[Met_help] RMSE for wind speed in MET

[John Halley Gotway]

Thomas,

Actually, yes, MET can be used to verify wind speed.  However, MET treats wind speed as it does any other continuous field of scalars.  If you're verifying with gridded observations, you'd use the
Grid-Stat tool.  If you're verifying with point observations, you'd use the Point-Stat tool.  In either case, you could verify wind speed with the following setting in the Grid-Stat or Point-Stat
config file:

fcst_field = [ "WIND/Z2" ];
obs_field  = [];

You'd use that to verify 2-meter wind speed.  There is some special logic for handling wind speed and here it is:
(1) If wind speed exists in the gridded input file(s), use it as is.
(2) If wind speed does NOT exist in the gridded input file(s), look for the corresponding U and V wind components, and derive wind speed from them.

This only applies to gridded input files.  If you're verifying using point observations, you'll need to provide the wind speed observations yourself.  The only caveat to that is if you're using
PREPBUFR point observation files, the PB2NC tool can be configured to derive wind speed for you.

Once you are successfully verifying wind speed, you'll find the RMSE values for it in the CNT (i.e. continuous statistics) output line type.

In addition, you could dump out the SL1L2 partial sums for wind speed, and use the STAT-Analysis tool aggregate those SL1L2 partial sums through time.

Hope that helps, and let me know if any more questions come up.

John

Thomas Schwitalla wrote:
> Hi John,
> 
> I am currently working on a publication. One of the editors suggested to
> also plot the RMSE of wind speed.
> As far as I see, this is currently not possible with MET - is this correct?
> 
> Best regards
> Thomas
>