20010410: GRIBDEC on ncep AVN fields

>From: Anthony James Wimmers <address@hidden>
>Organization: UVa
>Keywords: 200104101932.f3AJWCL20597 McIDAS-X gribdec.k AVN

Tony,

>I was wandering if you could take some time off of solving
>the compiler problem to help me out with this other thing.

???  I am not quite sure what you are referring to.

>I'm trying to decode some grib files from an ncep archive,
>and it's not working out correctly. I did an experiment to
>find out whether it was a problem in the files, or a problem
>related to the decoder, and I think I've narrowed the problem
>down to the decoder (gribdec).

OK.  I hate to remind you of this, but gribdec.k is not officially
supported.

>Here's the experiment that you can try over there, and maybe
>you can point out my mistake based on that. (BTW, I'm using
>mcidas version 7.600.)
>
>1. ftp a recent initialization of the AVN from the NCEP ftp server:
>
>ftp.ncep.noaa.gov/pub/data/grib/avn
>
>2. Decode using gribdec.k, for example:
>
>gribdec.k avn_010410_06_00 6000 MAX=6000

OK, I grabbed avn_010411_06_00 since that was what was available today.
I did the decoding:

gribdec.k avn_010411_06_00 6000 MAX=6000

and all seemed to go smoothly.  A GRDLIST of the GRID file produced looked
normal as well:

GRDLIST MYDATA/GRIDS.6001 NUM=ALL
Dataset position 6001   Directory Title=
PAR    LEVEL       DAY        TIME    SRC FHOUR     FDAY       FTIME  GRID  PRO
---- --------- ------------ -------- ---- ----- ------------ -------- ----- ----
Z    1000 MB   11 APR 01101 06:00:00  AVN     0 11 APR 01101 06:00:00     1 MERC
Z     975 MB   11 APR 01101 06:00:00  AVN     0 11 APR 01101 06:00:00     2 MERC
Z     950 MB   11 APR 01101 06:00:00  AVN     0 11 APR 01101 06:00:00     3 MERC
 ...

>3. Display one of the grids in mcidas, for example:
>
>GRDDISP LOCALDATA/GRIDS.6010 PAR=RH LEV=500

I chose T for 850:

EG;GRDDISP MYDATA/GRIDS.6001 PAR=T LEV=850 CINT=5 UNIT=C MAP=CHINA

>What it shows me at this point is a set of very horizontal
>contours that occupy rectangle on the map that goes from
>the equator to about 70 degrees north-south, and from east
>Asia to the prime meridion east-west. (Maybe it goes all the
>way north to the pole, but doesn't change enough to contour.)

I got a similar plot, but mine only went as far west as about -100
(this is why I did my plot over CHINA).

>Do you think it's a scaling problem with the lat/lon coordinates?

No, after checking to see that GEMPAK could decode the grid correctly, I
am left with the impression that the decoder doesn't seem to understand the
grids (type 3) in this grib file.  It is decoding it as if were a type
2 grid (2.5 degree spacing, etc.).

gribdec.k is creating output grids that are 2.5 degree, when the grids
are actually 1 x 1 degree.  Why this is the case, I can't answer.

>I think that would be strange, because I don't have a
>problem decoding ruc2 gribs from the same site.

I will have to see why gribdec.k is not reading the PDUS word that tells
it what grid type the grib messages contain.

>Thanks,

Sorry I couldn't come up with anything right off that would make things work.

Tom