA on at least 4 different
> > projections - is there a separate letter for each?  What happens
> > when they have more than 26 models?
> > 
> 
> The message said that the last *four* chars of the ID indicates the model.  
> That gives plenty of possibilities for specifying lots of stuff, including 
> the projection.  
> 
> Darien said Tom Kent has a list of all the IDs.  I'll contact him (I knew him 
> at FSL) and ask him for the list - maybe that will give us more info.
> 
> Anne
>