[Met_help] [rt.rap.ucar.edu #60859] History for grib/mode compatibility
John Halley Gotway via RT
met_help at ucar.edu
Fri May 24 10:28:45 MDT 2013
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
hello,
we encountered a problem while using specific grib files as mode input.
Since many months, we have done studies with mode using model grib files
in input (forecast and analyses), so hundred of runs have been done. We
have had no problem to configure mode, change config values several
time, interpret output file. All was ok.
Now we want to use precipitation analyses as observation input files.
The grib files are called ANTILOPE, produced by Météo-France/DSO, based
on mixing of précipitation data (Optimal interpolation) and stocked on a
fine regular grid (0.012 degres).
The ANTILOPE data could easyly been seen by drawing them with metview
software, and they match very well with the model data. All values,
local and statistical, as they could be read or computed inside metview,
are normal and in agreement with the model data.
When a mode run is carried out, for example with 2 ANTILOPE grib files
in input, all first seems to be normal. Except that the data used by
mode, as they can be seen on postscript figures produced by mode, don't
match with the original data. These figures show horizontal lines which
are not realistic. It looks like a noise, with a meridian
stratification. Moreover, the real physical structures, present in the
data, don't appears at all. But the data contains a mask, since at the
contrary to the model data, real precipitations are not available on
each point of the grid (no data when going out of France more than about
100 km). And this mask is perfectly seen by mode, since it is clearly
présent on the figures. Only the data inside the unmasked region are bad.
The ANTILOPE grib is build by using grib_handle_new_from_file (from
grib_api) without compressing.
At this step, we don't know if this problem is due to the grib file or
to the mode software, or an interaction between both. But we must solve
it in order to be able to further study the mode behaviour on
operational precipitation data.
Does have somebody an idea on whath is the origin of this problem ?
Thank you for your help and best regards,
--- jean ---
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #60859] grib/mode compatibility
From: John Halley Gotway
Time: Wed Apr 03 09:52:18 2013
Jean,
It would probably be easiest for you to just send me some sample data
so that I can take a look.
Please send me:
- The ANTILOPE GRIB files you're using as input to MODE.
- The MODE configuration file you're using.
- The command line you're using to run MODE.
- The version number of MET you're running.
I'll run it here to reproduce the problem and try to figure out what's
going on.
You can just post this data to our anonymous ftp site following these
instructions:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
Please write back to let me know when the data is on the ftp site, and
I'll go grab it.
Thanks,
John Halley Gotway
met_help at ucar.edu
On 04/03/2013 06:38 AM, Jean.Barckicke at meteo.fr via RT wrote:
>
> Wed Apr 03 06:38:20 2013: Request 60859 was acted upon.
> Transaction: Ticket created by Jean.Barckicke at meteo.fr
> Queue: met_help
> Subject: grib/mode compatibility
> Owner: Nobody
> Requestors: Jean.Barckicke at meteo.fr
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=60859 >
>
>
> hello,
> we encountered a problem while using specific grib files as mode
input.
> Since many months, we have done studies with mode using model grib
files
> in input (forecast and analyses), so hundred of runs have been done.
We
> have had no problem to configure mode, change config values several
> time, interpret output file. All was ok.
> Now we want to use precipitation analyses as observation input
files.
> The grib files are called ANTILOPE, produced by Météo-France/DSO,
based
> on mixing of précipitation data (Optimal interpolation) and stocked
on a
> fine regular grid (0.012 degres).
> The ANTILOPE data could easyly been seen by drawing them with
metview
> software, and they match very well with the model data. All values,
> local and statistical, as they could be read or computed inside
metview,
> are normal and in agreement with the model data.
> When a mode run is carried out, for example with 2 ANTILOPE grib
files
> in input, all first seems to be normal. Except that the data used by
> mode, as they can be seen on postscript figures produced by mode,
don't
> match with the original data. These figures show horizontal lines
which
> are not realistic. It looks like a noise, with a meridian
> stratification. Moreover, the real physical structures, present in
the
> data, don't appears at all. But the data contains a mask, since at
the
> contrary to the model data, real precipitations are not available on
> each point of the grid (no data when going out of France more than
about
> 100 km). And this mask is perfectly seen by mode, since it is
clearly
> présent on the figures. Only the data inside the unmasked region are
bad.
> The ANTILOPE grib is build by using grib_handle_new_from_file (from
> grib_api) without compressing.
> At this step, we don't know if this problem is due to the grib file
or
> to the mode software, or an interaction between both. But we must
solve
> it in order to be able to further study the mode behaviour on
> operational precipitation data.
> Does have somebody an idea on whath is the origin of this problem ?
> Thank you for your help and best regards,
> --- jean ---
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #60859] grib/mode compatibility
From: Jean.Barckicke at meteo.fr
Time: Thu Apr 04 04:13:17 2013
John Halley Gotway via RT a écrit :
> Jean,
>
> It would probably be easiest for you to just send me some sample
data so that I can take a look.
>
> Please send me:
> - The ANTILOPE GRIB files you're using as input to MODE.
> - The MODE configuration file you're using.
> - The command line you're using to run MODE.
> - The version number of MET you're running.
>
> I'll run it here to reproduce the problem and try to figure out
what's going on.
>
> You can just post this data to our anonymous ftp site following
these instructions:
> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>
> Please write back to let me know when the data is on the ftp site,
and I'll go grab it.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
Hi John,
thank you for your rapid answer. I have followed your instruction and
i
have done as indicated on the site. First by registering me with e-
mail
address, then by connecting with ftp (-p option is not know on my
mandriva system) and putting 4 files in "jean" directory:
- ANTILOPEQ0024_PRECIP24_FRAN0012_2010022706 for precipitation gridded
data (grib file)
- config_precip as configuration file
- masqueFRAN0012 as mask used in configuration file (limit of the grib
data)
- mode_240000L_20100228_060000V_240000A.ps.gz as output of mode,
showing
the problem. You only have to gunzip it.
This files were just used/produced in a special directory before
sending
them with ftp. So we can be sure that there are exactly the good files
producing/showing the problem.
Please don't hesitate if you have an idea about the problem, or if you
need other information to help you.
Thank you a lot,
--- Jean ---
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #60859] grib/mode compatibility
From: John Halley Gotway
Time: Fri Apr 05 17:13:53 2013
Jean,
Yes, I see the problem you're describing in the Postscript output you
sent.
When I run METv3.0.1, as originally released, with the data you sent,
I am able to reproduce the bad behavior you're seeing. However, when
I run the same data through METv3.0.1 with all the latest
bugfixes for that release, the output looks fine. I suspect that the
patch posted on 3/28/2011 is the one that resolves this problem.
Please follow the instructions on the following page to apply the
latest set of patches for METv3.0.1:
http://www.dtcenter.org/met/users/support/known_issues/METv3.0.1/index.php
Also, METv4.1 will be coming out in the next few weeks. When it does,
I'd suggest switching to that version because it will include a bug
fix that MODE was over-reporting object areas. Basically,
it's counting an extra row and column for each object and adding that
to the area.
Please let me know if you encounter additional problems.
Thanks,
John
On 04/04/2013 04:13 AM, Jean.Barckicke at meteo.fr via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=60859 >
>
> John Halley Gotway via RT a écrit :
>> Jean,
>>
>> It would probably be easiest for you to just send me some sample
data so that I can take a look.
>>
>> Please send me:
>> - The ANTILOPE GRIB files you're using as input to MODE.
>> - The MODE configuration file you're using.
>> - The command line you're using to run MODE.
>> - The version number of MET you're running.
>>
>> I'll run it here to reproduce the problem and try to figure out
what's going on.
>>
>> You can just post this data to our anonymous ftp site following
these instructions:
>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>
>> Please write back to let me know when the data is on the ftp site,
and I'll go grab it.
>>
>> Thanks,
>> John Halley Gotway
>> met_help at ucar.edu
>>
> Hi John,
> thank you for your rapid answer. I have followed your instruction
and i
> have done as indicated on the site. First by registering me with e-
mail
> address, then by connecting with ftp (-p option is not know on my
> mandriva system) and putting 4 files in "jean" directory:
>
> - ANTILOPEQ0024_PRECIP24_FRAN0012_2010022706 for precipitation
gridded
> data (grib file)
> - config_precip as configuration file
> - masqueFRAN0012 as mask used in configuration file (limit of the
grib data)
> - mode_240000L_20100228_060000V_240000A.ps.gz as output of mode,
showing
> the problem. You only have to gunzip it.
>
> This files were just used/produced in a special directory before
sending
> them with ftp. So we can be sure that there are exactly the good
files
> producing/showing the problem.
> Please don't hesitate if you have an idea about the problem, or if
you
> need other information to help you.
> Thank you a lot,
>
> --- Jean ---
>
------------------------------------------------
More information about the Met_help
mailing list