[Met_help] Problems with copygb

[John Halley Gotway]

Justin,

I am not aware of any other tools that people run on WRF output to regrid
it, but that isn't to say that they aren't out there.  MET is a relatively
new package.  The version that's available is still a beta version.  As
more people begin to use it, and we release updated versions, I hope to
learn about more about the types of reformatting issues users encounter.

Would you mind sending me one of these files?  I'd be interested in
playing around with it to see if I see the same behavior.

John

> Hi John,
>
> Thanks for your suggestion.  I tried it out, but unfortunately the result
> was a constant zero field centered over a different location...so even if
> my grid points were off, the problem of a constant zero field still
> existed.  A wrf-help response has suggested problems with the way my grib
> data is packed (binary vs. decimal packing)...something I would have never
> had guessed.
>
> This brings up one quick question.  Does anyone use anything other than
> copygb to create input to things like mode, grid_stats, etc?  I thought
> about using lats4d when I ran into this problem, but didn't have success
> with it either.
>
> Thanks again for your help!
> Justin
>
> ----- Original Message -----
> From: John Halley Gotway <johnhg at rap.ucar.edu>
> Date: Thursday, October 4, 2007 9:37 pm
> Subject: Re: [Met_help] Problems with copygb
>
>> Justin,
>>
>> I'd say that the WRF mailing list or an email to wrf_help at ucar.edu
>> wouldbe your best bet.  The MET development team doesn't support
>> copygb.
>> But I thought I'd take a stab at it anyway.
>>
>> Let me caution, I've only used copygb a few times.  But your grid
>> definition does look a little funny to me.  Here's the example from
>> theCPC website
>> (http://www.cpc.ncep.noaa.gov/products/wesley/copygb.html) for
>> specifying a lat/lon grid for copygb:
>>
>> NX = number of points on latitude cirle
>> NY = number of points on longitude meridian
>> LAT0 = latitude of origin * 1000  (lat = -90 .. 90)
>> LON0 = longitude of origin * 1000  (lon=-360 .. 360 ?)
>> LAT1 = latitude of extreme (last) point * 1000
>> LON1 = longitude of extreme (last) point * 1000
>> DX = latitudianal increment * 1000  (sign used)
>> DY = longitudinal increment * 1000  (sign used)
>> grid='255 0 NX NY LAT0 LON0 128 LAT1 LAT2 DX DY 64'
>>
>> In your case (255 0 86 43 40000 276000 128 45880 287900 140 140),
>> you've set:
>> NX   = 86
>> NY   = 43
>> LAT0 = 40000
>> LON0 = 276000
>> LAT1 = 45880
>> LON1 = 287900
>> DX   = 140
>> DY   = 140
>>
>> But what seems odd to me is that your lower left corner (40, 276)
>> plus the
>> increment*interval doesn't equal the upper right corner (45.88,
>> 287.90).
>> In your case,
>> LAT0 + NX * DX = 40000 + 86 * 140 = 52040 != 45880 (LAT1)
>> LON0 + NY * DY = 276000 + 43 * 140 = 282020 != 287900 (LON1)
>>
>> It seems like you'll need to work on the grid definition some more.
>> Also,
>> if you look at the example of on the CPC website, the last argument
>> in the
>> grid definition is the number "64".  I really don't know what the
>> significance of that is.
>>
>> Keeping the LAT0, LON0, NX, NY, DX, and DY are unchanged, I'd try
>> something like:
>> 255 0 86 43 40000 276000 128 52040 282020 140 140 64
>>
>> Or you could keep the lower-left and upper-right corners fixed and
>> adjustthe NX, NY, DX, and DY values appropriately.
>>
>> Good luck,
>> John Halley Gotway
>> johnhg at ucar.edu
>>
>> > Good Afternoon,
>> >
>> > I am using the copygb utility to regrid some WRF precip (variable:
>> > apcpsfc) output onto a common grid with another dataset (NCEP
>> Stage 4
>> > precip) to prepare it for use in the MODE verification software.  I
>> > noticed that for certain time periods (e.g. forecast hour 1) the
>> grib> output using bilinear interpolation to the new grid gives me
>> a constant
>> > "0" field, even though the apcpsfc field in the original grib
>> file is
>> > nonzero.  My command line input follows:
>> >
>> > copygb -g"255 0 86 43 40000 276000 128 45880 287900 140 140" -x
>> > 2007020700_BGM-ARW.F00100.grb out.grb
>> >
>> > The grb file is a 1 hour output file from WRF.  Interestingly, other
>> > fields, such as tmpprs or prmslmsl seem to go work fine.
>> > Thinking that the interpolation type might have caused the
>> problem, I
>> > asked for no interpolation using:
>> >
>> > copygb -x 2007020700_BGM-ARW.F00100.grb out.grb
>> >
>> > This still gives me a resultant precip field that is filled with
>> zeroes.>
>> > Does anyone see anything wrong with my input or how I'm using
>> copygb?  I
>> > appreciate your help in advance.
>> >
>> > I also sent this email along to the WRF email list as copygb is
>> part of
>> > the WRF  Post-Processor.
>> >
>> > Justin Arnott
>> > NWS Binghamton
>> > _______________________________________________
>> > Met_help mailing list
>> > Met_help at mailman.ucar.edu
>> > http://mailman.ucar.edu/mailman/listinfo/met_help
>> >
>>
>>
>>
>