[Met_help] Differences in point_stat output v2 vs. v1.1

[John Halley Gotway]

Jonathan,

Good luck.  Hope it goes well.

While you're welcome to use the tools however you like, let me offer a
word of advice.  The MPR line type was never really intended to be used on
a large scale basis.  It's a pretty inefficient way of storing individual
matched pair data.  So dumping the matched pairs from each run and then
aggregating them afterwards with the STAT-Analysis tool will take up a lot
of storage space!  We intended it more as a diagnostic tool to be used on
a case by case basis.

Rather than dumping the MPR lines for each case, you could consider just
dumping the scalar partial sums (SL1L2 lines) and aggregating those. 
While it's true that there are some statistics that can't be computed
after the fact for SL1L2 lines, a lot of them can.  And it'd take a lot
less storage space and probably run quicker - since it wouldn't have to
spend time writing and reading all those MPR lines.

Just an idea.

John

> John,
>
> I'm glad for the reduced book-keeping.  So, I agree with the decision by
> your team.
> I just want to confirm these differences.
>
> I'm going to run my pb2nc and point_stat scripts over the weekend on the
> 3-months of daily WRF runs I made last summer.
> I just started learning the stat_analysis tool yesterday and today, and
> it looks like the job "stat_job_aggr_mpr" will accomplish most of what I
> want (overall error summaries by forecast hour for period of record).
>
> I'll tackle grid_stat next week with intentions of computing traditional
> precip threat scores and then comparing them to MODE output down the
> road.
>
> Thanks for all the email tutorials!
>
> Enjoy your weekend,
> Jonathan
>
>> -----Original Message-----
>> From: John Halley Gotway [mailto:johnhg at rap.ucar.edu]
>> Sent: Friday, February 20, 2009 4:57 PM
>> To: Case, Jonathan (MSFC-VP61)[Other]
>> Cc: met_help at ucar.edu
>> Subject: Re: Differences in point_stat output v2 vs. v1.1
>>
>> Jonathan,
>>
>> Yes, that's correct.  We realized that putting out 5 different line
>> types
>> (FHO, CTC, CFP, COP, and CTP) for what's really the same data - just
>> the
>> counts from the contingency table - was a bit excessive.
>>
>> It made for a lot of extra book-keeping in the STAT-Analysis tool with
>> limited benefit.  Hopefully, that won't negatively impact you.
>> Converting
>> between CTC and COP, CFP, or CTP is pretty trivial.
>>
>> Thanks,
>> John
>>
>> > John,
>> >
>> >
>> >
>> > It looks like point_stat in MET v2 no longer outputs the CTP, CFP,
>> and
>> > COP proportions files.  Is this correct?
>> >
>> >
>> >
>> > Jonathan
>> >
>> >
>> >
>> > ***********************************************************
>> > Jonathan Case, ENSCO, Inc.
>> > Aerospace Sciences & Engineering Division
>> > Short-term Prediction Research and Transition Center
>> > 320 Sparkman Drive, Room 3062
>> > Huntsville, AL 35805-1912
>> > Voice: (256) 961-7504   Fax: (256) 961-7788
>> > Emails: Jonathan.Case-1 at nasa.gov
>> >
>> >              case.jonathan at ensco.com
>> >
>> > ***********************************************************
>> >
>> >
>> >
>> >
>>
>
>