[Met_help] [rt.rap.ucar.edu #89958] History for comparing models

[John Halley Gotway via RT]

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

Hello,
We would like to evaluate a newly developed model against established
counterparts.
Would Grid-Stat Tool be appropriate for this purpose?
The models come at different horizontal and vertical resolutions in netcdf
and we would like to arrive at a common resolution of e.g. 0.25 degree and
interpolate from native vertical levels to pressure, possibly in GRIB to
make application of Stat-Tool easier.
Are tools in MET available to do that or there are other ways?
Thanks for comments and suggestions to help us find a sound method to
accomplish our task.
Thanks,
Mariusz


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

Subject: comparing models
From: John Halley Gotway
Time: Mon Apr 29 09:54:09 2019

Mariusz,

Yes, I would definitely recommend using the Grid-Stat tool to inter-
compare
models.  You might also find the Series-Analysis tool to be useful.
Whereas Grid-Stat processes data for a single point in time and
computes
statistics aggregated over one or more spatial regions, Series-
Analysis
processes a time series of data and computes statistics separately for
each
grid point.  So Series-Analysis would help you determine how the
models
compare over different regions of the domain.

It's fine that the two input grids differ.  You can use the automated
regridding functionality in MET to put them on a common grid.  For
example,
in the Grid-Stat config file, you'd set:











*//// Verification grid// May be set separately in each "field"
entry//
regrid = {   to_grid      = "G193";   method     = BILIN;   width
=
2;   vld_thresh = 0.5;   shape        = SQUARE;}*

Where NCEP grid 193 is a global lat/lon grid with 0.25 degree spacing
(
https://www.nco.ncep.noaa.gov/pmb/docs/on388/tableb.html#GRID193).
And I
chose the regridding method as bilinear, but you can choose whatever
you'd
like.

Now, as for interpolating raw model output to standard pressure
levels,
there's nothing in MET that would help you do that.  The Unified Post
Processor code typically handles that functionality:
   https://dtcenter.org/community-code/unified-post-processor-upp

So it really depends on whether the model you're running is handled by
UPP.

Hope that helps.

Thanks,
John Halley Gotway

On Sun, Apr 28, 2019 at 10:24 PM Mariusz Pagowski via RT
<met_help at ucar.edu>
wrote:

>
> Sun Apr 28 22:24:35 2019: Request 89958 was acted upon.
> Transaction: Ticket created by Mariusz.Pagowski at noaa.gov
>        Queue: met_help
>      Subject: comparing models
>        Owner: Nobody
>   Requestors: Mariusz.Pagowski at noaa.gov
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89958 >
>
>
> Hello,
> We would like to evaluate a newly developed model against
established
> counterparts.
> Would Grid-Stat Tool be appropriate for this purpose?
> The models come at different horizontal and vertical resolutions in
netcdf
> and we would like to arrive at a common resolution of e.g. 0.25
degree and
> interpolate from native vertical levels to pressure, possibly in
GRIB to
> make application of Stat-Tool easier.
> Are tools in MET available to do that or there are other ways?
> Thanks for comments and suggestions to help us find a sound method
to
> accomplish our task.
> Thanks,
> Mariusz
>
>

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