[Met_help] [rt.rap.ucar.edu #68397] History for a question about rank histogram in Ensemble-Stat

John Halley Gotway via RT met_help at ucar.edu
Mon Jul 28 09:53:25 MDT 2014


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

Dear sir or madam,
  I have a question about the rank histogram computed by Ensemble-Stat. If there are several ensemble members forecast the same value as the observation, which is quite common in precipitation forecast, will MET add 1 to the first bin, or will it add 1 to a randomly picked bin among those equal members? Thanks!

J.Zhu





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

Subject: a question about rank histogram in Ensemble-Stat
From: John Halley Gotway
Time: Fri Jul 25 09:17:40 2014

J. Zhu,

Very good question!  When ensemble_stat encounters ties while
computing
ranked histograms, it randomly assigns the rank from the available
tied
ranks.  If you run ensemble_stat twice on a dataset with tied ranks,
you'll
likely get slightly different ranked histograms in the output.  The
ensemble_stat config file does contain an entry to specify the random
number generator to be using in this random assignment:

//
// Random number generator
//
rng = {
   type = "mt19937";
   seed = "";
}

The seed value is the "starting point" for the random number
generator.  By
default, the "seed" value is unset and the tool will choose a somewhat
random seed value based on a permutation of the current time.  If
you'd
like your results to be exactly repeatable from run to run, all you
need to
do is set that seed value.  For example, just set... seed = "1"; and
your
results should no longer change from run to run.

Thanks,
John Halley Gotway
met_help at ucar.edu


On Thu, Jul 24, 2014 at 7:35 PM, zhujiangshan via RT
<met_help at ucar.edu>
wrote:

>
> Thu Jul 24 19:35:52 2014: Request 68397 was acted upon.
> Transaction: Ticket created by zhujiangshan at mail.iap.ac.cn
>        Queue: met_help
>      Subject: a question about rank histogram in Ensemble-Stat
>        Owner: Nobody
>   Requestors: zhujiangshan at mail.iap.ac.cn
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=68397 >
>
>
> Dear sir or madam,
>   I have a question about the rank histogram computed by Ensemble-
Stat. If
> there are several ensemble members forecast the same value as the
> observation, which is quite common in precipitation forecast, will
MET add
> 1 to the first bin, or will it add 1 to a randomly picked bin among
those
> equal members? Thanks!
>
> J.Zhu
>
>
>
>
>

------------------------------------------------
Subject: a question about rank histogram in Ensemble-Stat
From: zhujiangshan
Time: Fri Jul 25 20:48:59 2014

John,
  That helps a lot, thanks!

Jiangshan


> -----原始邮件-----
> 发件人: "John Halley Gotway via RT" <met_help at ucar.edu>
> 发送时间: 2014年7月25日 星期五
> 收件人: zhujiangshan at mail.iap.ac.cn
> 抄送:
> 主题: Re: [rt.rap.ucar.edu #68397] a question about rank histogram in
Ensemble-Stat
>
> J. Zhu,
>
> Very good question!  When ensemble_stat encounters ties while
computing
> ranked histograms, it randomly assigns the rank from the available
tied
> ranks.  If you run ensemble_stat twice on a dataset with tied ranks,
you'll
> likely get slightly different ranked histograms in the output.  The
> ensemble_stat config file does contain an entry to specify the
random
> number generator to be using in this random assignment:
>
> //
> // Random number generator
> //
> rng = {
>    type = "mt19937";
>    seed = "";
> }
>
> The seed value is the "starting point" for the random number
generator.  By
> default, the "seed" value is unset and the tool will choose a
somewhat
> random seed value based on a permutation of the current time.  If
you'd
> like your results to be exactly repeatable from run to run, all you
need to
> do is set that seed value.  For example, just set... seed = "1"; and
your
> results should no longer change from run to run.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
>
> On Thu, Jul 24, 2014 at 7:35 PM, zhujiangshan via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Thu Jul 24 19:35:52 2014: Request 68397 was acted upon.
> > Transaction: Ticket created by zhujiangshan at mail.iap.ac.cn
> >        Queue: met_help
> >      Subject: a question about rank histogram in Ensemble-Stat
> >        Owner: Nobody
> >   Requestors: zhujiangshan at mail.iap.ac.cn
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=68397 >
> >
> >
> > Dear sir or madam,
> >   I have a question about the rank histogram computed by Ensemble-
Stat. If
> > there are several ensemble members forecast the same value as the
> > observation, which is quite common in precipitation forecast, will
MET add
> > 1 to the first bin, or will it add 1 to a randomly picked bin
among those
> > equal members? Thanks!
> >
> > J.Zhu
> >
> >
> >
> >
> >
>




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


More information about the Met_help mailing list