[Met_help] [rt.rap.ucar.edu #71367] History for using precipitation budget with copygb

Tue Jan 26 12:17:09 MST 2016

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

Hello,
I am using METv4.1 in order to verify precipitation from grib files.  These grib files are generated using copygb to interpolate the precipitation fields for models of multiple resolutions onto a common 4km grid.  I understand that copygb is not inherently part of the MET system, however I was wondering if you had any personal experience using the '-i3' option (precipitation budget option) with copygb in verifications?

I am looking to speed up my code, and copygb with the 'i3' option is about 7 times slower than without specifying the option.  I have only tested two days so far, but I have not seen a significant difference between specifying the 'i3' option as opposed to leaving it out.  So, I had considered performing the interpolation in the future without i3.  Could you comment on this with any experience/thoughts as to verifications using this interpolation method?

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

Subject: using precipitation budget with copygb
From: Randy Bullock
Time: Wed Apr 08 11:35:23 2015

Hello Andrew -

Thanks for your email.  As it happens, I just recently looked pretty
closely at the UPP code for budget interpolation.  In fact I
implemented it
in MET, and it'll be available in the next MET release, so you won't
have
to use copygb.

Anyway ... the UPP code (which copygb uses) allows you to set some
tunable
parameters.  One such is what the call they interpolation "width",
which by
default is set to 2.  This means that for each output grid point, (2*2
+
1)^2 = 25 points are actually interpolated (using bilinear
interpolation),
so as you can see, the amount of work being done for budget
interpolation
is quite large.

The UPP library code allows you to change that interpolation width,
but I'm
not sure if copygb does.  I'm afraid my knowledge of copygb is not
very
deep, but you might look into seeing if the copygb command-line allows
you
to change the interpolation width.  If it does, then try lowering it
from 2
(the default) to 1, and see how that goes.

Randy

On Wed, Apr 8, 2015 at 8:54 AM, Andrew J. via RT <met_help at ucar.edu>
wrote:

>
> Wed Apr 08 08:54:47 2015: Request 71367 was acted upon.
> Transaction: Ticket created by andrewwx at yahoo.com
>        Queue: met_help
>      Subject: using precipitation budget with copygb
>        Owner: Nobody
>   Requestors: andrewwx at yahoo.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71367 >
>
>
> Hello,
> I am using METv4.1 in order to verify precipitation from grib files.
> These grib files are generated using copygb to interpolate the
> precipitation fields for models of multiple resolutions onto a
common 4km
> grid.  I understand that copygb is not inherently part of the MET
system,
> however I was wondering if you had any personal experience using the
'-i3'
> option (precipitation budget option) with copygb in verifications?
>
> I am looking to speed up my code, and copygb with the 'i3' option is
about
> 7 times slower than without specifying the option.  I have only
tested two
> days so far, but I have not seen a significant difference between
> specifying the 'i3' option as opposed to leaving it out.  So, I had
> considered performing the interpolation in the future without i3.
Could
> you comment on this with any experience/thoughts as to verifications
using
> this interpolation method?
>
>

------------------------------------------------
Subject: using precipitation budget with copygb
From: John Halley Gotway
Time: Mon Apr 13 15:07:03 2015

Hi Andrew,

This is John Halley Gotway.  I also work on providing met-help
support.  I
wanted to mention one additional wrinkle to this.  It is generally
recommended that budget interpolation be used when regridding
accumulated
precipitation.  However, it is generally not recommended for other
variables such as temperature and winds.  But that's confusing because
you
typically regrid all the data in a GRIB file in a single call to
copygb and
you can only set the interpolation option (-i3) once.

To solve this issue, I wrote a PERL script a while back that we use
for our
internal testing.  It's called copygb_budget.pl and you call it with
the
same arguments that you use for copygb.  It calls wgrib to strip out
GRIB
records of ACPCP, NCPCP, APCP, PWAT and calls copygb using budget
interpolation.  Then it calls copgyb using the default interpolation
method
for all the other records.  And at the end, it slaps them back
together.

The output file has GRIB records in a different order than the input,
but
that generally is not a problem.

If you'd like to use it, you'd need to update some of the constants at
the
top.

Hope that helps.

Thanks,
John

On Wed, Apr 8, 2015 at 11:35 AM, Randy Bullock via RT
<met_help at ucar.edu>
wrote:

>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71367 >
>
> Hello Andrew -
>
> Thanks for your email.  As it happens, I just recently looked pretty
> closely at the UPP code for budget interpolation.  In fact I
implemented it
> in MET, and it'll be available in the next MET release, so you won't
have
> to use copygb.
>
> Anyway ... the UPP code (which copygb uses) allows you to set some
tunable
> parameters.  One such is what the call they interpolation "width",
which by
> default is set to 2.  This means that for each output grid point,
(2*2 +
> 1)^2 = 25 points are actually interpolated (using bilinear
interpolation),
> so as you can see, the amount of work being done for budget
interpolation
> is quite large.
>
> The UPP library code allows you to change that interpolation width,
but I'm
> not sure if copygb does.  I'm afraid my knowledge of copygb is not
very
> deep, but you might look into seeing if the copygb command-line
allows you
> to change the interpolation width.  If it does, then try lowering it
from 2
> (the default) to 1, and see how that goes.
>
> Randy
>
>
>
> On Wed, Apr 8, 2015 at 8:54 AM, Andrew J. via RT <met_help at ucar.edu>
> wrote:
>
> >
> > Wed Apr 08 08:54:47 2015: Request 71367 was acted upon.
> > Transaction: Ticket created by andrewwx at yahoo.com
> >        Queue: met_help
> >      Subject: using precipitation budget with copygb
> >        Owner: Nobody
> >   Requestors: andrewwx at yahoo.com
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71367 >
> >
> >
> > Hello,
> > I am using METv4.1 in order to verify precipitation from grib
files.
> > These grib files are generated using copygb to interpolate the
> > precipitation fields for models of multiple resolutions onto a
common 4km
> > grid.  I understand that copygb is not inherently part of the MET
system,
> > however I was wondering if you had any personal experience using
the
> '-i3'
> > option (precipitation budget option) with copygb in verifications?
> >
> > I am looking to speed up my code, and copygb with the 'i3' option
is
> about
> > 7 times slower than without specifying the option.  I have only
tested
> two
> > days so far, but I have not seen a significant difference between
> > specifying the 'i3' option as opposed to leaving it out.  So, I
had
> > considered performing the interpolation in the future without i3.
Could
> > you comment on this with any experience/thoughts as to
verifications
> using
> > this interpolation method?
> >
> >
>
>

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