[Met_help] met and WRF/ARW

Wed Feb 17 08:09:00 MST 2010

Julien,

Sure, happy to help.

MET is primarily set up to read GRIB files.  In the Point-Stat configuration file, you'll need to set the "fcst_field" and "obs_field" parameters to indicate what variables/levels that you'd like to
verify.  Here's an excerpt of the comments from that configuration file.  But I've added an example to the end of each line:

// Each verification field is specified as one of the following:
//    GC/ANNN for accumulation interval NNN             (APCP/A12 or 61/A12 for 12-hour accumulated precipitation)
//    GC/ZNNN for vertical level NNN                    (TMP/Z2 or 11/Z2 for 2-meter temperature)
//    GC/PNNN for pressure level NNN in hPa             (TMP/P850 or 11/P850 for temperature at 850 hPa)
//    GC/PNNN-NNN for a range of pressure levels in hPa (TMP/P850-700 or 11/P850-700 for temperature between 850 and 700 hPa)
//    GC/LNNN for a generic level type                  (TSOIL/L20 or 85/L20 for the soil temperature 20-cm below ground)
//    GC/RNNN for a specific GRIB record number         (TMP/R45 for the temperature record stored in the 45-th GRIB record)
//    Where GC is the number of or abbreviation for the grib code
//    to be verified.
// http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html

GRIB records may be defined at a variety of different types of levels.  For example, a record value of "10" may mean 10-meters above ground or 10 hPa.  We use the letter (A, Z, P, L, or R) to indicate
the type of level to be used.  A is for accumulation interval and is mostly for precipitation.  Z is for a vertical level, like 2-meters or 10-meters above ground.  P is for a pressure level.  We use
L to mean a "generic" level where we ignore the "type" of level.  Suppose you have temperature defined at 10-meters and 10 hPa in the same GRIB file.  If you request, TMP/L10, both of those records
would match.  Point-Stat will look through the GRIB file for the FIRST record it finds with a level value of 10 and use it.  We use R to specify a particular GRIB record number from a file.

Generally, we suggest using A, Z, or P to specify the level that you'd like to use.  But if your record is defined using an odd level type and A, Z, or P don't work, you can try using L.  If that
still doesn't work, and you just want to get it to run, you could explicitly specify the record number with R.  But using R isn't a good long-term solution, since you shouldn't rely on the ordering of
records to remain the same from one GRIB file to another.

Hope that helps.

John

Julien Pergaud wrote:
> Thanks a lot for your answer and your promptness.
> I will use WPP. Just to inform you, i have one time in each output file. So, i 
> think that the problem i have with WPP doesn't come from that.
> Just another question, could you explain to me what are the levels L and R 
> defined in the PointStatConfig file.
> We want to use MET in an operational context, i think that MET is a very 
> complete tool to verify our forecast.
> 
> Regards
> 
> Julien
> 
> NUMTECH <http://www.numtech.fr> 	Julien PERGAUD
> Ingénieur d'études et de recherche / SO
> 6, allée Alan Turing - BP 30242 - Parc Technologique de la Pardieu
> 63175 AUBIERE Cedex - FRANCE
> Tél: 33 (0)4 73.28.75.95 - Fax: 33 (0)4.73.28.75.99
> Web: http://www.numtech.fr
> 
> 
> Le 16/02/2010 18:51, John Halley Gotway a écrit :
>> Julien,
>>
>> Yes, we highly recommend running your WRF output through WPP prior to using it in MET.  Running it through WPP does three things:
>>
>> (1) Interpolates in the horizontal from the staggered model grid to a regular grid.
>> (2) Interpolates in the vertical from eta levels to pressure levels.  The MET Point-Stat tool does matching in the vertical using pressure levels, not eta levels, since observations are typically
>> reported at pressure levels.
>> (3) Derives several fields, including temperature from the ARW model output.
>>
>> Try switching your WRF output format back to NetCDF and rerunning it through WPP.  I mentioned this to the primary WPP support person here, and she guessed that since you're running ARW, you probably
>> have multiple output times in each NetCDF output file.  Because of that, you'll need to run the WPPV3.1/scripts/run_wrfpost_frames script.  That will read multiple times from the ARW NetCDF files and
>> write them out to separate GRIB output files.
>>
>> Please give that a shot, and if you continue to experience trouble running WPP, send an email to wrfhelp at ucar.edu.
>>
>> Thanks, and once you get up and running, let us know if any more questions arise in your use of MET.
>>
>> John Halley Gotway
>> johnhg at ucar.edu
>>
>> Julien Pergaud wrote:
>>   
>>> Hello
>>>
>>> I have some questions abous the use of MET with the version ARW of WRF.
>>> First, as explained in the tutorial, i have used WPP to transform nectdf output 
>>> in grib.
>>> But, if the first files (first time of the simulation) was tranformed in grib 
>>> without problem, it was not the case for other files (other times).
>>> So, i have decided to use the possibility of WRF/ARW to create as output grib 
>>> files directly.
>>> When i want to use the capability of MET to deal with UPA data, i meet some 
>>> problems.
>>> First, in grib file obtained directly from WRF/ARW, levels are defined as ETA 
>>> levels and not pressure levels.
>>>   - How can i read eta levels with MET? Is it possible?
>>>   - TMP can be extract at different level using LNNN or RNNN. Are L and R linked 
>>> to eta levels. For example, L1.2 doesn't run while L2.5 runs.
>>>    - This remark is linked to the previous. Temperature is an output variable of 
>>> WRF only for 2m level. WRF provides potentiel temperature at the different 
>>> vertical levels and not TMP.
>>>      Is MET able to compute temperature from the different variable of WRF 
>>> (potential temperature and pressure)? If not what is TMP/L2.5
>>>      if i put POT as a variable name in the PointStat config file, MET 
>>>
>>> I have checked the only grib file coming from WPP and this last one contains TMP 
>>> at different vertical pressure levels.  So WPP seems to compute temperature from 
>>> potential temperature.
>>> Do you think it's better to use WPP. If yes, how can i solve my problem with WPP.
>>> Thank you in advance for your answer.
>>>
>>> Julien  
>>> -- 
>>>
>>> NUMTECH <http://www.numtech.fr> 	Julien PERGAUD
>>> Ingénieur d'études et de recherche / SO
>>> 6, allée Alan Turing - BP 30242 - Parc Technologique de la Pardieu
>>> 63175 AUBIERE Cedex - FRANCE
>>> Tél: 33 (0)4 73.28.75.95 - Fax: 33 (0)4.73.28.75.99
>>> Web: http://www.numtech.fr
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> _______________________________________________
>>> Met_help mailing list
>>> Met_help at mailman.ucar.edu
>>> http://mailman.ucar.edu/mailman/listinfo/met_help
>>>     
>>   