[Met_help] [rt.rap.ucar.edu #45411] History for Neighborhood Method

[RAL HelpDesk {for Paul Oldenburg}]

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

Hi,

When I run grid-stat in the hopes of computing neighborhood method statistics, I end up with an odd problem... 100% of my forecast values are considered hits.  This obviously isn't correct, but I have yet to find the problem.  I've attached the needed files, but if you need anything else, let me know.  

Thanks,
Derek Stratman


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

Subject: Re: [rt.rap.ucar.edu #45411] Neighborhood Method
From: Paul Oldenburg
Time: Wed Mar 16 14:03:25 2011

Derek,

The reason that you see 100% hits for the NBRCTCs is that you
specified the following in the grid_stat config file:

cov_thresh[] = ["ge0.00" ];

This setting specifies a threshold for the coverage field, which has
the range [0,1].  Unfortunately, this feature is
not documented in the MET User's Guide and has been used largely for
research purposes only.  Typically, users who are
interested in using neighborhood method analysis focus on the Fraction
Skill Score (FSS) and Fractional Brier Score and
(FBS) in the NBRCTS output.  Interpreting verification results based
on the NBRCTC information can be difficult.

Please let us know if you have any questions.

Thanks,

Paul


On 03/16/2011 10:52 AM, RAL HelpDesk {for Derek.Stratman} wrote:
>
> Wed Mar 16 10:52:27 2011: Request 45411 was acted upon.
> Transaction: Ticket created by Derek.Stratman at noaa.gov
>        Queue: met_help
>      Subject: Neighborhood Method
>        Owner: Nobody
>   Requestors: Derek.Stratman at noaa.gov
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=45411 >
>
>
> Hi,
>
> When I run grid-stat in the hopes of computing neighborhood method
statistics, I end up with an odd problem... 100% of my forecast values
are considered hits.  This obviously isn't correct, but I have yet to
find the problem.  I've attached the needed files, but if you need
anything else, let me know.
>
> Thanks,
> Derek Stratman
>


------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #45411] Neighborhood Method
From: Derek.Stratman
Time: Wed Mar 16 14:49:35 2011

Paul,

I think I'll stay clear of the coverage thresholds and focus only on
the FBS and FSS.  I am getting reasonable values for those, so I don't
think I have anymore questions at this time.

Thanks for your help!

Derek

----- Original Message -----
From: RAL HelpDesk {for Paul Oldenburg} <met_help at ucar.edu>
Date: Wednesday, March 16, 2011 3:03 pm
Subject: Re: [rt.rap.ucar.edu #45411] Neighborhood Method

> Derek,
>
> The reason that you see 100% hits for the NBRCTCs is that you
> specified the following in the grid_stat config file:
>
> cov_thresh[] = ["ge0.00" ];
>
> This setting specifies a threshold for the coverage field, which
> has the range [0,1].  Unfortunately, this feature is
> not documented in the MET User's Guide and has been used largely
> for research purposes only.  Typically, users who are
> interested in using neighborhood method analysis focus on the
> Fraction Skill Score (FSS) and Fractional Brier Score and
> (FBS) in the NBRCTS output.  Interpreting verification results
> based on the NBRCTC information can be difficult.
>
> Please let us know if you have any questions.
>
> Thanks,
>
> Paul
>
>
> On 03/16/2011 10:52 AM, RAL HelpDesk {for Derek.Stratman} wrote:
> >
> > Wed Mar 16 10:52:27 2011: Request 45411 was acted upon.
> > Transaction: Ticket created by Derek.Stratman at noaa.gov
> >        Queue: met_help
> >      Subject: Neighborhood Method
> >        Owner: Nobody
> >   Requestors: Derek.Stratman at noaa.gov
> >       Status: new
> >  Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=45411 >
> >
> >
> > Hi,
> >
> > When I run grid-stat in the hopes of computing neighborhood
> method statistics, I end up with an odd problem... 100% of my
> forecast values are considered hits.  This obviously isn't correct,
> but I have yet to find the problem.  I've attached the needed
> files, but if you need anything else, let me know.
> >
> > Thanks,
> > Derek Stratman
> >
>
>
>

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