[Met_help] MET point stat problem with wind projection

John Halley Gotway johnhg at ucar.edu
Fri May 7 14:56:42 MDT 2010


Marlous,

OK, I see what's going on here.  There are two main issues - one that's easy to address and one that will take more time.

First, the easy one.  The GRIB file you sent me contains 4 records: UGRD, VGRD, TMP, DPT.  The records are defined on a lat/lon projection, and the winds fields are defined relative to the grid.  In
Point-Stat, the "process_grid()" subroutine is called to extract the grid definition from the file and setup masking regions and what not.  That "process_grid()" routine just reads the first record
from the GRIB file to extract that grid definition.  However, the first record in the file you sent is for UGRD.  And since your UGRD records are on a lat/lon grid and the winds are defined relative
to the grid, that's the problem.  When winds are defined to be grid relative, MET first rotates them to be earth relative before verifying them.  And for some reason, we haven't implemented that grid
rotation for lat/lon projections.

My fix for this first problem is, rather than reading the first record from the GRIB file, just use the first verification field requested in the config file to define the grid.  That's what we do in
the other MET tools anyway.  I'm not sure why the logic was different in Point-Stat.  The attached version of "point_stat.cc" does just that and should enable you to read these GRIB files with
Point-Stat.  I'll include this change in the development version of Point-Stat so it gets included in the next official release of MET.  But I won't post the change as a bug fix unless I get more
complaints about it.

So copy the attached version of "point_stat.cc" into METv2.0/src/point_stat/. and recompile MET.

Second, the harder issue.  We need to add support for rotating winds from grid relative to earth relative for lat/lon grids.  I'll talk to another developer here about this issue next week.  Until we
add that functionality, you won't be able to verify these wind fields using MET.

Hope that helps.

Thanks,
John Halley Gotway
met_help at ucar.edu

Marlous Jonker wrote:
> Hi John!
> 
>  
> 
> How are you doing? Since your last help with MET I've been able to
> produce PointStat statistics for temperature and dew point temperature
> using WRF output and it looks pretty good, so thanks again for your
> previous help! Next thing coming up is 10m wind verification.
> Implementing it in the config file, I encounter this error:
> 
>  
> 
> "ERROR: read_grib_record() -> rotation of non-conformal Plate Carree
> projection not yet implemented"
> 
>  
> 
> I even get this error when removing the wind from the config file as
> verification parameter. It's only when the U GRD and V GRD variables are
> removed from the grib file that I could get rid of this error.
> 
>  
> 
> I wonder what the meaning of this error is (something with the wind
> projection?) and whether there is something wrong with the grib file or
> some other reason I cannot think of.
> 
>  
> 
> In the attachments are the config file, MET ascii observation file and 2
> grib files: one _orig with the wind and one without the wind variables.
> 
>  
> 
> I hope you can help me with this problem too.
> 
>  
> 
> Cheers and thanks in advance,
> 
> Marlous
> 
>  
> 
> -------------------------------------------------------------------
> 
>   Marlous Jonker
> 
>   Meteorological Researcher
> 
>   Meteo Consult BV, Wageningen, the Netherlands
> 
>   Tel: +31 (0)317 399872
> 
>   Email: M.Jonker at weer.nl
> 
>   www: research.meteogroup.com
> 
> -------------------------------------------------------------------
> 
>  
> 
> 
> 
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: point_stat.cc
Type: text/x-c++src
Size: 60396 bytes
Desc: not available
Url : http://mailman.ucar.edu/pipermail/met_help/attachments/20100507/736fbf45/attachment-0001.bin 


More information about the Met_help mailing list