Fast Downloading of GRIB Files with get_gfs.pl

[Valerio Guerra]

Hi All,
I'm testing 'get_gfs.pl' on grib2 GFS files. It works very well on FreeBSD 6.1 with 'wgrib2' and 'g2ctl.pl' and I'm happy with it.
Anyway I discovered a "strangeness" which should be caused by the 'curl' multipart http response while getting bytes ranges.
Actually in the locally downloaded grib files there are some "extra-data" that is for each retrieved variable, before the 'GRIB' starting string there's something like the following:

> --47bc0e812627
> Content-type: text/plain; charset=iso-8859-1
> Content-range: bytes 31882619-31947871/47503363
>

For instance the following call:

> perl get_gfs.pl data 2008022000 6 9 3 PRMSL:APCP surface:mean_sea_level /usr/home/valerio/gribs

dowloads 2 grib2 files (gfs.t00z.pgrb2f06, gfs.t00z.pgrb2f09) each with 2 variables and each with "extra-data" before the 'GRIB' starting string (the multipart id is also placed at the end of each file, after the '7777' grib ending string).

This seems not to be a problem for wgrib/wgrib2 but other grib software like 'cnvgrib' do not work on such grib/grib2 files.

Do you know if such extra data in grib/grib2 files could cause problems or unpredictable results when using Grads or other grib software?

Valerio Rossi Guerra