[Met_help] [rt.rap.ucar.edu #42042] History for Help with using MET

Thu Nov 11 14:27:02 MST 2010

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

Dear MET_help,

I'm considering the use of MET for the verification of a 
series of forecasts, then aggregated on a custom netcdf 
file. I would, thus, appreciate if You could provide me 
some info regarding a couple of questions:

1) Is met able to read this "custom made" netcdf file? 
Basically, it is something like this:

ncdump -h fwi_output.nc
netcdf fwi_output {
dimensions:
	Time = 213 ;
	west_east = 110 ;
	south_north = 200 ;
variables:
	float Dates(Time) ;
	float lat(south_north, west_east) ;
	float lon(south_north, west_east) ;
	byte landmask(south_north, west_east) ;
	float T2(Time, south_north, west_east) ;
	float Wind(Time, south_north, west_east) ;
	float RH(Time, south_north, west_east) ;
	float RAIN(Time, south_north, west_east) ;
	float FFMC(Time, south_north, west_east) ;
	float DMC(Time, south_north, west_east) ;
	float DC(Time, south_north, west_east) ;
	float ISI(Time, south_north, west_east) ;
	float BUI(Time, south_north, west_east) ;
	float FWI(Time, south_north, west_east) ;
	float DSR(Time, south_north, west_east) ;

// global attributes:
		:description = "FFWI RESULTS FROM WRF GRID" ;
		:author = "Luis Carvalheiro" ;
		:date = "09-Nov-2010" ;
}

As you can see, it results from a postprocessing using 
wrfout files, containing latitude, longitude, Time and 
some meteorologycal output variables, as well as some 
"products", related with the canadian FWI system 
components. Is there a way of ingesting this file in MET 
and perform evaluation for each time? I'm considering 
using the point stat tool for 2m-temperature, 10m-wind 
magnitude and 2m-relative humidity, and perhaps the MODE 
tool to evaluate precipitation.

2) Could you "teach" me a quick way of performing my 
intentions? My observation datasets are text files with 
point observations from several stations around my domain.

3) Can i achieve a "time series" of forecast verification 
measures (RMSE,MBE, and so on)?

Thank You Very Much in Advance,
Best Regards,
Luis Carvalheiro

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

Subject: Re: [rt.rap.ucar.edu #42042] Help with using MET
From: John Halley Gotway
Time: Wed Nov 10 09:36:45 2010

Luis,

Yes, you should be able to use METv3.0 with this dataset, but it'll
take a little bit of tweaking.  Basically, we want to make your data
look like the output of the WRF-ARW pinterp tool.  So we want
to tell MET to interpret this data as such:
(1) You should add a global attribute and make sure it contains the
string "ON PRES LEVELS":
   : TITLE = "ON PRES LEVELS" ;
(2) You'll need to add several global attributes to define the grid on
which this data resides.  Looks like it can either be polar stereo,
lambert conformal, or mercator.  Do you have the definition
of your grid somewhere - perhaps in the NetCDF files output from WRF
directly?

(3) Next, you'll need to format your point observations so that MET
can read them.  Since you have text observations, I'd suggest
formatting them in such a way that the ASCII2NC tool can read them.
I'm guessing that all of you forecast variables and observations are
at the surface, and as such, I'd suggest encoding the observations
using the "ADPSFC" message type.

Once MET is able to read your NetCDF files and you're point
observations, we can help you set up the configuration file for Point-
Stat to perform the verification.  Basically, you'd run Point-Stat for
each valid time and verify it.  Then you could plot a time series of
the statistics using whatever plotting software you'd like.

Hope that helps.  You may find the MET online tutorial of use:
   http://www.dtcenter.org/met/users/support/online_tutorial/METv3.0/index.php

Thanks,
John Halley Gotway

On 11/10/2010 04:31 AM, RAL HelpDesk {for Luis Carlos Oliveira
Carvalheiro} wrote:
> 2) Could you "teach" me a quick way of performing my

------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #42042] Help with using MET
From: Luis Carlos Oliveira Carvalheiro
Time: Wed Nov 10 10:35:27 2010

Dear John,

Thank You for your very kind reply.
I'll add the attributes, and kind of copy the grid
definitions from the wrfout netcdf files to "my" file. In
the meantime, I'll continue reading thoroughly the MET
users manual.

Do you mind if I write you back once these processes are
accomplished? Thank You.

Best Regards,
Luis

Em Wed, 10 Nov 2010 09:36:46 -0700
  "RAL HelpDesk {for John Halley Gotway}"
<met_help at ucar.edu> escreveu:
> Luis,
>
> Yes, you should be able to use METv3.0 with this
>dataset, but it'll take a little bit of tweaking.
> Basically, we want to make your data look like the
>output of the WRF-ARW pinterp tool.  So we want
> to tell MET to interpret this data as such:
> (1) You should add a global attribute and make sure it
>contains the string "ON PRES LEVELS":
>   : TITLE = "ON PRES LEVELS" ;
> (2) You'll need to add several global attributes to
>define the grid on which this data resides.  Looks like
>it can either be polar stereo, lambert conformal, or
>mercator.  Do you have the definition
> of your grid somewhere - perhaps in the NetCDF files
>output from WRF directly?
>
> (3) Next, you'll need to format your point observations
>so that MET can read them.  Since you have text
>observations, I'd suggest formatting them in such a way
>that the ASCII2NC tool can read them.
> I'm guessing that all of you forecast variables and
>observations are at the surface, and as such, I'd suggest
>encoding the observations using the "ADPSFC" message
>type.
>
> Once MET is able to read your NetCDF files and you're
>point observations, we can help you set up the
>configuration file for Point-Stat to perform the
>verification.  Basically, you'd run Point-Stat for
> each valid time and verify it.  Then you could plot a
>time series of the statistics using whatever plotting
>software you'd like.
>
> Hope that helps.  You may find the MET online tutorial
>of use:
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv3.0/index.php
>
> Thanks,
> John Halley Gotway
>
> On 11/10/2010 04:31 AM, RAL HelpDesk {for Luis Carlos
>Oliveira Carvalheiro} wrote:
>> 2) Could you "teach" me a quick way of performing my
>

------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #42042] Help with using MET
From: John Halley Gotway
Time: Wed Nov 10 10:59:25 2010

Luis,

Sure, that's fine.  Just write us here at met_help at ucar.edu if any
issues come up in your use of MET.

John

On 11/10/2010 10:35 AM, RAL HelpDesk {for Luis Carlos Oliveira
Carvalheiro} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=42042 >
>
> Dear John,
>
> Thank You for your very kind reply.
> I'll add the attributes, and kind of copy the grid
> definitions from the wrfout netcdf files to "my" file. In
> the meantime, I'll continue reading thoroughly the MET
> users manual.
>
> Do you mind if I write you back once these processes are
> accomplished? Thank You.
>
> Best Regards,
> Luis
>
> Em Wed, 10 Nov 2010 09:36:46 -0700
>   "RAL HelpDesk {for John Halley Gotway}"
> <met_help at ucar.edu> escreveu:
>> Luis,
>>
>> Yes, you should be able to use METv3.0 with this
>> dataset, but it'll take a little bit of tweaking.
>> Basically, we want to make your data look like the
>> output of the WRF-ARW pinterp tool.  So we want
>> to tell MET to interpret this data as such:
>> (1) You should add a global attribute and make sure it
>> contains the string "ON PRES LEVELS":
>>   : TITLE = "ON PRES LEVELS" ;
>> (2) You'll need to add several global attributes to
>> define the grid on which this data resides.  Looks like
>> it can either be polar stereo, lambert conformal, or
>> mercator.  Do you have the definition
>> of your grid somewhere - perhaps in the NetCDF files
>> output from WRF directly?
>>
>> (3) Next, you'll need to format your point observations
>> so that MET can read them.  Since you have text
>> observations, I'd suggest formatting them in such a way
>> that the ASCII2NC tool can read them.
>> I'm guessing that all of you forecast variables and
>> observations are at the surface, and as such, I'd suggest
>> encoding the observations using the "ADPSFC" message
>> type.
>>
>> Once MET is able to read your NetCDF files and you're
>> point observations, we can help you set up the
>> configuration file for Point-Stat to perform the
>> verification.  Basically, you'd run Point-Stat for
>> each valid time and verify it.  Then you could plot a
>> time series of the statistics using whatever plotting
>> software you'd like.
>>
>> Hope that helps.  You may find the MET online tutorial
>> of use:
>>
http://www.dtcenter.org/met/users/support/online_tutorial/METv3.0/index.php
>>
>> Thanks,
>> John Halley Gotway
>>
>> On 11/10/2010 04:31 AM, RAL HelpDesk {for Luis Carlos
>> Oliveira Carvalheiro} wrote:
>>> 2) Could you "teach" me a quick way of performing my
>>
>

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