[Met_help] [rt.rap.ucar.edu #85956] History for how to improve computational efficiency of the point_stat an so on?

John Halley Gotway via RT met_help at ucar.edu
Tue Jul 9 12:03:29 MDT 2019


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

hello!


I have used Met4.1 long time to verificate, and there is a question annoyed me always.
I use point_stat for verificating  at obseration points at a region of interest.
there is about 1500 stations in region.
I use point_stat for interpolation using Nearest Neighbor for 72 hours forecast results, that will be cost 1 hour for computation time.
I donot how to improve computational efficiency of the point_stat?
can you give me suggestion?


best wishes!


Li Qiang

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

Subject: how to improve computational efficiency of the point_stat an so on?
From: John Halley Gotway
Time: Mon Jul 02 09:38:11 2018

Li,

I see you have a question about how to improve the runtime of the
Point-Stat tool.  Without running the exact set of commands you're
running,
I can't know for sure, but here are some ideas to consider:

(1) The current version of MET is 7.0, and we've made a lot of changes
since version 4.1.  It possible that some of those changes would speed
up
the runtime of Point-Stat.

Take a look in the Point-Stat configuration file...

(2) Turning off the computation of bootstrap confidence intervals
would
speed it up.  Just set "n_rep = 0":
   boot = {
     interval = PCTILE;
      rep_prop = 1.0;
      n_rep    = 0;
      rng      = "mt19937";
      seed     = "";
   };

(3) Disabling the computation of rank correlation coefficients would
also
speed it up.  Set "rank_corr_flag = FALSE":
   rank_corr_flag = FALSE;

(4) Take a look at how you have your masking regions defined.  If
you're
using any polyline mask files in the "poly" setting, consider using
gen_vx_mask to pre-generate the mask for your domain.  If there are a
lot
of points in your domain, recomputing the masking region each time
point-stat runs could be slow:
   mask = {
      grid = [ "FULL" ];
      poly = [];
      sid  = [];
   };

Hope that helps.

Thanks,
John Halley Gotway

On Mon, Jul 2, 2018 at 1:21 AM 李强 via RT <met_help at ucar.edu> wrote:

>
> Mon Jul 02 01:20:37 2018: Request 85956 was acted upon.
> Transaction: Ticket created by cqlqlq at 163.com
>        Queue: met_help
>      Subject: how to improve computational efficiency of the
point_stat an
> so on?
>        Owner: Nobody
>   Requestors: cqlqlq at 163.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=85956 >
>
>
> hello!
>
>
> I have used Met4.1 long time to verificate, and there is a question
> annoyed me always.
> I use point_stat for verificating  at obseration points at a region
of
> interest.
> there is about 1500 stations in region.
> I use point_stat for interpolation using Nearest Neighbor for 72
hours
> forecast results, that will be cost 1 hour for computation time.
> I donot how to improve computational efficiency of the point_stat?
> can you give me suggestion?
>
>
> best wishes!
>
>
> Li Qiang
>

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


More information about the Met_help mailing list