[Met_help] [rt.rap.ucar.edu #59053] History for Verifying winds with Point-Stat

[John Halley Gotway via RT]

----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

Hi, the other problem has been resolved. Thank you.
I made a python script to convert the file metar and synoptic to 10 columns
format.
Now I have another problem, I need to know how to compare components U10
and V10 (forecast file) versus  wind speed and wind direcction (obs file).

- Is there any way to indicate the type of units used for variables to
convert and compare them?
- Could you tell me where to find more information about the configuration
file?

----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: Verifying winds with Point-Stat
From: John Halley Gotway
Time: Thu Nov 01 09:33:30 2012

Don,

I resolved the ticket for your first issue and have opened a new one
for the question you recently sent.  You have some questions about
verifying 10-meter winds using the MET Point-Stat tool.

Regarding the question of units, the answer is no.  It is the user's
responsibility to ensure that the units used for the forecasts and
observations are comparable.  MET is doing no automated unit
conversions behind the scenes.  It's just reading the forecast and
observation values as-is and comparing them.

More information about the MET configuration files can be found in the
METv4.0/data/config/README file.

Next, verifying winds depends greatly on the forecast file format
you're using - either the GRIB output of the Unified Post-Processor
(UPP) or the NetCDF output of pinterp.  Basically, you have to have
the GRIB output of UPP in order to verify winds.

If you're verifying winds from the GRIB output of UPP with Point-Stat,
your configuration file might look something like this:

fcst = {
   wind_thresh  = [ NA ];
   message_type = [ "ADPSFC" ];

   field = [
      {
        name       = "UGRD";
        level      = [ "Z10" ];
        cat_thresh = [ >=5 ];
      },
      {
        name       = "VGRD";
        level      = [ "Z10" ];
        cat_thresh = [ >=5 ];
      },
      {
        name       = "WIND";
        level      = [ "Z10" ];
        cat_thresh = [ >=5 ];
      }
   ];

};
obs = fcst;

This says to verify the U and V components of wind and the wind speed
at 10 meters.  However, there is a bit of derivation that's going on
(which you can see by setting the verbosity higher on the command
line: -v 4).  Since observations of wind are always defined to be
"earth-relative" and forecasts are often "grid-relative", MET
automatically rotates the forecast winds from grid to earth-relative,
when necessary.  Also, if you request wind speed (GRIB abbrev = WIND)
from a GRIB file, MET will look for the record in the GRIB file, but
if it can't find it, it'll automatically derive it from the U and V
components.

So you can verify U and V winds plus wind speed directly using Point-
Stat.  However, wind direction is a bit trickier.  You run Point-Stat
and dump out the VL1L2 lines.  Then you run a STAT-Analysis job to
dump out information about wind direction errors.  Please refer to
slides 17-20 from this tutorial presentation:

http://www.dtcenter.org/met/users/docs/presentations/MET_Tutorial_20110627/session3.pdf

Hope that helps.

Thanks,
John

------------------------------------------------