[Met_help] [rt.rap.ucar.edu #69016] History for changing the intensity percentile comparison

John Halley Gotway via RT met_help at ucar.edu
Fri Sep 19 10:32:30 MDT 2014


----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

Greetings,

I'm successfully running MODE on custom data, consisting of forecast hail size and observed hail size.  My next question is - in addition to comparing the 50th and 90th percentiles of data within my matched cluster objects, I'd also like to compare the maximum of each (100th percentile).  Could you just point me to the code files I'd need to modify to do that?  Looking in MET/src/tools/core/mode/cluster_page.cc, I see where the current percentile values are output (lines 232 and 237).  Any chance it's as simple as changing out the ".p90" for a ".p100"?

I should note I'm still using METv4.1.

Thanks,

Dr. Becky Adams Selin
Staff Scientist, AER
16th Weather Squadron
Air Force Weather Agency


----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: changing the intensity percentile comparison
From: John Halley Gotway
Time: Thu Sep 18 11:46:05 2014

When you run MODE, you can configure what logic you'd like to use when
doing the matching/merging.  Please take a look at this sample MODE
configuration file for METv4.1:

http://www.dtcenter.org/met/users/support/online_tutorial/METv4.1/config/MODEConfig_default

Eight entries in the "weights" section allow you to specify relative
weights to be used for each of the different attributes used when
comparing
two objects.  The last weight listed is named "inten_perc_ratio", for
intensity percentile ratio.  The user can pick the percentile of
intensity
that they'd like to use in the comparisons.  The default
"inten_perc_value
= 50" means that the median intensity of the objects will be used in
the
fuzzy logic.  If you switch that to 100, then the 100th percentile, or
maximum value, will be used.

The MODE output will then contain information about the maximum value
within each object, along with the other intensity percentiles.

You actually don't need to recompile - you can just handle this in the
config file.

Please be aware though that there was an issue in METv4.1 where the
code
was only computing intensity percentiles for raw values >= 0.  This is
mentioned in the METv5.0 release notes:

http://www.dtcenter.org/met/users/support/release_notes/METv5.0_release_notes.php

If all of your data is non-negative, it won't impact you.

Hope that helps clarify.

Thanks,
John Halley Gotway
met_help at ucar.edu

On Thu, Sep 18, 2014 at 10:59 AM, SELIN, REBECCA D CTR USAF AFWA 16
WS/WXE
via RT <met_help at ucar.edu> wrote:

>
> Thu Sep 18 10:59:49 2014: Request 69016 was acted upon.
> Transaction: Ticket created by rebecca.selin.ctr at us.af.mil
>        Queue: met_help
>      Subject: changing the intensity percentile comparison
>        Owner: Nobody
>   Requestors: rebecca.selin.ctr at us.af.mil
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69016 >
>
>
> Greetings,
>
> I'm successfully running MODE on custom data, consisting of forecast
hail
> size and observed hail size.  My next question is - in addition to
> comparing the 50th and 90th percentiles of data within my matched
cluster
> objects, I'd also like to compare the maximum of each (100th
percentile).
> Could you just point me to the code files I'd need to modify to do
that?
> Looking in MET/src/tools/core/mode/cluster_page.cc, I see where the
current
> percentile values are output (lines 232 and 237).  Any chance it's
as
> simple as changing out the ".p90" for a ".p100"?
>
> I should note I'm still using METv4.1.
>
> Thanks,
>
> Dr. Becky Adams Selin
> Staff Scientist, AER
> 16th Weather Squadron
> Air Force Weather Agency
>
>

------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #69016] changing the intensity percentile comparison
From: SELIN, REBECCA D CTR USAF AFWA 16 WS/WXE
Time: Thu Sep 18 12:08:17 2014

John,

Fantastic news!  To clarify though - I *don't* want to use the maximum
value for matching.  I just want to compare maximum values of objects
*after* they are matched.  I believe I am doing that by setting
inten_perc_ratio = 0.0 and inten_perc_value = 100 within the weights
section, is that correct?

Dr. Becky Adams Selin
Staff Scientist, AER
16th Weather Squadron
Air Force Weather Agency

________________________________________
From: John Halley Gotway via RT [met_help at ucar.edu]
Sent: Thursday, September 18, 2014 12:46 PM
To: SELIN, REBECCA D CTR USAF AFWA 16 WS/WXE
Subject: Re: [rt.rap.ucar.edu #69016] changing the intensity
percentile comparison

When you run MODE, you can configure what logic you'd like to use when
doing the matching/merging.  Please take a look at this sample MODE
configuration file for METv4.1:

http://www.dtcenter.org/met/users/support/online_tutorial/METv4.1/config/MODEConfig_default

Eight entries in the "weights" section allow you to specify relative
weights to be used for each of the different attributes used when
comparing
two objects.  The last weight listed is named "inten_perc_ratio", for
intensity percentile ratio.  The user can pick the percentile of
intensity
that they'd like to use in the comparisons.  The default
"inten_perc_value
= 50" means that the median intensity of the objects will be used in
the
fuzzy logic.  If you switch that to 100, then the 100th percentile, or
maximum value, will be used.

The MODE output will then contain information about the maximum value
within each object, along with the other intensity percentiles.

You actually don't need to recompile - you can just handle this in the
config file.

Please be aware though that there was an issue in METv4.1 where the
code
was only computing intensity percentiles for raw values >= 0.  This is
mentioned in the METv5.0 release notes:

http://www.dtcenter.org/met/users/support/release_notes/METv5.0_release_notes.php

If all of your data is non-negative, it won't impact you.

Hope that helps clarify.

Thanks,
John Halley Gotway
met_help at ucar.edu

On Thu, Sep 18, 2014 at 10:59 AM, SELIN, REBECCA D CTR USAF AFWA 16
WS/WXE
via RT <met_help at ucar.edu> wrote:

>
> Thu Sep 18 10:59:49 2014: Request 69016 was acted upon.
> Transaction: Ticket created by rebecca.selin.ctr at us.af.mil
>        Queue: met_help
>      Subject: changing the intensity percentile comparison
>        Owner: Nobody
>   Requestors: rebecca.selin.ctr at us.af.mil
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69016 >
>
>
> Greetings,
>
> I'm successfully running MODE on custom data, consisting of forecast
hail
> size and observed hail size.  My next question is - in addition to
> comparing the 50th and 90th percentiles of data within my matched
cluster
> objects, I'd also like to compare the maximum of each (100th
percentile).
> Could you just point me to the code files I'd need to modify to do
that?
> Looking in MET/src/tools/core/mode/cluster_page.cc, I see where the
current
> percentile values are output (lines 232 and 237).  Any chance it's
as
> simple as changing out the ".p90" for a ".p100"?
>
> I should note I'm still using METv4.1.
>
> Thanks,
>
> Dr. Becky Adams Selin
> Staff Scientist, AER
> 16th Weather Squadron
> Air Force Weather Agency
>
>



------------------------------------------------
Subject: changing the intensity percentile comparison
From: John Halley Gotway
Time: Thu Sep 18 12:26:53 2014

Yes, that's correct.  Setting a weight to zero means that it is not
used in
the merging/matching.  In our applications of MODE, I've found the 8
weights to be overkill.  Typically using the centroid and/or boundary
distance, the intersection/union area ratio, and perhaps the angle
difference provides for good matching logic.

If for some reason, you did want to use the median intensity
percentile for
matching/merging, *AND* see the maximum value in the MODE output, then
yes,
you'd need to edit the source code.  You probably need to do it in at
least
a couple of places - to dump out the correct value and update the
column
names.

Thanks
John

On Thu, Sep 18, 2014 at 12:08 PM, SELIN, REBECCA D CTR USAF AFWA 16
WS/WXE
via RT <met_help at ucar.edu> wrote:

>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69016 >
>
> John,
>
> Fantastic news!  To clarify though - I *don't* want to use the
maximum
> value for matching.  I just want to compare maximum values of
objects
> *after* they are matched.  I believe I am doing that by setting
> inten_perc_ratio = 0.0 and inten_perc_value = 100 within the weights
> section, is that correct?
>
> Dr. Becky Adams Selin
> Staff Scientist, AER
> 16th Weather Squadron
> Air Force Weather Agency
>
> ________________________________________
> From: John Halley Gotway via RT [met_help at ucar.edu]
> Sent: Thursday, September 18, 2014 12:46 PM
> To: SELIN, REBECCA D CTR USAF AFWA 16 WS/WXE
> Subject: Re: [rt.rap.ucar.edu #69016] changing the intensity
percentile
> comparison
>
> When you run MODE, you can configure what logic you'd like to use
when
> doing the matching/merging.  Please take a look at this sample MODE
> configuration file for METv4.1:
>
>
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv4.1/config/MODEConfig_default
>
> Eight entries in the "weights" section allow you to specify relative
> weights to be used for each of the different attributes used when
comparing
> two objects.  The last weight listed is named "inten_perc_ratio",
for
> intensity percentile ratio.  The user can pick the percentile of
intensity
> that they'd like to use in the comparisons.  The default
"inten_perc_value
> = 50" means that the median intensity of the objects will be used in
the
> fuzzy logic.  If you switch that to 100, then the 100th percentile,
or
> maximum value, will be used.
>
> The MODE output will then contain information about the maximum
value
> within each object, along with the other intensity percentiles.
>
> You actually don't need to recompile - you can just handle this in
the
> config file.
>
> Please be aware though that there was an issue in METv4.1 where the
code
> was only computing intensity percentiles for raw values >= 0.  This
is
> mentioned in the METv5.0 release notes:
>
>
>
http://www.dtcenter.org/met/users/support/release_notes/METv5.0_release_notes.php
>
> If all of your data is non-negative, it won't impact you.
>
> Hope that helps clarify.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On Thu, Sep 18, 2014 at 10:59 AM, SELIN, REBECCA D CTR USAF AFWA 16
WS/WXE
> via RT <met_help at ucar.edu> wrote:
>
> >
> > Thu Sep 18 10:59:49 2014: Request 69016 was acted upon.
> > Transaction: Ticket created by rebecca.selin.ctr at us.af.mil
> >        Queue: met_help
> >      Subject: changing the intensity percentile comparison
> >        Owner: Nobody
> >   Requestors: rebecca.selin.ctr at us.af.mil
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69016 >
> >
> >
> > Greetings,
> >
> > I'm successfully running MODE on custom data, consisting of
forecast hail
> > size and observed hail size.  My next question is - in addition to
> > comparing the 50th and 90th percentiles of data within my matched
cluster
> > objects, I'd also like to compare the maximum of each (100th
percentile).
> > Could you just point me to the code files I'd need to modify to do
that?
> > Looking in MET/src/tools/core/mode/cluster_page.cc, I see where
the
> current
> > percentile values are output (lines 232 and 237).  Any chance it's
as
> > simple as changing out the ".p90" for a ".p100"?
> >
> > I should note I'm still using METv4.1.
> >
> > Thanks,
> >
> > Dr. Becky Adams Selin
> > Staff Scientist, AER
> > 16th Weather Squadron
> > Air Force Weather Agency
> >
> >
>
>
>
>

------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #69016] changing the intensity percentile comparison
From: SELIN, REBECCA D CTR USAF AFWA 16 WS/WXE
Time: Thu Sep 18 12:40:21 2014

Great, thanks for the help.

Dr. Becky Adams Selin
Staff Scientist, AER
16th Weather Squadron
Air Force Weather Agency

________________________________________
From: John Halley Gotway via RT [met_help at ucar.edu]
Sent: Thursday, September 18, 2014 1:26 PM
To: SELIN, REBECCA D CTR USAF AFWA 16 WS/WXE
Subject: Re: [rt.rap.ucar.edu #69016] changing the intensity
percentile comparison

Yes, that's correct.  Setting a weight to zero means that it is not
used in
the merging/matching.  In our applications of MODE, I've found the 8
weights to be overkill.  Typically using the centroid and/or boundary
distance, the intersection/union area ratio, and perhaps the angle
difference provides for good matching logic.

If for some reason, you did want to use the median intensity
percentile for
matching/merging, *AND* see the maximum value in the MODE output, then
yes,
you'd need to edit the source code.  You probably need to do it in at
least
a couple of places - to dump out the correct value and update the
column
names.

Thanks
John

On Thu, Sep 18, 2014 at 12:08 PM, SELIN, REBECCA D CTR USAF AFWA 16
WS/WXE
via RT <met_help at ucar.edu> wrote:

>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69016 >
>
> John,
>
> Fantastic news!  To clarify though - I *don't* want to use the
maximum
> value for matching.  I just want to compare maximum values of
objects
> *after* they are matched.  I believe I am doing that by setting
> inten_perc_ratio = 0.0 and inten_perc_value = 100 within the weights
> section, is that correct?
>
> Dr. Becky Adams Selin
> Staff Scientist, AER
> 16th Weather Squadron
> Air Force Weather Agency
>
> ________________________________________
> From: John Halley Gotway via RT [met_help at ucar.edu]
> Sent: Thursday, September 18, 2014 12:46 PM
> To: SELIN, REBECCA D CTR USAF AFWA 16 WS/WXE
> Subject: Re: [rt.rap.ucar.edu #69016] changing the intensity
percentile
> comparison
>
> When you run MODE, you can configure what logic you'd like to use
when
> doing the matching/merging.  Please take a look at this sample MODE
> configuration file for METv4.1:
>
>
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv4.1/config/MODEConfig_default
>
> Eight entries in the "weights" section allow you to specify relative
> weights to be used for each of the different attributes used when
comparing
> two objects.  The last weight listed is named "inten_perc_ratio",
for
> intensity percentile ratio.  The user can pick the percentile of
intensity
> that they'd like to use in the comparisons.  The default
"inten_perc_value
> = 50" means that the median intensity of the objects will be used in
the
> fuzzy logic.  If you switch that to 100, then the 100th percentile,
or
> maximum value, will be used.
>
> The MODE output will then contain information about the maximum
value
> within each object, along with the other intensity percentiles.
>
> You actually don't need to recompile - you can just handle this in
the
> config file.
>
> Please be aware though that there was an issue in METv4.1 where the
code
> was only computing intensity percentiles for raw values >= 0.  This
is
> mentioned in the METv5.0 release notes:
>
>
>
http://www.dtcenter.org/met/users/support/release_notes/METv5.0_release_notes.php
>
> If all of your data is non-negative, it won't impact you.
>
> Hope that helps clarify.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On Thu, Sep 18, 2014 at 10:59 AM, SELIN, REBECCA D CTR USAF AFWA 16
WS/WXE
> via RT <met_help at ucar.edu> wrote:
>
> >
> > Thu Sep 18 10:59:49 2014: Request 69016 was acted upon.
> > Transaction: Ticket created by rebecca.selin.ctr at us.af.mil
> >        Queue: met_help
> >      Subject: changing the intensity percentile comparison
> >        Owner: Nobody
> >   Requestors: rebecca.selin.ctr at us.af.mil
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69016 >
> >
> >
> > Greetings,
> >
> > I'm successfully running MODE on custom data, consisting of
forecast hail
> > size and observed hail size.  My next question is - in addition to
> > comparing the 50th and 90th percentiles of data within my matched
cluster
> > objects, I'd also like to compare the maximum of each (100th
percentile).
> > Could you just point me to the code files I'd need to modify to do
that?
> > Looking in MET/src/tools/core/mode/cluster_page.cc, I see where
the
> current
> > percentile values are output (lines 232 and 237).  Any chance it's
as
> > simple as changing out the ".p90" for a ".p100"?
> >
> > I should note I'm still using METv4.1.
> >
> > Thanks,
> >
> > Dr. Becky Adams Selin
> > Staff Scientist, AER
> > 16th Weather Squadron
> > Air Force Weather Agency
> >
> >
>
>
>
>



------------------------------------------------


More information about the Met_help mailing list