[Met_help] [rt.rap.ucar.edu #79851] History for question about Point-Stat (UNCLASSIFIED)

John Halley Gotway via RT met_help at ucar.edu
Wed Mar 22 10:34:47 MDT 2017

```----------------------------------------------------------------
Initial Request
----------------------------------------------------------------

CLASSIFICATION: UNCLASSIFIED

If I want to perform verification tasks as follows:

Set fcst -> level = [ "P100", "P200", "P300" ]; and obs -> level =
> [ "P50-150", "P150-250", "P250-350" ]; This will perform 3 different
> verification tasks.  For each, only one forecast level is used (100,
> 200, or 300), but the observation level  is defined as a range.  For
> example, all observations between 50 and 150 will be compared to the
> forecast values at 100mb.

Is it possible that the same observations used for the verification of the
P100 level will be used for the P200 level because the ranges have a common
value?

Considering vertical interpolation only, are the forecast values for the
matched pairs derived by vertical interpolation up and down in pressure
coordinate to the pressure level of the observations falling in the range of
pressure levels for which the obs levels were set?

Thanks.

R/
John

Mr John W. Raby, Meteorologist
U.S. Army Research Laboratory
White Sands Missile Range, NM 88002
(575) 678-2004 DSN 258-2004
FAX (575) 678-1230 DSN 258-1230
Email: john.w.raby2.civ at mail.mil

CLASSIFICATION: UNCLASSIFIED

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

From: John Halley Gotway
Time: Wed Mar 15 14:33:38 2017

John,

Yes, an observation falling exactly at 150mb would be used to verify
both
the P100 and the P200 forecast levels based on your current settings.
The
observation level ranges include their endpoints on both sides.  To
prevent
that from happening, you could compare forecast "P100" to observed
"P50-149" and "P200" to "P150-249"... or however you want to handle
the
endpoints.  Theoretically, we could put in logic to allow the user to
include/exclude endpoints, but my guess is that'd be more trouble than
it's
worth.

But to be clear, there is no vertical interpolation occurring using
this
setup.  You've told Point-Stat to verify at 3 separate, distinct
levels.
For forecast P100, it uses all observations falling between 50 and
150mb.
It interpolates the P100 forecast data *HORIZONTALLY* to the
observation
lat/lon, but does no interpolation in the vertical.  You've told it to
only
verify forecast level P100 and nothing else.  If you want to
interpolate in
the vertical, you need to give it a range of forecast levels.

For the sake of argument, let's say you have model data from 100 to
1000mb,
every 100mb.  Let's say you're interested in P500.  Here are two
similar,
but different, ways of configuring that:

(1) What you're currently doing with no vertical interpolation:  fcst
->
level = [ "P500" ]; and obs -> level = [ "P450-550" ];

(2) Give Point-Stat forecast level data so that it can do vertical
interpolation: fcst -> level = [ "P400-600" ]; and obs -> level = [
"P450-550" ];

In (2), Point-Stat has 3 forecast levels of data to use, 400, 500, and
600.  Observations falling between 450 and 500 will be compared to a
forecast that was vertically interpolated between 400 and 500.
Observations falling between 500 and 550 will be compared to a
forecast
that was vertically interpolated between 500 and 600.

Make sense?

Thanks,
John Halley Gotway

John

On Wed, Mar 15, 2017 at 1:49 PM, Raby, John W USA CIV via RT <
met_help at ucar.edu> wrote:

>
> Wed Mar 15 13:49:11 2017: Request 79851 was acted upon.
> Transaction: Ticket created by john.w.raby2.civ at mail.mil
>        Queue: met_help
>      Subject: question about Point-Stat (UNCLASSIFIED)
>        Owner: Nobody
>   Requestors: john.w.raby2.civ at mail.mil
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=79851 >
>
>
> CLASSIFICATION: UNCLASSIFIED
>
> If I want to perform verification tasks as follows:
>
> Set fcst -> level = [ "P100", "P200", "P300" ]; and obs -> level =
> > [ "P50-150", "P150-250", "P250-350" ]; This will perform 3
different
> > verification tasks.  For each, only one forecast level is used
(100,
> > 200, or 300), but the observation level  is defined as a range.
For
> > example, all observations between 50 and 150 will be compared to
the
> > forecast values at 100mb.
>
> Is it possible that the same observations used for the verification
of the
> P100 level will be used for the P200 level because the ranges have a
common
> value?
>
> Considering vertical interpolation only, are the forecast values for
the
> matched pairs derived by vertical interpolation up and down in
pressure
> coordinate to the pressure level of the observations falling in the
range
> of
> pressure levels for which the obs levels were set?
>
> Thanks.
>
> R/
> John
>
> Mr John W. Raby, Meteorologist
> U.S. Army Research Laboratory
> White Sands Missile Range, NM 88002
> (575) 678-2004 DSN 258-2004
> FAX (575) 678-1230 DSN 258-1230
> Email: john.w.raby2.civ at mail.mil
>
>
>
> CLASSIFICATION: UNCLASSIFIED
>
>

------------------------------------------------
From: Raby, John W USA CIV
Time: Wed Mar 15 14:50:00 2017

CLASSIFICATION: UNCLASSIFIED

John -

Thanks for clarifying this important difference for me. I am glad I
question, because the answer helps a lot. This makes sense. I like the
option
to verify against a single pressure level with no vertical
interpolation. That
way I can tell the modelers that the only interpolation is in the
horizontal.
It turns out that I modified the UPP post-processor to output 70
pressure
levels instead of 25, so the vertical spacing between pressure levels
is now
much smaller than before and the ranges of the observations for a
given
forecast level is significantly smaller than before and therefore the
observations are more representative of that smaller layer since they
are not
that distant (in pressure) from the forecast value.

I'll experiment with changes in the endpoints to make sure I don't
include
observations in two different levels as you described.

R/
John

-----Original Message-----
From: John Halley Gotway via RT [mailto:met_help at ucar.edu]
Sent: Wednesday, March 15, 2017 2:34 PM
To: Raby, John W CIV USARMY RDECOM ARL (US)
<john.w.raby2.civ at mail.mil>
Subject: [Non-DoD Source] Re: [rt.rap.ucar.edu #79851] question about
Point-Stat (UNCLASSIFIED)

the
identity of the sender, and confirm the authenticity of all links
contained
within the message prior to copying and pasting the address to a Web
browser.

----

John,

Yes, an observation falling exactly at 150mb would be used to verify
both the
P100 and the P200 forecast levels based on your current settings.  The
observation level ranges include their endpoints on both sides.  To
prevent
that from happening, you could compare forecast "P100" to observed
"P50-149"
and "P200" to "P150-249"... or however you want to handle the
endpoints.
Theoretically, we could put in logic to allow the user to
include/exclude
endpoints, but my guess is that'd be more trouble than it's worth.

But to be clear, there is no vertical interpolation occurring using
this
setup.  You've told Point-Stat to verify at 3 separate, distinct
levels.
For forecast P100, it uses all observations falling between 50 and
150mb.
It interpolates the P100 forecast data *HORIZONTALLY* to the
observation
lat/lon, but does no interpolation in the vertical.  You've told it to
only
verify forecast level P100 and nothing else.  If you want to
interpolate in
the vertical, you need to give it a range of forecast levels.

For the sake of argument, let's say you have model data from 100 to
1000mb,
every 100mb.  Let's say you're interested in P500.  Here are two
similar, but
different, ways of configuring that:

(1) What you're currently doing with no vertical interpolation:  fcst
-> level
= [ "P500" ]; and obs -> level = [ "P450-550" ];

(2) Give Point-Stat forecast level data so that it can do vertical
interpolation: fcst -> level = [ "P400-600" ]; and obs -> level = [
"P450-550" ];

In (2), Point-Stat has 3 forecast levels of data to use, 400, 500, and
600.
Observations falling between 450 and 500 will be compared to a
forecast that
was vertically interpolated between 400 and 500.
Observations falling between 500 and 550 will be compared to a
forecast that
was vertically interpolated between 500 and 600.

Make sense?

Thanks,
John Halley Gotway

John

On Wed, Mar 15, 2017 at 1:49 PM, Raby, John W USA CIV via RT <
met_help at ucar.edu> wrote:

>
> Wed Mar 15 13:49:11 2017: Request 79851 was acted upon.
> Transaction: Ticket created by john.w.raby2.civ at mail.mil
>        Queue: met_help
>      Subject: question about Point-Stat (UNCLASSIFIED)
>        Owner: Nobody
>   Requestors: john.w.raby2.civ at mail.mil
>       Status: new
>  Ticket <Caution-url:
> Caution-https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=79851 >
>
>
> CLASSIFICATION: UNCLASSIFIED
>
> If I want to perform verification tasks as follows:
>
> Set fcst -> level = [ "P100", "P200", "P300" ]; and obs -> level =
> > [ "P50-150", "P150-250", "P250-350" ]; This will perform 3
different
> > verification tasks.  For each, only one forecast level is used
(100,
> > 200, or 300), but the observation level  is defined as a range.
For
> > example, all observations between 50 and 150 will be compared to
the
> > forecast values at 100mb.
>
> Is it possible that the same observations used for the verification
of
> the
> P100 level will be used for the P200 level because the ranges have a
> common value?
>
> Considering vertical interpolation only, are the forecast values for
> the matched pairs derived by vertical interpolation up and down in
> pressure coordinate to the pressure level of the observations
falling
> in the range of pressure levels for which the obs levels were set?
>
> Thanks.
>
> R/
> John
>
> Mr John W. Raby, Meteorologist
> U.S. Army Research Laboratory
> White Sands Missile Range, NM 88002
> (575) 678-2004 DSN 258-2004
> FAX (575) 678-1230 DSN 258-1230
> Email: john.w.raby2.civ at mail.mil
>
>
>
> CLASSIFICATION: UNCLASSIFIED
>
>

CLASSIFICATION: UNCLASSIFIED

------------------------------------------------
From: John Halley Gotway
Time: Wed Mar 15 15:07:26 2017

John,

Great, glad that helped.  Lots of little details to consider!

Thanks,
John

On Wed, Mar 15, 2017 at 2:50 PM, Raby, John W USA CIV via RT <
met_help at ucar.edu> wrote:

>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=79851 >
>
> CLASSIFICATION: UNCLASSIFIED
>
> John -
>
> Thanks for clarifying this important difference for me. I am glad I
> the
> question, because the answer helps a lot. This makes sense. I like
the
> option
> to verify against a single pressure level with no vertical
interpolation.
> That
> way I can tell the modelers that the only interpolation is in the
> horizontal.
> It turns out that I modified the UPP post-processor to output 70
pressure
> levels instead of 25, so the vertical spacing between pressure
levels is
> now
> much smaller than before and the ranges of the observations for a
given
> forecast level is significantly smaller than before and therefore
the
> observations are more representative of that smaller layer since
they are
> not
> that distant (in pressure) from the forecast value.
>
> I'll experiment with changes in the endpoints to make sure I don't
include
> observations in two different levels as you described.
>
> R/
> John
>
>
> -----Original Message-----
> From: John Halley Gotway via RT [mailto:met_help at ucar.edu]
> Sent: Wednesday, March 15, 2017 2:34 PM
> To: Raby, John W CIV USARMY RDECOM ARL (US)
<john.w.raby2.civ at mail.mil>
> Subject: [Non-DoD Source] Re: [rt.rap.ucar.edu #79851] question
> Point-Stat (UNCLASSIFIED)
>
verify the
> identity of the sender, and confirm the authenticity of all links
contained
> within the message prior to copying and pasting the address to a Web
> browser.
>
>
>
>
> ----
>
> John,
>
> Yes, an observation falling exactly at 150mb would be used to verify
both
> the
> P100 and the P200 forecast levels based on your current settings.
The
> observation level ranges include their endpoints on both sides.  To
prevent
> that from happening, you could compare forecast "P100" to observed
> "P50-149"
> and "P200" to "P150-249"... or however you want to handle the
endpoints.
> Theoretically, we could put in logic to allow the user to
include/exclude
> endpoints, but my guess is that'd be more trouble than it's worth.
>
> But to be clear, there is no vertical interpolation occurring using
this
> setup.  You've told Point-Stat to verify at 3 separate, distinct
levels.
> For forecast P100, it uses all observations falling between 50 and
150mb.
> It interpolates the P100 forecast data *HORIZONTALLY* to the
observation
> lat/lon, but does no interpolation in the vertical.  You've told it
to only
> verify forecast level P100 and nothing else.  If you want to
interpolate in
> the vertical, you need to give it a range of forecast levels.
>
> For the sake of argument, let's say you have model data from 100 to
1000mb,
> every 100mb.  Let's say you're interested in P500.  Here are two
similar,
> but
> different, ways of configuring that:
>
> (1) What you're currently doing with no vertical interpolation:
fcst ->
> level
> = [ "P500" ]; and obs -> level = [ "P450-550" ];
>
> (2) Give Point-Stat forecast level data so that it can do vertical
> interpolation: fcst -> level = [ "P400-600" ]; and obs -> level = [
> "P450-550" ];
>
> In (2), Point-Stat has 3 forecast levels of data to use, 400, 500,
and 600.
> Observations falling between 450 and 500 will be compared to a
forecast
> that
> was vertically interpolated between 400 and 500.
> Observations falling between 500 and 550 will be compared to a
forecast
> that
> was vertically interpolated between 500 and 600.
>
> Make sense?
>
> Thanks,
> John Halley Gotway
>
>
> John
>
> On Wed, Mar 15, 2017 at 1:49 PM, Raby, John W USA CIV via RT <
> met_help at ucar.edu> wrote:
>
> >
> > Wed Mar 15 13:49:11 2017: Request 79851 was acted upon.
> > Transaction: Ticket created by john.w.raby2.civ at mail.mil
> >        Queue: met_help
> >      Subject: question about Point-Stat (UNCLASSIFIED)
> >        Owner: Nobody
> >   Requestors: john.w.raby2.civ at mail.mil
> >       Status: new
> >  Ticket <Caution-url:
> > Caution-https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=79851 >
> >
> >
> > CLASSIFICATION: UNCLASSIFIED
> >
> > If I want to perform verification tasks as follows:
> >
> > Set fcst -> level = [ "P100", "P200", "P300" ]; and obs -> level =
> > > [ "P50-150", "P150-250", "P250-350" ]; This will perform 3
different
> > > verification tasks.  For each, only one forecast level is used
(100,
> > > 200, or 300), but the observation level  is defined as a range.
For
> > > example, all observations between 50 and 150 will be compared to
the
> > > forecast values at 100mb.
> >
> > Is it possible that the same observations used for the
verification of
> > the
> > P100 level will be used for the P200 level because the ranges have
a
> > common value?
> >
> > Considering vertical interpolation only, are the forecast values
for
> > the matched pairs derived by vertical interpolation up and down in
> > pressure coordinate to the pressure level of the observations
falling
> > in the range of pressure levels for which the obs levels were set?
> >
> > Thanks.
> >
> > R/
> > John
> >
> > Mr John W. Raby, Meteorologist
> > U.S. Army Research Laboratory
> > White Sands Missile Range, NM 88002
> > (575) 678-2004 DSN 258-2004
> > FAX (575) 678-1230 DSN 258-1230
> > Email: john.w.raby2.civ at mail.mil
> >
> >
> >
> > CLASSIFICATION: UNCLASSIFIED
> >
> >
>
>
> CLASSIFICATION: UNCLASSIFIED
>
>

------------------------------------------------
```