NOAA / NWS's Meteorological Development Laboratory
Degrib (NDFD GRIB2 Decoder)
Subject: Degrib version 1.18
Date: Fri, 06 Jun 2003 18:59:55 -0400
Organization: DOC/NOAA/NWS - National Weather Service
To: Arthur Taylor
You all have the unique distinction of being registered on my degrib
(aka NDFD_GRIB2Decoder) mailing list (all 63 of us) (I'd like to
welcome: mmuttray and apiradee_s059)
Please let me know (email me) if you want to unregister.
First, some news...
1) As some of you may have noticed, Temp and Dew Point for the CONUS
started appearing on somewhere around 6/3/2003. I think it was
officially announced that there would be Temp and Dew Point starting
today. (Note: the http option in version 1.17 timed out on the CONUS
Temp and Dew Point files, I have now fixed that).
2) On June 16, files are also expected to be available with forecasts
for PoP12, Weather, Sky Cover, QPF, Snow Amount, Wave Height, Wind
Direction, and Wind Speed. These forecasts will be for the CONUS as
well as the 16 Overlapping CONUS Sectors and Puerto Rico.
In order to take advantage of 2), you will have to go to
"File->configure", and then Highlight the variables that you want to
make active, and press the "Activate" button. Then you will have to go
to the Sectors Tab, Highlight the Sectors that you want to make active,
and press the "Activate" button. (Note: In previous versions of
tkdegrib I did not have variables called "Wx" or "WaveHeight", I have
now fixed that)
3) ***** I now have limited experimental GrADS support ***** (keep
I just wanted to let you all know that I have just finished putting
together version "1.18" (last one that I announced was "1.17" on
The latest version of the program is on:
***** GrADS support *****
1) GrADS support: GrADS is a software package that has been around for a
while, and is available for free from http://Grads.iges.org/. There is
a PC version (the one I used), and several Unix versions and even a mac
version. GrADS can take for input either GRIB1, NetCDF, or a binary
file (it even claims to be able to convert to NetCDF). The binary file
is the same as what I termed a ".flt file". The only thing that was
lacking was a GrADS control file (or .ctl), which I am now producing.
NOTE: GrADS usually uses unprojected data. When it starts dealing with
what it considers "pre-projected" data, it interpolates the data to a
lat/lon grid specified in the ASCII .ctl file. Almost all GRIB and
GRIB2 grids are what GrADS considers "pre-projected", so one needs to
provide information for GrADS to perform the projection. (for a better
explanation see: http://Grads.iges.org/grads/gadoc/gadocindex.html ::
THE FIRST CATCH: GrADS Hardwired a Radius of Earth = 6371.2 km. Most
GRIB files use 6367.47 km, and NDFD is no exception. It is not
currently possible to tell GrADS to use a different Radius of Earth.
(For ArcView, I wrote an .ave script, which if used, sets the radius of
earth correctly. Note you don't need it for the .shp files, just the
THE SECOND CATCH: GrADS does not have a mercator projection definition,
and its north polar stereographic definition is prone to round off
errors. I have set it to try using the lambert conformal conic
projection for the mercator projection (since mercator is a special case
of a lambert conformal conic projection).
THE THIRD CATCH: GrADS on my PC version, when it tried projecting the
NDFD lambert conformal conic grid, took forever and then ran out of
SOLUTION: Use degrib (and tkddegrib's) Interpolate option. This will
create a lat/lon grid with approximately the same coverage and
resolution as the original grid, and GrADS works quite well with the
interpolated grid. Example usage: "degrib ndfd.grib -C -msg 1 -Flt
-Interp -GrADS" should create an interpolated .flt file and a .ctl file
which GrADS can use. Remember you are interpolating here so you are not
entirely true to the data (but you wouldn't be true to the data if GrADS
interpolated for you.)
Future: GrADS prefers to have all the data in a big data cube (similar
to what NetCDF wants). Eventually I will get around to dumping the
whole GRIB2 file to a data cube, instead of just wrapping the .flt file
which is just a slice of the data cube. (Note potentially one could
simply cat the .flt files together to create the cube, and modify the
.ctl file a little).
----------- In other news...
2) I now allow one to call the program using "degrib <filename>" instead
of requiring a -in option.
3) Fixed the timeout issue with regards to the http download method.
4) Added Wx and WaveHeight to tkdegrib.
5) Worked on Wave Height meta file support.
As always, I tested this release on my PC (win 2000), a Linux machine, a
HP-Unix machine, and an AIX-Unix (machine). Although I may have missed
something in my testing, it should at least compile, and run for some of
the test cases. For those of you with access to the NWS NCEP machine,
my current version is in: /nfsuser/g06/we25at/degrib (on snow/frost, I'm
no longer using asp/bsp). For those with access to MDL's GDP or NHDW
machines, see: ~taylor/degrib/
Please let me know if you have problems.