[Met_help] [rt.rap.ucar.edu #69307] History for changes in object identification between METv2.0 and METv4.1
John Halley Gotway via RT
met_help at ucar.edu
Mon Oct 13 11:06:05 MDT 2014
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hello,
I'm in the process of upgrading from METv2.0 to METv4.1 and I've noticed
some differences in the way objects are identified between the two
versions. Specifically, METv4.1 has a tendency to not identify as many
small objects, and the interest value seems to be slightly lower overall
for most matched pairs. I know METv4.1 included a fix to a problem that
was discovered with the way object areas were being calculated. Is this
most likely the cause of the differences I'm seeing, or is something else
going on?
Thanks,
Faye
--
Faye Barthold
Testbed Meteorologist
I.M. Systems Group, Inc.
NOAA/NWS/Weather Prediction Center
301-683-1475
faye.barthold at noaa.gov
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: changes in object identification between METv2.0 and METv4.1
From: John Halley Gotway
Time: Wed Oct 08 09:56:56 2014
Faye,
I looked back through the release notes since METv2.0, listed here:
http://www.dtcenter.org/met/users/support/release_notes/index.php
METv4.1 was the only one I could find with a substantive change for
MODE
that would affect how objects are defined. Here's an excerpt from the
METv4.1 release notes:
- Fixed two major bugs in MODE:
- The object areas being reported were artificially inflated. See
explanation: mode_areas.pdf
<http://www.dtcenter.org/met/users/docs/write_ups/mode_areas.pdf>
- Areas of bad data were being treated as zero's in the
convolution
step.
Correcting the artificially inflated object areas would cause the
overall
distribution of object areas to decrease. We were basically counting
an
extra row and column in the object area. Relatively speaking, this
would
affect smaller objects more than larger objects. But I don't know if
this
explains all the differences you're seeing.
If you'd like to send me (or point me to) some sample data you're
using,
along with your MODE configuration file, I'd be happy to run it
through
METv2.0 and METv4.1 and reconcile any differences we find.
Thanks,
John Halley Gotway
met_help at ucar.edu
On Wed, Oct 8, 2014 at 7:11 AM, Faye Barthold via RT
<met_help at ucar.edu>
wrote:
>
> Wed Oct 08 07:11:46 2014: Request 69307 was acted upon.
> Transaction: Ticket created by Faye.Barthold at noaa.gov
> Queue: met_help
> Subject: changes in object identification between METv2.0 and
METv4.1
> Owner: Nobody
> Requestors: Faye.Barthold at noaa.gov
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307 >
>
>
> Hello,
>
> I'm in the process of upgrading from METv2.0 to METv4.1 and I've
noticed
> some differences in the way objects are identified between the two
> versions. Specifically, METv4.1 has a tendency to not identify as
many
> small objects, and the interest value seems to be slightly lower
overall
> for most matched pairs. I know METv4.1 included a fix to a problem
that
> was discovered with the way object areas were being calculated. Is
this
> most likely the cause of the differences I'm seeing, or is something
else
> going on?
>
> Thanks,
> Faye
>
> --
> Faye Barthold
> Testbed Meteorologist
> I.M. Systems Group, Inc.
> NOAA/NWS/Weather Prediction Center
> 301-683-1475
> faye.barthold at noaa.gov
>
>
------------------------------------------------
Subject: changes in object identification between METv2.0 and METv4.1
From: Faye Barthold
Time: Wed Oct 08 11:05:51 2014
Hi John,
I just put some files on the ftp site for a case I was looking at from
October 1st, verifying a 36hr NAM forecast of 24hr precipitation at
the
0.5" threshold. I also included the configuration files I'm using in
both
versions, which should be identical other than formatting differences,
as
well as the postscript output I'm getting. Let me know if you find
anything amiss beyond the change in the area definition.
Thanks,
Faye
On Wed, Oct 8, 2014 at 11:56 AM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:
> Faye,
>
> I looked back through the release notes since METv2.0, listed here:
> http://www.dtcenter.org/met/users/support/release_notes/index.php
>
> METv4.1 was the only one I could find with a substantive change for
MODE
> that would affect how objects are defined. Here's an excerpt from
the
> METv4.1 release notes:
> - Fixed two major bugs in MODE:
> - The object areas being reported were artificially inflated.
See
> explanation: mode_areas.pdf
> <http://www.dtcenter.org/met/users/docs/write_ups/mode_areas.pdf>
> - Areas of bad data were being treated as zero's in the
convolution
> step.
>
> Correcting the artificially inflated object areas would cause the
overall
> distribution of object areas to decrease. We were basically
counting an
> extra row and column in the object area. Relatively speaking, this
would
> affect smaller objects more than larger objects. But I don't know
if this
> explains all the differences you're seeing.
>
> If you'd like to send me (or point me to) some sample data you're
using,
> along with your MODE configuration file, I'd be happy to run it
through
> METv2.0 and METv4.1 and reconcile any differences we find.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On Wed, Oct 8, 2014 at 7:11 AM, Faye Barthold via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Wed Oct 08 07:11:46 2014: Request 69307 was acted upon.
> > Transaction: Ticket created by Faye.Barthold at noaa.gov
> > Queue: met_help
> > Subject: changes in object identification between METv2.0 and
> METv4.1
> > Owner: Nobody
> > Requestors: Faye.Barthold at noaa.gov
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307 >
> >
> >
> > Hello,
> >
> > I'm in the process of upgrading from METv2.0 to METv4.1 and I've
noticed
> > some differences in the way objects are identified between the two
> > versions. Specifically, METv4.1 has a tendency to not identify as
many
> > small objects, and the interest value seems to be slightly lower
overall
> > for most matched pairs. I know METv4.1 included a fix to a
problem that
> > was discovered with the way object areas were being calculated.
Is this
> > most likely the cause of the differences I'm seeing, or is
something else
> > going on?
> >
> > Thanks,
> > Faye
> >
> > --
> > Faye Barthold
> > Testbed Meteorologist
> > I.M. Systems Group, Inc.
> > NOAA/NWS/Weather Prediction Center
> > 301-683-1475
> > faye.barthold at noaa.gov
> >
> >
>
>
--
Faye Barthold
Testbed Meteorologist
I.M. Systems Group, Inc.
NOAA/NWS/Weather Prediction Center
301-683-1475
faye.barthold at noaa.gov
------------------------------------------------
Subject: changes in object identification between METv2.0 and METv4.1
From: John Halley Gotway
Time: Wed Oct 08 14:34:36 2014
Faye,
Thanks for sending the sample data. I took a look and think that,
yes,
these differences are all explainable by the bug fixes to MODE
included in
METv4.1. Note that your configuration is using a minimum area of 20
grid
squares, meaning that it's tossing objects smaller than that area. I
re-ran this case through METv2.0 and METv4.1 with no minimum area
requirement.
Here's how the object counts come out...
METv2.0 -> area_thresh >= 20 -> 8 fcst objects and 7 obs objects
METv4.1 -> area_thresh >= 20 -> 6 fcst objects and 4 obs objects
METv2.0 -> area_thresh >= 0 -> 9 fcst objects and 9 obs objects ->
total
fcst area of 1340 and obs area of 1147
METv4.1 -> area_thresh >= 0 -> 9 fcst objects and 9 obs objects ->
total
fcst area of 1081 and obs area of 947
Looking at the area_thresh >= 0, here are how the individual object
areas
change from METv2.0 to METv4.1:
ID AREA_2.0 AREA_4.1
F001 477 400
F002 156 123
F003 45 30
F004 29 17
F005 329 285
F006 29 18
F007 8 3
F008 56 37
F009 211 168
O001 517 436
O002 33 21
O003 21 11
O004 6 2
O005 21 12
O006 467 413
O007 46 32
O008 32 19
O009 4 1
As you can see objects F004, F006, O003, O005, and O008 went from an
area
greater than 20 to an area less than 20. Using an area threshold of
20
grid squares, they'd show up in the METv2.0 output but not in the
METv4.1
output. That's why we went from 8 to 6 forecast objects and 7 to 4
observation objects. So I'm pretty confident that the differences are
all
explained by the bug fix.
Thanks,
John
On Wed, Oct 8, 2014 at 11:05 AM, Faye Barthold via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307 >
>
> Hi John,
>
> I just put some files on the ftp site for a case I was looking at
from
> October 1st, verifying a 36hr NAM forecast of 24hr precipitation at
the
> 0.5" threshold. I also included the configuration files I'm using
in both
> versions, which should be identical other than formatting
differences, as
> well as the postscript output I'm getting. Let me know if you find
> anything amiss beyond the change in the area definition.
>
> Thanks,
> Faye
>
> On Wed, Oct 8, 2014 at 11:56 AM, John Halley Gotway via RT <
> met_help at ucar.edu> wrote:
>
> > Faye,
> >
> > I looked back through the release notes since METv2.0, listed
here:
> >
http://www.dtcenter.org/met/users/support/release_notes/index.php
> >
> > METv4.1 was the only one I could find with a substantive change
for MODE
> > that would affect how objects are defined. Here's an excerpt from
the
> > METv4.1 release notes:
> > - Fixed two major bugs in MODE:
> > - The object areas being reported were artificially inflated.
See
> > explanation: mode_areas.pdf
> > <http://www.dtcenter.org/met/users/docs/write_ups/mode_areas.pdf>
> > - Areas of bad data were being treated as zero's in the
convolution
> > step.
> >
> > Correcting the artificially inflated object areas would cause the
overall
> > distribution of object areas to decrease. We were basically
counting an
> > extra row and column in the object area. Relatively speaking,
this would
> > affect smaller objects more than larger objects. But I don't know
if
> this
> > explains all the differences you're seeing.
> >
> > If you'd like to send me (or point me to) some sample data you're
using,
> > along with your MODE configuration file, I'd be happy to run it
through
> > METv2.0 and METv4.1 and reconcile any differences we find.
> >
> > Thanks,
> > John Halley Gotway
> > met_help at ucar.edu
> >
> > On Wed, Oct 8, 2014 at 7:11 AM, Faye Barthold via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Wed Oct 08 07:11:46 2014: Request 69307 was acted upon.
> > > Transaction: Ticket created by Faye.Barthold at noaa.gov
> > > Queue: met_help
> > > Subject: changes in object identification between METv2.0
and
> > METv4.1
> > > Owner: Nobody
> > > Requestors: Faye.Barthold at noaa.gov
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307
> >
> > >
> > >
> > > Hello,
> > >
> > > I'm in the process of upgrading from METv2.0 to METv4.1 and I've
> noticed
> > > some differences in the way objects are identified between the
two
> > > versions. Specifically, METv4.1 has a tendency to not identify
as many
> > > small objects, and the interest value seems to be slightly lower
> overall
> > > for most matched pairs. I know METv4.1 included a fix to a
problem
> that
> > > was discovered with the way object areas were being calculated.
Is
> this
> > > most likely the cause of the differences I'm seeing, or is
something
> else
> > > going on?
> > >
> > > Thanks,
> > > Faye
> > >
> > > --
> > > Faye Barthold
> > > Testbed Meteorologist
> > > I.M. Systems Group, Inc.
> > > NOAA/NWS/Weather Prediction Center
> > > 301-683-1475
> > > faye.barthold at noaa.gov
> > >
> > >
> >
> >
>
>
> --
> Faye Barthold
> Testbed Meteorologist
> I.M. Systems Group, Inc.
> NOAA/NWS/Weather Prediction Center
> 301-683-1475
> faye.barthold at noaa.gov
>
>
------------------------------------------------
Subject: changes in object identification between METv2.0 and METv4.1
From: Faye Barthold
Time: Thu Oct 09 07:23:58 2014
Hi John,
Great, that's what I was hoping the answer would be! I'm thinking
about
lowering the area threshold somewhat to recapture some of those
objects
that are now being rejected.
Thanks again for your help,
Faye
On Wed, Oct 8, 2014 at 4:34 PM, John Halley Gotway via RT
<met_help at ucar.edu
> wrote:
> Faye,
>
> Thanks for sending the sample data. I took a look and think that,
yes,
> these differences are all explainable by the bug fixes to MODE
included in
> METv4.1. Note that your configuration is using a minimum area of 20
grid
> squares, meaning that it's tossing objects smaller than that area.
I
> re-ran this case through METv2.0 and METv4.1 with no minimum area
> requirement.
>
> Here's how the object counts come out...
>
> METv2.0 -> area_thresh >= 20 -> 8 fcst objects and 7 obs objects
> METv4.1 -> area_thresh >= 20 -> 6 fcst objects and 4 obs objects
>
> METv2.0 -> area_thresh >= 0 -> 9 fcst objects and 9 obs objects ->
total
> fcst area of 1340 and obs area of 1147
> METv4.1 -> area_thresh >= 0 -> 9 fcst objects and 9 obs objects ->
total
> fcst area of 1081 and obs area of 947
>
> Looking at the area_thresh >= 0, here are how the individual object
areas
> change from METv2.0 to METv4.1:
>
> ID AREA_2.0 AREA_4.1
> F001 477 400
> F002 156 123
> F003 45 30
> F004 29 17
> F005 329 285
> F006 29 18
> F007 8 3
> F008 56 37
> F009 211 168
> O001 517 436
> O002 33 21
> O003 21 11
> O004 6 2
> O005 21 12
> O006 467 413
> O007 46 32
> O008 32 19
> O009 4 1
>
> As you can see objects F004, F006, O003, O005, and O008 went from an
area
> greater than 20 to an area less than 20. Using an area threshold of
20
> grid squares, they'd show up in the METv2.0 output but not in the
METv4.1
> output. That's why we went from 8 to 6 forecast objects and 7 to 4
> observation objects. So I'm pretty confident that the differences
are all
> explained by the bug fix.
>
> Thanks,
> John
>
>
>
>
>
>
>
>
> On Wed, Oct 8, 2014 at 11:05 AM, Faye Barthold via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307 >
> >
> > Hi John,
> >
> > I just put some files on the ftp site for a case I was looking at
from
> > October 1st, verifying a 36hr NAM forecast of 24hr precipitation
at the
> > 0.5" threshold. I also included the configuration files I'm using
in
> both
> > versions, which should be identical other than formatting
differences, as
> > well as the postscript output I'm getting. Let me know if you
find
> > anything amiss beyond the change in the area definition.
> >
> > Thanks,
> > Faye
> >
> > On Wed, Oct 8, 2014 at 11:56 AM, John Halley Gotway via RT <
> > met_help at ucar.edu> wrote:
> >
> > > Faye,
> > >
> > > I looked back through the release notes since METv2.0, listed
here:
> > >
http://www.dtcenter.org/met/users/support/release_notes/index.php
> > >
> > > METv4.1 was the only one I could find with a substantive change
for
> MODE
> > > that would affect how objects are defined. Here's an excerpt
from the
> > > METv4.1 release notes:
> > > - Fixed two major bugs in MODE:
> > > - The object areas being reported were artificially
inflated. See
> > > explanation: mode_areas.pdf
> > >
<http://www.dtcenter.org/met/users/docs/write_ups/mode_areas.pdf>
> > > - Areas of bad data were being treated as zero's in the
> convolution
> > > step.
> > >
> > > Correcting the artificially inflated object areas would cause
the
> overall
> > > distribution of object areas to decrease. We were basically
counting
> an
> > > extra row and column in the object area. Relatively speaking,
this
> would
> > > affect smaller objects more than larger objects. But I don't
know if
> > this
> > > explains all the differences you're seeing.
> > >
> > > If you'd like to send me (or point me to) some sample data
you're
> using,
> > > along with your MODE configuration file, I'd be happy to run it
through
> > > METv2.0 and METv4.1 and reconcile any differences we find.
> > >
> > > Thanks,
> > > John Halley Gotway
> > > met_help at ucar.edu
> > >
> > > On Wed, Oct 8, 2014 at 7:11 AM, Faye Barthold via RT <
> met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Wed Oct 08 07:11:46 2014: Request 69307 was acted upon.
> > > > Transaction: Ticket created by Faye.Barthold at noaa.gov
> > > > Queue: met_help
> > > > Subject: changes in object identification between METv2.0
and
> > > METv4.1
> > > > Owner: Nobody
> > > > Requestors: Faye.Barthold at noaa.gov
> > > > Status: new
> > > > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307
> > >
> > > >
> > > >
> > > > Hello,
> > > >
> > > > I'm in the process of upgrading from METv2.0 to METv4.1 and
I've
> > noticed
> > > > some differences in the way objects are identified between the
two
> > > > versions. Specifically, METv4.1 has a tendency to not
identify as
> many
> > > > small objects, and the interest value seems to be slightly
lower
> > overall
> > > > for most matched pairs. I know METv4.1 included a fix to a
problem
> > that
> > > > was discovered with the way object areas were being
calculated. Is
> > this
> > > > most likely the cause of the differences I'm seeing, or is
something
> > else
> > > > going on?
> > > >
> > > > Thanks,
> > > > Faye
> > > >
> > > > --
> > > > Faye Barthold
> > > > Testbed Meteorologist
> > > > I.M. Systems Group, Inc.
> > > > NOAA/NWS/Weather Prediction Center
> > > > 301-683-1475
> > > > faye.barthold at noaa.gov
> > > >
> > > >
> > >
> > >
> >
> >
> > --
> > Faye Barthold
> > Testbed Meteorologist
> > I.M. Systems Group, Inc.
> > NOAA/NWS/Weather Prediction Center
> > 301-683-1475
> > faye.barthold at noaa.gov
> >
> >
>
>
--
Faye Barthold
Testbed Meteorologist
I.M. Systems Group, Inc.
NOAA/NWS/Weather Prediction Center
301-683-1475
faye.barthold at noaa.gov
------------------------------------------------
Subject: changes in object identification between METv2.0 and METv4.1
From: John Halley Gotway
Time: Thu Oct 09 09:45:12 2014
Faye,
Great, sounds good. I'll resolve this ticket. Just let us know if
any
more issues or questions arise in your use of the MET tools.
Thanks,
John
On Thu, Oct 9, 2014 at 7:23 AM, Faye Barthold via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307 >
>
> Hi John,
>
> Great, that's what I was hoping the answer would be! I'm thinking
about
> lowering the area threshold somewhat to recapture some of those
objects
> that are now being rejected.
>
> Thanks again for your help,
> Faye
>
> On Wed, Oct 8, 2014 at 4:34 PM, John Halley Gotway via RT <
> met_help at ucar.edu
> > wrote:
>
> > Faye,
> >
> > Thanks for sending the sample data. I took a look and think that,
yes,
> > these differences are all explainable by the bug fixes to MODE
included
> in
> > METv4.1. Note that your configuration is using a minimum area of
20 grid
> > squares, meaning that it's tossing objects smaller than that area.
I
> > re-ran this case through METv2.0 and METv4.1 with no minimum area
> > requirement.
> >
> > Here's how the object counts come out...
> >
> > METv2.0 -> area_thresh >= 20 -> 8 fcst objects and 7 obs objects
> > METv4.1 -> area_thresh >= 20 -> 6 fcst objects and 4 obs objects
> >
> > METv2.0 -> area_thresh >= 0 -> 9 fcst objects and 9 obs objects
-> total
> > fcst area of 1340 and obs area of 1147
> > METv4.1 -> area_thresh >= 0 -> 9 fcst objects and 9 obs objects
-> total
> > fcst area of 1081 and obs area of 947
> >
> > Looking at the area_thresh >= 0, here are how the individual
object areas
> > change from METv2.0 to METv4.1:
> >
> > ID AREA_2.0 AREA_4.1
> > F001 477 400
> > F002 156 123
> > F003 45 30
> > F004 29 17
> > F005 329 285
> > F006 29 18
> > F007 8 3
> > F008 56 37
> > F009 211 168
> > O001 517 436
> > O002 33 21
> > O003 21 11
> > O004 6 2
> > O005 21 12
> > O006 467 413
> > O007 46 32
> > O008 32 19
> > O009 4 1
> >
> > As you can see objects F004, F006, O003, O005, and O008 went from
an area
> > greater than 20 to an area less than 20. Using an area threshold
of 20
> > grid squares, they'd show up in the METv2.0 output but not in the
METv4.1
> > output. That's why we went from 8 to 6 forecast objects and 7 to
4
> > observation objects. So I'm pretty confident that the differences
are
> all
> > explained by the bug fix.
> >
> > Thanks,
> > John
> >
> >
> >
> >
> >
> >
> >
> >
> > On Wed, Oct 8, 2014 at 11:05 AM, Faye Barthold via RT
<met_help at ucar.edu
> >
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307 >
> > >
> > > Hi John,
> > >
> > > I just put some files on the ftp site for a case I was looking
at from
> > > October 1st, verifying a 36hr NAM forecast of 24hr precipitation
at the
> > > 0.5" threshold. I also included the configuration files I'm
using in
> > both
> > > versions, which should be identical other than formatting
differences,
> as
> > > well as the postscript output I'm getting. Let me know if you
find
> > > anything amiss beyond the change in the area definition.
> > >
> > > Thanks,
> > > Faye
> > >
> > > On Wed, Oct 8, 2014 at 11:56 AM, John Halley Gotway via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > > Faye,
> > > >
> > > > I looked back through the release notes since METv2.0, listed
here:
> > > >
http://www.dtcenter.org/met/users/support/release_notes/index.php
> > > >
> > > > METv4.1 was the only one I could find with a substantive
change for
> > MODE
> > > > that would affect how objects are defined. Here's an excerpt
from
> the
> > > > METv4.1 release notes:
> > > > - Fixed two major bugs in MODE:
> > > > - The object areas being reported were artificially
inflated.
> See
> > > > explanation: mode_areas.pdf
> > > >
<http://www.dtcenter.org/met/users/docs/write_ups/mode_areas.pdf>
> > > > - Areas of bad data were being treated as zero's in the
> > convolution
> > > > step.
> > > >
> > > > Correcting the artificially inflated object areas would cause
the
> > overall
> > > > distribution of object areas to decrease. We were basically
counting
> > an
> > > > extra row and column in the object area. Relatively speaking,
this
> > would
> > > > affect smaller objects more than larger objects. But I don't
know if
> > > this
> > > > explains all the differences you're seeing.
> > > >
> > > > If you'd like to send me (or point me to) some sample data
you're
> > using,
> > > > along with your MODE configuration file, I'd be happy to run
it
> through
> > > > METv2.0 and METv4.1 and reconcile any differences we find.
> > > >
> > > > Thanks,
> > > > John Halley Gotway
> > > > met_help at ucar.edu
> > > >
> > > > On Wed, Oct 8, 2014 at 7:11 AM, Faye Barthold via RT <
> > met_help at ucar.edu>
> > > > wrote:
> > > >
> > > > >
> > > > > Wed Oct 08 07:11:46 2014: Request 69307 was acted upon.
> > > > > Transaction: Ticket created by Faye.Barthold at noaa.gov
> > > > > Queue: met_help
> > > > > Subject: changes in object identification between
METv2.0 and
> > > > METv4.1
> > > > > Owner: Nobody
> > > > > Requestors: Faye.Barthold at noaa.gov
> > > > > Status: new
> > > > > Ticket <URL:
> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=69307
> > > >
> > > > >
> > > > >
> > > > > Hello,
> > > > >
> > > > > I'm in the process of upgrading from METv2.0 to METv4.1 and
I've
> > > noticed
> > > > > some differences in the way objects are identified between
the two
> > > > > versions. Specifically, METv4.1 has a tendency to not
identify as
> > many
> > > > > small objects, and the interest value seems to be slightly
lower
> > > overall
> > > > > for most matched pairs. I know METv4.1 included a fix to a
problem
> > > that
> > > > > was discovered with the way object areas were being
calculated. Is
> > > this
> > > > > most likely the cause of the differences I'm seeing, or is
> something
> > > else
> > > > > going on?
> > > > >
> > > > > Thanks,
> > > > > Faye
> > > > >
> > > > > --
> > > > > Faye Barthold
> > > > > Testbed Meteorologist
> > > > > I.M. Systems Group, Inc.
> > > > > NOAA/NWS/Weather Prediction Center
> > > > > 301-683-1475
> > > > > faye.barthold at noaa.gov
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> > > --
> > > Faye Barthold
> > > Testbed Meteorologist
> > > I.M. Systems Group, Inc.
> > > NOAA/NWS/Weather Prediction Center
> > > 301-683-1475
> > > faye.barthold at noaa.gov
> > >
> > >
> >
> >
>
>
> --
> Faye Barthold
> Testbed Meteorologist
> I.M. Systems Group, Inc.
> NOAA/NWS/Weather Prediction Center
> 301-683-1475
> faye.barthold at noaa.gov
>
>
------------------------------------------------
More information about the Met_help
mailing list