[Met_help] MODE problem

[John Halley Gotway]

Jeff,

OK.  I think I know what's going on here.  Your forecast and observation
grids have the same dimension but are not identical.  There is a problem
in your call to copygb somewhere.

When MODE reads the forecast and observation data, it only checks to see
that NX and NY are the same.  It doesn't do a robust check of the grids to
make sure they're identical.  Because of the problem you're having, and a
similar problem another user had a couple weeks ago, we're planning to add
more robust checking of the grids for the next release of MET.

When you pass it the forecast and observation fields, MODE reads the grid
definition info from the forecast data.  It checks that the NX and NY of
the observation field match and assumes that the observation grid is the
same as the forecast grid.  It then uses the forecast grid the rest of the
way.

Here's what you should do.  Run both your forecast and observation GRIB
files through wgrib using the -V option:

wgrib -V forecast_file.grib
wgrib -V observation_file.grib

That'll dump out the grid definition information for the data.  I'm
guessing you'll find that they're very similar but off by a very small
amount.  You'll need to modify your call to copygb very slightly to get
them to be identical.  And once they are identical, your problem should be
solved.

Hope that helps.

John

> John,
> The version of MODE I'm using is in a place where I don't have permission
> to
> mess around with files.  So I talked to our network administrator and he
> installed the patches for MODE.  I reran it the same way I had been before
> but got the same incorrect output.  I tried matching the Stage IV data to
> itself like you did and the position was correct.  But then I decided to
> try
> running mode by reversing the order of the files in the call so that I put
> the Stage IV data first (as if it were the forecast) and my WRF data
> second
> (as if it were the observation), and wouldn't you know it...it plotted
> both
> of them correctly!  I compared the MODE output (object information)
> between
> the two sets of MODE runs (switched and unswitched), and everything I saw
> was the same between the two (interest values, number of objects
> identified
> and matched etc.) except for they were switched around.  So it almost
> seems
> I could just use MODE flipped around.  Strange.  Thanks for looking into
> this for me.
>
> Jeff
>
> On Tue, Feb 16, 2010 at 3:45 PM, John Halley Gotway
> <johnhg at rap.ucar.edu>wrote:
>
>> Jeff,
>>
>> Thanks for sending the data.  I took the Stage4 file you sent and did
>> two
>> things with it:
>>
>> (1) I plotted it using IDV and found that the image matches the GRADS
>> image you sent.  I've attached that image from IDV.
>>
>> (2) Next, I ran that Stage4 file you sent through MODE, comparing it to
>> itself, and found that IT TOO matches the the plots produced by IDV and
>> GRADS.  I've also attached that image.
>>
>> I'm not sure what's going on here.  Perhaps there's been some confusion
>> in
>> file names or time stamps.  One thing I'll recommend is that you be sure
>> you've downloaded all the latest patches from METv2.0 from:
>>   http://www.dtcenter.org/met/users/support/known_issues/METv2.0/index.php
>> Just follow the instructions at the top of the page to apply the
>> patches.
>>
>> So I've attached 4 files which show plots of the data you sent.  The
>> only
>> one that doesn't match is the MODE output you sent me, which I named
>> "mode_results_duda.png".  I'm guessing there was some problem in your
>> script that ran MODE.
>>
>> Just let me know if more questions come up.
>>
>> Thanks,
>> John
>>
>>
>
>
> --
> Jeff Duda
> Iowa State University
> Meteorology Graduate Student
> 3134 Agronomy Hall
> www.meteor.iastate.edu/~jdduda
>