[Met_help] Re: Problem with point-stat procedure

[John Halley Gotway]

Valerio,

Sorry for the delay in getting back to you.  I was out of the office last Friday.

I took a look at this issue today, and I think I know what's going on.  I believe that the grid in your GRIB file is defined in a way that we haven't encountered before.  And the library code needs to
be updated to handle your data.

Typically when defining a lat/lon grid, the GRIB file specifies the following:
- lat/lon of the lower-left corner
- lat/lon of the upper-right corner
- number of points in the lat direction
- number of points in the lon direction
- step size in the lat direction
- step size in the lon direction

In your GRIB file, it appears that the last two are omitted.  Instead they should be computed from the other values.  But I think that according to the GRIB specification, that's fine.  We just
haven't encountered this variant of GRIB yet.  So I'll try to put together a fix that will address this and send you an updated file to try.

Thanks,
John


capecchi wrote:
> John, I'm writing (again) to you to overtake a problem with the point-stat
> procedure.
> Syntax of the command:
> point_stat meteosalute_2007012100-2007012100_d01.grb obs.ncPointStatConfigMETEOS
> (in attach you can find input files)
> It produce, among the others, the attached continuos statistics (cnt file)
> but as you can see
> the forecast value 72.2 for variable pressure tendency is not the same as
> you can dump from
> the grib file 96.6998....
> I guess there's some problem with the definition of masking region (I tried
> proper(?) definition for mask_grids[] and mask_polys[] and mask_stations)
> but I had the same result.
> Where am I wrong??
> Thanks, Valerio
> 
>