[Met_help] [rt.rap.ucar.edu #42721] History for Re: threshold values
RAL HelpDesk {for John Halley Gotway}
met_help at ucar.edu
Wed Feb 23 14:44:36 MST 2011
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hello.
Another threshold question...sorta.
It looks like when I threshold a mask for "just compare where obs values are greater than 0 for precip" and compare to two forecast field cases, MET only uses pairs where both the forecast and the obs are >0, so I have different TOTAL counts of pairs for the two cases, even though I would like to compare to the same number of obs each time. This is the case whether or not the forecast threshold is set, of course.
How can can I compare to the exact same obs grid points in MET2, whether or not forecast values are "0"?
Thanks,
Mike
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #42721] Re: threshold values
From: John Halley Gotway
Time: Mon Dec 13 16:12:24 2010
Mike,
I went ahead and created a MET-Help ticket to track your question.
I'm sorry, but can you remind me of the specifics? I see that you're
running METv2.0, but is this question in relation to the Grid-Stat or
the Point-Stat tool?
It sounds like you only want to consider matched pairs who's observed
value is > 0. If you're running Grid-Stat, you can perform "data
masking" in the configuration file by using the "mask_poly"
parameter to pass in a gridded data field and threshold value to
define which grid points you'd like to use. Just pass in the the
gridded observation field as your data mask - since that's how you'd
like to condition it.
If you're running Point-Stat, there isn't a direct way to filter the
points when you run Point-Stat. Instead, you could run Point-Stat
using all the grid points in your domain (or some subset by
defining a masking region), but be sure to dump out the matched pair
(MPR) line type. Then you could use the Stat-Analysis tool to perform
any sort of filtering of those MPR lines that you'd like.
For example, the following type of job would look for a bunch of MPR
lines for 6-hour accumulated precip, select out only those who's
observation value is > 0, and then recompute continuous statistics:
stat_analysis -lookin METv2.0/out/point_stat -job aggregate_stat
-line_type MPR -out_line_type CNT -fcst_var APCP_06 -column_min OBS
0.001
The "-column_min" and "-column_max" values may be used to define
filtering criteria - which is what you're looking for here I believe.
Please do know that I've added your request to enhance Grid-Stat and
Point-Stat to perform this type of filtering directly to our list of
feature requests. We've begun development for the next
version of MET, but have not yet discussed which individual features
we'll be able to add.
Hope that helps. Please let me know if you continue to have problems
or questions.
Thanks,
John
On 12/13/2010 04:03 PM, RAL HelpDesk {for John Halley Gotway} wrote:
>
> Mon Dec 13 16:03:07 2010: Request 42721 was acted upon.
> Transaction: Ticket created by johnhg
> Queue: met_help
> Subject: Re: threshold values
> Owner: johnhg
> Requestors: Michael.Shaw.Ctr at offutt.af.mil
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=42721 >
>
>
> Hello.
>
> Another threshold question...sorta.
>
> It looks like when I threshold a mask for "just compare where obs
values are greater than 0 for precip" and compare to two forecast
field cases, MET only uses pairs where both the forecast and the obs
are >0, so I have different TOTAL counts of pairs for the two cases,
even though I would like to compare to the same number of obs each
time. This is the case whether or not the forecast threshold is set,
of course.
>
> How can can I compare to the exact same obs grid points in MET2,
whether or not forecast values are "0"?
>
> Thanks,
> Mike
>
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #42721] Re: threshold values
From: Michael.Shaw.Ctr at offutt.af.mil
Time: Mon Dec 13 16:21:16 2010
Thanks.
It seems that even when a mask of "only >0 values within
the obs field" is applied in MET, different counts and stats for the
obs field are still possible; I think it's throwing out obs grid point
values when forecast values are "0". (This is grid_stat)
______________________________________________
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
-----Original Message-----
From: RAL
HelpDesk {for John Halley Gotway} [mailto:met_help at ucar.edu]
Sent:
Monday, December 13, 2010 5:12 PM
To: Shaw, Michael J CTR USAF AFWA
16 WS/WXE
Subject: Re: [rt.rap.ucar.edu #42721] Re: threshold values
Mike,
I went ahead and created a MET-Help ticket to track your
question.
I'm sorry, but can you remind me of the specifics? I see
that you're running METv2.0, but is this question in relation to the
Grid-Stat or the Point-Stat tool?
It sounds like you only want to
consider matched pairs who's observed value is > 0. If you're running
Grid-Stat, you can perform "data masking" in the configuration file by
using the "mask_poly"
parameter to pass in a gridded data field and
threshold value to define which grid points you'd like to use. Just
pass in the the gridded observation field as your data mask - since
that's how you'd
like to condition it.
If you're running Point-
Stat, there isn't a direct way to filter the points when you run
Point-Stat. Instead, you could run Point-Stat using all the grid
points in your domain (or some subset by
defining a masking region),
but be sure to dump out the matched pair (MPR) line type. Then you
could use the Stat-Analysis tool to perform any sort of filtering of
those MPR lines that you'd like.
For example, the following type of
job would look for a bunch of MPR lines for 6-hour accumulated precip,
select out only those who's observation value is > 0, and then
recompute continuous statistics:
stat_analysis -lookin
METv2.0/out/point_stat -job aggregate_stat -line_type MPR
-out_line_type CNT -fcst_var APCP_06 -column_min OBS 0.001
The "-
column_min" and "-column_max" values may be used to define filtering
criteria - which is what you're looking for here I believe.
Please
do know that I've added your request to enhance Grid-Stat and Point-
Stat to perform this type of filtering directly to our list of feature
requests. We've begun development for the next
version of MET, but
have not yet discussed which individual features we'll be able to add.
Hope that helps. Please let me know if you continue to have problems
or questions.
Thanks,
John
On 12/13/2010 04:03 PM, RAL
HelpDesk {for John Halley Gotway} wrote:
>
> Mon Dec 13 16:03:07
2010: Request 42721 was acted upon.
> Transaction: Ticket created by
johnhg
> Queue: met_help
> Subject: Re: threshold values
> Owner: johnhg
> Requestors: Michael.Shaw.Ctr at offutt.af.mil
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=42721 >
>
>
>
Hello.
>
> Another threshold question...sorta.
>
> It looks like
when I threshold a mask for "just compare where obs values are greater
than 0 for precip" and compare to two forecast field cases, MET only
uses pairs where both the forecast and the obs are >0, so I have
different TOTAL counts of pairs for the two cases, even though I would
like to compare to the same number of obs each time. This is the case
whether or not the forecast threshold is set, of course.
>
> How
can can I compare to the exact same obs grid points in MET2, whether
or not forecast values are "0"?
>
> Thanks,
> Mike
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #42721] Re: threshold values
From: John Halley Gotway
Time: Mon Dec 13 16:49:17 2010
Mike,
OK, now I understand what you're saying. I tried doing what you're
doing using the sample data that's distributed with the MET tarball.
But I did not see the behavior you're describing. Instead, I
see the expected behavior - only those points whose observed value > 0
were used. And when I run 2 different forecast fields against the
same observation, I get the same number of matched pairs.
There are other possible reasons for why the number of matched pairs
wouldn't match up. For example, if one of your forecast fields
contains some missing data, those matched pairs will not be used.
But if you think there is a problem, I'd be happy to take a look.
Could you please send me some data that demonstrates it? I'd need:
(1) Your 2 forecast files.
(2) Your 1 observation file.
(3) Your Grid-Stat config file.
(4) The command line you use to run Grid-Stat.
You can post it to our anonymous ftp site following these directions:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
Please let me know when you post data so that I can go grab it.
Thanks,
John
On 12/13/2010 04:21 PM, RAL HelpDesk {for
Michael.Shaw.Ctr at offutt.af.mil} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=42721 >
>
> Thanks.
>
> It seems that even when a mask of "only >0 values within the obs
field" is applied in MET, different counts and stats for the obs field
are still possible; I think it's throwing out obs grid point values
when forecast values are "0". (This is grid_stat)
>
> ______________________________________________
>
> 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
>
>
> -----Original Message-----
> From: RAL HelpDesk {for John Halley Gotway}
[mailto:met_help at ucar.edu]
> Sent: Monday, December 13, 2010 5:12 PM
> To: Shaw, Michael J CTR USAF AFWA 16 WS/WXE
> Subject: Re: [rt.rap.ucar.edu #42721] Re: threshold values
>
> Mike,
>
> I went ahead and created a MET-Help ticket to track your question.
>
> I'm sorry, but can you remind me of the specifics? I see that
you're running METv2.0, but is this question in relation to the Grid-
Stat or the Point-Stat tool?
>
> It sounds like you only want to consider matched pairs who's
observed value is > 0. If you're running Grid-Stat, you can perform
"data masking" in the configuration file by using the "mask_poly"
> parameter to pass in a gridded data field and threshold value to
define which grid points you'd like to use. Just pass in the the
gridded observation field as your data mask - since that's how you'd
> like to condition it.
>
> If you're running Point-Stat, there isn't a direct way to filter the
points when you run Point-Stat. Instead, you could run Point-Stat
using all the grid points in your domain (or some subset by
> defining a masking region), but be sure to dump out the matched pair
(MPR) line type. Then you could use the Stat-Analysis tool to perform
any sort of filtering of those MPR lines that you'd like.
> For example, the following type of job would look for a bunch of MPR
lines for 6-hour accumulated precip, select out only those who's
observation value is > 0, and then recompute continuous statistics:
> stat_analysis -lookin METv2.0/out/point_stat -job aggregate_stat
-line_type MPR -out_line_type CNT -fcst_var APCP_06 -column_min OBS
0.001
>
> The "-column_min" and "-column_max" values may be used to define
filtering criteria - which is what you're looking for here I believe.
>
> Please do know that I've added your request to enhance Grid-Stat and
Point-Stat to perform this type of filtering directly to our list of
feature requests. We've begun development for the next
> version of MET, but have not yet discussed which individual features
we'll be able to add.
>
> Hope that helps. Please let me know if you continue to have
problems or questions.
>
> Thanks,
> John
>
>
>
> On 12/13/2010 04:03 PM, RAL HelpDesk {for John Halley Gotway} wrote:
>>
>> Mon Dec 13 16:03:07 2010: Request 42721 was acted upon.
>> Transaction: Ticket created by johnhg
>> Queue: met_help
>> Subject: Re: threshold values
>> Owner: johnhg
>> Requestors: Michael.Shaw.Ctr at offutt.af.mil
>> Status: new
>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=42721 >
>>
>>
>> Hello.
>>
>> Another threshold question...sorta.
>>
>> It looks like when I threshold a mask for "just compare where obs
values are greater than 0 for precip" and compare to two forecast
field cases, MET only uses pairs where both the forecast and the obs
are >0, so I have different TOTAL counts of pairs for the two cases,
even though I would like to compare to the same number of obs each
time. This is the case whether or not the forecast threshold is set,
of course.
>>
>> How can can I compare to the exact same obs grid points in MET2,
whether or not forecast values are "0"?
>>
>> Thanks,
>> Mike
>>
>
------------------------------------------------
More information about the Met_help
mailing list