this is a forward of an old email, discussing a problem encountered during the
conversion of a MM5 grib with a SCC (southern hemisphere) projection centered
over Rio de Janeiro, Brazil. The whole problem is described below!
---------- Forwarded Message -----------
From: "Guilherme O. Chagas" <goc@xxxxxxx>
Sent: Thu, 15 Sep 2005 16:09:47 -0200
Subject: almost there...
well, back with the MM5 SCC conversion that we discussed in the workshop...
I've got to actually decode the grib here (like we did before), but it
occurred to me that there's still something wrong with the output... Take a
look at the picture attached, comparing the GrADS output (left) with GEMPAK
(right). The red circles indicate an island as a reference point, and, as you
can see, the plot is pretty deviated. The UR corner is misplaced by a few
During the conversion, theres this warning in the log (using -v 4):
 050915/1429[DCGRIB 4] Warning, SCC grid with unexpected GDS.flag2 0
I looked at the source code, and there's a subroutine that "gathers" the grid
size - dcogfil.c . I tried to change it, in order to set the lat and lon
values manually, but it didn't work out the way I expected (I assumed that
another routine is also working with lat and lon variables).
Can I set the grid size manually when using dcgrib2?
I'll try to find another SCC model output (WRF perhaps) to see if the same
thing happens, I'll keep you posted!
Guilherme O. Chagas