[Met_help] Re: MET question

[John Halley Gotway]

Ed,

I'll do my best here.

That difference between WRF-NMM and WRF-ARW is exactly why we put in the "subtract" option to the pcp_combine tool.  What you're describing is the same type of massaging we had to do in the DTC for
the CORETEST runs

(1) I checked some sample data from the CORETEST output.  The precip values at initialization time appear to be equal to 0 (which is what you'd want I think).

(2) The accumulation times you've listed appear correct.  And you will need to specify them.  Those arguments tell the pcp-combine tool what GRIB records to search for.  If you tell it to search for
an accumulation of 48 hours but the file you pass it only contains an accumulation of 24 hours, it will error out.

(3) 61 is the GRIB code for accumulated precipitation and that should never change.  There are other GRIB codes other types of precip (eg. 62=Large Scale Precip, and 63=Convective Precip), but your
WRF output probably doesn't contain those anyway.  I think you just want 61.  For a complete list of GRIB codes please refer to:
http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html

Hope that helps.

John

Ed Tollerud wrote:
> John,
> 
> Yet another pesky question:
> 
> I am planning to verify WRF-NMM and WRF-ARW runs against gage
> precipitation observations that are 24h totals between 1200-1200. I
> believe that the NMM resets total accumulated  precipitation
> occasionally (every 6h?) so I am designing scripts that simply sum up
> the appropriate 4 6h periods. For the ARW, I believe that the
> precipitation variable is accumulated precipitation between run start
> and run end (in our case, 5 days). To produce 24h totals to match the
> observations for the 5 days, I thought to use ARW runs that begin at
> 1200 UTC and then subtract the previous 1200 total value from each
> proceeding day. My questions are these:
> 
> 1) Are model precipitation values at initialization time equal to 0, or
> missing?
> 2) In the pcp_combine subtract argument list, what would I use for
> accum1 and accum2? Would that be 0 for day1 for accum2 and 24 for
> accum1, then 24 for day2 for accum2 and 48 for accum1, etc.? Since we
> are designating the file name for each model valid time, would it make
> any difference if these values were not set at all?
> 3) Is the appropriate GRIB variable number always 61 for total
> precipitation for ARW and NMM?
> 
> Ed
> 
> John Halley Gotway wrote:
>> Ed,
>>
>> Wow, thanks for catching that!  The example on page 3-15 is correct while the example on page 9-1 is not.
>>
>> I see two differences in these examples:
>> (1) On page 3-15, the "-sum" command line option is specified, while on page 9-1 it is not.  This actually isn't a problem because summing is the default behavior, so "-sum" doesn't need to be
>> explicitly stated.
>>
>> (2) The date/time format on page 3-15 is YYYYMMDD_HHMMSS while on page 9-1, it's YYYY:MM:DD_HH:MM:SS.  Page 3-15 is correct while page 9-1 is wrong.  We just neglected to update this when that change
>> was made.
>>
>> The usage statement for pcp_combine correctly lists the expected date/time format.  You can see the usage statement by just running the pcp_combine executable with no arguments.
>>
>> We'll make a note to fix the page 9-1 of the documentation for the next release.
>>
>> Thanks,
>> John
>>
>> Edward Tollerud wrote:
>>   
>>> John,
>>>
>>> For pcp_combine, the argument formats look different on your script
>>> example (9-1) and the examples in 3-15. Do these differences matter?
>>>
>>> Ed
>>>
>>> John Halley Gotway wrote:
>>>     
>>>> Ed,
>>>>
>>>> Specifying P1000 should work just fine.  The NNN is just meant to
>>>> indicate a number, but perhaps that's misleading.  Let me know if you
>>>> have any problems using P1000.
>>>>
>>>> Thanks,
>>>> John
>>>>
>>>> Edward Tollerud wrote:
>>>>  
>>>>       
>>>>> John,
>>>>>
>>>>> A quick question about MET: tools lik grid-stat and point-stat that need
>>>>> level information are assigned a form such as that for vx_grib_code (eg.
>>>>> PNNN). This suggests a three-character pressure; is there a way to
>>>>> specify (for instance) 1000 mb?
>>>>>
>>>>> Ed
>>>>>     
>>>>>         
>