[Met_help] [rt.rap.ucar.edu #41630] History for Fwd: Problema with point stat
RAL HelpDesk {for John Halley Gotway}
met_help at ucar.edu
Fri Oct 29 12:52:58 MDT 2010
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi my name is Andrea and I am an Italian student of Physics of the Atmosphere. I have a problem using
point_stat application that get an error reading from a grb file create by WPP:
ERROR: GCInfo::set_gcinfo() -> unsupported input file type "TMP/P500".
I have attacched the result of wgrib application.
Thank's for the help.
Andrea Chini
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: John Halley Gotway
Time: Wed Oct 20 09:33:44 2010
Andrea,
I see that you're having trouble running the Point-Stat tool. I'll be
happy to help you try to resolve this issue.
I realize that that error message you're getting is a bit cryptic.
However, based on that error message, I'm guessing that there's some
problem with the command line arguments you've passed to
Point-Stat. Here's the usage statement for Point-Stat:
Usage: point_stat
fcst_file
obs_file
config_file
[-climo climo_file]
[-point_obs file]
[-fcst_valid time]
[-fcst_lead time]
[-obs_valid_beg time]
[-obs_valid_end time]
[-outdir path]
[-v level]
So at a minimum you need to pass it the forecast file (your GRIB
data), the observation file (point observations as the NetCDF output
of PB2NC or ASCII2NC), and the configuration file.
You can see examples of running Point-Stat by looking in:
(1) The test scripts:
METv3.0/scripts/test_point_stat.sh
(2) The online tutorial:
http://www.dtcenter.org/met/users/support/online_tutorial/METv2.0/point_stat/index.php
I also see that you're trying to verify 500mb temperature. But when I
look at the wgrib output you sent me, I don't see any records for
that. Here are the records I see for temperature in that wgrib
output:
6:470:d=10101906:TMP:kpds5=11:kpds6=109:kpds7=1:TR=10:P1=0:P2=20:TimeU=1:hybrid
lev 1:20hr fcst:NAve=0
20:17572:d=10101906:TMP:kpds5=11:kpds6=109:kpds7=2:TR=10:P1=0:P2=20:TimeU=1:hybrid
lev 2:20hr fcst:NAve=0
248:67538:d=10101906:TMP:kpds5=11:kpds6=3:kpds7=0:TR=10:P1=0:P2=20:TimeU=1:cld
top:20hr fcst:NAve=0
256:71174:d=10101906:TMP:kpds5=11:kpds6=7:kpds7=0:TR=10:P1=0:P2=20:TimeU=1:tropopause:20hr
fcst:NAve=0
If you'd like to verify 500mb temperature, you'll need to make sure
that the WRF PostProcessor is writing it out. Please refer to the WRF
NMM User's Guide for instructions on configuring/running WPP:
http://www.dtcenter.org/wrf-
nmm/users/docs/user_guide/V3/users_guide_nmm_chap1-7.pdf
Lastly, if you continue to experience problems running Point-Stat, you
can send us your data, and I'll try to run it here. If that's the
case, you can follow these instructions for posting data to
our ftp site:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
We would need your (1) forecast GRIB file, (2) NetCDF observation
file, (3) Point-Stat configuration file, and (4) a text file showing
how you're calling Point-Stat on the command line.
Hope that helps.
Thanks,
John Halley Gotway
met_help at ucar.edu
On 10/20/2010 03:11 AM, RAL HelpDesk {for Andrea Chini} wrote:
>
> Wed Oct 20 03:11:05 2010: Request 41630 was acted upon.
> Transaction: Ticket created by andrea.chini at studio.unibo.it
> Queue: met_help
> Subject: Fwd: Problema with point stat
> Owner: Nobody
> Requestors: andrea.chini at studio.unibo.it
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=41630 >
>
>
>
> Hi my name is Andrea and I am an Italian student of Physics of the
Atmosphere. I have a problem using
> point_stat application that get an error reading from a grb file
create by WPP:
>
> ERROR: GCInfo::set_gcinfo() -> unsupported input file type
"TMP/P500".
>
>
> I have attacched the result of wgrib application.
>
> Thank's for the help.
>
> Andrea Chini
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: Andrea Chini
Time: Wed Oct 20 11:06:44 2010
In data mercoledì 20 ottobre 2010 17:33:44, hai scritto:
> Andrea,
>
> I see that you're having trouble running the Point-Stat tool. I'll
be
> happy to help you try to resolve this issue.
>
> I realize that that error message you're getting is a bit cryptic.
> However, based on that error message, I'm guessing that there's some
> problem with the command line arguments you've passed to Point-Stat.
> Here's the usage statement for Point-Stat:
>
> Usage: point_stat
> fcst_file
> obs_file
> config_file
> [-climo climo_file]
> [-point_obs file]
> [-fcst_valid time]
> [-fcst_lead time]
> [-obs_valid_beg time]
> [-obs_valid_end time]
> [-outdir path]
> [-v level]
>
> So at a minimum you need to pass it the forecast file (your GRIB
data), the
> observation file (point observations as the NetCDF output of PB2NC
or
> ASCII2NC), and the configuration file.
>
> You can see examples of running Point-Stat by looking in:
> (1) The test scripts:
> METv3.0/scripts/test_point_stat.sh
> (2) The online tutorial:
>
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv2.0/point_st
> at/index.php
>
> I also see that you're trying to verify 500mb temperature. But when
I look
> at the wgrib output you sent me, I don't see any records for that.
Here
> are the records I see for temperature in that wgrib output:
>
6:470:d=10101906:TMP:kpds5=11:kpds6=109:kpds7=1:TR=10:P1=0:P2=20:TimeU=1:hy
> brid lev 1:20hr fcst:NAve=0
>
20:17572:d=10101906:TMP:kpds5=11:kpds6=109:kpds7=2:TR=10:P1=0:P2=20:TimeU=
> 1:hybrid lev 2:20hr fcst:NAve=0
>
248:67538:d=10101906:TMP:kpds5=11:kpds6=3:kpds7=0:TR=10:P1=0:P2=20:TimeU=1
> :cld top:20hr fcst:NAve=0
>
256:71174:d=10101906:TMP:kpds5=11:kpds6=7:kpds7=0:TR=10:P1=0:P2=20:TimeU=1
> :tropopause:20hr fcst:NAve=0
>
> If you'd like to verify 500mb temperature, you'll need to make sure
that
> the WRF PostProcessor is writing it out. Please refer to the WRF
NMM
> User's Guide for instructions on configuring/running WPP:
> http://www.dtcenter.org/wrf-
nmm/users/docs/user_guide/V3/users_guide_nmm_c
> hap1-7.pdf
>
> Lastly, if you continue to experience problems running Point-Stat,
you can
> send us your data, and I'll try to run it here. If that's the case,
you
> can follow these instructions for posting data to our ftp site:
> http://www.dtcenter.org/met/users/support/met_help.php#ftp
> We would need your (1) forecast GRIB file, (2) NetCDF observation
file, (3)
> Point-Stat configuration file, and (4) a text file showing how
you're
> calling Point-Stat on the command line.
>
> Hope that helps.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On 10/20/2010 03:11 AM, RAL HelpDesk {for Andrea Chini} wrote:
> > Wed Oct 20 03:11:05 2010: Request 41630 was acted upon.
> > Transaction: Ticket created by andrea.chini at studio.unibo.it
> >
> > Queue: met_help
> >
> > Subject: Fwd: Problema with point stat
> >
> > Owner: Nobody
> >
> > Requestors: andrea.chini at studio.unibo.it
> >
> > Status: new
> >
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=41630 >
> >
> > Hi my name is Andrea and I am an Italian student of Physics of the
> > Atmosphere. I have a problem using point_stat application that get
an
> > error reading from a grb file create by WPP:
> >
> > ERROR: GCInfo::set_gcinfo() -> unsupported input file type
"TMP/P500".
> >
> >
> > I have attacched the result of wgrib application.
> >
> > Thank's for the help.
> >
> > Andrea Chini
Hi Mr. Gotway,
I can resolve the trouble: there was a problem on WPP because it
didn't write some varaibles in grib files.
With new grib files of WRF-ARW today's forecast there isn't any
problem!
I want ask you another thing about another argument if it is possible:
can I plot .nc files created by ascii2nc
with a NCL script? Because the lat lon variables are separated by
observations.
Thanks a lot.
Andrea Chini
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: John Halley Gotway
Time: Wed Oct 20 11:15:36 2010
Andrea,
Great. Glad you got it working.
It sounds like you'd like to plot the locations of the NetCDF point
observations. Since you downloaded MET on 10/15/2010, I assume that
you're running METv3.0. We have actually included a utility in
METv3.0 that will plot those locations for you. However, by default,
it is not compiled.
Please do the following:
(1) Edit the top-level MET Makefile, by setting "DISABLE_TOOLS" to 0.
By default, it's set to 1.
(2) Then rebuild MET. You should just need to type "make" (meaning,
you don't need to do a clean first).
(3) Look for the newly built executable METv3.0/bin/plot_point_obs
Listed below is the usage statement for this tool. You pass it the
NetCDF point observation file to be plotted and the name of the
PostScript file it should create. By default, it'll create a global
plot. You can zoom in to your data using the "-dim" argument. It
just puts a red dot at each observing location. You can filter which
observation locations are plotted using the "-gc" and
"-msg_typ" command line arguments. Please note that this is meant to
be a diagnostic tool, not a highly configurable plotting tool. But
hopefully, it'll give you what you need.
Usage: plot_point_obs
nc_file
ps_file
[-gc code]
[-msg_typ name]
[-data_dir path]
[-dim lat_ll lon_ll lat_ur lon_ur]
where "nc_file" is the netCDF point observation file to be
plotted.
"ps_file" is the name for the output postscript file
to be generated.
"-gc code" specifies the GRIB code(s) to be plotted
(optional).
"-msg_typ name" is the message type(s) to be plotted
(optional).
"-data_dir path" indicates the MET data directory to
be used (optional).
"-dim lat_ll lon_ll width" defines the lat/lon box to
be plotted (optional).
Thanks,
John
On 10/20/2010 11:06 AM, RAL HelpDesk {for Andrea Chini} wrote:
> another argument
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: Andrea Chini
Time: Wed Oct 20 11:54:01 2010
In data mercoledì 20 ottobre 2010 19:15:37, hai scritto:
> Andrea,
>
> Great. Glad you got it working.
>
> It sounds like you'd like to plot the locations of the NetCDF point
> observations. Since you downloaded MET on 10/15/2010, I assume that
> you're running METv3.0. We have actually included a utility in
METv3.0
> that will plot those locations for you. However, by default, it is
not
> compiled.
>
> Please do the following:
> (1) Edit the top-level MET Makefile, by setting "DISABLE_TOOLS" to
0. By
> default, it's set to 1. (2) Then rebuild MET. You should just need
to
> type "make" (meaning, you don't need to do a clean first). (3) Look
for
> the newly built executable METv3.0/bin/plot_point_obs
>
> Listed below is the usage statement for this tool. You pass it the
NetCDF
> point observation file to be plotted and the name of the PostScript
file
> it should create. By default, it'll create a global plot. You can
zoom
> in to your data using the "-dim" argument. It just puts a red dot
at each
> observing location. You can filter which observation locations are
> plotted using the "-gc" and "-msg_typ" command line arguments.
Please
> note that this is meant to be a diagnostic tool, not a highly
configurable
> plotting tool. But hopefully, it'll give you what you need.
>
> Usage: plot_point_obs
> nc_file
> ps_file
> [-gc code]
> [-msg_typ name]
> [-data_dir path]
> [-dim lat_ll lon_ll lat_ur lon_ur]
>
> where "nc_file" is the netCDF point observation file to be
> plotted. "ps_file" is the name for the output postscript file to be
> generated. "-gc code" specifies the GRIB code(s) to be plotted
(optional).
> "-msg_typ name" is the message type(s) to be plotted (optional).
> "-data_dir path" indicates the MET data directory to be used
(optional).
> "-dim lat_ll lon_ll width" defines the lat/lon box to be plotted
> (optional).
>
> Thanks,
> John
>
> On 10/20/2010 11:06 AM, RAL HelpDesk {for Andrea Chini} wrote:
> > another argument
Hi,
i have just seen plot_point_obs, but I want to plot maps like those
that I have with WRF-ARW on NCL. It's
possible in your opinion?
Thanks a lot.
Andrea Chini
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: John Halley Gotway
Time: Wed Oct 20 11:58:20 2010
Andrea,
Sure, I don't see why not. You should be able to use NCL to open up
the NetCDF files, and plot the lat/lon locations of your observations.
Look in the "hdr_arr" variable for the lat/lon/elevation
values for each observing location.
If you actually want to plot the values of the observations, rather
than just their location, you'd use the data in the "obs_arr"
variable. The first element of each obs_arr entry is the index of the
hdr_arr to which that observation corresponds.
So all the data you need to make such a plot is there. You'd just
need to figure out how to do so in NCL.
John
On 10/20/2010 11:54 AM, RAL HelpDesk {for Andrea Chini} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=41630 >
>
> In data mercoledì 20 ottobre 2010 19:15:37, hai scritto:
>> Andrea,
>>
>> Great. Glad you got it working.
>>
>> It sounds like you'd like to plot the locations of the NetCDF point
>> observations. Since you downloaded MET on 10/15/2010, I assume
that
>> you're running METv3.0. We have actually included a utility in
METv3.0
>> that will plot those locations for you. However, by default, it is
not
>> compiled.
>>
>> Please do the following:
>> (1) Edit the top-level MET Makefile, by setting "DISABLE_TOOLS" to
0. By
>> default, it's set to 1. (2) Then rebuild MET. You should just need
to
>> type "make" (meaning, you don't need to do a clean first). (3) Look
for
>> the newly built executable METv3.0/bin/plot_point_obs
>>
>> Listed below is the usage statement for this tool. You pass it the
NetCDF
>> point observation file to be plotted and the name of the PostScript
file
>> it should create. By default, it'll create a global plot. You can
zoom
>> in to your data using the "-dim" argument. It just puts a red dot
at each
>> observing location. You can filter which observation locations are
>> plotted using the "-gc" and "-msg_typ" command line arguments.
Please
>> note that this is meant to be a diagnostic tool, not a highly
configurable
>> plotting tool. But hopefully, it'll give you what you need.
>>
>> Usage: plot_point_obs
>> nc_file
>> ps_file
>> [-gc code]
>> [-msg_typ name]
>> [-data_dir path]
>> [-dim lat_ll lon_ll lat_ur lon_ur]
>>
>> where "nc_file" is the netCDF point observation file to
be
>> plotted. "ps_file" is the name for the output postscript file to be
>> generated. "-gc code" specifies the GRIB code(s) to be plotted
(optional).
>> "-msg_typ name" is the message type(s) to be plotted (optional).
>> "-data_dir path" indicates the MET data directory to be used
(optional).
>> "-dim lat_ll lon_ll width" defines the lat/lon box to be plotted
>> (optional).
>>
>> Thanks,
>> John
>>
>> On 10/20/2010 11:06 AM, RAL HelpDesk {for Andrea Chini} wrote:
>>> another argument
>
> Hi,
> i have just seen plot_point_obs, but I want to plot maps like those
that I have with WRF-ARW on NCL. It's
> possible in your opinion?
>
> Thanks a lot.
>
> Andrea Chini
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: Andrea Chini
Time: Thu Oct 21 09:32:33 2010
Thank's for the answer! Another thing: why the point_Stat tool make a
file with double number of matched pairs
between fct and obs? I have 223 T2M obs but there are 443 in file
.mpr.
Thank's a lot.
Andrea Chini
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #41630] Fwd: Problema with point stat
From: John Halley Gotway
Time: Thu Oct 21 11:31:28 2010
Andrea,
If you look very closely in the MPR file, I think you will actually
find some differences between them. The MPR data is dumped out for
each verification task you perform. Perhaps you're verifying
over multiple regions or using multiple interpolation methods. Both
would lead to multiple verification tasks being defined and more
output than you expect in the MPR file. Look closely at the first
21 header columns in that file, and I'll bet you'll find one column
that changes.
Also, dumping out the MPR lines produces a lot of output. Keep in
mind that all of the ascii output is written to the "*.stat" file,
including the MPR lines. Point-Stat also creates the "*.txt"
files with the data sorted by line type. So the MPR output is showing
up twice, once in the ".stat" file and once in the "_mpr.txt" file.
In the Point-Stat config file, you could modify the
output_flag setting to tell Point-Stat to only write out the MPR data
in the ".stat" file to save some space. To do so, you'd change the 2
to a 1 in the output_flag entry for the MPR line type.
Thanks,
John
On 10/21/2010 09:32 AM, RAL HelpDesk {for Andrea Chini} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=41630 >
>
> Thank's for the answer! Another thing: why the point_Stat tool make
a file with double number of matched pairs
> between fct and obs? I have 223 T2M obs but there are 443 in file
.mpr.
>
>
> Thank's a lot.
>
>
> Andrea Chini
------------------------------------------------
More information about the Met_help
mailing list