[Met_help] threshold values

John Halley Gotway johnhg at ucar.edu
Wed Apr 21 09:54:18 MDT 2010


Mike,

Yes, that's correct.  You can currently only apply masks one at a time and there is no mechanism for taking the intersection (or union) of multiple masks.  However, adding that type of functionality
would be feasible.  Is that something that would be useful in your work?

Thanks,
John

Shaw, Michael J CTR USAF AFWA 16 WS/WXE wrote:
> Thanks.
> 
> And...looks like can't mask with the intersection or union of two masks, right?  (E.g., mask the intersection of EAST poly and obs > 20.)
> 
> -----Original Message-----
> From: John Halley Gotway [mailto:johnhg at ucar.edu] 
> Sent: Tuesday, April 20, 2010 5:16 PM
> To: Shaw, Michael J CTR USAF AFWA 16 WS/WXE; met_help
> Subject: Re: threshold values
> 
> Mike,
> 
> Yep, I was just writing you a message about that.  Let me know if that's sufficient for you.  Or if there's an enhancement you'd like to see to "filter" the raw values using a threshold before
> computing continuous statistics, we could add that the list of requested upgrades for future development.
> 
> Thanks,
> John
> 
> Shaw, Michael J CTR USAF AFWA 16 WS/WXE wrote:
>> Ah - I'm seeing I can also threshold the masks.  I.e., use the mask for obs, e.g., for values greater than 20mm and compare the model...
>>
>> Thanks,
>> Mike
>>
>> -----Original Message-----
>> From: John Halley Gotway [mailto:johnhg at ucar.edu] 
>> Sent: Tuesday, April 20, 2010 5:05 PM
>> To: Shaw, Michael J CTR USAF AFWA 16 WS/WXE
>> Cc: 'met_help'
>> Subject: Re: threshold values
>>
>> Mike,
>>
>> I think there's some confusion about how the thresholds you set in the "fcst_thresh" and "obs_thresh" parameters of the configuration file are actually used by MET.  The Grid-Stat and Point-Stat tools
>> compute a variety of different statistics, and those statistics are grouped together by type.  Two of those group are the "continuous" and "categorical" statistics.  The continuous type statistics are
>> all grouped together into the CNT lines.  The categorical type statistics are all grouped together into the CTS lines.
>>
>> Continuous statistics are computed using the raw data values in the fields.  Whereas categorical statistics are computed by applying thresholds to the forecast and observation fields to create fields
>> of 0's and 1's, defining contingency tables, and deriving statistics.  As you change the threshold values, those contingency table counts and corresponding statistics will change.
>>
>> However FBAR is a continuous statistic computed over the raw data values.  It's just the mean of the forecast values in the verification region.  There is no thresholding involved.  When you look at
>> an output CNT line from MET, the "fcst_thresh" and "obs_thresh" values in that line should be set "NA", meaning that thresholds do not apply.
>>
>> Hope that helps straighten it out.  You can find more information about this in the MET User's Guide.
>>
>> John Halley Gotway
>> met_help at ucar.edu
>>
>> Shaw, Michael J CTR USAF AFWA 16 WS/WXE wrote:
>>> Hi.
>>>
>>> When I include a forecast thresh of 0 and fcst thresh of 20, masked over a region (EAST poly), I get the same average of precip (accumulated over a week or so) of about 13.5.  As in identical FBAR.  I'm obviously missing something - I wouldn't think a mean of 13.5 would be possible if I'm using only values >20.
>>>
>>> What am I missing?
>>>
>>> Thanks,
>>> Mike
>>>
>>> ______________________________________________
>>>
>>> Michael Shaw, Contractor
>>> SAIC/NASA Support Scientist
>>> michael.shaw.ctr at offutt.af.mil
>>> 16WS Environmental Characterization
>>> HQ AFWA-SAIC
>>> 101 Nelson Drive
>>> Offutt AFB, NE 68113-1023
>>> 402-232-7690 Comm * 402-272-7690 DSN * 402-294-8230 Fax
>>>
>>>


More information about the Met_help mailing list