[Met_help] [rt.rap.ucar.edu #72102] History for FSS, UFSS, AFSS
John Halley Gotway via RT
met_help at ucar.edu
Tue Jan 26 12:17:46 MST 2016
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi John. I have plotted the FSS for various neighbourhood widths starting from 1, 3,5,7,9,11,13 and 15 and for various thresholds (0.1,2.5,7.5, and so on.). This is for One case of thunderstorm event. I have taken FSS column from my output. But there are three, FSS, UFSS and AFSS. I presume that UFSS is the UNIFORM FSS which is the base line for a good forecast. I am sending My plot. (attached). for lower threshold, and lowest neighbourhood, I get lowest FSS which increases very gradually. But it is less as compared to the UFSS (I plotted UFSS for threshold of 0.1). But when I look at my model forecast and observed data (trmm), Plots are matching so nicely. I m sending my grid_stat output file. can u help me.
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: FSS, UFSS, AFSS
From: John Halley Gotway
Time: Fri May 29 16:43:33 2015
Geeta,
Good questions. Unfortunately, I don't have a good answer for you
yet.
I'm headed out of town all next week, but I referred your question to
a
scientist and statistician here.
Hopefully you'll hear back from them about this next week.
Thanks,
John
On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> Transaction: Ticket created by geeta124 at hotmail.com
> Queue: met_help
> Subject: FSS, UFSS, AFSS
> Owner: Nobody
> Requestors: geeta124 at hotmail.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
>
>
> Hi John. I have plotted the FSS for various neighbourhood widths
starting
> from 1, 3,5,7,9,11,13 and 15 and for various thresholds
(0.1,2.5,7.5, and
> so on.). This is for One case of thunderstorm event. I have taken
FSS
> column from my output. But there are three, FSS, UFSS and AFSS. I
presume
> that UFSS is the UNIFORM FSS which is the base line for a good
forecast. I
> am sending My plot. (attached). for lower threshold, and lowest
> neighbourhood, I get lowest FSS which increases very gradually. But
it is
> less as compared to the UFSS (I plotted UFSS for threshold of 0.1).
But
> when I look at my model forecast and observed data (trmm), Plots are
> matching so nicely. I m sending my grid_stat output file. can u
help me.
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Sat May 30 04:04:37 2015
Thanks John.
Also I was wondering why I should get bad FSS /low FSS for Lowest
Neighbourhood and lower threshold????. This looks a bit confusing.
I should get low FSS for Lower neighbourhood and high threshold!!!!.
OK????
I shall be waiting for your reply eagerly.
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 29 May 2015 16:43:34 -0600
>
> Geeta,
>
> Good questions. Unfortunately, I don't have a good answer for you
yet.
> I'm headed out of town all next week, but I referred your question
to a
> scientist and statistician here.
>
> Hopefully you'll hear back from them about this next week.
>
> Thanks,
> John
>
> On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> > Queue: met_help
> > Subject: FSS, UFSS, AFSS
> > Owner: Nobody
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> >
> > Hi John. I have plotted the FSS for various neighbourhood widths
starting
> > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
(0.1,2.5,7.5, and
> > so on.). This is for One case of thunderstorm event. I have taken
FSS
> > column from my output. But there are three, FSS, UFSS and AFSS. I
presume
> > that UFSS is the UNIFORM FSS which is the base line for a good
forecast. I
> > am sending My plot. (attached). for lower threshold, and lowest
> > neighbourhood, I get lowest FSS which increases very gradually.
But it is
> > less as compared to the UFSS (I plotted UFSS for threshold of
0.1). But
> > when I look at my model forecast and observed data (trmm), Plots
are
> > matching so nicely. I m sending my grid_stat output file. can u
help me.
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Tue Jun 02 00:37:41 2015
Hi John,. I have another question. Does any tool in MET 5.0 can plot
the Binary fields (OBSERVATION and FCST) for FSS depending on the
Threshold. I have a image to see of both the fields.
I shall be waiting for your reply. Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Sat, 30 May 2015 15:34:33 +0530
Thanks John.
Also I was wondering why I should get bad FSS /low FSS for Lowest
Neighbourhood and lower threshold????. This looks a bit confusing.
I should get low FSS for Lower neighbourhood and high threshold!!!!.
OK????
I shall be waiting for your reply eagerly.
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 29 May 2015 16:43:34 -0600
>
> Geeta,
>
> Good questions. Unfortunately, I don't have a good answer for you
yet.
> I'm headed out of town all next week, but I referred your question
to a
> scientist and statistician here.
>
> Hopefully you'll hear back from them about this next week.
>
> Thanks,
> John
>
> On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > Transaction: Ticket created by geeta124 at hotmail.com
> > Queue: met_help
> > Subject: FSS, UFSS, AFSS
> > Owner: Nobody
> > Requestors: geeta124 at hotmail.com
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> >
> > Hi John. I have plotted the FSS for various neighbourhood widths
starting
> > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
(0.1,2.5,7.5, and
> > so on.). This is for One case of thunderstorm event. I have taken
FSS
> > column from my output. But there are three, FSS, UFSS and AFSS. I
presume
> > that UFSS is the UNIFORM FSS which is the base line for a good
forecast. I
> > am sending My plot. (attached). for lower threshold, and lowest
> > neighbourhood, I get lowest FSS which increases very gradually.
But it is
> > less as compared to the UFSS (I plotted UFSS for threshold of
0.1). But
> > when I look at my model forecast and observed data (trmm), Plots
are
> > matching so nicely. I m sending my grid_stat output file. can u
help me.
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Julie Prestopnik
Time: Tue Jun 02 09:28:14 2015
Hi Geeta. As John indicated in his email from May 29th, he is out of
town
this week.
METv5.0 has three plotting tools: plot_point_obs, plot_data_plane, and
plot_mode field
I'll provide links to their sections in the online tutorial below:
http://www.dtcenter.org/met/users/support/online_tutorial/METv5.0/plot_point_obs/index.php
http://www.dtcenter.org/met/users/support/online_tutorial/METv5.0/plot_data_plane/index.php
http://www.dtcenter.org/met/users/support/online_tutorial/METv5.0/plot_mode_field/index.php
There are also some sample R scripts for plotting referred to on this
page:
http://www.dtcenter.org/met/users/downloads/analysis_scripts.php
John is much more familiar with MET. If you don't find what you're
looking
for on the above pages, John may have more resources for you when he
returns.
Thanks,
Julie
On Tue, Jun 2, 2015 at 12:37 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
>
> Hi John,. I have another question. Does any tool in MET 5.0 can plot
the
> Binary fields (OBSERVATION and FCST) for FSS depending on the
Threshold. I
> have a image to see of both the fields.
> I shall be waiting for your reply. Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> Date: Sat, 30 May 2015 15:34:33 +0530
>
>
>
>
> Thanks John.
> Also I was wondering why I should get bad FSS /low FSS for Lowest
> Neighbourhood and lower threshold????. This looks a bit confusing.
> I should get low FSS for Lower neighbourhood and high threshold!!!!.
OK????
> I shall be waiting for your reply eagerly.
>
> Geeta
>
> > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Fri, 29 May 2015 16:43:34 -0600
> >
> > Geeta,
> >
> > Good questions. Unfortunately, I don't have a good answer for you
yet.
> > I'm headed out of town all next week, but I referred your question
to a
> > scientist and statistician here.
> >
> > Hopefully you'll hear back from them about this next week.
> >
> > Thanks,
> > John
> >
> > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > Transaction: Ticket created by geeta124 at hotmail.com
> > > Queue: met_help
> > > Subject: FSS, UFSS, AFSS
> > > Owner: Nobody
> > > Requestors: geeta124 at hotmail.com
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
> >
> > >
> > >
> > > Hi John. I have plotted the FSS for various neighbourhood widths
> starting
> > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
(0.1,2.5,7.5,
> and
> > > so on.). This is for One case of thunderstorm event. I have
taken FSS
> > > column from my output. But there are three, FSS, UFSS and AFSS.
I
> presume
> > > that UFSS is the UNIFORM FSS which is the base line for a good
> forecast. I
> > > am sending My plot. (attached). for lower threshold, and lowest
> > > neighbourhood, I get lowest FSS which increases very gradually.
But it
> is
> > > less as compared to the UFSS (I plotted UFSS for threshold of
0.1). But
> > > when I look at my model forecast and observed data (trmm), Plots
are
> > > matching so nicely. I m sending my grid_stat output file. can u
help
> me.
> > >
> >
>
>
--
Julie Prestopnik
National Center for Atmospheric Research
Research Applications Laboratory
Phone: 303.497.8399
Email: jpresto at ucar.edu
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Tue Jun 02 09:36:29 2015
thanks Julie.
I will wait for John to come. I m interested to know if I can plot the
BINARY field Created after using the THRESHOLD criterion.
But I also wish to know about the FSS/UFSS and AFSS.
I had sent my output of case1 to Ur Office and John said he will refer
my question to a statistician and a Scientist.
This is only for your information. Thank U so much for providing so
much help..........
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 2 Jun 2015 09:28:14 -0600
>
> Hi Geeta. As John indicated in his email from May 29th, he is out
of town
> this week.
>
> METv5.0 has three plotting tools: plot_point_obs, plot_data_plane,
and
> plot_mode field
>
> I'll provide links to their sections in the online tutorial below:
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv5.0/plot_point_obs/index.php
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv5.0/plot_data_plane/index.php
>
http://www.dtcenter.org/met/users/support/online_tutorial/METv5.0/plot_mode_field/index.php
>
> There are also some sample R scripts for plotting referred to on
this page:
> http://www.dtcenter.org/met/users/downloads/analysis_scripts.php
>
> John is much more familiar with MET. If you don't find what you're
looking
> for on the above pages, John may have more resources for you when he
> returns.
>
> Thanks,
> Julie
>
> On Tue, Jun 2, 2015 at 12:37 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > Hi John,. I have another question. Does any tool in MET 5.0 can
plot the
> > Binary fields (OBSERVATION and FCST) for FSS depending on the
Threshold. I
> > have a image to see of both the fields.
> > I shall be waiting for your reply. Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Sat, 30 May 2015 15:34:33 +0530
> >
> >
> >
> >
> > Thanks John.
> > Also I was wondering why I should get bad FSS /low FSS for Lowest
> > Neighbourhood and lower threshold????. This looks a bit confusing.
> > I should get low FSS for Lower neighbourhood and high
threshold!!!!. OK????
> > I shall be waiting for your reply eagerly.
> >
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Fri, 29 May 2015 16:43:34 -0600
> > >
> > > Geeta,
> > >
> > > Good questions. Unfortunately, I don't have a good answer for
you yet.
> > > I'm headed out of town all next week, but I referred your
question to a
> > > scientist and statistician here.
> > >
> > > Hopefully you'll hear back from them about this next week.
> > >
> > > Thanks,
> > > John
> > >
> > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: FSS, UFSS, AFSS
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
> > >
> > > >
> > > >
> > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > starting
> > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
(0.1,2.5,7.5,
> > and
> > > > so on.). This is for One case of thunderstorm event. I have
taken FSS
> > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > presume
> > > > that UFSS is the UNIFORM FSS which is the base line for a good
> > forecast. I
> > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > neighbourhood, I get lowest FSS which increases very
gradually. But it
> > is
> > > > less as compared to the UFSS (I plotted UFSS for threshold of
0.1). But
> > > > when I look at my model forecast and observed data (trmm),
Plots are
> > > > matching so nicely. I m sending my grid_stat output file. can
u help
> > me.
> > > >
> > >
> >
> >
>
>
> --
> Julie Prestopnik
> National Center for Atmospheric Research
> Research Applications Laboratory
> Phone: 303.497.8399
> Email: jpresto at ucar.edu
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Tara Jensen
Time: Tue Jun 02 09:52:37 2015
Geeta,
Sorry for the delay in answering your question about FSS etc.
Yesterday was a very busy day here. One of our statisticians put
together the following answer:
According to the Roberts and Lean (2008) paper, the AFSS (asymptotic
FSS) is a useful descriptor of the bias because it relates the bias to
the spatial accuracy of a forecast and is linked to the conventional
frequency bias, but is less sensitive to small frequencies. Compared
to the frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4, and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS = 1, indicating perfect skill over the domain.
I will send more information as I or others pull it together.
As for plotting the NetCDF files that come from MET. We typically use
ncview or NCL for plotting the files, however the plotting tools Julie
mentioned should also be able to read the files.
Thanks for your patience as I prepare more information about FSS, AFSS
etc.
Tara
On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> Thanks John.
> Also I was wondering why I should get bad FSS /low FSS for Lowest
> Neighbourhood and lower threshold????. This looks a bit confusing.
> I should get low FSS for Lower neighbourhood and high threshold!!!!.
> OK????
> I shall be waiting for your reply eagerly.
>
> Geeta
>
> > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Fri, 29 May 2015 16:43:34 -0600
> >
> > Geeta,
> >
> > Good questions. Unfortunately, I don't have a good answer for you
> yet.
> > I'm headed out of town all next week, but I referred your question
> to a
> > scientist and statistician here.
> >
> > Hopefully you'll hear back from them about this next week.
> >
> > Thanks,
> > John
> >
> > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> <met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > Transaction: Ticket created by geeta124 at hotmail.com
> > > Queue: met_help
> > > Subject: FSS, UFSS, AFSS
> > > Owner: Nobody
> > > Requestors: geeta124 at hotmail.com
> > > Status: new
> > > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > >
> > >
> > > Hi John. I have plotted the FSS for various neighbourhood widths
> starting
> > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> (0.1,2.5,7.5, and
> > > so on.). This is for One case of thunderstorm event. I have
taken
> FSS
> > > column from my output. But there are three, FSS, UFSS and AFSS.
I
> presume
> > > that UFSS is the UNIFORM FSS which is the base line for a good
> forecast. I
> > > am sending My plot. (attached). for lower threshold, and lowest
> > > neighbourhood, I get lowest FSS which increases very gradually.
> But it is
> > > less as compared to the UFSS (I plotted UFSS for threshold of
> 0.1). But
> > > when I look at my model forecast and observed data (trmm), Plots
> are
> > > matching so nicely. I m sending my grid_stat output file. can u
> help me.
> > >
> >
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Tue Jun 02 23:17:34 2015
thanks Tara.
Nice to hear from U. I had met you at Verification workshop last year
in New Delhi.
I didnt understand Ur explanation fully.
It would be great if U can elaborate upon it in a little more detail.
See I got the output from Grid stat and now I need to understand what
is my FSS for this case??.
U must see my MODEL FCST and Observations. They are so close but I am
unable to understand why FSS is so low????????.
If U want, I can send my plots (FSS and RF fcst and observation) and
grid_stat output to you.
Geeta
> Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 2 Jun 2015 09:52:38 -0600
>
> Geeta,
>
> Sorry for the delay in answering your question about FSS etc.
Yesterday was a very busy day here. One of our statisticians put
together the following answer:
>
> According to the Roberts and Lean (2008) paper, the AFSS (asymptotic
FSS) is a useful descriptor of the bias because it relates the bias to
the spatial accuracy of a forecast and is linked to the conventional
frequency bias, but is less sensitive to small frequencies. Compared
to the frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4, and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS = 1, indicating perfect skill over the domain.
>
> I will send more information as I or others pull it together.
>
> As for plotting the NetCDF files that come from MET. We typically
use ncview or NCL for plotting the files, however the plotting tools
Julie mentioned should also be able to read the files.
>
> Thanks for your patience as I prepare more information about FSS,
AFSS etc.
> Tara
>
>
> On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > Thanks John.
> > Also I was wondering why I should get bad FSS /low FSS for Lowest
> > Neighbourhood and lower threshold????. This looks a bit confusing.
> > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > OK????
> > I shall be waiting for your reply eagerly.
> >
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Fri, 29 May 2015 16:43:34 -0600
> > >
> > > Geeta,
> > >
> > > Good questions. Unfortunately, I don't have a good answer for
you
> > yet.
> > > I'm headed out of town all next week, but I referred your
question
> > to a
> > > scientist and statistician here.
> > >
> > > Hopefully you'll hear back from them about this next week.
> > >
> > > Thanks,
> > > John
> > >
> > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > <met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: FSS, UFSS, AFSS
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > >
> > > >
> > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > starting
> > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > (0.1,2.5,7.5, and
> > > > so on.). This is for One case of thunderstorm event. I have
taken
> > FSS
> > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > presume
> > > > that UFSS is the UNIFORM FSS which is the base line for a good
> > forecast. I
> > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > But it is
> > > > less as compared to the UFSS (I plotted UFSS for threshold of
> > 0.1). But
> > > > when I look at my model forecast and observed data (trmm),
Plots
> > are
> > > > matching so nicely. I m sending my grid_stat output file. can
u
> > help me.
> > > >
> > >
>
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Wed Jun 03 00:30:44 2015
Hi Tara,
I was NOT asking about the plotting of the NETCDF data files
(observations n FCST). Those can be plotted with plot_data_plane and
ncview etc.
But a step further.
I want to know if I can plot the binary data after thresholding.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 3 Jun 2015 10:47:30 +0530
thanks Tara.
Nice to hear from U. I had met you at Verification workshop last year
in New Delhi.
I didnt understand Ur explanation fully.
It would be great if U can elaborate upon it in a little more detail.
See I got the output from Grid stat and now I need to understand what
is my FSS for this case??.
U must see my MODEL FCST and Observations. They are so close but I am
unable to understand why FSS is so low????????.
If U want, I can send my plots (FSS and RF fcst and observation) and
grid_stat output to you.
Geeta
> Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 2 Jun 2015 09:52:38 -0600
>
> Geeta,
>
> Sorry for the delay in answering your question about FSS etc.
Yesterday was a very busy day here. One of our statisticians put
together the following answer:
>
> According to the Roberts and Lean (2008) paper, the AFSS (asymptotic
FSS) is a useful descriptor of the bias because it relates the bias to
the spatial accuracy of a forecast and is linked to the conventional
frequency bias, but is less sensitive to small frequencies. Compared
to the frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4, and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS = 1, indicating perfect skill over the domain.
>
> I will send more information as I or others pull it together.
>
> As for plotting the NetCDF files that come from MET. We typically
use ncview or NCL for plotting the files, however the plotting tools
Julie mentioned should also be able to read the files.
>
> Thanks for your patience as I prepare more information about FSS,
AFSS etc.
> Tara
>
>
> On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > Thanks John.
> > Also I was wondering why I should get bad FSS /low FSS for Lowest
> > Neighbourhood and lower threshold????. This looks a bit confusing.
> > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > OK????
> > I shall be waiting for your reply eagerly.
> >
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Fri, 29 May 2015 16:43:34 -0600
> > >
> > > Geeta,
> > >
> > > Good questions. Unfortunately, I don't have a good answer for
you
> > yet.
> > > I'm headed out of town all next week, but I referred your
question
> > to a
> > > scientist and statistician here.
> > >
> > > Hopefully you'll hear back from them about this next week.
> > >
> > > Thanks,
> > > John
> > >
> > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > <met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: FSS, UFSS, AFSS
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > >
> > > >
> > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > starting
> > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > (0.1,2.5,7.5, and
> > > > so on.). This is for One case of thunderstorm event. I have
taken
> > FSS
> > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > presume
> > > > that UFSS is the UNIFORM FSS which is the base line for a good
> > forecast. I
> > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > But it is
> > > > less as compared to the UFSS (I plotted UFSS for threshold of
> > 0.1). But
> > > > when I look at my model forecast and observed data (trmm),
Plots
> > are
> > > > matching so nicely. I m sending my grid_stat output file. can
u
> > help me.
> > > >
> > >
>
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Sun Jun 07 06:24:30 2015
Hi John and Tara. I was waiting for your reply. Kindly help me
further. ThanksGeeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 3 Jun 2015 11:59:35 +0530
Hi Tara,
I was NOT asking about the plotting of the NETCDF data files
(observations n FCST). Those can be plotted with plot_data_plane and
ncview etc.
But a step further.
I want to know if I can plot the binary data after thresholding.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 3 Jun 2015 10:47:30 +0530
thanks Tara.
Nice to hear from U. I had met you at Verification workshop last year
in New Delhi.
I didnt understand Ur explanation fully.
It would be great if U can elaborate upon it in a little more detail.
See I got the output from Grid stat and now I need to understand what
is my FSS for this case??.
U must see my MODEL FCST and Observations. They are so close but I am
unable to understand why FSS is so low????????.
If U want, I can send my plots (FSS and RF fcst and observation) and
grid_stat output to you.
Geeta
> Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 2 Jun 2015 09:52:38 -0600
>
> Geeta,
>
> Sorry for the delay in answering your question about FSS etc.
Yesterday was a very busy day here. One of our statisticians put
together the following answer:
>
> According to the Roberts and Lean (2008) paper, the AFSS (asymptotic
FSS) is a useful descriptor of the bias because it relates the bias to
the spatial accuracy of a forecast and is linked to the conventional
frequency bias, but is less sensitive to small frequencies. Compared
to the frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4, and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS = 1, indicating perfect skill over the domain.
>
> I will send more information as I or others pull it together.
>
> As for plotting the NetCDF files that come from MET. We typically
use ncview or NCL for plotting the files, however the plotting tools
Julie mentioned should also be able to read the files.
>
> Thanks for your patience as I prepare more information about FSS,
AFSS etc.
> Tara
>
>
> On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > Thanks John.
> > Also I was wondering why I should get bad FSS /low FSS for Lowest
> > Neighbourhood and lower threshold????. This looks a bit confusing.
> > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > OK????
> > I shall be waiting for your reply eagerly.
> >
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Fri, 29 May 2015 16:43:34 -0600
> > >
> > > Geeta,
> > >
> > > Good questions. Unfortunately, I don't have a good answer for
you
> > yet.
> > > I'm headed out of town all next week, but I referred your
question
> > to a
> > > scientist and statistician here.
> > >
> > > Hopefully you'll hear back from them about this next week.
> > >
> > > Thanks,
> > > John
> > >
> > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > <met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: FSS, UFSS, AFSS
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > >
> > > >
> > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > starting
> > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > (0.1,2.5,7.5, and
> > > > so on.). This is for One case of thunderstorm event. I have
taken
> > FSS
> > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > presume
> > > > that UFSS is the UNIFORM FSS which is the base line for a good
> > forecast. I
> > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > But it is
> > > > less as compared to the UFSS (I plotted UFSS for threshold of
> > 0.1). But
> > > > when I look at my model forecast and observed data (trmm),
Plots
> > are
> > > > matching so nicely. I m sending my grid_stat output file. can
u
> > help me.
> > > >
> > >
>
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Tue Jun 09 05:41:55 2015
Hi John and tara. I have tried to run grid_stat with TRMM daily files.
hence I extracted the model data for 24 hours starting 00 UTC of 22
April 2011.
I could not make the Quilt plot, but just for understanding I plotted
data in EXCEL. (results attatched). I want to know, the details about
UFSS. It is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2) where f0 is the observation rate.
Kindly clarify these points. I m waiting.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Sun, 7 Jun 2015 17:54:25 +0530
Hi John and Tara. I was waiting for your reply. Kindly help me
further. ThanksGeeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 3 Jun 2015 11:59:35 +0530
Hi Tara,
I was NOT asking about the plotting of the NETCDF data files
(observations n FCST). Those can be plotted with plot_data_plane and
ncview etc.
But a step further.
I want to know if I can plot the binary data after thresholding.
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 3 Jun 2015 10:47:30 +0530
thanks Tara.
Nice to hear from U. I had met you at Verification workshop last year
in New Delhi.
I didnt understand Ur explanation fully.
It would be great if U can elaborate upon it in a little more detail.
See I got the output from Grid stat and now I need to understand what
is my FSS for this case??.
U must see my MODEL FCST and Observations. They are so close but I am
unable to understand why FSS is so low????????.
If U want, I can send my plots (FSS and RF fcst and observation) and
grid_stat output to you.
Geeta
> Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 2 Jun 2015 09:52:38 -0600
>
> Geeta,
>
> Sorry for the delay in answering your question about FSS etc.
Yesterday was a very busy day here. One of our statisticians put
together the following answer:
>
> According to the Roberts and Lean (2008) paper, the AFSS (asymptotic
FSS) is a useful descriptor of the bias because it relates the bias to
the spatial accuracy of a forecast and is linked to the conventional
frequency bias, but is less sensitive to small frequencies. Compared
to the frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4, and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS = 1, indicating perfect skill over the domain.
>
> I will send more information as I or others pull it together.
>
> As for plotting the NetCDF files that come from MET. We typically
use ncview or NCL for plotting the files, however the plotting tools
Julie mentioned should also be able to read the files.
>
> Thanks for your patience as I prepare more information about FSS,
AFSS etc.
> Tara
>
>
> On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > Thanks John.
> > Also I was wondering why I should get bad FSS /low FSS for Lowest
> > Neighbourhood and lower threshold????. This looks a bit confusing.
> > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > OK????
> > I shall be waiting for your reply eagerly.
> >
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Fri, 29 May 2015 16:43:34 -0600
> > >
> > > Geeta,
> > >
> > > Good questions. Unfortunately, I don't have a good answer for
you
> > yet.
> > > I'm headed out of town all next week, but I referred your
question
> > to a
> > > scientist and statistician here.
> > >
> > > Hopefully you'll hear back from them about this next week.
> > >
> > > Thanks,
> > > John
> > >
> > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > <met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > Queue: met_help
> > > > Subject: FSS, UFSS, AFSS
> > > > Owner: Nobody
> > > > Requestors: geeta124 at hotmail.com
> > > > Status: new
> > > > Ticket <URL:
> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > >
> > > >
> > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > starting
> > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > (0.1,2.5,7.5, and
> > > > so on.). This is for One case of thunderstorm event. I have
taken
> > FSS
> > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > presume
> > > > that UFSS is the UNIFORM FSS which is the base line for a good
> > forecast. I
> > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > But it is
> > > > less as compared to the UFSS (I plotted UFSS for threshold of
> > 0.1). But
> > > > when I look at my model forecast and observed data (trmm),
Plots
> > are
> > > > matching so nicely. I m sending my grid_stat output file. can
u
> > help me.
> > > >
> > >
>
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Tue Jun 09 05:41:55 2015
VERSION MODEL FCST_LEAD FCST_VALID_BEG FCST_VALID_END OBS_LEAD
OBS_VALID_BEG OBS_VALID_END FCST_VAR FCST_LEV OBS_VAR OBS_LEV
OBTYPE VX_MASK INTERP_MTHD INTERP_PNTS FCST_THRESH OBS_THRESH
COV_THRESH ALPHA LINE_TYPE TOTAL FBS FBS_BCL FBS_BCU FSS
FSS_BCL FSS_BCU AFSS AFSS_BCL AFSS_BCU UFSS UFSS_BCL UFSS_BCU
F_RATE F_RATE_BCL F_RATE_BCU O_RATE O_RATE_BCL O_RATE_BCU
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=0.100 >=0.100
NA 0.05000 NBRCNT 644 0.25621 NA NA 0.83808 NA
NA 0.99575 NA NA 0.91382 NA NA 0.75466
NA NA 0.82764 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=2.400 >=2.400
NA 0.05000 NBRCNT 644 0.33230 NA NA 0.70765 NA
NA 0.99748 NA NA 0.79425 NA NA 0.54814
NA NA 0.58851 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=7.500 >=7.500
NA 0.05000 NBRCNT 644 0.36180 NA NA 0.48337 NA
NA 0.90449 NA NA 0.63587 NA NA 0.42857
NA NA 0.27174 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=15.600 >=15.600
NA 0.05000 NBRCNT 644 0.29814 NA NA 0.17241 NA
NA 0.31201 NA NA 0.52484 NA NA 0.31056
NA NA 0.04969 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=25.600 >=25.600
NA 0.05000 NBRCNT 644 0.22360 NA NA 0.00000 NA
NA 0.02816 NA NA 0.50155 NA NA 0.22050
NA NA 0.00311 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=35.600 >=35.600
NA 0.05000 NBRCNT 644 0.16460 NA NA 0.00000 NA
NA 0.03845 NA NA 0.50155 NA NA 0.16149
NA NA 0.00311 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=45.600 >=45.600
NA 0.05000 NBRCNT 644 0.12112 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12112
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=55.600 >=55.600
NA 0.05000 NBRCNT 644 0.07764 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.07764
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=64.500 >=64.500
NA 0.05000 NBRCNT 644 0.05745 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.05745
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 1 >=124.500 >=124.500
NA 0.05000 NBRCNT 644 0.00311 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00311
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=0.100 >=0.100
NA 0.05000 NBRCNT 640 0.13004 NA NA 0.91136 NA
NA 0.99608 NA NA 0.91406 NA NA 0.75781
NA NA 0.82812 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=2.400 >=2.400
NA 0.05000 NBRCNT 640 0.14444 NA NA 0.84736 NA
NA 0.99728 NA NA 0.79609 NA NA 0.55000
NA NA 0.59219 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=7.500 >=7.500
NA 0.05000 NBRCNT 640 0.16501 NA NA 0.66565 NA
NA 0.90449 NA NA 0.63672 NA NA 0.43125
NA NA 0.27344 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=15.600 >=15.600
NA 0.05000 NBRCNT 640 0.14343 NA NA 0.29149 NA
NA 0.31201 NA NA 0.52500 NA NA 0.31250
NA NA 0.05000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=25.600 >=25.600
NA 0.05000 NBRCNT 640 0.11214 NA NA 0.00513 NA
NA 0.02816 NA NA 0.50156 NA NA 0.22187
NA NA 0.00313 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=35.600 >=35.600
NA 0.05000 NBRCNT 640 0.07626 NA NA 0.00753 NA
NA 0.03845 NA NA 0.50156 NA NA 0.16250
NA NA 0.00313 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=45.600 >=45.600
NA 0.05000 NBRCNT 640 0.04789 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12187
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=55.600 >=55.600
NA 0.05000 NBRCNT 640 0.02538 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.07812
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=64.500 >=64.500
NA 0.05000 NBRCNT 640 0.01619 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.05781
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 9 >=124.500 >=124.500
NA 0.05000 NBRCNT 640 0.00042 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00313
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=0.100 >=0.100
NA 0.05000 NBRCNT 632 0.09160 NA NA 0.93581 NA
NA 0.99617 NA NA 0.91456 NA NA 0.75949
NA NA 0.82911 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=2.400 >=2.400
NA 0.05000 NBRCNT 632 0.10592 NA NA 0.88236 NA
NA 0.99745 NA NA 0.79826 NA NA 0.55538
NA NA 0.59652 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=7.500 >=7.500
NA 0.05000 NBRCNT 632 0.12199 NA NA 0.72730 NA
NA 0.90588 NA NA 0.63845 NA NA 0.43513
NA NA 0.27690 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=15.600 >=15.600
NA 0.05000 NBRCNT 632 0.11270 NA NA 0.33641 NA
NA 0.31201 NA NA 0.52532 NA NA 0.31646
NA NA 0.05063 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=25.600 >=25.600
NA 0.05000 NBRCNT 632 0.08992 NA NA 0.01081 NA
NA 0.02816 NA NA 0.50158 NA NA 0.22468
NA NA 0.00316 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=35.600 >=35.600
NA 0.05000 NBRCNT 632 0.05968 NA NA 0.01579 NA
NA 0.03845 NA NA 0.50158 NA NA 0.16456
NA NA 0.00316 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=45.600 >=45.600
NA 0.05000 NBRCNT 632 0.03595 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12342
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=55.600 >=55.600
NA 0.05000 NBRCNT 632 0.01744 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.07911
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=64.500 >=64.500
NA 0.05000 NBRCNT 632 0.01007 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.05854
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 25 >=124.500 >=124.500
NA 0.05000 NBRCNT 632 0.00014 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00316
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=0.100 >=0.100
NA 0.05000 NBRCNT 624 0.07060 NA NA 0.94967 NA
NA 0.99607 NA NA 0.91506 NA NA 0.75962
NA NA 0.83013 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=2.400 >=2.400
NA 0.05000 NBRCNT 624 0.08243 NA NA 0.90483 NA
NA 0.99797 NA NA 0.79808 NA NA 0.55929
NA NA 0.59615 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=7.500 >=7.500
NA 0.05000 NBRCNT 624 0.09919 NA NA 0.76607 NA
NA 0.90867 NA NA 0.64022 NA NA 0.43750
NA NA 0.28045 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=15.600 >=15.600
NA 0.05000 NBRCNT 624 0.09993 NA NA 0.35534 NA
NA 0.31350 NA NA 0.52564 NA NA 0.31891
NA NA 0.05128 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=25.600 >=25.600
NA 0.05000 NBRCNT 624 0.08169 NA NA 0.01657 NA
NA 0.02816 NA NA 0.50160 NA NA 0.22756
NA NA 0.00321 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=35.600 >=35.600
NA 0.05000 NBRCNT 624 0.05341 NA NA 0.02386 NA
NA 0.03845 NA NA 0.50160 NA NA 0.16667
NA NA 0.00321 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=45.600 >=45.600
NA 0.05000 NBRCNT 624 0.03229 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12500
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=55.600 >=55.600
NA 0.05000 NBRCNT 624 0.01480 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.08013
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=64.500 >=64.500
NA 0.05000 NBRCNT 624 0.00835 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.05929
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 49 >=124.500 >=124.500
NA 0.05000 NBRCNT 624 0.00007 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00321
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=0.100 >=0.100
NA 0.05000 NBRCNT 612 0.05620 NA NA 0.95953 NA
NA 0.99648 NA NA 0.91585 NA NA 0.76471
NA NA 0.83170 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=2.400 >=2.400
NA 0.05000 NBRCNT 612 0.06512 NA NA 0.92291 NA
NA 0.99856 NA NA 0.79739 NA NA 0.56373
NA NA 0.59477 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=7.500 >=7.500
NA 0.05000 NBRCNT 612 0.08777 NA NA 0.78718 NA
NA 0.91207 NA NA 0.64216 NA NA 0.43954
NA NA 0.28431 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=15.600 >=15.600
NA 0.05000 NBRCNT 612 0.09611 NA NA 0.36130 NA
NA 0.31805 NA NA 0.52614 NA NA 0.32026
NA NA 0.05229 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=25.600 >=25.600
NA 0.05000 NBRCNT 612 0.07905 NA NA 0.01940 NA
NA 0.02877 NA NA 0.50163 NA NA 0.22712
NA NA 0.00327 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=35.600 >=35.600
NA 0.05000 NBRCNT 612 0.05055 NA NA 0.02808 NA
NA 0.03882 NA NA 0.50163 NA NA 0.16830
NA NA 0.00327 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=45.600 >=45.600
NA 0.05000 NBRCNT 612 0.03064 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12582
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=55.600 >=55.600
NA 0.05000 NBRCNT 612 0.01347 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.08007
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=64.500 >=64.500
NA 0.05000 NBRCNT 612 0.00759 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.06046
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 81 >=124.500 >=124.500
NA 0.05000 NBRCNT 612 0.00005 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00327
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=0.100 >=0.100
NA 0.05000 NBRCNT 596 0.04501 NA NA 0.96734 NA
NA 0.99685 NA NA 0.91695 NA NA 0.77013
NA NA 0.83389 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=2.400 >=2.400
NA 0.05000 NBRCNT 596 0.05328 NA NA 0.93569 NA
NA 0.99906 NA NA 0.79698 NA NA 0.56879
NA NA 0.59396 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=7.500 >=7.500
NA 0.05000 NBRCNT 596 0.08194 NA NA 0.79703 NA
NA 0.91688 NA NA 0.64513 NA NA 0.44295
NA NA 0.29027 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=15.600 >=15.600
NA 0.05000 NBRCNT 596 0.09535 NA NA 0.35928 NA
NA 0.32432 NA NA 0.52685 NA NA 0.32215
NA NA 0.05369 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=25.600 >=25.600
NA 0.05000 NBRCNT 596 0.07742 NA NA 0.02001 NA
NA 0.02962 NA NA 0.50168 NA NA 0.22651
NA NA 0.00336 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=35.600 >=35.600
NA 0.05000 NBRCNT 596 0.04882 NA NA 0.02840 NA
NA 0.03959 NA NA 0.50168 NA NA 0.16946
NA NA 0.00336 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=45.600 >=45.600
NA 0.05000 NBRCNT 596 0.02908 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12752
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=55.600 >=55.600
NA 0.05000 NBRCNT 596 0.01238 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.08221
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=64.500 >=64.500
NA 0.05000 NBRCNT 596 0.00697 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.06208
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 121 >=124.500 >=124.500
NA 0.05000 NBRCNT 596 0.00004 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00336
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=0.100 >=0.100
NA 0.05000 NBRCNT 576 0.03676 NA NA 0.97314 NA
NA 0.99749 NA NA 0.91840 NA NA 0.77951
NA NA 0.83681 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=2.400 >=2.400
NA 0.05000 NBRCNT 576 0.04543 NA NA 0.94418 NA
NA 0.99956 NA NA 0.79774 NA NA 0.57812
NA NA 0.59549 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=7.500 >=7.500
NA 0.05000 NBRCNT 576 0.07767 NA NA 0.80311 NA
NA 0.91888 NA NA 0.64757 NA NA 0.44792
NA NA 0.29514 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=15.600 >=15.600
NA 0.05000 NBRCNT 576 0.09561 NA NA 0.35117 NA
NA 0.32432 NA NA 0.52691 NA NA 0.32292
NA NA 0.05382 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=25.600 >=25.600
NA 0.05000 NBRCNT 576 0.07624 NA NA 0.02011 NA
NA 0.03076 NA NA 0.50174 NA NA 0.22569
NA NA 0.00347 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=35.600 >=35.600
NA 0.05000 NBRCNT 576 0.04743 NA NA 0.02797 NA
NA 0.04122 NA NA 0.50174 NA NA 0.16840
NA NA 0.00347 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=45.600 >=45.600
NA 0.05000 NBRCNT 576 0.02798 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12674
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=55.600 >=55.600
NA 0.05000 NBRCNT 576 0.01138 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.08160
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=64.500 >=64.500
NA 0.05000 NBRCNT 576 0.00636 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.06076
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 169 >=124.500 >=124.500
NA 0.05000 NBRCNT 576 0.00003 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00347
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=0.100 >=0.100
NA 0.05000 NBRCNT 552 0.02979 NA NA 0.97810 NA
NA 0.99859 NA NA 0.91938 NA NA 0.79529
NA NA 0.83877 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=2.400 >=2.400
NA 0.05000 NBRCNT 552 0.03929 NA NA 0.95081 NA
NA 0.99998 NA NA 0.79620 NA NA 0.58877
NA NA 0.59239 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=7.500 >=7.500
NA 0.05000 NBRCNT 552 0.07296 NA NA 0.80931 NA
NA 0.92022 NA NA 0.65036 NA NA 0.45471
NA NA 0.30072 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=15.600 >=15.600
NA 0.05000 NBRCNT 552 0.09410 NA NA 0.34399 NA
NA 0.33452 NA NA 0.52808 NA NA 0.32609
NA NA 0.05616 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=25.600 >=25.600
NA 0.05000 NBRCNT 552 0.07371 NA NA 0.01963 NA
NA 0.03149 NA NA 0.50181 NA NA 0.23007
NA NA 0.00362 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=35.600 >=35.600
NA 0.05000 NBRCNT 552 0.04496 NA NA 0.02720 NA
NA 0.04253 NA NA 0.50181 NA NA 0.17029
NA NA 0.00362 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=45.600 >=45.600
NA 0.05000 NBRCNT 552 0.02639 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.12862
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=55.600 >=55.600
NA 0.05000 NBRCNT 552 0.01045 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.08152
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=64.500 >=64.500
NA 0.05000 NBRCNT 552 0.00588 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.05978
NA NA 0.00000 NA NA
V5.0 WRF 240000 20110423_000000 20110423_000000 240000
20110423_000000 20110423_000000 APCP A24 APCP A24
MC_PCP FULL NBRHD 225 >=124.500 >=124.500
NA 0.05000 NBRCNT 552 0.00003 NA NA 0.00000 NA
NA 0.00000 NA NA 0.50000 NA NA 0.00362
NA NA 0.00000 NA NA
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: John Halley Gotway
Time: Fri Jun 12 15:06:01 2015
Geeta,
Yes, you are correct. UFSS = 0.5 + base_rate/2
For lower thresholds, the base rate will be higher, making UFSS
larger. As
you increase the threshold, UFSS should decrease down to close to 0.5.
Looking at the grid_stat NBRCNT file you sent, you'll notice that the
UFSS
column = 0.5 + O_RATE column divided by 2.
Thanks,
John
On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
>
> Hi John and tara. I have tried to run grid_stat with TRMM daily
files.
> hence I extracted the model data for 24 hours starting 00 UTC of 22
April
> 2011.
> I could not make the Quilt plot, but just for understanding I
plotted data
> in EXCEL. (results attatched). I want to know, the details about
UFSS. It
> is varying for each threshold and NBRHD SIZE (becuz UFSS=0.5+f0/2)
where f0
> is the observation rate.
> Kindly clarify these points. I m waiting.
>
> Geeta
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> Date: Sun, 7 Jun 2015 17:54:25 +0530
>
>
>
>
> Hi John and Tara. I was waiting for your reply. Kindly help me
further.
> ThanksGeeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> Date: Wed, 3 Jun 2015 11:59:35 +0530
>
>
>
>
> Hi Tara,
> I was NOT asking about the plotting of the NETCDF data files
(observations
> n FCST). Those can be plotted with plot_data_plane and ncview etc.
>
> But a step further.
> I want to know if I can plot the binary data after thresholding.
>
> Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> Date: Wed, 3 Jun 2015 10:47:30 +0530
>
>
>
>
> thanks Tara.
> Nice to hear from U. I had met you at Verification workshop last
year in
> New Delhi.
> I didnt understand Ur explanation fully.
> It would be great if U can elaborate upon it in a little more
detail.
>
> See I got the output from Grid stat and now I need to understand
what is
> my FSS for this case??.
> U must see my MODEL FCST and Observations. They are so close but I
am
> unable to understand why FSS is so low????????.
> If U want, I can send my plots (FSS and RF fcst and observation) and
> grid_stat output to you.
>
> Geeta
>
> > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Tue, 2 Jun 2015 09:52:38 -0600
> >
> > Geeta,
> >
> > Sorry for the delay in answering your question about FSS etc.
Yesterday
> was a very busy day here. One of our statisticians put together the
> following answer:
> >
> > According to the Roberts and Lean (2008) paper, the AFSS
(asymptotic
> FSS) is a useful descriptor of the bias because it relates the bias
to the
> spatial accuracy of a forecast and is linked to the conventional
frequency
> bias, but is less sensitive to small frequencies. Compared to the
> frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a factor
of 4,
> and AFSS = 0.2 is a factor of 10. If there is no bias, then AFSS =
1,
> indicating perfect skill over the domain.
> >
> > I will send more information as I or others pull it together.
> >
> > As for plotting the NetCDF files that come from MET. We typically
use
> ncview or NCL for plotting the files, however the plotting tools
Julie
> mentioned should also be able to read the files.
> >
> > Thanks for your patience as I prepare more information about FSS,
AFSS
> etc.
> > Tara
> >
> >
> > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > Thanks John.
> > > Also I was wondering why I should get bad FSS /low FSS for
Lowest
> > > Neighbourhood and lower threshold????. This looks a bit
confusing.
> > > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > > OK????
> > > I shall be waiting for your reply eagerly.
> > >
> > > Geeta
> > >
> > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > From: met_help at ucar.edu
> > > > To: geeta124 at hotmail.com
> > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > >
> > > > Geeta,
> > > >
> > > > Good questions. Unfortunately, I don't have a good answer for
you
> > > yet.
> > > > I'm headed out of town all next week, but I referred your
question
> > > to a
> > > > scientist and statistician here.
> > > >
> > > > Hopefully you'll hear back from them about this next week.
> > > >
> > > > Thanks,
> > > > John
> > > >
> > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > <met_help at ucar.edu>
> > > > wrote:
> > > >
> > > > >
> > > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > Queue: met_help
> > > > > Subject: FSS, UFSS, AFSS
> > > > > Owner: Nobody
> > > > > Requestors: geeta124 at hotmail.com
> > > > > Status: new
> > > > > Ticket <URL:
> > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > >
> > > > >
> > > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > > starting
> > > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > > (0.1,2.5,7.5, and
> > > > > so on.). This is for One case of thunderstorm event. I have
taken
> > > FSS
> > > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > > presume
> > > > > that UFSS is the UNIFORM FSS which is the base line for a
good
> > > forecast. I
> > > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > > But it is
> > > > > less as compared to the UFSS (I plotted UFSS for threshold
of
> > > 0.1). But
> > > > > when I look at my model forecast and observed data (trmm),
Plots
> > > are
> > > > > matching so nicely. I m sending my grid_stat output file.
can u
> > > help me.
> > > > >
> > > >
> >
> >
> >
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Sun Jun 14 07:41:25 2015
thanks John for you clarification.
But I didnt understand the AFSS. Actually, FSS increases
Asymptotically with increase in NBRHD width.
Then what's the Differerence b/w AFSS FSS???? and How is it (AFSS)
calculated.
Kindly explain. . I m not sure abt the AFSS????
2nd Ques. Is is correct when I say that UFSS is NOT constant. It
varies.
So for a useful FCST, I will compare if the FSS (unique for 1 NBRHD
and 1 THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD, FCST HAS SKILL!!!!!!!!!!.
Is my understanding correct????.
3. If U can guide me a little abt the input files Required for running
the Rsript for quilt plot, It would be great............
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
>
> Geeta,
>
> Yes, you are correct. UFSS = 0.5 + base_rate/2
>
> For lower thresholds, the base rate will be higher, making UFSS
larger. As
> you increase the threshold, UFSS should decrease down to close to
0.5.
>
> Looking at the grid_stat NBRCNT file you sent, you'll notice that
the UFSS
> column = 0.5 + O_RATE column divided by 2.
>
> Thanks,
> John
>
>
>
>
> On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > Hi John and tara. I have tried to run grid_stat with TRMM daily
files.
> > hence I extracted the model data for 24 hours starting 00 UTC of
22 April
> > 2011.
> > I could not make the Quilt plot, but just for understanding I
plotted data
> > in EXCEL. (results attatched). I want to know, the details about
UFSS. It
> > is varying for each threshold and NBRHD SIZE (becuz UFSS=0.5+f0/2)
where f0
> > is the observation rate.
> > Kindly clarify these points. I m waiting.
> >
> > Geeta
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Sun, 7 Jun 2015 17:54:25 +0530
> >
> >
> >
> >
> > Hi John and Tara. I was waiting for your reply. Kindly help me
further.
> > ThanksGeeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Wed, 3 Jun 2015 11:59:35 +0530
> >
> >
> >
> >
> > Hi Tara,
> > I was NOT asking about the plotting of the NETCDF data files
(observations
> > n FCST). Those can be plotted with plot_data_plane and ncview etc.
> >
> > But a step further.
> > I want to know if I can plot the binary data after thresholding.
> >
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Wed, 3 Jun 2015 10:47:30 +0530
> >
> >
> >
> >
> > thanks Tara.
> > Nice to hear from U. I had met you at Verification workshop last
year in
> > New Delhi.
> > I didnt understand Ur explanation fully.
> > It would be great if U can elaborate upon it in a little more
detail.
> >
> > See I got the output from Grid stat and now I need to understand
what is
> > my FSS for this case??.
> > U must see my MODEL FCST and Observations. They are so close but I
am
> > unable to understand why FSS is so low????????.
> > If U want, I can send my plots (FSS and RF fcst and observation)
and
> > grid_stat output to you.
> >
> > Geeta
> >
> > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > >
> > > Geeta,
> > >
> > > Sorry for the delay in answering your question about FSS etc.
Yesterday
> > was a very busy day here. One of our statisticians put together
the
> > following answer:
> > >
> > > According to the Roberts and Lean (2008) paper, the AFSS
(asymptotic
> > FSS) is a useful descriptor of the bias because it relates the
bias to the
> > spatial accuracy of a forecast and is linked to the conventional
frequency
> > bias, but is less sensitive to small frequencies. Compared to the
> > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4,
> > and AFSS = 0.2 is a factor of 10. If there is no bias, then AFSS
= 1,
> > indicating perfect skill over the domain.
> > >
> > > I will send more information as I or others pull it together.
> > >
> > > As for plotting the NetCDF files that come from MET. We
typically use
> > ncview or NCL for plotting the files, however the plotting tools
Julie
> > mentioned should also be able to read the files.
> > >
> > > Thanks for your patience as I prepare more information about
FSS, AFSS
> > etc.
> > > Tara
> > >
> > >
> > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > Thanks John.
> > > > Also I was wondering why I should get bad FSS /low FSS for
Lowest
> > > > Neighbourhood and lower threshold????. This looks a bit
confusing.
> > > > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > > > OK????
> > > > I shall be waiting for your reply eagerly.
> > > >
> > > > Geeta
> > > >
> > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > From: met_help at ucar.edu
> > > > > To: geeta124 at hotmail.com
> > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > >
> > > > > Geeta,
> > > > >
> > > > > Good questions. Unfortunately, I don't have a good answer
for you
> > > > yet.
> > > > > I'm headed out of town all next week, but I referred your
question
> > > > to a
> > > > > scientist and statistician here.
> > > > >
> > > > > Hopefully you'll hear back from them about this next week.
> > > > >
> > > > > Thanks,
> > > > > John
> > > > >
> > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > <met_help at ucar.edu>
> > > > > wrote:
> > > > >
> > > > > >
> > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > > Queue: met_help
> > > > > > Subject: FSS, UFSS, AFSS
> > > > > > Owner: Nobody
> > > > > > Requestors: geeta124 at hotmail.com
> > > > > > Status: new
> > > > > > Ticket <URL:
> > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > >
> > > > > >
> > > > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > > > starting
> > > > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > > > (0.1,2.5,7.5, and
> > > > > > so on.). This is for One case of thunderstorm event. I
have taken
> > > > FSS
> > > > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > > > presume
> > > > > > that UFSS is the UNIFORM FSS which is the base line for a
good
> > > > forecast. I
> > > > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > > > But it is
> > > > > > less as compared to the UFSS (I plotted UFSS for threshold
of
> > > > 0.1). But
> > > > > > when I look at my model forecast and observed data (trmm),
Plots
> > > > are
> > > > > > matching so nicely. I m sending my grid_stat output file.
can u
> > > > help me.
> > > > > >
> > > > >
> > >
> > >
> > >
> >
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Wed Jun 17 08:56:31 2015
Hi John.
Kindly help me in getting to know about the UFSS and AFSS. I was
waiting for your response.
I need a little more insight into these ideas.
thanks
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Sun, 14 Jun 2015 19:11:21 +0530
thanks John for you clarification.
But I didnt understand the AFSS. Actually, FSS increases
Asymptotically with increase in NBRHD width.
Then what's the Differerence b/w AFSS FSS???? and How is it (AFSS)
calculated.
Kindly explain. . I m not sure abt the AFSS????
2nd Ques. Is is correct when I say that UFSS is NOT constant. It
varies.
So for a useful FCST, I will compare if the FSS (unique for 1 NBRHD
and 1 THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD, FCST HAS SKILL!!!!!!!!!!.
Is my understanding correct????.
3. If U can guide me a little abt the input files Required for running
the Rsript for quilt plot, It would be great............
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
>
> Geeta,
>
> Yes, you are correct. UFSS = 0.5 + base_rate/2
>
> For lower thresholds, the base rate will be higher, making UFSS
larger. As
> you increase the threshold, UFSS should decrease down to close to
0.5.
>
> Looking at the grid_stat NBRCNT file you sent, you'll notice that
the UFSS
> column = 0.5 + O_RATE column divided by 2.
>
> Thanks,
> John
>
>
>
>
> On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > Hi John and tara. I have tried to run grid_stat with TRMM daily
files.
> > hence I extracted the model data for 24 hours starting 00 UTC of
22 April
> > 2011.
> > I could not make the Quilt plot, but just for understanding I
plotted data
> > in EXCEL. (results attatched). I want to know, the details about
UFSS. It
> > is varying for each threshold and NBRHD SIZE (becuz UFSS=0.5+f0/2)
where f0
> > is the observation rate.
> > Kindly clarify these points. I m waiting.
> >
> > Geeta
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Sun, 7 Jun 2015 17:54:25 +0530
> >
> >
> >
> >
> > Hi John and Tara. I was waiting for your reply. Kindly help me
further.
> > ThanksGeeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Wed, 3 Jun 2015 11:59:35 +0530
> >
> >
> >
> >
> > Hi Tara,
> > I was NOT asking about the plotting of the NETCDF data files
(observations
> > n FCST). Those can be plotted with plot_data_plane and ncview etc.
> >
> > But a step further.
> > I want to know if I can plot the binary data after thresholding.
> >
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Wed, 3 Jun 2015 10:47:30 +0530
> >
> >
> >
> >
> > thanks Tara.
> > Nice to hear from U. I had met you at Verification workshop last
year in
> > New Delhi.
> > I didnt understand Ur explanation fully.
> > It would be great if U can elaborate upon it in a little more
detail.
> >
> > See I got the output from Grid stat and now I need to understand
what is
> > my FSS for this case??.
> > U must see my MODEL FCST and Observations. They are so close but I
am
> > unable to understand why FSS is so low????????.
> > If U want, I can send my plots (FSS and RF fcst and observation)
and
> > grid_stat output to you.
> >
> > Geeta
> >
> > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > >
> > > Geeta,
> > >
> > > Sorry for the delay in answering your question about FSS etc.
Yesterday
> > was a very busy day here. One of our statisticians put together
the
> > following answer:
> > >
> > > According to the Roberts and Lean (2008) paper, the AFSS
(asymptotic
> > FSS) is a useful descriptor of the bias because it relates the
bias to the
> > spatial accuracy of a forecast and is linked to the conventional
frequency
> > bias, but is less sensitive to small frequencies. Compared to the
> > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of 4,
> > and AFSS = 0.2 is a factor of 10. If there is no bias, then AFSS
= 1,
> > indicating perfect skill over the domain.
> > >
> > > I will send more information as I or others pull it together.
> > >
> > > As for plotting the NetCDF files that come from MET. We
typically use
> > ncview or NCL for plotting the files, however the plotting tools
Julie
> > mentioned should also be able to read the files.
> > >
> > > Thanks for your patience as I prepare more information about
FSS, AFSS
> > etc.
> > > Tara
> > >
> > >
> > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > Thanks John.
> > > > Also I was wondering why I should get bad FSS /low FSS for
Lowest
> > > > Neighbourhood and lower threshold????. This looks a bit
confusing.
> > > > I should get low FSS for Lower neighbourhood and high
threshold!!!!.
> > > > OK????
> > > > I shall be waiting for your reply eagerly.
> > > >
> > > > Geeta
> > > >
> > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > From: met_help at ucar.edu
> > > > > To: geeta124 at hotmail.com
> > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > >
> > > > > Geeta,
> > > > >
> > > > > Good questions. Unfortunately, I don't have a good answer
for you
> > > > yet.
> > > > > I'm headed out of town all next week, but I referred your
question
> > > > to a
> > > > > scientist and statistician here.
> > > > >
> > > > > Hopefully you'll hear back from them about this next week.
> > > > >
> > > > > Thanks,
> > > > > John
> > > > >
> > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > <met_help at ucar.edu>
> > > > > wrote:
> > > > >
> > > > > >
> > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > > Queue: met_help
> > > > > > Subject: FSS, UFSS, AFSS
> > > > > > Owner: Nobody
> > > > > > Requestors: geeta124 at hotmail.com
> > > > > > Status: new
> > > > > > Ticket <URL:
> > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > >
> > > > > >
> > > > > > Hi John. I have plotted the FSS for various neighbourhood
widths
> > > > starting
> > > > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > > > (0.1,2.5,7.5, and
> > > > > > so on.). This is for One case of thunderstorm event. I
have taken
> > > > FSS
> > > > > > column from my output. But there are three, FSS, UFSS and
AFSS. I
> > > > presume
> > > > > > that UFSS is the UNIFORM FSS which is the base line for a
good
> > > > forecast. I
> > > > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > > > But it is
> > > > > > less as compared to the UFSS (I plotted UFSS for threshold
of
> > > > 0.1). But
> > > > > > when I look at my model forecast and observed data (trmm),
Plots
> > > > are
> > > > > > matching so nicely. I m sending my grid_stat output file.
can u
> > > > help me.
> > > > > >
> > > > >
> > >
> > >
> > >
> >
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: John Halley Gotway
Time: Thu Jun 18 13:55:06 2015
Geeta,
These quantities are computed in
met-5.0/src/libcode/vx_statistics/met_stats.cc
Here are the relevant lines of code:
2215 //
2216 // Compute Asymptotic Fractions Skill Score
2217 //
2218 num = 2.0*f_rate*o_rate;
2219 den = f_rate*f_rate + o_rate*o_rate;
2220
2221 if(is_bad_data(f_rate) || is_bad_data(o_rate) || is_eq(den,
0.0)) {
2222 afss = bad_data_double;
2223 }
2224 else {
2225 afss = num/den;
2226 }
2236 //
2237 // Compute Uniform Fractions Skill Score
2238 //
2239 if(is_bad_data(o_rate)) ufss = bad_data_double;
2240 else ufss = 0.5 + o_rate/2.0;
In here, f_rate is the ratio (0 to 1) of forecast grid squares where
the
event (as defined by your threshold) is occurring. o_rate is the
ratio of
observation grid squares where the event is occurring.
Take a look at this picture:
http://www.cawcr.gov.au/projects/verification/FSSplot.gif
This is taken from this website:
http://www.cawcr.gov.au/projects/verification/
AFSS (asymptotic fractions skill score) is the dotted line in that
picture
and, as computed above, is based on the bias for the forecast
threshold
you've chosen. Mathematically speaking, FSS can never exceed that
AFSS
value.
UFSS (useful or uniform fraction skill score) is just 0.5 + the base
rate.
A forecast performs well when FSS > UFSS and, of course, FSS < AFSS
because
it can't exceed that value.
Hopefully that helps.
Thanks,
John
On Wed, Jun 17, 2015 at 8:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
>
> Hi John.
> Kindly help me in getting to know about the UFSS and AFSS. I was
waiting
> for your response.
> I need a little more insight into these ideas.
>
> thanks
> Geeta
>
> From: geeta124 at hotmail.com
> To: met_help at ucar.edu
> Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> Date: Sun, 14 Jun 2015 19:11:21 +0530
>
>
>
>
> thanks John for you clarification.
>
> But I didnt understand the AFSS. Actually, FSS increases
Asymptotically
> with increase in NBRHD width.
> Then what's the Differerence b/w AFSS FSS???? and How is it (AFSS)
> calculated.
> Kindly explain. . I m not sure abt the AFSS????
>
> 2nd Ques. Is is correct when I say that UFSS is NOT constant. It
varies.
> So for a useful FCST, I will compare if the FSS (unique for 1 NBRHD
and 1
> THRESHOLD) with that of UFSS. If it's MORE then at that scale/NBRHD,
FCST
> HAS SKILL!!!!!!!!!!.
> Is my understanding correct????.
>
> 3. If U can guide me a little abt the input files Required for
running the
> Rsript for quilt plot, It would be great............
>
> Geeta
>
> > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
> >
> > Geeta,
> >
> > Yes, you are correct. UFSS = 0.5 + base_rate/2
> >
> > For lower thresholds, the base rate will be higher, making UFSS
larger.
> As
> > you increase the threshold, UFSS should decrease down to close to
0.5.
> >
> > Looking at the grid_stat NBRCNT file you sent, you'll notice that
the
> UFSS
> > column = 0.5 + O_RATE column divided by 2.
> >
> > Thanks,
> > John
> >
> >
> >
> >
> > On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > >
> > > Hi John and tara. I have tried to run grid_stat with TRMM daily
files.
> > > hence I extracted the model data for 24 hours starting 00 UTC of
22
> April
> > > 2011.
> > > I could not make the Quilt plot, but just for understanding I
plotted
> data
> > > in EXCEL. (results attatched). I want to know, the details
about
> UFSS. It
> > > is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2)
> where f0
> > > is the observation rate.
> > > Kindly clarify these points. I m waiting.
> > >
> > > Geeta
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > Date: Sun, 7 Jun 2015 17:54:25 +0530
> > >
> > >
> > >
> > >
> > > Hi John and Tara. I was waiting for your reply. Kindly help me
further.
> > > ThanksGeeta
> > >
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > Date: Wed, 3 Jun 2015 11:59:35 +0530
> > >
> > >
> > >
> > >
> > > Hi Tara,
> > > I was NOT asking about the plotting of the NETCDF data files
> (observations
> > > n FCST). Those can be plotted with plot_data_plane and ncview
etc.
> > >
> > > But a step further.
> > > I want to know if I can plot the binary data after thresholding.
> > >
> > > Geeta
> > >
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > Date: Wed, 3 Jun 2015 10:47:30 +0530
> > >
> > >
> > >
> > >
> > > thanks Tara.
> > > Nice to hear from U. I had met you at Verification workshop last
year
> in
> > > New Delhi.
> > > I didnt understand Ur explanation fully.
> > > It would be great if U can elaborate upon it in a little more
detail.
> > >
> > > See I got the output from Grid stat and now I need to understand
what
> is
> > > my FSS for this case??.
> > > U must see my MODEL FCST and Observations. They are so close but
I am
> > > unable to understand why FSS is so low????????.
> > > If U want, I can send my plots (FSS and RF fcst and observation)
and
> > > grid_stat output to you.
> > >
> > > Geeta
> > >
> > > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > From: met_help at ucar.edu
> > > > To: geeta124 at hotmail.com
> > > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > > >
> > > > Geeta,
> > > >
> > > > Sorry for the delay in answering your question about FSS etc.
> Yesterday
> > > was a very busy day here. One of our statisticians put together
the
> > > following answer:
> > > >
> > > > According to the Roberts and Lean (2008) paper, the AFSS
(asymptotic
> > > FSS) is a useful descriptor of the bias because it relates the
bias to
> the
> > > spatial accuracy of a forecast and is linked to the conventional
> frequency
> > > bias, but is less sensitive to small frequencies. Compared to
the
> > > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of
> 4,
> > > and AFSS = 0.2 is a factor of 10. If there is no bias, then
AFSS = 1,
> > > indicating perfect skill over the domain.
> > > >
> > > > I will send more information as I or others pull it together.
> > > >
> > > > As for plotting the NetCDF files that come from MET. We
typically
> use
> > > ncview or NCL for plotting the files, however the plotting tools
Julie
> > > mentioned should also be able to read the files.
> > > >
> > > > Thanks for your patience as I prepare more information about
FSS,
> AFSS
> > > etc.
> > > > Tara
> > > >
> > > >
> > > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > > Thanks John.
> > > > > Also I was wondering why I should get bad FSS /low FSS for
Lowest
> > > > > Neighbourhood and lower threshold????. This looks a bit
confusing.
> > > > > I should get low FSS for Lower neighbourhood and high
> threshold!!!!.
> > > > > OK????
> > > > > I shall be waiting for your reply eagerly.
> > > > >
> > > > > Geeta
> > > > >
> > > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > From: met_help at ucar.edu
> > > > > > To: geeta124 at hotmail.com
> > > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > > >
> > > > > > Geeta,
> > > > > >
> > > > > > Good questions. Unfortunately, I don't have a good answer
for
> you
> > > > > yet.
> > > > > > I'm headed out of town all next week, but I referred your
> question
> > > > > to a
> > > > > > scientist and statistician here.
> > > > > >
> > > > > > Hopefully you'll hear back from them about this next week.
> > > > > >
> > > > > > Thanks,
> > > > > > John
> > > > > >
> > > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > > <met_help at ucar.edu>
> > > > > > wrote:
> > > > > >
> > > > > > >
> > > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted upon.
> > > > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > > > Queue: met_help
> > > > > > > Subject: FSS, UFSS, AFSS
> > > > > > > Owner: Nobody
> > > > > > > Requestors: geeta124 at hotmail.com
> > > > > > > Status: new
> > > > > > > Ticket <URL:
> > > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > > >
> > > > > > >
> > > > > > > Hi John. I have plotted the FSS for various
neighbourhood
> widths
> > > > > starting
> > > > > > > from 1, 3,5,7,9,11,13 and 15 and for various thresholds
> > > > > (0.1,2.5,7.5, and
> > > > > > > so on.). This is for One case of thunderstorm event. I
have
> taken
> > > > > FSS
> > > > > > > column from my output. But there are three, FSS, UFSS
and
> AFSS. I
> > > > > presume
> > > > > > > that UFSS is the UNIFORM FSS which is the base line for
a good
> > > > > forecast. I
> > > > > > > am sending My plot. (attached). for lower threshold, and
lowest
> > > > > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > > > > But it is
> > > > > > > less as compared to the UFSS (I plotted UFSS for
threshold of
> > > > > 0.1). But
> > > > > > > when I look at my model forecast and observed data
(trmm),
> Plots
> > > > > are
> > > > > > > matching so nicely. I m sending my grid_stat output
file. can
> u
> > > > > help me.
> > > > > > >
> > > > > >
> > > >
> > > >
> > > >
> > >
> > >
> >
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Sun Jul 12 06:59:20 2015
hi John.
I have been reading about the use of FSS in a few publications.
Kindly explain me the difference between "Random and Uniform FSS".
Didnot understand from the publications (Mittermair et al 2011).
what's is Random FSS??.
2. If I have questions regarding the running of R esp the quilt plot
of FSS whom should I ask.
where I should adrress. I m not familiar with R. I mean what should be
the input file, where it should be kept etc????
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Thu, 18 Jun 2015 13:55:07 -0600
>
> Geeta,
>
> These quantities are computed in
> met-5.0/src/libcode/vx_statistics/met_stats.cc
>
> Here are the relevant lines of code:
>
> 2215 //
> 2216 // Compute Asymptotic Fractions Skill Score
> 2217 //
> 2218 num = 2.0*f_rate*o_rate;
> 2219 den = f_rate*f_rate + o_rate*o_rate;
> 2220
> 2221 if(is_bad_data(f_rate) || is_bad_data(o_rate) ||
is_eq(den,
> 0.0)) {
> 2222 afss = bad_data_double;
> 2223 }
> 2224 else {
> 2225 afss = num/den;
> 2226 }
>
> 2236 //
> 2237 // Compute Uniform Fractions Skill Score
> 2238 //
> 2239 if(is_bad_data(o_rate)) ufss = bad_data_double;
> 2240 else ufss = 0.5 + o_rate/2.0;
>
> In here, f_rate is the ratio (0 to 1) of forecast grid squares where
the
> event (as defined by your threshold) is occurring. o_rate is the
ratio of
> observation grid squares where the event is occurring.
>
> Take a look at this picture:
> http://www.cawcr.gov.au/projects/verification/FSSplot.gif
>
> This is taken from this website:
> http://www.cawcr.gov.au/projects/verification/
>
> AFSS (asymptotic fractions skill score) is the dotted line in that
picture
> and, as computed above, is based on the bias for the forecast
threshold
> you've chosen. Mathematically speaking, FSS can never exceed that
AFSS
> value.
>
> UFSS (useful or uniform fraction skill score) is just 0.5 + the base
rate.
>
> A forecast performs well when FSS > UFSS and, of course, FSS < AFSS
because
> it can't exceed that value.
>
> Hopefully that helps.
>
> Thanks,
> John
>
>
> On Wed, Jun 17, 2015 at 8:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > Hi John.
> > Kindly help me in getting to know about the UFSS and AFSS. I was
waiting
> > for your response.
> > I need a little more insight into these ideas.
> >
> > thanks
> > Geeta
> >
> > From: geeta124 at hotmail.com
> > To: met_help at ucar.edu
> > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > Date: Sun, 14 Jun 2015 19:11:21 +0530
> >
> >
> >
> >
> > thanks John for you clarification.
> >
> > But I didnt understand the AFSS. Actually, FSS increases
Asymptotically
> > with increase in NBRHD width.
> > Then what's the Differerence b/w AFSS FSS???? and How is it (AFSS)
> > calculated.
> > Kindly explain. . I m not sure abt the AFSS????
> >
> > 2nd Ques. Is is correct when I say that UFSS is NOT constant. It
varies.
> > So for a useful FCST, I will compare if the FSS (unique for 1
NBRHD and 1
> > THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD, FCST
> > HAS SKILL!!!!!!!!!!.
> > Is my understanding correct????.
> >
> > 3. If U can guide me a little abt the input files Required for
running the
> > Rsript for quilt plot, It would be great............
> >
> > Geeta
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
> > >
> > > Geeta,
> > >
> > > Yes, you are correct. UFSS = 0.5 + base_rate/2
> > >
> > > For lower thresholds, the base rate will be higher, making UFSS
larger.
> > As
> > > you increase the threshold, UFSS should decrease down to close
to 0.5.
> > >
> > > Looking at the grid_stat NBRCNT file you sent, you'll notice
that the
> > UFSS
> > > column = 0.5 + O_RATE column divided by 2.
> > >
> > > Thanks,
> > > John
> > >
> > >
> > >
> > >
> > > On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
>
> > > >
> > > > Hi John and tara. I have tried to run grid_stat with TRMM
daily files.
> > > > hence I extracted the model data for 24 hours starting 00 UTC
of 22
> > April
> > > > 2011.
> > > > I could not make the Quilt plot, but just for understanding I
plotted
> > data
> > > > in EXCEL. (results attatched). I want to know, the details
about
> > UFSS. It
> > > > is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2)
> > where f0
> > > > is the observation rate.
> > > > Kindly clarify these points. I m waiting.
> > > >
> > > > Geeta
> > > > From: geeta124 at hotmail.com
> > > > To: met_help at ucar.edu
> > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > Date: Sun, 7 Jun 2015 17:54:25 +0530
> > > >
> > > >
> > > >
> > > >
> > > > Hi John and Tara. I was waiting for your reply. Kindly help me
further.
> > > > ThanksGeeta
> > > >
> > > > From: geeta124 at hotmail.com
> > > > To: met_help at ucar.edu
> > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > Date: Wed, 3 Jun 2015 11:59:35 +0530
> > > >
> > > >
> > > >
> > > >
> > > > Hi Tara,
> > > > I was NOT asking about the plotting of the NETCDF data files
> > (observations
> > > > n FCST). Those can be plotted with plot_data_plane and ncview
etc.
> > > >
> > > > But a step further.
> > > > I want to know if I can plot the binary data after
thresholding.
> > > >
> > > > Geeta
> > > >
> > > > From: geeta124 at hotmail.com
> > > > To: met_help at ucar.edu
> > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > Date: Wed, 3 Jun 2015 10:47:30 +0530
> > > >
> > > >
> > > >
> > > >
> > > > thanks Tara.
> > > > Nice to hear from U. I had met you at Verification workshop
last year
> > in
> > > > New Delhi.
> > > > I didnt understand Ur explanation fully.
> > > > It would be great if U can elaborate upon it in a little more
detail.
> > > >
> > > > See I got the output from Grid stat and now I need to
understand what
> > is
> > > > my FSS for this case??.
> > > > U must see my MODEL FCST and Observations. They are so close
but I am
> > > > unable to understand why FSS is so low????????.
> > > > If U want, I can send my plots (FSS and RF fcst and
observation) and
> > > > grid_stat output to you.
> > > >
> > > > Geeta
> > > >
> > > > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > From: met_help at ucar.edu
> > > > > To: geeta124 at hotmail.com
> > > > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > > > >
> > > > > Geeta,
> > > > >
> > > > > Sorry for the delay in answering your question about FSS
etc.
> > Yesterday
> > > > was a very busy day here. One of our statisticians put
together the
> > > > following answer:
> > > > >
> > > > > According to the Roberts and Lean (2008) paper, the AFSS
(asymptotic
> > > > FSS) is a useful descriptor of the bias because it relates the
bias to
> > the
> > > > spatial accuracy of a forecast and is linked to the
conventional
> > frequency
> > > > bias, but is less sensitive to small frequencies. Compared to
the
> > > > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
factor of
> > 4,
> > > > and AFSS = 0.2 is a factor of 10. If there is no bias, then
AFSS = 1,
> > > > indicating perfect skill over the domain.
> > > > >
> > > > > I will send more information as I or others pull it
together.
> > > > >
> > > > > As for plotting the NetCDF files that come from MET. We
typically
> > use
> > > > ncview or NCL for plotting the files, however the plotting
tools Julie
> > > > mentioned should also be able to read the files.
> > > > >
> > > > > Thanks for your patience as I prepare more information about
FSS,
> > AFSS
> > > > etc.
> > > > > Tara
> > > > >
> > > > >
> > > > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > > > Thanks John.
> > > > > > Also I was wondering why I should get bad FSS /low FSS for
Lowest
> > > > > > Neighbourhood and lower threshold????. This looks a bit
confusing.
> > > > > > I should get low FSS for Lower neighbourhood and high
> > threshold!!!!.
> > > > > > OK????
> > > > > > I shall be waiting for your reply eagerly.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > > From: met_help at ucar.edu
> > > > > > > To: geeta124 at hotmail.com
> > > > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > > > >
> > > > > > > Geeta,
> > > > > > >
> > > > > > > Good questions. Unfortunately, I don't have a good
answer for
> > you
> > > > > > yet.
> > > > > > > I'm headed out of town all next week, but I referred
your
> > question
> > > > > > to a
> > > > > > > scientist and statistician here.
> > > > > > >
> > > > > > > Hopefully you'll hear back from them about this next
week.
> > > > > > >
> > > > > > > Thanks,
> > > > > > > John
> > > > > > >
> > > > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > > > <met_help at ucar.edu>
> > > > > > > wrote:
> > > > > > >
> > > > > > > >
> > > > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted
upon.
> > > > > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > > > > Queue: met_help
> > > > > > > > Subject: FSS, UFSS, AFSS
> > > > > > > > Owner: Nobody
> > > > > > > > Requestors: geeta124 at hotmail.com
> > > > > > > > Status: new
> > > > > > > > Ticket <URL:
> > > > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > > > >
> > > > > > > >
> > > > > > > > Hi John. I have plotted the FSS for various
neighbourhood
> > widths
> > > > > > starting
> > > > > > > > from 1, 3,5,7,9,11,13 and 15 and for various
thresholds
> > > > > > (0.1,2.5,7.5, and
> > > > > > > > so on.). This is for One case of thunderstorm event. I
have
> > taken
> > > > > > FSS
> > > > > > > > column from my output. But there are three, FSS, UFSS
and
> > AFSS. I
> > > > > > presume
> > > > > > > > that UFSS is the UNIFORM FSS which is the base line
for a good
> > > > > > forecast. I
> > > > > > > > am sending My plot. (attached). for lower threshold,
and lowest
> > > > > > > > neighbourhood, I get lowest FSS which increases very
gradually.
> > > > > > But it is
> > > > > > > > less as compared to the UFSS (I plotted UFSS for
threshold of
> > > > > > 0.1). But
> > > > > > > > when I look at my model forecast and observed data
(trmm),
> > Plots
> > > > > > are
> > > > > > > > matching so nicely. I m sending my grid_stat output
file. can
> > u
> > > > > > help me.
> > > > > > > >
> > > > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: John Halley Gotway
Time: Tue Jul 21 13:26:11 2015
Geeta,
Sorry, I'm not able to provide any more help on this issue. For
questions
about R, I'd suggest posting it to an R users group.
John
On Sun, Jul 12, 2015 at 6:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
>
> hi John.
> I have been reading about the use of FSS in a few publications.
> Kindly explain me the difference between "Random and Uniform FSS".
Didnot
> understand from the publications (Mittermair et al 2011).
>
> what's is Random FSS??.
> 2. If I have questions regarding the running of R esp the quilt plot
of
> FSS whom should I ask.
> where I should adrress. I m not familiar with R. I mean what should
be the
> input file, where it should be kept etc????
>
>
>
> > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > From: met_help at ucar.edu
> > To: geeta124 at hotmail.com
> > Date: Thu, 18 Jun 2015 13:55:07 -0600
> >
> > Geeta,
> >
> > These quantities are computed in
> > met-5.0/src/libcode/vx_statistics/met_stats.cc
> >
> > Here are the relevant lines of code:
> >
> > 2215 //
> > 2216 // Compute Asymptotic Fractions Skill Score
> > 2217 //
> > 2218 num = 2.0*f_rate*o_rate;
> > 2219 den = f_rate*f_rate + o_rate*o_rate;
> > 2220
> > 2221 if(is_bad_data(f_rate) || is_bad_data(o_rate) ||
is_eq(den,
> > 0.0)) {
> > 2222 afss = bad_data_double;
> > 2223 }
> > 2224 else {
> > 2225 afss = num/den;
> > 2226 }
> >
> > 2236 //
> > 2237 // Compute Uniform Fractions Skill Score
> > 2238 //
> > 2239 if(is_bad_data(o_rate)) ufss = bad_data_double;
> > 2240 else ufss = 0.5 + o_rate/2.0;
> >
> > In here, f_rate is the ratio (0 to 1) of forecast grid squares
where the
> > event (as defined by your threshold) is occurring. o_rate is the
ratio
> of
> > observation grid squares where the event is occurring.
> >
> > Take a look at this picture:
> > http://www.cawcr.gov.au/projects/verification/FSSplot.gif
> >
> > This is taken from this website:
> > http://www.cawcr.gov.au/projects/verification/
> >
> > AFSS (asymptotic fractions skill score) is the dotted line in that
> picture
> > and, as computed above, is based on the bias for the forecast
threshold
> > you've chosen. Mathematically speaking, FSS can never exceed that
AFSS
> > value.
> >
> > UFSS (useful or uniform fraction skill score) is just 0.5 + the
base
> rate.
> >
> > A forecast performs well when FSS > UFSS and, of course, FSS <
AFSS
> because
> > it can't exceed that value.
> >
> > Hopefully that helps.
> >
> > Thanks,
> > John
> >
> >
> > On Wed, Jun 17, 2015 at 8:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > >
> > > Hi John.
> > > Kindly help me in getting to know about the UFSS and AFSS. I was
> waiting
> > > for your response.
> > > I need a little more insight into these ideas.
> > >
> > > thanks
> > > Geeta
> > >
> > > From: geeta124 at hotmail.com
> > > To: met_help at ucar.edu
> > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > Date: Sun, 14 Jun 2015 19:11:21 +0530
> > >
> > >
> > >
> > >
> > > thanks John for you clarification.
> > >
> > > But I didnt understand the AFSS. Actually, FSS increases
Asymptotically
> > > with increase in NBRHD width.
> > > Then what's the Differerence b/w AFSS FSS???? and How is it
(AFSS)
> > > calculated.
> > > Kindly explain. . I m not sure abt the AFSS????
> > >
> > > 2nd Ques. Is is correct when I say that UFSS is NOT constant.
It
> varies.
> > > So for a useful FCST, I will compare if the FSS (unique for 1
NBRHD
> and 1
> > > THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD,
> FCST
> > > HAS SKILL!!!!!!!!!!.
> > > Is my understanding correct????.
> > >
> > > 3. If U can guide me a little abt the input files Required for
running
> the
> > > Rsript for quilt plot, It would be great............
> > >
> > > Geeta
> > >
> > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > From: met_help at ucar.edu
> > > > To: geeta124 at hotmail.com
> > > > Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
> > > >
> > > > Geeta,
> > > >
> > > > Yes, you are correct. UFSS = 0.5 + base_rate/2
> > > >
> > > > For lower thresholds, the base rate will be higher, making
UFSS
> larger.
> > > As
> > > > you increase the threshold, UFSS should decrease down to close
to
> 0.5.
> > > >
> > > > Looking at the grid_stat NBRCNT file you sent, you'll notice
that the
> > > UFSS
> > > > column = 0.5 + O_RATE column divided by 2.
> > > >
> > > > Thanks,
> > > > John
> > > >
> > > >
> > > >
> > > >
> > > > On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT <
> met_help at ucar.edu>
> > > > wrote:
> > > >
> > > > >
> > > > > <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > >
> > > > > Hi John and tara. I have tried to run grid_stat with TRMM
daily
> files.
> > > > > hence I extracted the model data for 24 hours starting 00
UTC of 22
> > > April
> > > > > 2011.
> > > > > I could not make the Quilt plot, but just for understanding
I
> plotted
> > > data
> > > > > in EXCEL. (results attatched). I want to know, the details
about
> > > UFSS. It
> > > > > is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2)
> > > where f0
> > > > > is the observation rate.
> > > > > Kindly clarify these points. I m waiting.
> > > > >
> > > > > Geeta
> > > > > From: geeta124 at hotmail.com
> > > > > To: met_help at ucar.edu
> > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > Date: Sun, 7 Jun 2015 17:54:25 +0530
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Hi John and Tara. I was waiting for your reply. Kindly help
me
> further.
> > > > > ThanksGeeta
> > > > >
> > > > > From: geeta124 at hotmail.com
> > > > > To: met_help at ucar.edu
> > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > Date: Wed, 3 Jun 2015 11:59:35 +0530
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > Hi Tara,
> > > > > I was NOT asking about the plotting of the NETCDF data files
> > > (observations
> > > > > n FCST). Those can be plotted with plot_data_plane and
ncview etc.
> > > > >
> > > > > But a step further.
> > > > > I want to know if I can plot the binary data after
thresholding.
> > > > >
> > > > > Geeta
> > > > >
> > > > > From: geeta124 at hotmail.com
> > > > > To: met_help at ucar.edu
> > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > Date: Wed, 3 Jun 2015 10:47:30 +0530
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > thanks Tara.
> > > > > Nice to hear from U. I had met you at Verification workshop
last
> year
> > > in
> > > > > New Delhi.
> > > > > I didnt understand Ur explanation fully.
> > > > > It would be great if U can elaborate upon it in a little
more
> detail.
> > > > >
> > > > > See I got the output from Grid stat and now I need to
understand
> what
> > > is
> > > > > my FSS for this case??.
> > > > > U must see my MODEL FCST and Observations. They are so close
but I
> am
> > > > > unable to understand why FSS is so low????????.
> > > > > If U want, I can send my plots (FSS and RF fcst and
observation)
> and
> > > > > grid_stat output to you.
> > > > >
> > > > > Geeta
> > > > >
> > > > > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > From: met_help at ucar.edu
> > > > > > To: geeta124 at hotmail.com
> > > > > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > > > > >
> > > > > > Geeta,
> > > > > >
> > > > > > Sorry for the delay in answering your question about FSS
etc.
> > > Yesterday
> > > > > was a very busy day here. One of our statisticians put
together the
> > > > > following answer:
> > > > > >
> > > > > > According to the Roberts and Lean (2008) paper, the AFSS
> (asymptotic
> > > > > FSS) is a useful descriptor of the bias because it relates
the
> bias to
> > > the
> > > > > spatial accuracy of a forecast and is linked to the
conventional
> > > frequency
> > > > > bias, but is less sensitive to small frequencies. Compared
to the
> > > > > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is a
> factor of
> > > 4,
> > > > > and AFSS = 0.2 is a factor of 10. If there is no bias, then
AFSS
> = 1,
> > > > > indicating perfect skill over the domain.
> > > > > >
> > > > > > I will send more information as I or others pull it
together.
> > > > > >
> > > > > > As for plotting the NetCDF files that come from MET. We
> typically
> > > use
> > > > > ncview or NCL for plotting the files, however the plotting
tools
> Julie
> > > > > mentioned should also be able to read the files.
> > > > > >
> > > > > > Thanks for your patience as I prepare more information
about FSS,
> > > AFSS
> > > > > etc.
> > > > > > Tara
> > > > > >
> > > > > >
> > > > > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > > > > Thanks John.
> > > > > > > Also I was wondering why I should get bad FSS /low FSS
for
> Lowest
> > > > > > > Neighbourhood and lower threshold????. This looks a bit
> confusing.
> > > > > > > I should get low FSS for Lower neighbourhood and high
> > > threshold!!!!.
> > > > > > > OK????
> > > > > > > I shall be waiting for your reply eagerly.
> > > > > > >
> > > > > > > Geeta
> > > > > > >
> > > > > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > > > From: met_help at ucar.edu
> > > > > > > > To: geeta124 at hotmail.com
> > > > > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > > > > >
> > > > > > > > Geeta,
> > > > > > > >
> > > > > > > > Good questions. Unfortunately, I don't have a good
answer
> for
> > > you
> > > > > > > yet.
> > > > > > > > I'm headed out of town all next week, but I referred
your
> > > question
> > > > > > > to a
> > > > > > > > scientist and statistician here.
> > > > > > > >
> > > > > > > > Hopefully you'll hear back from them about this next
week.
> > > > > > > >
> > > > > > > > Thanks,
> > > > > > > > John
> > > > > > > >
> > > > > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > > > > <met_help at ucar.edu>
> > > > > > > > wrote:
> > > > > > > >
> > > > > > > > >
> > > > > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted
upon.
> > > > > > > > > Transaction: Ticket created by geeta124 at hotmail.com
> > > > > > > > > Queue: met_help
> > > > > > > > > Subject: FSS, UFSS, AFSS
> > > > > > > > > Owner: Nobody
> > > > > > > > > Requestors: geeta124 at hotmail.com
> > > > > > > > > Status: new
> > > > > > > > > Ticket <URL:
> > > > > > > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
>
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > Hi John. I have plotted the FSS for various
neighbourhood
> > > widths
> > > > > > > starting
> > > > > > > > > from 1, 3,5,7,9,11,13 and 15 and for various
thresholds
> > > > > > > (0.1,2.5,7.5, and
> > > > > > > > > so on.). This is for One case of thunderstorm event.
I have
> > > taken
> > > > > > > FSS
> > > > > > > > > column from my output. But there are three, FSS,
UFSS and
> > > AFSS. I
> > > > > > > presume
> > > > > > > > > that UFSS is the UNIFORM FSS which is the base line
for a
> good
> > > > > > > forecast. I
> > > > > > > > > am sending My plot. (attached). for lower threshold,
and
> lowest
> > > > > > > > > neighbourhood, I get lowest FSS which increases very
> gradually.
> > > > > > > But it is
> > > > > > > > > less as compared to the UFSS (I plotted UFSS for
threshold
> of
> > > > > > > 0.1). But
> > > > > > > > > when I look at my model forecast and observed data
(trmm),
> > > Plots
> > > > > > > are
> > > > > > > > > matching so nicely. I m sending my grid_stat output
file.
> can
> > > u
> > > > > > > help me.
> > > > > > > > >
> > > > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > >
> > >
> >
>
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Wed Jul 22 04:58:13 2015
Hi John.
I have been able to install the packages required for R.
when I run the FSS quilt script,
Rscript ../R_scripts/fss_quilt_plot.R ./out-22apr2011-00to00UTC/ WRF
240000 -valid_time 20110423_000000 -bw
Loading required package: methods
Loading required package: spam
Loading required package: grid
Spam version 1.0-1 (2014-09-09) is loaded.
Type 'help( Spam)' or 'demo( spam)' for a short introduction
and overview of this package.
Help for individual functions is also obtained by adding the
suffix '.spam' to the function name, e.g. 'help( chol.spam)'.
Attaching package: ‘spam’
The following objects are masked from ‘package:base’:
backsolve, forwardsolve
Loading required package: maps
Attaching package: ‘fields’
The following object is masked from ‘package:maps’:
ozone
gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
Attaching package: ‘gdata’
The following object is masked from ‘package:stats’:
nobs
The following object is masked from ‘package:utils’:
object.size
Error in read.table(NBR_FILE, header = TRUE) :
more columns than column names
Execution halted
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$
the NBR_WRF file looks like this.
can U tell me if the usage of scrpt is OK???
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 21 Jul 2015 13:26:11 -0600
>
> Geeta,
>
> Sorry, I'm not able to provide any more help on this issue. For
questions
> about R, I'd suggest posting it to an R users group.
>
> John
>
> On Sun, Jul 12, 2015 at 6:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > hi John.
> > I have been reading about the use of FSS in a few publications.
> > Kindly explain me the difference between "Random and Uniform FSS".
Didnot
> > understand from the publications (Mittermair et al 2011).
> >
> > what's is Random FSS??.
> > 2. If I have questions regarding the running of R esp the quilt
plot of
> > FSS whom should I ask.
> > where I should adrress. I m not familiar with R. I mean what
should be the
> > input file, where it should be kept etc????
> >
> >
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Thu, 18 Jun 2015 13:55:07 -0600
> > >
> > > Geeta,
> > >
> > > These quantities are computed in
> > > met-5.0/src/libcode/vx_statistics/met_stats.cc
> > >
> > > Here are the relevant lines of code:
> > >
> > > 2215 //
> > > 2216 // Compute Asymptotic Fractions Skill Score
> > > 2217 //
> > > 2218 num = 2.0*f_rate*o_rate;
> > > 2219 den = f_rate*f_rate + o_rate*o_rate;
> > > 2220
> > > 2221 if(is_bad_data(f_rate) || is_bad_data(o_rate) ||
is_eq(den,
> > > 0.0)) {
> > > 2222 afss = bad_data_double;
> > > 2223 }
> > > 2224 else {
> > > 2225 afss = num/den;
> > > 2226 }
> > >
> > > 2236 //
> > > 2237 // Compute Uniform Fractions Skill Score
> > > 2238 //
> > > 2239 if(is_bad_data(o_rate)) ufss = bad_data_double;
> > > 2240 else ufss = 0.5 +
o_rate/2.0;
> > >
> > > In here, f_rate is the ratio (0 to 1) of forecast grid squares
where the
> > > event (as defined by your threshold) is occurring. o_rate is
the ratio
> > of
> > > observation grid squares where the event is occurring.
> > >
> > > Take a look at this picture:
> > > http://www.cawcr.gov.au/projects/verification/FSSplot.gif
> > >
> > > This is taken from this website:
> > > http://www.cawcr.gov.au/projects/verification/
> > >
> > > AFSS (asymptotic fractions skill score) is the dotted line in
that
> > picture
> > > and, as computed above, is based on the bias for the forecast
threshold
> > > you've chosen. Mathematically speaking, FSS can never exceed
that AFSS
> > > value.
> > >
> > > UFSS (useful or uniform fraction skill score) is just 0.5 + the
base
> > rate.
> > >
> > > A forecast performs well when FSS > UFSS and, of course, FSS <
AFSS
> > because
> > > it can't exceed that value.
> > >
> > > Hopefully that helps.
> > >
> > > Thanks,
> > > John
> > >
> > >
> > > On Wed, Jun 17, 2015 at 8:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
>
> > > >
> > > > Hi John.
> > > > Kindly help me in getting to know about the UFSS and AFSS. I
was
> > waiting
> > > > for your response.
> > > > I need a little more insight into these ideas.
> > > >
> > > > thanks
> > > > Geeta
> > > >
> > > > From: geeta124 at hotmail.com
> > > > To: met_help at ucar.edu
> > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > Date: Sun, 14 Jun 2015 19:11:21 +0530
> > > >
> > > >
> > > >
> > > >
> > > > thanks John for you clarification.
> > > >
> > > > But I didnt understand the AFSS. Actually, FSS increases
Asymptotically
> > > > with increase in NBRHD width.
> > > > Then what's the Differerence b/w AFSS FSS???? and How is it
(AFSS)
> > > > calculated.
> > > > Kindly explain. . I m not sure abt the AFSS????
> > > >
> > > > 2nd Ques. Is is correct when I say that UFSS is NOT
constant. It
> > varies.
> > > > So for a useful FCST, I will compare if the FSS (unique for 1
NBRHD
> > and 1
> > > > THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD,
> > FCST
> > > > HAS SKILL!!!!!!!!!!.
> > > > Is my understanding correct????.
> > > >
> > > > 3. If U can guide me a little abt the input files Required for
running
> > the
> > > > Rsript for quilt plot, It would be great............
> > > >
> > > > Geeta
> > > >
> > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > From: met_help at ucar.edu
> > > > > To: geeta124 at hotmail.com
> > > > > Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
> > > > >
> > > > > Geeta,
> > > > >
> > > > > Yes, you are correct. UFSS = 0.5 + base_rate/2
> > > > >
> > > > > For lower thresholds, the base rate will be higher, making
UFSS
> > larger.
> > > > As
> > > > > you increase the threshold, UFSS should decrease down to
close to
> > 0.5.
> > > > >
> > > > > Looking at the grid_stat NBRCNT file you sent, you'll notice
that the
> > > > UFSS
> > > > > column = 0.5 + O_RATE column divided by 2.
> > > > >
> > > > > Thanks,
> > > > > John
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT <
> > met_help at ucar.edu>
> > > > > wrote:
> > > > >
> > > > > >
> > > > > > <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > >
> > > > > > Hi John and tara. I have tried to run grid_stat with TRMM
daily
> > files.
> > > > > > hence I extracted the model data for 24 hours starting 00
UTC of 22
> > > > April
> > > > > > 2011.
> > > > > > I could not make the Quilt plot, but just for
understanding I
> > plotted
> > > > data
> > > > > > in EXCEL. (results attatched). I want to know, the
details about
> > > > UFSS. It
> > > > > > is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2)
> > > > where f0
> > > > > > is the observation rate.
> > > > > > Kindly clarify these points. I m waiting.
> > > > > >
> > > > > > Geeta
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Sun, 7 Jun 2015 17:54:25 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi John and Tara. I was waiting for your reply. Kindly
help me
> > further.
> > > > > > ThanksGeeta
> > > > > >
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Wed, 3 Jun 2015 11:59:35 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi Tara,
> > > > > > I was NOT asking about the plotting of the NETCDF data
files
> > > > (observations
> > > > > > n FCST). Those can be plotted with plot_data_plane and
ncview etc.
> > > > > >
> > > > > > But a step further.
> > > > > > I want to know if I can plot the binary data after
thresholding.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Wed, 3 Jun 2015 10:47:30 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > thanks Tara.
> > > > > > Nice to hear from U. I had met you at Verification
workshop last
> > year
> > > > in
> > > > > > New Delhi.
> > > > > > I didnt understand Ur explanation fully.
> > > > > > It would be great if U can elaborate upon it in a little
more
> > detail.
> > > > > >
> > > > > > See I got the output from Grid stat and now I need to
understand
> > what
> > > > is
> > > > > > my FSS for this case??.
> > > > > > U must see my MODEL FCST and Observations. They are so
close but I
> > am
> > > > > > unable to understand why FSS is so low????????.
> > > > > > If U want, I can send my plots (FSS and RF fcst and
observation)
> > and
> > > > > > grid_stat output to you.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > > From: met_help at ucar.edu
> > > > > > > To: geeta124 at hotmail.com
> > > > > > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > > > > > >
> > > > > > > Geeta,
> > > > > > >
> > > > > > > Sorry for the delay in answering your question about FSS
etc.
> > > > Yesterday
> > > > > > was a very busy day here. One of our statisticians put
together the
> > > > > > following answer:
> > > > > > >
> > > > > > > According to the Roberts and Lean (2008) paper, the AFSS
> > (asymptotic
> > > > > > FSS) is a useful descriptor of the bias because it relates
the
> > bias to
> > > > the
> > > > > > spatial accuracy of a forecast and is linked to the
conventional
> > > > frequency
> > > > > > bias, but is less sensitive to small frequencies.
Compared to the
> > > > > > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is
a
> > factor of
> > > > 4,
> > > > > > and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS
> > = 1,
> > > > > > indicating perfect skill over the domain.
> > > > > > >
> > > > > > > I will send more information as I or others pull it
together.
> > > > > > >
> > > > > > > As for plotting the NetCDF files that come from MET. We
> > typically
> > > > use
> > > > > > ncview or NCL for plotting the files, however the plotting
tools
> > Julie
> > > > > > mentioned should also be able to read the files.
> > > > > > >
> > > > > > > Thanks for your patience as I prepare more information
about FSS,
> > > > AFSS
> > > > > > etc.
> > > > > > > Tara
> > > > > > >
> > > > > > >
> > > > > > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > > > > > Thanks John.
> > > > > > > > Also I was wondering why I should get bad FSS /low FSS
for
> > Lowest
> > > > > > > > Neighbourhood and lower threshold????. This looks a
bit
> > confusing.
> > > > > > > > I should get low FSS for Lower neighbourhood and high
> > > > threshold!!!!.
> > > > > > > > OK????
> > > > > > > > I shall be waiting for your reply eagerly.
> > > > > > > >
> > > > > > > > Geeta
> > > > > > > >
> > > > > > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS,
AFSS
> > > > > > > > > From: met_help at ucar.edu
> > > > > > > > > To: geeta124 at hotmail.com
> > > > > > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > > > > > >
> > > > > > > > > Geeta,
> > > > > > > > >
> > > > > > > > > Good questions. Unfortunately, I don't have a good
answer
> > for
> > > > you
> > > > > > > > yet.
> > > > > > > > > I'm headed out of town all next week, but I referred
your
> > > > question
> > > > > > > > to a
> > > > > > > > > scientist and statistician here.
> > > > > > > > >
> > > > > > > > > Hopefully you'll hear back from them about this next
week.
> > > > > > > > >
> > > > > > > > > Thanks,
> > > > > > > > > John
> > > > > > > > >
> > > > > > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > > > > > <met_help at ucar.edu>
> > > > > > > > > wrote:
> > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted
upon.
> > > > > > > > > > Transaction: Ticket created by
geeta124 at hotmail.com
> > > > > > > > > > Queue: met_help
> > > > > > > > > > Subject: FSS, UFSS, AFSS
> > > > > > > > > > Owner: Nobody
> > > > > > > > > > Requestors: geeta124 at hotmail.com
> > > > > > > > > > Status: new
> > > > > > > > > > Ticket <URL:
> > > > > > > >
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Hi John. I have plotted the FSS for various
neighbourhood
> > > > widths
> > > > > > > > starting
> > > > > > > > > > from 1, 3,5,7,9,11,13 and 15 and for various
thresholds
> > > > > > > > (0.1,2.5,7.5, and
> > > > > > > > > > so on.). This is for One case of thunderstorm
event. I have
> > > > taken
> > > > > > > > FSS
> > > > > > > > > > column from my output. But there are three, FSS,
UFSS and
> > > > AFSS. I
> > > > > > > > presume
> > > > > > > > > > that UFSS is the UNIFORM FSS which is the base
line for a
> > good
> > > > > > > > forecast. I
> > > > > > > > > > am sending My plot. (attached). for lower
threshold, and
> > lowest
> > > > > > > > > > neighbourhood, I get lowest FSS which increases
very
> > gradually.
> > > > > > > > But it is
> > > > > > > > > > less as compared to the UFSS (I plotted UFSS for
threshold
> > of
> > > > > > > > 0.1). But
> > > > > > > > > > when I look at my model forecast and observed data
(trmm),
> > > > Plots
> > > > > > > > are
> > > > > > > > > > matching so nicely. I m sending my grid_stat
output file.
> > can
> > > > u
> > > > > > > > help me.
> > > > > > > > > >
> > > > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > >
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Wed Jul 22 05:43:20 2015
Hi John.
I tried,
Rscript ../R_scripts/fss_quilt_plot.R ./out-22apr2011-00to00UTC/ WRF
240000 -valid_time 20110423_000000 -bw
and got the folowing error.
Loading required package: methods
Loading required package: spam
Loading required package: grid
Spam version 1.0-1 (2014-09-09) is loaded.
Type 'help( Spam)' or 'demo( spam)' for a short introduction
and overview of this package.
Help for individual functions is also obtained by adding the
suffix '.spam' to the function name, e.g. 'help( chol.spam)'.
Attaching package: ‘spam’
The following objects are masked from ‘package:base’:
backsolve, forwardsolve
Loading required package: maps
Attaching package: ‘fields’
The following object is masked from ‘package:maps’:
ozone
gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
Attaching package: ‘gdata’
The following object is masked from ‘package:stats’:
nobs
The following object is masked from ‘package:utils’:
object.size
Error in read.table(NBR_FILE, header = TRUE) :
more columns than column names
Execution halted
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 22 Jul 2015 16:28:02 +0530
Hi John.
I have been able to install the packages required for R.
when I run the FSS quilt script,
Rscript ../R_scripts/fss_quilt_plot.R ./out-22apr2011-00to00UTC/ WRF
240000 -valid_time 20110423_000000 -bw
Loading required package: methods
Loading required package: spam
Loading required package: grid
Spam version 1.0-1 (2014-09-09) is loaded.
Type 'help( Spam)' or 'demo( spam)' for a short introduction
and overview of this package.
Help for individual functions is also obtained by adding the
suffix '.spam' to the function name, e.g. 'help( chol.spam)'.
Attaching package: ‘spam’
The following objects are masked from ‘package:base’:
backsolve, forwardsolve
Loading required package: maps
Attaching package: ‘fields’
The following object is masked from ‘package:maps’:
ozone
gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
Attaching package: ‘gdata’
The following object is masked from ‘package:stats’:
nobs
The following object is masked from ‘package:utils’:
object.size
Error in read.table(NBR_FILE, header = TRUE) :
more columns than column names
Execution halted
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$
the NBR_WRF file looks like this.
can U tell me if the usage of scrpt is OK???
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 21 Jul 2015 13:26:11 -0600
>
> Geeta,
>
> Sorry, I'm not able to provide any more help on this issue. For
questions
> about R, I'd suggest posting it to an R users group.
>
> John
>
> On Sun, Jul 12, 2015 at 6:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > hi John.
> > I have been reading about the use of FSS in a few publications.
> > Kindly explain me the difference between "Random and Uniform FSS".
Didnot
> > understand from the publications (Mittermair et al 2011).
> >
> > what's is Random FSS??.
> > 2. If I have questions regarding the running of R esp the quilt
plot of
> > FSS whom should I ask.
> > where I should adrress. I m not familiar with R. I mean what
should be the
> > input file, where it should be kept etc????
> >
> >
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Thu, 18 Jun 2015 13:55:07 -0600
> > >
> > > Geeta,
> > >
> > > These quantities are computed in
> > > met-5.0/src/libcode/vx_statistics/met_stats.cc
> > >
> > > Here are the relevant lines of code:
> > >
> > > 2215 //
> > > 2216 // Compute Asymptotic Fractions Skill Score
> > > 2217 //
> > > 2218 num = 2.0*f_rate*o_rate;
> > > 2219 den = f_rate*f_rate + o_rate*o_rate;
> > > 2220
> > > 2221 if(is_bad_data(f_rate) || is_bad_data(o_rate) ||
is_eq(den,
> > > 0.0)) {
> > > 2222 afss = bad_data_double;
> > > 2223 }
> > > 2224 else {
> > > 2225 afss = num/den;
> > > 2226 }
> > >
> > > 2236 //
> > > 2237 // Compute Uniform Fractions Skill Score
> > > 2238 //
> > > 2239 if(is_bad_data(o_rate)) ufss = bad_data_double;
> > > 2240 else ufss = 0.5 +
o_rate/2.0;
> > >
> > > In here, f_rate is the ratio (0 to 1) of forecast grid squares
where the
> > > event (as defined by your threshold) is occurring. o_rate is
the ratio
> > of
> > > observation grid squares where the event is occurring.
> > >
> > > Take a look at this picture:
> > > http://www.cawcr.gov.au/projects/verification/FSSplot.gif
> > >
> > > This is taken from this website:
> > > http://www.cawcr.gov.au/projects/verification/
> > >
> > > AFSS (asymptotic fractions skill score) is the dotted line in
that
> > picture
> > > and, as computed above, is based on the bias for the forecast
threshold
> > > you've chosen. Mathematically speaking, FSS can never exceed
that AFSS
> > > value.
> > >
> > > UFSS (useful or uniform fraction skill score) is just 0.5 + the
base
> > rate.
> > >
> > > A forecast performs well when FSS > UFSS and, of course, FSS <
AFSS
> > because
> > > it can't exceed that value.
> > >
> > > Hopefully that helps.
> > >
> > > Thanks,
> > > John
> > >
> > >
> > > On Wed, Jun 17, 2015 at 8:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
>
> > > >
> > > > Hi John.
> > > > Kindly help me in getting to know about the UFSS and AFSS. I
was
> > waiting
> > > > for your response.
> > > > I need a little more insight into these ideas.
> > > >
> > > > thanks
> > > > Geeta
> > > >
> > > > From: geeta124 at hotmail.com
> > > > To: met_help at ucar.edu
> > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > Date: Sun, 14 Jun 2015 19:11:21 +0530
> > > >
> > > >
> > > >
> > > >
> > > > thanks John for you clarification.
> > > >
> > > > But I didnt understand the AFSS. Actually, FSS increases
Asymptotically
> > > > with increase in NBRHD width.
> > > > Then what's the Differerence b/w AFSS FSS???? and How is it
(AFSS)
> > > > calculated.
> > > > Kindly explain. . I m not sure abt the AFSS????
> > > >
> > > > 2nd Ques. Is is correct when I say that UFSS is NOT
constant. It
> > varies.
> > > > So for a useful FCST, I will compare if the FSS (unique for 1
NBRHD
> > and 1
> > > > THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD,
> > FCST
> > > > HAS SKILL!!!!!!!!!!.
> > > > Is my understanding correct????.
> > > >
> > > > 3. If U can guide me a little abt the input files Required for
running
> > the
> > > > Rsript for quilt plot, It would be great............
> > > >
> > > > Geeta
> > > >
> > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > From: met_help at ucar.edu
> > > > > To: geeta124 at hotmail.com
> > > > > Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
> > > > >
> > > > > Geeta,
> > > > >
> > > > > Yes, you are correct. UFSS = 0.5 + base_rate/2
> > > > >
> > > > > For lower thresholds, the base rate will be higher, making
UFSS
> > larger.
> > > > As
> > > > > you increase the threshold, UFSS should decrease down to
close to
> > 0.5.
> > > > >
> > > > > Looking at the grid_stat NBRCNT file you sent, you'll notice
that the
> > > > UFSS
> > > > > column = 0.5 + O_RATE column divided by 2.
> > > > >
> > > > > Thanks,
> > > > > John
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT <
> > met_help at ucar.edu>
> > > > > wrote:
> > > > >
> > > > > >
> > > > > > <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > >
> > > > > > Hi John and tara. I have tried to run grid_stat with TRMM
daily
> > files.
> > > > > > hence I extracted the model data for 24 hours starting 00
UTC of 22
> > > > April
> > > > > > 2011.
> > > > > > I could not make the Quilt plot, but just for
understanding I
> > plotted
> > > > data
> > > > > > in EXCEL. (results attatched). I want to know, the
details about
> > > > UFSS. It
> > > > > > is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2)
> > > > where f0
> > > > > > is the observation rate.
> > > > > > Kindly clarify these points. I m waiting.
> > > > > >
> > > > > > Geeta
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Sun, 7 Jun 2015 17:54:25 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi John and Tara. I was waiting for your reply. Kindly
help me
> > further.
> > > > > > ThanksGeeta
> > > > > >
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Wed, 3 Jun 2015 11:59:35 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi Tara,
> > > > > > I was NOT asking about the plotting of the NETCDF data
files
> > > > (observations
> > > > > > n FCST). Those can be plotted with plot_data_plane and
ncview etc.
> > > > > >
> > > > > > But a step further.
> > > > > > I want to know if I can plot the binary data after
thresholding.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Wed, 3 Jun 2015 10:47:30 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > thanks Tara.
> > > > > > Nice to hear from U. I had met you at Verification
workshop last
> > year
> > > > in
> > > > > > New Delhi.
> > > > > > I didnt understand Ur explanation fully.
> > > > > > It would be great if U can elaborate upon it in a little
more
> > detail.
> > > > > >
> > > > > > See I got the output from Grid stat and now I need to
understand
> > what
> > > > is
> > > > > > my FSS for this case??.
> > > > > > U must see my MODEL FCST and Observations. They are so
close but I
> > am
> > > > > > unable to understand why FSS is so low????????.
> > > > > > If U want, I can send my plots (FSS and RF fcst and
observation)
> > and
> > > > > > grid_stat output to you.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > > From: met_help at ucar.edu
> > > > > > > To: geeta124 at hotmail.com
> > > > > > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > > > > > >
> > > > > > > Geeta,
> > > > > > >
> > > > > > > Sorry for the delay in answering your question about FSS
etc.
> > > > Yesterday
> > > > > > was a very busy day here. One of our statisticians put
together the
> > > > > > following answer:
> > > > > > >
> > > > > > > According to the Roberts and Lean (2008) paper, the AFSS
> > (asymptotic
> > > > > > FSS) is a useful descriptor of the bias because it relates
the
> > bias to
> > > > the
> > > > > > spatial accuracy of a forecast and is linked to the
conventional
> > > > frequency
> > > > > > bias, but is less sensitive to small frequencies.
Compared to the
> > > > > > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is
a
> > factor of
> > > > 4,
> > > > > > and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS
> > = 1,
> > > > > > indicating perfect skill over the domain.
> > > > > > >
> > > > > > > I will send more information as I or others pull it
together.
> > > > > > >
> > > > > > > As for plotting the NetCDF files that come from MET. We
> > typically
> > > > use
> > > > > > ncview or NCL for plotting the files, however the plotting
tools
> > Julie
> > > > > > mentioned should also be able to read the files.
> > > > > > >
> > > > > > > Thanks for your patience as I prepare more information
about FSS,
> > > > AFSS
> > > > > > etc.
> > > > > > > Tara
> > > > > > >
> > > > > > >
> > > > > > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > > > > > Thanks John.
> > > > > > > > Also I was wondering why I should get bad FSS /low FSS
for
> > Lowest
> > > > > > > > Neighbourhood and lower threshold????. This looks a
bit
> > confusing.
> > > > > > > > I should get low FSS for Lower neighbourhood and high
> > > > threshold!!!!.
> > > > > > > > OK????
> > > > > > > > I shall be waiting for your reply eagerly.
> > > > > > > >
> > > > > > > > Geeta
> > > > > > > >
> > > > > > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS,
AFSS
> > > > > > > > > From: met_help at ucar.edu
> > > > > > > > > To: geeta124 at hotmail.com
> > > > > > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > > > > > >
> > > > > > > > > Geeta,
> > > > > > > > >
> > > > > > > > > Good questions. Unfortunately, I don't have a good
answer
> > for
> > > > you
> > > > > > > > yet.
> > > > > > > > > I'm headed out of town all next week, but I referred
your
> > > > question
> > > > > > > > to a
> > > > > > > > > scientist and statistician here.
> > > > > > > > >
> > > > > > > > > Hopefully you'll hear back from them about this next
week.
> > > > > > > > >
> > > > > > > > > Thanks,
> > > > > > > > > John
> > > > > > > > >
> > > > > > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > > > > > <met_help at ucar.edu>
> > > > > > > > > wrote:
> > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted
upon.
> > > > > > > > > > Transaction: Ticket created by
geeta124 at hotmail.com
> > > > > > > > > > Queue: met_help
> > > > > > > > > > Subject: FSS, UFSS, AFSS
> > > > > > > > > > Owner: Nobody
> > > > > > > > > > Requestors: geeta124 at hotmail.com
> > > > > > > > > > Status: new
> > > > > > > > > > Ticket <URL:
> > > > > > > >
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Hi John. I have plotted the FSS for various
neighbourhood
> > > > widths
> > > > > > > > starting
> > > > > > > > > > from 1, 3,5,7,9,11,13 and 15 and for various
thresholds
> > > > > > > > (0.1,2.5,7.5, and
> > > > > > > > > > so on.). This is for One case of thunderstorm
event. I have
> > > > taken
> > > > > > > > FSS
> > > > > > > > > > column from my output. But there are three, FSS,
UFSS and
> > > > AFSS. I
> > > > > > > > presume
> > > > > > > > > > that UFSS is the UNIFORM FSS which is the base
line for a
> > good
> > > > > > > > forecast. I
> > > > > > > > > > am sending My plot. (attached). for lower
threshold, and
> > lowest
> > > > > > > > > > neighbourhood, I get lowest FSS which increases
very
> > gradually.
> > > > > > > > But it is
> > > > > > > > > > less as compared to the UFSS (I plotted UFSS for
threshold
> > of
> > > > > > > > 0.1). But
> > > > > > > > > > when I look at my model forecast and observed data
(trmm),
> > > > Plots
> > > > > > > > are
> > > > > > > > > > matching so nicely. I m sending my grid_stat
output file.
> > can
> > > > u
> > > > > > > > help me.
> > > > > > > > > >
> > > > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > >
> > > >
> > >
> >
> >
>
------------------------------------------------
Subject: FSS, UFSS, AFSS
From: Geeta Geeta
Time: Thu Jul 23 00:10:18 2015
Hi John.
I copied the following files
cp grid_stat_240000L_20110423_000000V_ctc.txt CTC_WRF.stat
cp grid_stat_240000L_20110423_000000V_nbrcts.txt NBRCTS_WRF.stat
cp grid_stat_240000L_20110423_000000V_nbrcnt.txt NBRCNT_WRF.stat
then I run the run script,
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$ Rscript
../R_scripts/fss_quilt_plot.R ./out-22apr2011-00to00UTC/ WRF 240000
but it did not run fully.
Loading required package: methods
Loading required package: spam
Loading required package: grid
Spam version 1.0-1 (2014-09-09) is loaded.
Type 'help( Spam)' or 'demo( spam)' for a short introduction
and overview of this package.
Help for individual functions is also obtained by adding the
suffix '.spam' to the function name, e.g. 'help( chol.spam)'.
Attaching package: ‘spam’
The following objects are masked from ‘package:base’:
backsolve, forwardsolve
Loading required package: maps
Attaching package: ‘fields’
The following object is masked from ‘package:maps’:
ozone
gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
Attaching package: ‘gdata’
The following object is masked from ‘package:stats’:
nobs
The following object is masked from ‘package:utils’:
object.size
Read 80 lines from STAT file: ./out-22apr2011-
00to00UTC//NBRCNT_WRF.stat
Read 10 lines from STAT file: ./out-22apr2011-00to00UTC//CTC_WRF.stat
Plotting 80 lines
Warning message:
NAs introduced by coercion
Warning message:
NAs introduced by coercion
Thresholds Found [ 0 ]:
Neighborhood Sizes Found [ 8 ]:
1 9 25 49 81 121 169 225
Error in ctc_scores$FY_OY + ctc_scores$FN_OY :
non-numeric argument to binary operator
Execution halted
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$
can you suggest????
Geeta
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 22 Jul 2015 17:13:15 +0530
Hi John.
I tried,
Rscript ../R_scripts/fss_quilt_plot.R ./out-22apr2011-00to00UTC/ WRF
240000 -valid_time 20110423_000000 -bw
and got the folowing error.
Loading required package: methods
Loading required package: spam
Loading required package: grid
Spam version 1.0-1 (2014-09-09) is loaded.
Type 'help( Spam)' or 'demo( spam)' for a short introduction
and overview of this package.
Help for individual functions is also obtained by adding the
suffix '.spam' to the function name, e.g. 'help( chol.spam)'.
Attaching package: ‘spam’
The following objects are masked from ‘package:base’:
backsolve, forwardsolve
Loading required package: maps
Attaching package: ‘fields’
The following object is masked from ‘package:maps’:
ozone
gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
Attaching package: ‘gdata’
The following object is masked from ‘package:stats’:
nobs
The following object is masked from ‘package:utils’:
object.size
Error in read.table(NBR_FILE, header = TRUE) :
more columns than column names
Execution halted
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$
From: geeta124 at hotmail.com
To: met_help at ucar.edu
Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
Date: Wed, 22 Jul 2015 16:28:02 +0530
Hi John.
I have been able to install the packages required for R.
when I run the FSS quilt script,
Rscript ../R_scripts/fss_quilt_plot.R ./out-22apr2011-00to00UTC/ WRF
240000 -valid_time 20110423_000000 -bw
Loading required package: methods
Loading required package: spam
Loading required package: grid
Spam version 1.0-1 (2014-09-09) is loaded.
Type 'help( Spam)' or 'demo( spam)' for a short introduction
and overview of this package.
Help for individual functions is also obtained by adding the
suffix '.spam' to the function name, e.g. 'help( chol.spam)'.
Attaching package: ‘spam’
The following objects are masked from ‘package:base’:
backsolve, forwardsolve
Loading required package: maps
Attaching package: ‘fields’
The following object is masked from ‘package:maps’:
ozone
gdata: read.xls support for 'XLS' (Excel 97-2004) files ENABLED.
gdata: read.xls support for 'XLSX' (Excel 2007+) files ENABLED.
Attaching package: ‘gdata’
The following object is masked from ‘package:stats’:
nobs
The following object is masked from ‘package:utils’:
object.size
Error in read.table(NBR_FILE, header = TRUE) :
more columns than column names
Execution halted
aditya at agniilap:~/geeta/installs/met/RUN_MET-22apr2011-BNG$
the NBR_WRF file looks like this.
can U tell me if the usage of scrpt is OK???
Geeta
> Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> From: met_help at ucar.edu
> To: geeta124 at hotmail.com
> Date: Tue, 21 Jul 2015 13:26:11 -0600
>
> Geeta,
>
> Sorry, I'm not able to provide any more help on this issue. For
questions
> about R, I'd suggest posting it to an R users group.
>
> John
>
> On Sun, Jul 12, 2015 at 6:59 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> >
> > hi John.
> > I have been reading about the use of FSS in a few publications.
> > Kindly explain me the difference between "Random and Uniform FSS".
Didnot
> > understand from the publications (Mittermair et al 2011).
> >
> > what's is Random FSS??.
> > 2. If I have questions regarding the running of R esp the quilt
plot of
> > FSS whom should I ask.
> > where I should adrress. I m not familiar with R. I mean what
should be the
> > input file, where it should be kept etc????
> >
> >
> >
> > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > From: met_help at ucar.edu
> > > To: geeta124 at hotmail.com
> > > Date: Thu, 18 Jun 2015 13:55:07 -0600
> > >
> > > Geeta,
> > >
> > > These quantities are computed in
> > > met-5.0/src/libcode/vx_statistics/met_stats.cc
> > >
> > > Here are the relevant lines of code:
> > >
> > > 2215 //
> > > 2216 // Compute Asymptotic Fractions Skill Score
> > > 2217 //
> > > 2218 num = 2.0*f_rate*o_rate;
> > > 2219 den = f_rate*f_rate + o_rate*o_rate;
> > > 2220
> > > 2221 if(is_bad_data(f_rate) || is_bad_data(o_rate) ||
is_eq(den,
> > > 0.0)) {
> > > 2222 afss = bad_data_double;
> > > 2223 }
> > > 2224 else {
> > > 2225 afss = num/den;
> > > 2226 }
> > >
> > > 2236 //
> > > 2237 // Compute Uniform Fractions Skill Score
> > > 2238 //
> > > 2239 if(is_bad_data(o_rate)) ufss = bad_data_double;
> > > 2240 else ufss = 0.5 +
o_rate/2.0;
> > >
> > > In here, f_rate is the ratio (0 to 1) of forecast grid squares
where the
> > > event (as defined by your threshold) is occurring. o_rate is
the ratio
> > of
> > > observation grid squares where the event is occurring.
> > >
> > > Take a look at this picture:
> > > http://www.cawcr.gov.au/projects/verification/FSSplot.gif
> > >
> > > This is taken from this website:
> > > http://www.cawcr.gov.au/projects/verification/
> > >
> > > AFSS (asymptotic fractions skill score) is the dotted line in
that
> > picture
> > > and, as computed above, is based on the bias for the forecast
threshold
> > > you've chosen. Mathematically speaking, FSS can never exceed
that AFSS
> > > value.
> > >
> > > UFSS (useful or uniform fraction skill score) is just 0.5 + the
base
> > rate.
> > >
> > > A forecast performs well when FSS > UFSS and, of course, FSS <
AFSS
> > because
> > > it can't exceed that value.
> > >
> > > Hopefully that helps.
> > >
> > > Thanks,
> > > John
> > >
> > >
> > > On Wed, Jun 17, 2015 at 8:56 AM, Geeta Geeta via RT
<met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102
>
> > > >
> > > > Hi John.
> > > > Kindly help me in getting to know about the UFSS and AFSS. I
was
> > waiting
> > > > for your response.
> > > > I need a little more insight into these ideas.
> > > >
> > > > thanks
> > > > Geeta
> > > >
> > > > From: geeta124 at hotmail.com
> > > > To: met_help at ucar.edu
> > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > Date: Sun, 14 Jun 2015 19:11:21 +0530
> > > >
> > > >
> > > >
> > > >
> > > > thanks John for you clarification.
> > > >
> > > > But I didnt understand the AFSS. Actually, FSS increases
Asymptotically
> > > > with increase in NBRHD width.
> > > > Then what's the Differerence b/w AFSS FSS???? and How is it
(AFSS)
> > > > calculated.
> > > > Kindly explain. . I m not sure abt the AFSS????
> > > >
> > > > 2nd Ques. Is is correct when I say that UFSS is NOT
constant. It
> > varies.
> > > > So for a useful FCST, I will compare if the FSS (unique for 1
NBRHD
> > and 1
> > > > THRESHOLD) with that of UFSS. If it's MORE then at that
scale/NBRHD,
> > FCST
> > > > HAS SKILL!!!!!!!!!!.
> > > > Is my understanding correct????.
> > > >
> > > > 3. If U can guide me a little abt the input files Required for
running
> > the
> > > > Rsript for quilt plot, It would be great............
> > > >
> > > > Geeta
> > > >
> > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > From: met_help at ucar.edu
> > > > > To: geeta124 at hotmail.com
> > > > > Date: Fri, 12 Jun 2015 15:06:02 -0600ies.
> > > > >
> > > > > Geeta,
> > > > >
> > > > > Yes, you are correct. UFSS = 0.5 + base_rate/2
> > > > >
> > > > > For lower thresholds, the base rate will be higher, making
UFSS
> > larger.
> > > > As
> > > > > you increase the threshold, UFSS should decrease down to
close to
> > 0.5.
> > > > >
> > > > > Looking at the grid_stat NBRCNT file you sent, you'll notice
that the
> > > > UFSS
> > > > > column = 0.5 + O_RATE column divided by 2.
> > > > >
> > > > > Thanks,
> > > > > John
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On Tue, Jun 9, 2015 at 5:41 AM, Geeta Geeta via RT <
> > met_help at ucar.edu>
> > > > > wrote:
> > > > >
> > > > > >
> > > > > > <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > >
> > > > > > Hi John and tara. I have tried to run grid_stat with TRMM
daily
> > files.
> > > > > > hence I extracted the model data for 24 hours starting 00
UTC of 22
> > > > April
> > > > > > 2011.
> > > > > > I could not make the Quilt plot, but just for
understanding I
> > plotted
> > > > data
> > > > > > in EXCEL. (results attatched). I want to know, the
details about
> > > > UFSS. It
> > > > > > is varying for each threshold and NBRHD SIZE (becuz
UFSS=0.5+f0/2)
> > > > where f0
> > > > > > is the observation rate.
> > > > > > Kindly clarify these points. I m waiting.
> > > > > >
> > > > > > Geeta
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Sun, 7 Jun 2015 17:54:25 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi John and Tara. I was waiting for your reply. Kindly
help me
> > further.
> > > > > > ThanksGeeta
> > > > > >
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Wed, 3 Jun 2015 11:59:35 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > Hi Tara,
> > > > > > I was NOT asking about the plotting of the NETCDF data
files
> > > > (observations
> > > > > > n FCST). Those can be plotted with plot_data_plane and
ncview etc.
> > > > > >
> > > > > > But a step further.
> > > > > > I want to know if I can plot the binary data after
thresholding.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > From: geeta124 at hotmail.com
> > > > > > To: met_help at ucar.edu
> > > > > > Subject: RE: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > Date: Wed, 3 Jun 2015 10:47:30 +0530
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > thanks Tara.
> > > > > > Nice to hear from U. I had met you at Verification
workshop last
> > year
> > > > in
> > > > > > New Delhi.
> > > > > > I didnt understand Ur explanation fully.
> > > > > > It would be great if U can elaborate upon it in a little
more
> > detail.
> > > > > >
> > > > > > See I got the output from Grid stat and now I need to
understand
> > what
> > > > is
> > > > > > my FSS for this case??.
> > > > > > U must see my MODEL FCST and Observations. They are so
close but I
> > am
> > > > > > unable to understand why FSS is so low????????.
> > > > > > If U want, I can send my plots (FSS and RF fcst and
observation)
> > and
> > > > > > grid_stat output to you.
> > > > > >
> > > > > > Geeta
> > > > > >
> > > > > > > Subject: [rt.rap.ucar.edu #72102] FSS, UFSS, AFSS
> > > > > > > From: met_help at ucar.edu
> > > > > > > To: geeta124 at hotmail.com
> > > > > > > Date: Tue, 2 Jun 2015 09:52:38 -0600
> > > > > > >
> > > > > > > Geeta,
> > > > > > >
> > > > > > > Sorry for the delay in answering your question about FSS
etc.
> > > > Yesterday
> > > > > > was a very busy day here. One of our statisticians put
together the
> > > > > > following answer:
> > > > > > >
> > > > > > > According to the Roberts and Lean (2008) paper, the AFSS
> > (asymptotic
> > > > > > FSS) is a useful descriptor of the bias because it relates
the
> > bias to
> > > > the
> > > > > > spatial accuracy of a forecast and is linked to the
conventional
> > > > frequency
> > > > > > bias, but is less sensitive to small frequencies.
Compared to the
> > > > > > frequency bias: AFSS = 0.8 is a factor of 2, AFSS = 0.5 is
a
> > factor of
> > > > 4,
> > > > > > and AFSS = 0.2 is a factor of 10. If there is no bias,
then AFSS
> > = 1,
> > > > > > indicating perfect skill over the domain.
> > > > > > >
> > > > > > > I will send more information as I or others pull it
together.
> > > > > > >
> > > > > > > As for plotting the NetCDF files that come from MET. We
> > typically
> > > > use
> > > > > > ncview or NCL for plotting the files, however the plotting
tools
> > Julie
> > > > > > mentioned should also be able to read the files.
> > > > > > >
> > > > > > > Thanks for your patience as I prepare more information
about FSS,
> > > > AFSS
> > > > > > etc.
> > > > > > > Tara
> > > > > > >
> > > > > > >
> > > > > > > On Sat May 30 04:04:37 2015, geeta124 at hotmail.com wrote:
> > > > > > > > Thanks John.
> > > > > > > > Also I was wondering why I should get bad FSS /low FSS
for
> > Lowest
> > > > > > > > Neighbourhood and lower threshold????. This looks a
bit
> > confusing.
> > > > > > > > I should get low FSS for Lower neighbourhood and high
> > > > threshold!!!!.
> > > > > > > > OK????
> > > > > > > > I shall be waiting for your reply eagerly.
> > > > > > > >
> > > > > > > > Geeta
> > > > > > > >
> > > > > > > > > Subject: Re: [rt.rap.ucar.edu #72102] FSS, UFSS,
AFSS
> > > > > > > > > From: met_help at ucar.edu
> > > > > > > > > To: geeta124 at hotmail.com
> > > > > > > > > Date: Fri, 29 May 2015 16:43:34 -0600
> > > > > > > > >
> > > > > > > > > Geeta,
> > > > > > > > >
> > > > > > > > > Good questions. Unfortunately, I don't have a good
answer
> > for
> > > > you
> > > > > > > > yet.
> > > > > > > > > I'm headed out of town all next week, but I referred
your
> > > > question
> > > > > > > > to a
> > > > > > > > > scientist and statistician here.
> > > > > > > > >
> > > > > > > > > Hopefully you'll hear back from them about this next
week.
> > > > > > > > >
> > > > > > > > > Thanks,
> > > > > > > > > John
> > > > > > > > >
> > > > > > > > > On Thu, May 28, 2015 at 6:26 AM, Geeta Geeta via RT
> > > > > > > > <met_help at ucar.edu>
> > > > > > > > > wrote:
> > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Thu May 28 06:26:14 2015: Request 72102 was acted
upon.
> > > > > > > > > > Transaction: Ticket created by
geeta124 at hotmail.com
> > > > > > > > > > Queue: met_help
> > > > > > > > > > Subject: FSS, UFSS, AFSS
> > > > > > > > > > Owner: Nobody
> > > > > > > > > > Requestors: geeta124 at hotmail.com
> > > > > > > > > > Status: new
> > > > > > > > > > Ticket <URL:
> > > > > > > >
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=72102 >
> > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > Hi John. I have plotted the FSS for various
neighbourhood
> > > > widths
> > > > > > > > starting
> > > > > > > > > > from 1, 3,5,7,9,11,13 and 15 and for various
thresholds
> > > > > > > > (0.1,2.5,7.5, and
> > > > > > > > > > so on.). This is for One case of thunderstorm
event. I have
> > > > taken
> > > > > > > > FSS
> > > > > > > > > > column from my output. But there are three, FSS,
UFSS and
> > > > AFSS. I
> > > > > > > > presume
> > > > > > > > > > that UFSS is the UNIFORM FSS which is the base
line for a
> > good
> > > > > > > > forecast. I
> > > > > > > > > > am sending My plot. (attached). for lower
threshold, and
> > lowest
> > > > > > > > > > neighbourhood, I get lowest FSS which increases
very
> > gradually.
> > > > > > > > But it is
> > > > > > > > > > less as compared to the UFSS (I plotted UFSS for
threshold
> > of
> > > > > > > > 0.1). But
> > > > > > > > > > when I look at my model forecast and observed data
(trmm),
> > > > Plots
> > > > > > > > are
> > > > > > > > > > matching so nicely. I m sending my grid_stat
output file.
> > can
> > > > u
> > > > > > > > help me.
> > > > > > > > > >
> > > > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > >
> > > >
> > >
> >
> >
>
------------------------------------------------
More information about the Met_help
mailing list