[Met_help] wind direction or other parameters in MET (UNCLASSIFIED)

[John Halley Gotway]

Barb,

You're correct, in METv1.1, we do not verify wind direction directly.  METv1.1 can be used to process the U and V components of wind separately as scalars and can be used to compute vector L1 and L2
partial sums (VL1L2) line.

We've been requested to put in some better methods for handling the verification of wind direction.  However, it is a pretty tricky variable to handle.  And the statisticians in our group are not
satisfied with the way it's typically been verified in the past.  So here's our basic plan for how we'll handle the verification of wind direction in ME

(1) Short-term solution.  For METv2.0 we've made a few changes that will enable users to do some verification of wind direction.
   (A) First, we've added the computation of the VL1L2 partial sums to the Grid-Stat tool.  So now, Point-Stat and Grid-Stat create VL1L2 output lines.
   (B) Second, for Point-Stat and Grid-Stat, we've added parameters to the config file that enable the user set one or more wind speed thresholds.  These are used when computing the VL1L2 lines.  For
example, you can compute vector wind partial sums for all winds (wind speed >= 0), light wind (wind speed >= 2m/s), or strong wind (wind speed >= 5m/s) - please note that I'm a software engineer, so I
just made those thresholds up!  The idea is to only look at those points with a reasonably strong wind so that the wind speed is well-defined.
   (C) Third, we've added a job to the STAT-Analysis (formerly call VSDB-Analysis) tool to convert VL1L2 lines into information about errors in wind direction.

(2) Long-term solution.  Some of the people working on the MET project are also working in NCAR-RAL on the wind projects for Excel energy.  That project includes the development of appropriate
verification methods for wind direction.  The goal is to take the methods developed for Excel and include them in a future release of MET.

To answer you're questions directly...

No, unfortunately Point-Stat will not handle observations of wind direction directly.  Point-Stat and Grid-Stat work with the U and V components of the wind since those are what the models typically
produce and how observations are typically available.

Thanks for sending the verification tables from 4DWX.  I believe that MET could be configured to produce results stratified in a similar way, and one could write a script to generate a similar html
table.  While none of the 4DWX people work directly on MET, we did meet with at least one 4DWX person to discuss specifically how they handle the verification of wind direction!

I hope that answers your questions.  We're planning to release METv2.0 in the first week of April.  We're finishing up documentation and testing on it now.  We realize that verifying wind direction is
important, and our goal is to provide both useful and appropriate methods to do so.  If you have more questions about methods, let me refer you to Tressa Fowler (tressa at ucar.edu).  She's the
statistician who's leading the MET development effort.

Thanks Barb,
John


Sauter, Barbara (Civ, ARL/CISD) wrote:
> Classification:  UNCLASSIFIED 
> Caveats: NONE
>  
> Wind Direction doesn't seem to be a choice for either PointStat or the
> WRF Post Processor. Is that correct?
> 
> If I have both an observed parameter and the forecast value from the
> post-processed WRF file for that parameter other than the 10 listed in
> the PB2NCConfig and VSDB_Analysis documentation, will PointStat (and
> VSDB_Analysis) handle it?
> 
> I like the verification tables listed on the White Sands Missile Range
> 4DWX web page (example attached), although there's not an equivalent
> table for surface data, just charts.  I believe the site-specific 4DWX
> system was developed by NCAR before the MET was developed.  Were some of
> the same people involved?
> 
> Thanks,
> Barb
> 
> Barb Sauter
> US Army Research Laboratory
> White Sands Missile Range, NM
> (575) 678-2840
>  <<2009032000_d1.html>> 
>  
> Classification:  UNCLASSIFIED 
> Caveats: NONE
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help