Hello,
There are known problems with the GEMPAK nagrib2 decoder for proper
GRIB2 messages containing probabilities of exceeding thresholds for some
parameters, specifically snow accumulations in the case of HPC's new
experimental probabilistic winter precipitation forecasts (PWPF). The
problem lies in the conversion of the threshold value stored in the
GRIB2 PDS to a character string interpolated into the parameter name.
For snowfall, GRIB2 requires units of meters, requiring a scaling factor
of 4 for the threshold, resulting in values too small to appear as
nonzero in the parameter name. Hence, all the exceedance probability
grids are assigned the same parameter name. Since HPC is producing
valid GRIB2 messages, we plan no changes to our product generation.
This problem does not occur for freezing rain. Here is a suggested
work-around for GEMPAK users (I admit, I have not actually tried it.):
1. Break the monolithic grid file into individual GRIB2 messages
using wgrib2 (http://www.cpc.ncep.noaa.gov/products/wesley/wgrib2/)
2. Run each GRIB2 message separately through nagrib2, producing
separate GEMPAK grid files.
3. Run gddiag to change the parameter names and write the grids to a
single GEMPAK grid file.
When running nagrib2, set the G2TBLS as follows: G2TBLS =
g2varlcl.tbl;g2varlcl.tbl, where g2varlcl.tbl is a file having these
contents:
!
! Snowfall (thresholds in m) exceedance probabilities:
!
!D# CT# ID# PDT#
000 001 029 009 Snowfall Prob > 1 inch 0->1
PS-- 0 -9999.00
!
! Freezing rain (thresholds in mm) accumulation exceedance probabilities:
!
000 001 225 009 ZR Accum Prob > .01 inch 0->1
PZ-- 0 -9999.00
I hope this information is helpful.
Keith
=============================
david.novak@xxxxxxxx wrote:
Not sure if you're on the gembud mailing list, but saw this post about the snow
probs. I don't think it was resolved.
------------------------------------------------------------------------
Subject:
[gembud] adding info to tables
From:
Jeff Lake - Admin <admin@xxxxxxxxxxxxxxxxxxxx>
Date:
Sun, 16 Jan 2011 07:47:17 -0500
To:
gembud@xxxxxxxxxxxxxxxx
To:
gembud@xxxxxxxxxxxxxxxx
good morning,
I'm trying to muddle my way creating images using the experimental
snow probability gribs from the HPC
ftp://ftp.hpc.ncep.noaa.gov/pwpf/conus/pwpf_24hr/
using dcgrib2 to convert to gem I get the following
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[GB 1] No GEMPAK parameter name defined for this grid.
[2213] 110116/1224[DECODE_GRIB2 -34] Could not determine parameter
name 0 1 29 9 [0]
[2213] 110116/1224[DC 5] Normal termination.
[2213] 110116/1224[DC 2] Number of bulletins read and processed: 7
[2213] 110116/1224[DC 6] Shutting down.
looking at that, studying the 'tables/grid/g2varswmo2.tbl' using
DeGRIB I added the following to the g2varswmo2.tbl
000 001 029 009 24 hr Prob of Total snowfall %
ProbASNOW24 ? -9999.00
well now no errors in the log, it does create the gem file
but when I do a gdinfo on it... the output file is empty
so something isn't correct
do I need to add the definition somewhere else ??
thanks
-Jeff
_______________________________________________
gembud mailing list
gembud@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit:
http://www.unidata.ucar.edu/mailing_lists/
