[Met_help] [rt.rap.ucar.edu #74897] History for doubt reg Upscaling in Grid_stat tool.
John Halley Gotway via RT
met_help at ucar.edu
Thu Feb 4 11:59:16 MST 2016
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi John. Dear met_help.
I have some doubt regarding "Upscaling" in the grid_stat tool. I have defined different thresholds (for rainfall) in below sections of the Configuration file. fcst = { wind_thresh = [ NA ]; field = [ { name = "APCP_24"; level = [ "(*,*)" ]; cat_thresh = [ >=0.1, >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5]; } ];};obs = { wind_thresh = [ NA ]; field = [ { name = "APCP_24"; level = [ "(*,*)" ]; cat_thresh = [ >=0.1, >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5]; } ];interp = { field = FCST; // if BOTH then both will be upscaled vld_thresh = 1.0; type = [ { method = UW_MEAN; width = 1; }, { method = UW_MEAN; width = 3; }, { method = UW_MEAN; width = 5; }; ];};I want to know the procedure followed by grid_stat tool of MET to do UPSCALING. What I mea!
n is that both the OBS and FCST fields are converted to binary fields (depending on the threshold) and then more number of grid points are incorporated and then ETS is calculated??. I mean are the binary fields averaged???.
Or the actual rainfall sums are computed for different neighbourhoods???Kindly clarify. Geeta
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: doubt reg Upscaling in Grid_stat tool.
From: John Halley Gotway
Time: Thu Jan 28 16:39:13 2016
Hello Geeta,
The smoothing done by the Grid-Stat tool is done as follows...
(1) Read the raw forecast and observation fields into memory.
(2) Apply the smoothing method defined by the user. For example, for
UW_MEAN with a width of 5, replace the raw data value at each grid
point by
the average of the 25 points in the 5x5 box centered on that point.
In
your settings, you've applied the smoothing to only the forecast field
(field = FCST;).
(3) The raw forecast field has now replaced by it's smoothed version.
And
that smoothed version is compared to the raw observation field. The
categorical thresholds are applied to those fields and contingency
table
counts and statistics are computed.
To put it simply... the smoothing defined in the "interp" section is
done
on the raw fields, not the thresholded binary fields.
Thanks,
John Halley Gotway
On Thu, Jan 28, 2016 at 8:53 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> Thu Jan 28 08:53:00 2016: Request 74897 was acted upon.
> Transaction: Ticket created by geeta124 at hotmail.com
> Queue: met_help
> Subject: doubt reg Upscaling in Grid_stat tool.
> Owner: Nobody
> Requestors: geeta124 at hotmail.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
>
>
> Hi John. Dear met_help.
> I have some doubt regarding "Upscaling" in the grid_stat tool. I
have
> defined different thresholds (for rainfall) in below sections of the
> Configuration file. fcst = { wind_thresh = [ NA ]; field = [
{
> name = "APCP_24"; level = [ "(*,*)" ];
> cat_thresh = [ >=0.1,
> >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
}
> ];};obs = { wind_thresh = [ NA ]; field = [ { name
> = "APCP_24"; level = [ "(*,*)" ]; cat_thresh = [
>=0.1,
> >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
}
> ];interp = { field = FCST; // if BOTH then both will be
upscaled
> vld_thresh = 1.0; type = [ { method = UW_MEAN;
> width = 1; }, { method = UW_MEAN;
width =
> 3; }, { method = UW_MEAN; width = 5;
};
> ];};I want to know the procedure followed by grid_stat tool of MET
to do
> UPSCALING. What I mea!
> n is that both the OBS and FCST fields are converted to binary
fields
> (depending on the threshold) and then more number of grid points are
> incorporated and then ETS is calculated??. I mean are the binary
fields
> averaged???.
> Or the actual rainfall sums are computed for different
> neighbourhoods???Kindly clarify. Geeta
>
------------------------------------------------
Subject: doubt reg Upscaling in Grid_stat tool.
From: Geeta Geeta
Time: Fri Jan 29 04:59:02 2016
thanks John. this doubt arose when I was reading the paper by Dr Ebert
(2008) in Met applications. In the Table 1 mentioned there,
Upscaling (Zepeda-Arce et al.,2000; Weygandt et al., 2004;Yates et
al., 2006) NO-NF Useful forecast resembles the
observations when averaged to coarser scales
It uses the <Ix>s and <Iy>s terminology for the binary fields of OBS
and FCST NEIGHBOUR hoods. It says that <Ix>s and <Iy>s are compared in
upscaling.
I also did not understand the 3 point in the TABLE (Damrath 2004)
which says that the decision model is " Useful Forecast is more
correct than incorrect".
Can u pls suggest.
Geeta
> Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat tool.
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Thu, 28 Jan 2016 16:39:14 -0700
>
> Hello Geeta,
>
> The smoothing done by the Grid-Stat tool is done as follows...
>
> (1) Read the raw forecast and observation fields into memory.
>
> (2) Apply the smoothing method defined by the user. For example,
for
> UW_MEAN with a width of 5, replace the raw data value at each grid
point by
> the average of the 25 points in the 5x5 box centered on that point.
In
> your settings, you've applied the smoothing to only the forecast
field
> (field = FCST;).
>
> (3) The raw forecast field has now replaced by it's smoothed
version. And
> that smoothed version is compared to the raw observation field. The
> categorical thresholds are applied to those fields and contingency
table
> counts and statistics are computed.
>
> To put it simply... the smoothing defined in the "interp" section is
done
> on the raw fields, not the thresholded binary fields.
>
> Thanks,
> John Halley Gotway
>
>
>
> On Thu, Jan 28, 2016 at 8:53 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Thu Jan 28 08:53:00 2016: Request 74897 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> > Queue: met_help
> > Subject: doubt reg Upscaling in Grid_stat tool.
> > Owner: Nobody
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
> >
> >
> > Hi John. Dear met_help.
> > I have some doubt regarding "Upscaling" in the grid_stat tool. I
have
> > defined different thresholds (for rainfall) in below sections of
the
> > Configuration file. fcst = { wind_thresh = [ NA ]; field = [
{
> > name = "APCP_24"; level = [ "(*,*)" ];
> > cat_thresh = [ >=0.1,
> > >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
}
> > ];};obs = { wind_thresh = [ NA ]; field = [ {
name
> > = "APCP_24"; level = [ "(*,*)" ]; cat_thresh =
[ >=0.1,
> > >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
}
> > ];interp = { field = FCST; // if BOTH then both will be
upscaled
> > vld_thresh = 1.0; type = [ { method = UW_MEAN;
> > width = 1; }, { method = UW_MEAN;
width =
> > 3; }, { method = UW_MEAN; width = 5;
};
> > ];};I want to know the procedure followed by grid_stat tool of
MET to do
> > UPSCALING. What I mea!
> > n is that both the OBS and FCST fields are converted to binary
fields
> > (depending on the threshold) and then more number of grid points
are
> > incorporated and then ETS is calculated??. I mean are the binary
fields
> > averaged???.
> > Or the actual rainfall sums are computed for different
> > neighbourhoods???Kindly clarify. Geeta
> >
>
------------------------------------------------
Subject: doubt reg Upscaling in Grid_stat tool.
From: John Halley Gotway
Time: Fri Jan 29 12:14:21 2016
Geeta,
I am not able to provide any clarification on those papers. I don't
know
them well enough to comment.
In my previous email, I described the logic by which the interpolation
methods defined in the "interp" section of the Grid-Stat configuration
file
are applied. Those smoothing methods affect the statistical output
found
in the traditional statistics lines type (SL1L2, CNT, CTC, and CTS).
The logic for applying the "nbrhd" configuration settings is
different, and
they control the output found in the neighborhood statistics line
types
(NBRCNT, NBRCTC, and NBRCTS). I'm sure I've gone over this in the
past...
but simply put:
(1) Grid-Stat reads the raw forecast and observation fields.
(2) It applies the categorical threshold (cat_thresh) to convert those
raw
fields into binary 0/1 fields.
(3) It applies the neighborhood size to convert the binary field into
a
fractional coverage field. The fractional coverage value at a grid
point
is ratio for the number of grid points in that neighborhood that have
a
value of 1.
(4) Fractions Skill Score (FSS) and Fractions Brier Score (FBS) are
computed using these fractional coverage fields.
(5) This same method is applied for each combination of categorical
threshold and neighborhood size requested by the user.
Thanks,
John
On Fri, Jan 29, 2016 at 4:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
>
> thanks John. this doubt arose when I was reading the paper by Dr
Ebert
> (2008) in Met applications. In the Table 1 mentioned there,
> Upscaling (Zepeda-Arce et al.,2000; Weygandt et al., 2004;Yates et
al.,
> 2006) NO-NF Useful forecast resembles the
observations when
> averaged to coarser scales
> It uses the <Ix>s and <Iy>s terminology for the binary fields of OBS
and
> FCST NEIGHBOUR hoods. It says that <Ix>s and <Iy>s are compared in
> upscaling.
> I also did not understand the 3 point in the TABLE (Damrath 2004)
which
> says that the decision model is " Useful Forecast is more correct
than
> incorrect".
> Can u pls suggest.
> Geeta
> > Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat
> tool.
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Thu, 28 Jan 2016 16:39:14 -0700
> >
> > Hello Geeta,
> >
> > The smoothing done by the Grid-Stat tool is done as follows...
> >
> > (1) Read the raw forecast and observation fields into memory.
> >
> > (2) Apply the smoothing method defined by the user. For example,
for
> > UW_MEAN with a width of 5, replace the raw data value at each grid
point
> by
> > the average of the 25 points in the 5x5 box centered on that
point. In
> > your settings, you've applied the smoothing to only the forecast
field
> > (field = FCST;).
> >
> > (3) The raw forecast field has now replaced by it's smoothed
version.
> And
> > that smoothed version is compared to the raw observation field.
The
> > categorical thresholds are applied to those fields and contingency
table
> > counts and statistics are computed.
> >
> > To put it simply... the smoothing defined in the "interp" section
is done
> > on the raw fields, not the thresholded binary fields.
> >
> > Thanks,
> > John Halley Gotway
> >
> >
> >
> > On Thu, Jan 28, 2016 at 8:53 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Thu Jan 28 08:53:00 2016: Request 74897 was acted upon.
> > > Transaction: Ticket created by geeta124 at hotmail.com
> > > Queue: met_help
> > > Subject: doubt reg Upscaling in Grid_stat tool.
> > > Owner: Nobody
> > > Requestors: geeta124 at hotmail.com
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897
> >
> > >
> > >
> > > Hi John. Dear met_help.
> > > I have some doubt regarding "Upscaling" in the grid_stat tool. I
have
> > > defined different thresholds (for rainfall) in below sections of
the
> > > Configuration file. fcst = { wind_thresh = [ NA ]; field =
[
> {
> > > name = "APCP_24"; level = [ "(*,*)" ];
> > > cat_thresh = [ >=0.1,
> > > >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
}
> > > ];};obs = { wind_thresh = [ NA ]; field = [ {
name
> > > = "APCP_24"; level = [ "(*,*)" ]; cat_thresh
= [
> >=0.1,
> > > >=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
}
> > > ];interp = { field = FCST; // if BOTH then both will be
> upscaled
> > > vld_thresh = 1.0; type = [ { method = UW_MEAN;
> > > width = 1; }, { method = UW_MEAN;
width
> =
> > > 3; }, { method = UW_MEAN; width =
5;
> };
> > > ];};I want to know the procedure followed by grid_stat tool of
MET to
> do
> > > UPSCALING. What I mea!
> > > n is that both the OBS and FCST fields are converted to binary
fields
> > > (depending on the threshold) and then more number of grid points
are
> > > incorporated and then ETS is calculated??. I mean are the binary
fields
> > > averaged???.
> > > Or the actual rainfall sums are computed for different
> > > neighbourhoods???Kindly clarify. Geeta
> > >
> >
>
>
------------------------------------------------
Subject: doubt reg Upscaling in Grid_stat tool.
From: Geeta Geeta
Time: Tue Feb 02 02:23:07 2016
thanks John. In the grid_stat_270000L_20110423_030000V_nbrcnt.txt, FSS
and AFSS are calculated. what is the difference between the two??.
Geeta
> Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat tool.
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 29 Jan 2016 12:14:22 -0700
>
> Geeta,
>
> I am not able to provide any clarification on those papers. I don't
know
> them well enough to comment.
>
> In my previous email, I described the logic by which the
interpolation
> methods defined in the "interp" section of the Grid-Stat
configuration file
> are applied. Those smoothing methods affect the statistical output
found
> in the traditional statistics lines type (SL1L2, CNT, CTC, and CTS).
>
> The logic for applying the "nbrhd" configuration settings is
different, and
> they control the output found in the neighborhood statistics line
types
> (NBRCNT, NBRCTC, and NBRCTS). I'm sure I've gone over this in the
past...
> but simply put:
>
> (1) Grid-Stat reads the raw forecast and observation fields.
>
> (2) It applies the categorical threshold (cat_thresh) to convert
those raw
> fields into binary 0/1 fields.
>
> (3) It applies the neighborhood size to convert the binary field
into a
> fractional coverage field. The fractional coverage value at a grid
point
> is ratio for the number of grid points in that neighborhood that
have a
> value of 1.
>
> (4) Fractions Skill Score (FSS) and Fractions Brier Score (FBS) are
> computed using these fractional coverage fields.
>
> (5) This same method is applied for each combination of categorical
> threshold and neighborhood size requested by the user.
>
> Thanks,
> John
>
> On Fri, Jan 29, 2016 at 4:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
> >
> > thanks John. this doubt arose when I was reading the paper by Dr
Ebert
> > (2008) in Met applications. In the Table 1 mentioned there,
> > Upscaling (Zepeda-Arce et al.,2000; Weygandt et al., 2004;Yates et
al.,
> > 2006) NO-NF Useful forecast resembles the
observations when
> > averaged to coarser scales
> > It uses the <Ix>s and <Iy>s terminology for the binary fields of
OBS and
> > FCST NEIGHBOUR hoods. It says that <Ix>s and <Iy>s are compared in
> > upscaling.
> > I also did not understand the 3 point in the TABLE (Damrath 2004)
which
> > says that the decision model is " Useful Forecast is more correct
than
> > incorrect".
> > Can u pls suggest.
> > Geeta
> > > Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat
> > tool.
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Thu, 28 Jan 2016 16:39:14 -0700
> > >
> > > Hello Geeta,
> > >
> > > The smoothing done by the Grid-Stat tool is done as follows...
> > >
> > > (1) Read the raw forecast and observation fields into memory.
> > >
> > > (2) Apply the smoothing method defined by the user. For
example, for
> > > UW_MEAN with a width of 5, replace the raw data value at each
grid point
> > by
> > > the average of the 25 points in the 5x5 box centered on that
point. In
> > > your settings, you've applied the smoothing to only the forecast
field
> > > (field = FCST;).
> > >
> > > (3) The raw forecast field has now replaced by it's smoothed
version.
> > And
> > > that smoothed version is compared to the raw observation field.
The
> > > categorical thresholds are applied to those fields and
contingency table
> > > counts and statistics are computed.
> > >
> > > To put it simply... the smoothing defined in the "interp"
section is done
> > > on the raw fields, not the thresholded binary fields.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > >
> > > On Thu, Jan 28, 2016 at 8:53 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu Jan 28 08:53:00 2016: Request 74897 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: doubt reg Upscaling in Grid_stat tool.
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897
> > >
> > > >
> > > >
> > > > Hi John. Dear met_help.
> > > > I have some doubt regarding "Upscaling" in the grid_stat tool.
I have
> > > > defined different thresholds (for rainfall) in below sections
of the
> > > > Configuration file. fcst = { wind_thresh = [ NA ]; field
= [
> > {
> > > > name = "APCP_24"; level = [ "(*,*)" ];
> > > > cat_thresh = [ >=0.1,
> > > >
>=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5]; }
> > > > ];};obs = { wind_thresh = [ NA ]; field = [ {
name
> > > > = "APCP_24"; level = [ "(*,*)" ];
cat_thresh = [
> > >=0.1,
> > > >
>=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5]; }
> > > > ];interp = { field = FCST; // if BOTH then both will
be
> > upscaled
> > > > vld_thresh = 1.0; type = [ { method =
UW_MEAN;
> > > > width = 1; }, { method = UW_MEAN;
width
> > =
> > > > 3; }, { method = UW_MEAN; width =
5;
> > };
> > > > ];};I want to know the procedure followed by grid_stat tool
of MET to
> > do
> > > > UPSCALING. What I mea!
> > > > n is that both the OBS and FCST fields are converted to
binary fields
> > > > (depending on the threshold) and then more number of grid
points are
> > > > incorporated and then ETS is calculated??. I mean are the
binary fields
> > > > averaged???.
> > > > Or the actual rainfall sums are computed for different
> > > > neighbourhoods???Kindly clarify. Geeta
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: doubt reg Upscaling in Grid_stat tool.
From: Geeta Geeta
Time: Tue Feb 02 02:26:26 2016
I also want to know how is the UFSS calculated.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #74897] doubt reg Upscaling in Grid_stat
tool.
Date: Tue, 2 Feb 2016 14:53:03 +0530
thanks John. In the grid_stat_270000L_20110423_030000V_nbrcnt.txt, FSS
and AFSS are calculated. what is the difference between the two??.
Geeta
> Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat tool.
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 29 Jan 2016 12:14:22 -0700
>
> Geeta,
>
> I am not able to provide any clarification on those papers. I don't
know
> them well enough to comment.
>
> In my previous email, I described the logic by which the
interpolation
> methods defined in the "interp" section of the Grid-Stat
configuration file
> are applied. Those smoothing methods affect the statistical output
found
> in the traditional statistics lines type (SL1L2, CNT, CTC, and CTS).
>
> The logic for applying the "nbrhd" configuration settings is
different, and
> they control the output found in the neighborhood statistics line
types
> (NBRCNT, NBRCTC, and NBRCTS). I'm sure I've gone over this in the
past...
> but simply put:
>
> (1) Grid-Stat reads the raw forecast and observation fields.
>
> (2) It applies the categorical threshold (cat_thresh) to convert
those raw
> fields into binary 0/1 fields.
>
> (3) It applies the neighborhood size to convert the binary field
into a
> fractional coverage field. The fractional coverage value at a grid
point
> is ratio for the number of grid points in that neighborhood that
have a
> value of 1.
>
> (4) Fractions Skill Score (FSS) and Fractions Brier Score (FBS) are
> computed using these fractional coverage fields.
>
> (5) This same method is applied for each combination of categorical
> threshold and neighborhood size requested by the user.
>
> Thanks,
> John
>
> On Fri, Jan 29, 2016 at 4:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
> >
> > thanks John. this doubt arose when I was reading the paper by Dr
Ebert
> > (2008) in Met applications. In the Table 1 mentioned there,
> > Upscaling (Zepeda-Arce et al.,2000; Weygandt et al., 2004;Yates et
al.,
> > 2006) NO-NF Useful forecast resembles the
observations when
> > averaged to coarser scales
> > It uses the <Ix>s and <Iy>s terminology for the binary fields of
OBS and
> > FCST NEIGHBOUR hoods. It says that <Ix>s and <Iy>s are compared in
> > upscaling.
> > I also did not understand the 3 point in the TABLE (Damrath 2004)
which
> > says that the decision model is " Useful Forecast is more correct
than
> > incorrect".
> > Can u pls suggest.
> > Geeta
> > > Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat
> > tool.
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Thu, 28 Jan 2016 16:39:14 -0700
> > >
> > > Hello Geeta,
> > >
> > > The smoothing done by the Grid-Stat tool is done as follows...
> > >
> > > (1) Read the raw forecast and observation fields into memory.
> > >
> > > (2) Apply the smoothing method defined by the user. For
example, for
> > > UW_MEAN with a width of 5, replace the raw data value at each
grid point
> > by
> > > the average of the 25 points in the 5x5 box centered on that
point. In
> > > your settings, you've applied the smoothing to only the forecast
field
> > > (field = FCST;).
> > >
> > > (3) The raw forecast field has now replaced by it's smoothed
version.
> > And
> > > that smoothed version is compared to the raw observation field.
The
> > > categorical thresholds are applied to those fields and
contingency table
> > > counts and statistics are computed.
> > >
> > > To put it simply... the smoothing defined in the "interp"
section is done
> > > on the raw fields, not the thresholded binary fields.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > >
> > > On Thu, Jan 28, 2016 at 8:53 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu Jan 28 08:53:00 2016: Request 74897 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: doubt reg Upscaling in Grid_stat tool.
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897
> > >
> > > >
> > > >
> > > > Hi John. Dear met_help.
> > > > I have some doubt regarding "Upscaling" in the grid_stat tool.
I have
> > > > defined different thresholds (for rainfall) in below sections
of the
> > > > Configuration file. fcst = { wind_thresh = [ NA ]; field
= [
> > {
> > > > name = "APCP_24"; level = [ "(*,*)" ];
> > > > cat_thresh = [ >=0.1,
> > > >
>=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5]; }
> > > > ];};obs = { wind_thresh = [ NA ]; field = [ {
name
> > > > = "APCP_24"; level = [ "(*,*)" ];
cat_thresh = [
> > >=0.1,
> > > >
>=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5]; }
> > > > ];interp = { field = FCST; // if BOTH then both will
be
> > upscaled
> > > > vld_thresh = 1.0; type = [ { method =
UW_MEAN;
> > > > width = 1; }, { method = UW_MEAN;
width
> > =
> > > > 3; }, { method = UW_MEAN; width =
5;
> > };
> > > > ];};I want to know the procedure followed by grid_stat tool
of MET to
> > do
> > > > UPSCALING. What I mea!
> > > > n is that both the OBS and FCST fields are converted to
binary fields
> > > > (depending on the threshold) and then more number of grid
points are
> > > > incorporated and then ETS is calculated??. I mean are the
binary fields
> > > > averaged???.
> > > > Or the actual rainfall sums are computed for different
> > > > neighbourhoods???Kindly clarify. Geeta
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: doubt reg Upscaling in Grid_stat tool.
From: John Halley Gotway
Time: Tue Feb 02 07:50:01 2016
These statistics are described on pages 311 and 312 of the MET Users's
Guide:
http://www.dtcenter.org/met/users/docs/users_guide/MET_Users_Guide_v5.1.pdf
John
On Tue, Feb 2, 2016 at 2:26 AM, Geeta Geeta via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
>
> I also want to know how is the UFSS calculated.
> Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat
> tool.
> Date: Tue, 2 Feb 2016 14:53:03 +0530
>
>
>
>
> thanks John. In the grid_stat_270000L_20110423_030000V_nbrcnt.txt,
FSS and
> AFSS are calculated. what is the difference between the two??.
>
> Geeta
>
> > Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
Grid_stat
> tool.
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Fri, 29 Jan 2016 12:14:22 -0700
> >
> > Geeta,
> >
> > I am not able to provide any clarification on those papers. I
don't know
> > them well enough to comment.
> >
> > In my previous email, I described the logic by which the
interpolation
> > methods defined in the "interp" section of the Grid-Stat
configuration
> file
> > are applied. Those smoothing methods affect the statistical
output found
> > in the traditional statistics lines type (SL1L2, CNT, CTC, and
CTS).
> >
> > The logic for applying the "nbrhd" configuration settings is
different,
> and
> > they control the output found in the neighborhood statistics line
types
> > (NBRCNT, NBRCTC, and NBRCTS). I'm sure I've gone over this in the
> past...
> > but simply put:
> >
> > (1) Grid-Stat reads the raw forecast and observation fields.
> >
> > (2) It applies the categorical threshold (cat_thresh) to convert
those
> raw
> > fields into binary 0/1 fields.
> >
> > (3) It applies the neighborhood size to convert the binary field
into a
> > fractional coverage field. The fractional coverage value at a
grid point
> > is ratio for the number of grid points in that neighborhood that
have a
> > value of 1.
> >
> > (4) Fractions Skill Score (FSS) and Fractions Brier Score (FBS)
are
> > computed using these fractional coverage fields.
> >
> > (5) This same method is applied for each combination of
categorical
> > threshold and neighborhood size requested by the user.
> >
> > Thanks,
> > John
> >
> > On Fri, Jan 29, 2016 at 4:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897 >
> > >
> > > thanks John. this doubt arose when I was reading the paper by Dr
Ebert
> > > (2008) in Met applications. In the Table 1 mentioned there,
> > > Upscaling (Zepeda-Arce et al.,2000; Weygandt et al., 2004;Yates
et al.,
> > > 2006) NO-NF Useful forecast resembles the
observations
> when
> > > averaged to coarser scales
> > > It uses the <Ix>s and <Iy>s terminology for the binary fields of
OBS
> and
> > > FCST NEIGHBOUR hoods. It says that <Ix>s and <Iy>s are compared
in
> > > upscaling.
> > > I also did not understand the 3 point in the TABLE (Damrath
2004) which
> > > says that the decision model is " Useful Forecast is more
correct than
> > > incorrect".
> > > Can u pls suggest.
> > > Geeta
> > > > Subject: Re: [rt.rap.ucar.edu #74897] doubt reg Upscaling in
> Grid_stat
> > > tool.
> > > > From: met_help at ucar.edu
> > > > To: geeta124 at hotmail.com
> > > > Date: Thu, 28 Jan 2016 16:39:14 -0700
> > > >
> > > > Hello Geeta,
> > > >
> > > > The smoothing done by the Grid-Stat tool is done as follows...
> > > >
> > > > (1) Read the raw forecast and observation fields into memory.
> > > >
> > > > (2) Apply the smoothing method defined by the user. For
example, for
> > > > UW_MEAN with a width of 5, replace the raw data value at each
grid
> point
> > > by
> > > > the average of the 25 points in the 5x5 box centered on that
point.
> In
> > > > your settings, you've applied the smoothing to only the
forecast
> field
> > > > (field = FCST;).
> > > >
> > > > (3) The raw forecast field has now replaced by it's smoothed
version.
> > > And
> > > > that smoothed version is compared to the raw observation
field. The
> > > > categorical thresholds are applied to those fields and
contingency
> table
> > > > counts and statistics are computed.
> > > >
> > > > To put it simply... the smoothing defined in the "interp"
section is
> done
> > > > on the raw fields, not the thresholded binary fields.
> > > >
> > > > Thanks,
> > > > John Halley Gotway
> > > >
> > > >
> > > >
> > > > On Thu, Jan 28, 2016 at 8:53 AM, Geeta Geeta via RT <
> met_help at ucar.edu>
> > > > wrote:
> > > >
> > > > >
> > > > > Thu Jan 28 08:53:00 2016: Request 74897 was acted upon.
> > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > Queue: met_help
> > > > > Subject: doubt reg Upscaling in Grid_stat tool.
> > > > > Owner: Nobody
> > > > > Requestors: geeta124 at hotmail.com
> > > > > Status: new
> > > > > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=74897
> > > >
> > > > >
> > > > >
> > > > > Hi John. Dear met_help.
> > > > > I have some doubt regarding "Upscaling" in the grid_stat
tool. I
> have
> > > > > defined different thresholds (for rainfall) in below
sections of
> the
> > > > > Configuration file. fcst = { wind_thresh = [ NA ];
field = [
> > > {
> > > > > name = "APCP_24"; level = [ "(*,*)" ];
> > > > > cat_thresh = [ >=0.1,
> > > > >
>=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
> }
> > > > > ];};obs = { wind_thresh = [ NA ]; field = [ {
> name
> > > > > = "APCP_24"; level = [ "(*,*)" ];
cat_thresh =
> [
> > > >=0.1,
> > > > >
>=2.4,>=7.5,>=15.6,>=25.6,>=35.6,>=45.6,>=55.6,>=64.5,>=124.5];
> }
> > > > > ];interp = { field = FCST; // if BOTH then both
will be
> > > upscaled
> > > > > vld_thresh = 1.0; type = [ { method =
UW_MEAN;
> > > > > width = 1; }, { method = UW_MEAN;
> width
> > > =
> > > > > 3; }, { method = UW_MEAN; width
= 5;
> > > };
> > > > > ];};I want to know the procedure followed by grid_stat tool
of
> MET to
> > > do
> > > > > UPSCALING. What I mea!
> > > > > n is that both the OBS and FCST fields are converted to
binary
> fields
> > > > > (depending on the threshold) and then more number of grid
points
> are
> > > > > incorporated and then ETS is calculated??. I mean are the
binary
> fields
> > > > > averaged???.
> > > > > Or the actual rainfall sums are computed for different
> > > > > neighbourhoods???Kindly clarify. Geeta
> > > > >
> > > >
> > >
> > >
> >
>
>
------------------------------------------------
More information about the Met_help
mailing list