[Met_help] [rt.rap.ucar.edu #95398] History for Running Grid_Stat for a specific location
John Halley Gotway via RT
met_help at ucar.edu
Fri Jun 5 15:15:04 MDT 2020
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Dear helpdesk,
Please I would like to know whether I can run the grid_stat function on a single location of a netcdf file using level = [ "(*,*)" ];in the configure file. For example, I want to perform the statistics for Latitude 5 degrees North and Longitude 7 degrees E. Can I set the level in the configure file to level = [ "(5,7)" ];? Thank you.
Best regards,Imoleayo
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: David Fillmore
Time: Fri May 29 12:33:14 2020
Hi Imoleayo -
Let me check this and get back to you shortly,
thanks,
David
On Fri, May 29, 2020 at 5:07 AM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> Fri May 29 05:06:29 2020: Request 95398 was acted upon.
> Transaction: Ticket created by gbodeimoleayo at yahoo.com
> Queue: met_help
> Subject: Running Grid_Stat for a specific location
> Owner: Nobody
> Requestors: gbodeimoleayo at yahoo.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
>
> Dear helpdesk,
> Please I would like to know whether I can run the grid_stat function
on a
> single location of a netcdf file using level = [ "(*,*)"
];in
> the configure file. For example, I want to perform the statistics
for
> Latitude 5 degrees North and Longitude 7 degrees E. Can I set the
level in
> the configure file to level = [ "(5,7)" ];? Thank you.
> Best regards,Imoleayo
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Fri May 29 14:52:06 2020
Dear David,Thanks.I will be expecting to hear from you.
Best regards,Imoleayo On Friday, May 29, 2020, 08:33:17 PM GMT+2,
David Fillmore via RT <met_help at ucar.edu> wrote:
Hi Imoleayo -
Let me check this and get back to you shortly,
thanks,
David
On Fri, May 29, 2020 at 5:07 AM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> Fri May 29 05:06:29 2020: Request 95398 was acted upon.
> Transaction: Ticket created by gbodeimoleayo at yahoo.com
> Queue: met_help
> Subject: Running Grid_Stat for a specific location
> Owner: Nobody
> Requestors: gbodeimoleayo at yahoo.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
>
> Dear helpdesk,
> Please I would like to know whether I can run the grid_stat function
on a
> single location of a netcdf file using level = [ "(*,*)"
];in
> the configure file. For example, I want to perform the statistics
for
> Latitude 5 degrees North and Longitude 7 degrees E. Can I set the
level in
> the configure file to level = [ "(5,7)" ];? Thank you.
> Best regards,Imoleayo
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: John Halley Gotway
Time: Mon Jun 01 22:41:18 2020
Hello Imoleayo,
Sorry for the delay in getting back to you. I see that you're running
the
Grid-Stat tool and would like to know the results at a single
location,
namely 5 degrees latitude north and 7 degrees longitude east.
Before answering that question, let me take a step back to describe
the
purpose of the Grid-Stat tool. It is run to compare one or more fields
of
gridded forecast data to one or more fields of gridded analysis data.
Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
verifying 2-m temperature from a model run which was initialized at
00Z
with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
each of those output times: 00-hour, 06-hour, 12-hour, and so on. For
each
time, Grid-Stat computes spatially aggregated statistics. The user can
configure whatever regions they'd like for the computation of
statistics.
For example, perhaps you want stats computed over the northern
hemisphere
land points. You could define that region (using the gen_vx_mask tool)
and
configure Grid-Stat to compute statistics over all the northern
hemisphere
land grid points.
To answer your question, technically, yes, you could define a very
tiny
masking region that only includes the grid point for the single
lat/lon
location. But that region will only contain 1 matched pair! And MET
can't
compute meaningful statistics from a single matched pair.
I don't know what you're really trying to do here. Perhaps what you'd
really like to do is compute stats at that single location over a time
series of data? If that's the case, then you should run the Series-
Analysis
tool instead, which is described here:
https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
If you can give me a better idea of what you're trying to do, I might
have
better suggestions for how to do it with MET.
Thanks,
John Halley Gotway
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Wed Jun 03 17:34:09 2020
Dear John,
Thanks for this very detailed and comprehensive response.Please kindly
permit met to check the series-analysis and get back to you.
Thank you.
Best regards,Imoleayo On Tuesday, June 2, 2020, 06:41:21 AM GMT+2,
John Halley Gotway via RT <met_help at ucar.edu> wrote:
Hello Imoleayo,
Sorry for the delay in getting back to you. I see that you're running
the
Grid-Stat tool and would like to know the results at a single
location,
namely 5 degrees latitude north and 7 degrees longitude east.
Before answering that question, let me take a step back to describe
the
purpose of the Grid-Stat tool. It is run to compare one or more fields
of
gridded forecast data to one or more fields of gridded analysis data.
Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
verifying 2-m temperature from a model run which was initialized at
00Z
with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
each of those output times: 00-hour, 06-hour, 12-hour, and so on. For
each
time, Grid-Stat computes spatially aggregated statistics. The user can
configure whatever regions they'd like for the computation of
statistics.
For example, perhaps you want stats computed over the northern
hemisphere
land points. You could define that region (using the gen_vx_mask tool)
and
configure Grid-Stat to compute statistics over all the northern
hemisphere
land grid points.
To answer your question, technically, yes, you could define a very
tiny
masking region that only includes the grid point for the single
lat/lon
location. But that region will only contain 1 matched pair! And MET
can't
compute meaningful statistics from a single matched pair.
I don't know what you're really trying to do here. Perhaps what you'd
really like to do is compute stats at that single location over a time
series of data? If that's the case, then you should run the Series-
Analysis
tool instead, which is described here:
https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
If you can give me a better idea of what you're trying to do, I might
have
better suggestions for how to do it with MET.
Thanks,
John Halley Gotway
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Wed Jun 03 18:29:50 2020
Dear John,Thanks for your suggestions.
I am trying to run the series_analysis on my Cheyenne environment but
I get error messages (see attached log.series). Please also find
attached the run script and configure files for examination.
Thank you.
Best regards,Imoleayo
On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT <met_help at ucar.edu> wrote:
Hello Imoleayo,
Sorry for the delay in getting back to you. I see that you're running
the
Grid-Stat tool and would like to know the results at a single
location,
namely 5 degrees latitude north and 7 degrees longitude east.
Before answering that question, let me take a step back to describe
the
purpose of the Grid-Stat tool. It is run to compare one or more fields
of
gridded forecast data to one or more fields of gridded analysis data.
Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
verifying 2-m temperature from a model run which was initialized at
00Z
with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
each of those output times: 00-hour, 06-hour, 12-hour, and so on. For
each
time, Grid-Stat computes spatially aggregated statistics. The user can
configure whatever regions they'd like for the computation of
statistics.
For example, perhaps you want stats computed over the northern
hemisphere
land points. You could define that region (using the gen_vx_mask tool)
and
configure Grid-Stat to compute statistics over all the northern
hemisphere
land grid points.
To answer your question, technically, yes, you could define a very
tiny
masking region that only includes the grid point for the single
lat/lon
location. But that region will only contain 1 matched pair! And MET
can't
compute meaningful statistics from a single matched pair.
I don't know what you're really trying to do here. Perhaps what you'd
really like to do is compute stats at that single location over a time
series of data? If that's the case, then you should run the Series-
Analysis
tool instead, which is described here:
https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
If you can give me a better idea of what you're trying to do, I might
have
better suggestions for how to do it with MET.
Thanks,
John Halley Gotway
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Thu Jun 04 16:17:41 2020
Dear John,
Please I would like to know whether you saw my message?
Thank you.
Best regards,Imoleayo
On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo
<gbodeimoleayo at yahoo.com> wrote:
Dear John,Thanks for your suggestions.
I am trying to run the series_analysis on my Cheyenne environment but
I get error messages (see attached log.series). Please also find
attached the run script and configure files for examination.
Thank you.
Best regards,Imoleayo
On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT <met_help at ucar.edu> wrote:
Hello Imoleayo,
Sorry for the delay in getting back to you. I see that you're running
the
Grid-Stat tool and would like to know the results at a single
location,
namely 5 degrees latitude north and 7 degrees longitude east.
Before answering that question, let me take a step back to describe
the
purpose of the Grid-Stat tool. It is run to compare one or more fields
of
gridded forecast data to one or more fields of gridded analysis data.
Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
verifying 2-m temperature from a model run which was initialized at
00Z
with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
each of those output times: 00-hour, 06-hour, 12-hour, and so on. For
each
time, Grid-Stat computes spatially aggregated statistics. The user can
configure whatever regions they'd like for the computation of
statistics.
For example, perhaps you want stats computed over the northern
hemisphere
land points. You could define that region (using the gen_vx_mask tool)
and
configure Grid-Stat to compute statistics over all the northern
hemisphere
land grid points.
To answer your question, technically, yes, you could define a very
tiny
masking region that only includes the grid point for the single
lat/lon
location. But that region will only contain 1 matched pair! And MET
can't
compute meaningful statistics from a single matched pair.
I don't know what you're really trying to do here. Perhaps what you'd
really like to do is compute stats at that single location over a time
series of data? If that's the case, then you should run the Series-
Analysis
tool instead, which is described here:
https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
If you can give me a better idea of what you're trying to do, I might
have
better suggestions for how to do it with MET.
Thanks,
John Halley Gotway
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: John Halley Gotway
Time: Thu Jun 04 17:54:54 2020
This is a problem with the script you wrote (runSeriesAnalysis.sh).
You
have some trailing spaces after the line continuation marks, and those
wreak havoc in the script.
I tested this manually by create fcst_file_list and obs_file_list
containing those long lists of files and running:
*cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
*/glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc -config
mySeriesAnalysisConfig -v 2*
Note that in my local config file, I updated:
//
// Number of grid points to be processed concurrently. Set smaller to
use
// less memory but increase the number of passes through the data.
//
block_size = 10000;
//
// Ratio of valid matched pairs to compute statistics for a grid point
//
vld_thresh = 0.1;
See attached screenshot showing ncview of the FBAR and OBAR fields.
Thanks,
John
On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Dear John,
> Please I would like to know whether you saw my message?
> Thank you.
> Best regards,Imoleayo
> On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo <
> gbodeimoleayo at yahoo.com> wrote:
>
> Dear John,Thanks for your suggestions.
> I am trying to run the series_analysis on my Cheyenne environment
but I
> get error messages (see attached log.series). Please also find
attached the
> run script and configure files for examination.
> Thank you.
> Best regards,Imoleayo
>
> On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT
> <met_help at ucar.edu> wrote:
>
> Hello Imoleayo,
>
> Sorry for the delay in getting back to you. I see that you're
running the
> Grid-Stat tool and would like to know the results at a single
location,
> namely 5 degrees latitude north and 7 degrees longitude east.
>
> Before answering that question, let me take a step back to describe
the
> purpose of the Grid-Stat tool. It is run to compare one or more
fields of
> gridded forecast data to one or more fields of gridded analysis
data.
> Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
> verifying 2-m temperature from a model run which was initialized at
00Z
> with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
> each of those output times: 00-hour, 06-hour, 12-hour, and so on.
For each
> time, Grid-Stat computes spatially aggregated statistics. The user
can
> configure whatever regions they'd like for the computation of
statistics.
> For example, perhaps you want stats computed over the northern
hemisphere
> land points. You could define that region (using the gen_vx_mask
tool) and
> configure Grid-Stat to compute statistics over all the northern
hemisphere
> land grid points.
>
> To answer your question, technically, yes, you could define a very
tiny
> masking region that only includes the grid point for the single
lat/lon
> location. But that region will only contain 1 matched pair! And MET
can't
> compute meaningful statistics from a single matched pair.
>
> I don't know what you're really trying to do here. Perhaps what
you'd
> really like to do is compute stats at that single location over a
time
> series of data? If that's the case, then you should run the Series-
Analysis
> tool instead, which is described here:
> https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
>
> If you can give me a better idea of what you're trying to do, I
might have
> better suggestions for how to do it with MET.
>
> Thanks,
> John Halley Gotway
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Thu Jun 04 18:32:41 2020
Dear John,
Thanks for pointing out my mistakes and thanks for helping to fix the
script.
Please I would like to know why you choose to change the below portion
in the configure script? //
// Number of grid points to be processed concurrently. Set smaller to
use
// less memory but increase the number of passes through the data.
//
block_size = 10000;
//
// Ratio of valid matched pairs to compute statistics for a grid point
//
vld_thresh = 0.1;.Why did you use 10000 and 0.1?
Thank you.
Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM GMT+2,
John Halley Gotway via RT <met_help at ucar.edu> wrote:
This is a problem with the script you wrote (runSeriesAnalysis.sh).
You
have some trailing spaces after the line continuation marks, and those
wreak havoc in the script.
I tested this manually by create fcst_file_list and obs_file_list
containing those long lists of files and running:
*cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
*/glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc -config
mySeriesAnalysisConfig -v 2*
Note that in my local config file, I updated:
//
// Number of grid points to be processed concurrently. Set smaller to
use
// less memory but increase the number of passes through the data.
//
block_size = 10000;
//
// Ratio of valid matched pairs to compute statistics for a grid point
//
vld_thresh = 0.1;
See attached screenshot showing ncview of the FBAR and OBAR fields.
Thanks,
John
On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Dear John,
> Please I would like to know whether you saw my message?
> Thank you.
> Best regards,Imoleayo
> On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo <
> gbodeimoleayo at yahoo.com> wrote:
>
> Dear John,Thanks for your suggestions.
> I am trying to run the series_analysis on my Cheyenne environment
but I
> get error messages (see attached log.series). Please also find
attached the
> run script and configure files for examination.
> Thank you.
> Best regards,Imoleayo
>
> On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT
> <met_help at ucar.edu> wrote:
>
> Hello Imoleayo,
>
> Sorry for the delay in getting back to you. I see that you're
running the
> Grid-Stat tool and would like to know the results at a single
location,
> namely 5 degrees latitude north and 7 degrees longitude east.
>
> Before answering that question, let me take a step back to describe
the
> purpose of the Grid-Stat tool. It is run to compare one or more
fields of
> gridded forecast data to one or more fields of gridded analysis
data.
> Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
> verifying 2-m temperature from a model run which was initialized at
00Z
> with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
> each of those output times: 00-hour, 06-hour, 12-hour, and so on.
For each
> time, Grid-Stat computes spatially aggregated statistics. The user
can
> configure whatever regions they'd like for the computation of
statistics.
> For example, perhaps you want stats computed over the northern
hemisphere
> land points. You could define that region (using the gen_vx_mask
tool) and
> configure Grid-Stat to compute statistics over all the northern
hemisphere
> land grid points.
>
> To answer your question, technically, yes, you could define a very
tiny
> masking region that only includes the grid point for the single
lat/lon
> location. But that region will only contain 1 matched pair! And MET
can't
> compute meaningful statistics from a single matched pair.
>
> I don't know what you're really trying to do here. Perhaps what
you'd
> really like to do is compute stats at that single location over a
time
> series of data? If that's the case, then you should run the Series-
Analysis
> tool instead, which is described here:
> https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
>
> If you can give me a better idea of what you're trying to do, I
might have
> better suggestions for how to do it with MET.
>
> Thanks,
> John Halley Gotway
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Thu Jun 04 18:37:59 2020
Dear John,
Again, I specified three thresholds of precipitation but I can't find
that in the netcdf output.
//
// Forecast and observation fields to be verified
//
fcst = {
field = [
{
name = "pr";
level = "(0,*,*)";
cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
}
];
}
Also, the min and max values of CSI, GSS and some other statistics are
the same. Please what could be the cause?
Thank you.
Best regards,Imoleayo
On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo
<gbodeimoleayo at yahoo.com> wrote:
Dear John,
Thanks for pointing out my mistakes and thanks for helping to fix the
script.
Please I would like to know why you choose to change the below portion
in the configure script? //
// Number of grid points to be processed concurrently. Set smaller to
use
// less memory but increase the number of passes through the data.
//
block_size = 10000;
//
// Ratio of valid matched pairs to compute statistics for a grid point
//
vld_thresh = 0.1;.Why did you use 10000 and 0.1?
Thank you.
Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM GMT+2,
John Halley Gotway via RT <met_help at ucar.edu> wrote:
This is a problem with the script you wrote (runSeriesAnalysis.sh).
You
have some trailing spaces after the line continuation marks, and those
wreak havoc in the script.
I tested this manually by create fcst_file_list and obs_file_list
containing those long lists of files and running:
*cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
*/glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc -config
mySeriesAnalysisConfig -v 2*
Note that in my local config file, I updated:
//
// Number of grid points to be processed concurrently. Set smaller to
use
// less memory but increase the number of passes through the data.
//
block_size = 10000;
//
// Ratio of valid matched pairs to compute statistics for a grid point
//
vld_thresh = 0.1;
See attached screenshot showing ncview of the FBAR and OBAR fields.
Thanks,
John
On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Dear John,
> Please I would like to know whether you saw my message?
> Thank you.
> Best regards,Imoleayo
> On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo <
> gbodeimoleayo at yahoo.com> wrote:
>
> Dear John,Thanks for your suggestions.
> I am trying to run the series_analysis on my Cheyenne environment
but I
> get error messages (see attached log.series). Please also find
attached the
> run script and configure files for examination.
> Thank you.
> Best regards,Imoleayo
>
> On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT
> <met_help at ucar.edu> wrote:
>
> Hello Imoleayo,
>
> Sorry for the delay in getting back to you. I see that you're
running the
> Grid-Stat tool and would like to know the results at a single
location,
> namely 5 degrees latitude north and 7 degrees longitude east.
>
> Before answering that question, let me take a step back to describe
the
> purpose of the Grid-Stat tool. It is run to compare one or more
fields of
> gridded forecast data to one or more fields of gridded analysis
data.
> Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
> verifying 2-m temperature from a model run which was initialized at
00Z
> with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times to
verify
> each of those output times: 00-hour, 06-hour, 12-hour, and so on.
For each
> time, Grid-Stat computes spatially aggregated statistics. The user
can
> configure whatever regions they'd like for the computation of
statistics.
> For example, perhaps you want stats computed over the northern
hemisphere
> land points. You could define that region (using the gen_vx_mask
tool) and
> configure Grid-Stat to compute statistics over all the northern
hemisphere
> land grid points.
>
> To answer your question, technically, yes, you could define a very
tiny
> masking region that only includes the grid point for the single
lat/lon
> location. But that region will only contain 1 matched pair! And MET
can't
> compute meaningful statistics from a single matched pair.
>
> I don't know what you're really trying to do here. Perhaps what
you'd
> really like to do is compute stats at that single location over a
time
> series of data? If that's the case, then you should run the Series-
Analysis
> tool instead, which is described here:
> https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
>
> If you can give me a better idea of what you're trying to do, I
might have
> better suggestions for how to do it with MET.
>
> Thanks,
> John Halley Gotway
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: John Halley Gotway
Time: Fri Jun 05 12:44:43 2020
Imoleayo,
OK, I see what's going on here, and I realize why it's confusing.
To make it work, move the cat_thresh setting out of the "field"
dictionary
entry and up to the top-level of context, as I've done in the attached
version of your config file.
Looking in the source code (
https://github.com/NCAR/MET/blob/master_v9.0/met/src/tools/core/series_analysis/series_analysis_conf_info.cc)
at line 304, I see that we're parsing cat_thresh from the "fcst" and
"obs"
dictionary level of context. That means, if it's defined at that
level, or
higher, it'll be applied. In the attached file, I defined it at the
global
level of context, so that'll work.
And now, why is it set up this way? Series-Analysis can only be run on
a
single series at a time? That's usually a time-series, where you
define a
single field and pull it from multiple input files for different
times.
However, you could also define the series as multiple vertical levels
from
the same input file... in which case you'd have multiple fields
listed.
However, you can only define a single threshold for the entire series.
The
tool doesn't support different thresholds for each field entry.
I realize that this is confusing since it differs slightly from the
logic
in Point-Stat and Grid-Stat, but hopefully it at least makes sense
now.
Rerunning with these settings produces the following output variables,
where the threshold applied is included in the variable name:
*> ncdump -h seriesAnalysisOutWRF.nc | grep float float lat(lat) ;
float
lon(lon) ; float series_fho_F_RATE_gt0.0(lat, lon) ; float
series_fho_O_RATE_gt0.0(lat, lon) ; float series_ctc_FY_OY_gt0.0(lat,
lon)
; float series_ctc_FN_ON_gt0.0(lat, lon) ; float
series_cts_CSI_gt0.0(lat,
lon) ; float series_cts_GSS_gt0.0(lat, lon) ; float
series_fho_F_RATE_ge5.0(lat, lon) ; float series_fho_O_RATE_ge5.0(lat,
lon)
; float series_ctc_FY_OY_ge5.0(lat, lon) ; float
series_ctc_FN_ON_ge5.0(lat, lon) ; float series_cts_CSI_ge5.0(lat,
lon) ;
float series_cts_GSS_ge5.0(lat, lon) ; float
series_fho_F_RATE_ge7.0(lat,
lon) ; float series_fho_O_RATE_ge7.0(lat, lon) ; float
series_ctc_FY_OY_ge7.0(lat, lon) ; float series_ctc_FN_ON_ge7.0(lat,
lon) ;
float series_cts_CSI_ge7.0(lat, lon) ; float series_cts_GSS_ge7.0(lat,
lon)
; float series_fho_F_RATE_ge10.0(lat, lon) ; float
series_fho_O_RATE_ge10.0(lat, lon) ; float
series_ctc_FY_OY_ge10.0(lat,
lon) ; float series_ctc_FN_ON_ge10.0(lat, lon) ; float
series_cts_CSI_ge10.0(lat, lon) ; float series_cts_GSS_ge10.0(lat,
lon) ;
float series_cnt_TOTAL(lat, lon) ; float series_cnt_RMSE(lat, lon) ;
float
series_cnt_FBAR(lat, lon) ;*
*float series_cnt_OBAR(lat, lon) ;*
Also, note that increasing the "block_size" setting would make this
run
faster... as long as you have sufficient memory available.
Thanks,
John
On Thu, Jun 4, 2020 at 6:38 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Dear John,
> Again, I specified three thresholds of precipitation but I can't
find that
> in the netcdf output.
> //
>
> // Forecast and observation fields to be verified
>
> //
>
> fcst = {
>
>
>
>
> field = [
>
> {
>
> name = "pr";
>
> level = "(0,*,*)";
>
> cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
>
> }
>
> ];
>
>
>
>
> }
>
> Also, the min and max values of CSI, GSS and some other statistics
are the
> same. Please what could be the cause?
> Thank you.
> Best regards,Imoleayo
> On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo <
> gbodeimoleayo at yahoo.com> wrote:
>
> Dear John,
> Thanks for pointing out my mistakes and thanks for helping to fix
the
> script.
> Please I would like to know why you choose to change the below
portion in
> the configure script? //
> // Number of grid points to be processed concurrently. Set smaller
to use
> // less memory but increase the number of passes through the data.
> //
> block_size = 10000;
>
> //
> // Ratio of valid matched pairs to compute statistics for a grid
point
> //
> vld_thresh = 0.1;.Why did you use 10000 and 0.1?
> Thank you.
> Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM GMT+2,
John
> Halley Gotway via RT <met_help at ucar.edu> wrote:
>
> This is a problem with the script you wrote (runSeriesAnalysis.sh).
You
> have some trailing spaces after the line continuation marks, and
those
> wreak havoc in the script.
>
> I tested this manually by create fcst_file_list and obs_file_list
> containing those long lists of files and running:
>
> *cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
> */glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
> fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc
-config
> mySeriesAnalysisConfig -v 2*
>
> Note that in my local config file, I updated:
>
> //
> // Number of grid points to be processed concurrently. Set smaller
to use
> // less memory but increase the number of passes through the data.
> //
> block_size = 10000;
>
> //
> // Ratio of valid matched pairs to compute statistics for a grid
point
> //
> vld_thresh = 0.1;
>
> See attached screenshot showing ncview of the FBAR and OBAR fields.
>
> Thanks,
> John
>
> On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
> >
> > Dear John,
> > Please I would like to know whether you saw my message?
> > Thank you.
> > Best regards,Imoleayo
> > On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo <
> > gbodeimoleayo at yahoo.com> wrote:
> >
> > Dear John,Thanks for your suggestions.
> > I am trying to run the series_analysis on my Cheyenne environment
but I
> > get error messages (see attached log.series). Please also find
attached
> the
> > run script and configure files for examination.
> > Thank you.
> > Best regards,Imoleayo
> >
> > On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT
> > <met_help at ucar.edu> wrote:
> >
> > Hello Imoleayo,
> >
> > Sorry for the delay in getting back to you. I see that you're
running the
> > Grid-Stat tool and would like to know the results at a single
location,
> > namely 5 degrees latitude north and 7 degrees longitude east.
> >
> > Before answering that question, let me take a step back to
describe the
> > purpose of the Grid-Stat tool. It is run to compare one or more
fields of
> > gridded forecast data to one or more fields of gridded analysis
data.
> > Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
> > verifying 2-m temperature from a model run which was initialized
at 00Z
> > with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times
to
> verify
> > each of those output times: 00-hour, 06-hour, 12-hour, and so on.
For
> each
> > time, Grid-Stat computes spatially aggregated statistics. The user
can
> > configure whatever regions they'd like for the computation of
statistics.
> > For example, perhaps you want stats computed over the northern
hemisphere
> > land points. You could define that region (using the gen_vx_mask
tool)
> and
> > configure Grid-Stat to compute statistics over all the northern
> hemisphere
> > land grid points.
> >
> > To answer your question, technically, yes, you could define a very
tiny
> > masking region that only includes the grid point for the single
lat/lon
> > location. But that region will only contain 1 matched pair! And
MET can't
> > compute meaningful statistics from a single matched pair.
> >
> > I don't know what you're really trying to do here. Perhaps what
you'd
> > really like to do is compute stats at that single location over a
time
> > series of data? If that's the case, then you should run the
> Series-Analysis
> > tool instead, which is described here:
> > https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
> >
> > If you can give me a better idea of what you're trying to do, I
might
> have
> > better suggestions for how to do it with MET.
> >
> > Thanks,
> > John Halley Gotway
> >
> >
> >
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Fri Jun 05 14:47:25 2020
Great! Thanks John.
I am happy to now find values for the CSI and GSS. Thanks for your
support.
Please another thing, I want to compute Fractional Skill Score in Met
Tool. Is this possible?
Thank you.
Best regards,Imoleayo
On Friday, June 5, 2020, 08:44:48 PM GMT+2, John Halley Gotway via
RT <met_help at ucar.edu> wrote:
Imoleayo,
OK, I see what's going on here, and I realize why it's confusing.
To make it work, move the cat_thresh setting out of the "field"
dictionary
entry and up to the top-level of context, as I've done in the attached
version of your config file.
Looking in the source code (
https://github.com/NCAR/MET/blob/master_v9.0/met/src/tools/core/series_analysis/series_analysis_conf_info.cc)
at line 304, I see that we're parsing cat_thresh from the "fcst" and
"obs"
dictionary level of context. That means, if it's defined at that
level, or
higher, it'll be applied. In the attached file, I defined it at the
global
level of context, so that'll work.
And now, why is it set up this way? Series-Analysis can only be run on
a
single series at a time? That's usually a time-series, where you
define a
single field and pull it from multiple input files for different
times.
However, you could also define the series as multiple vertical levels
from
the same input file... in which case you'd have multiple fields
listed.
However, you can only define a single threshold for the entire series.
The
tool doesn't support different thresholds for each field entry.
I realize that this is confusing since it differs slightly from the
logic
in Point-Stat and Grid-Stat, but hopefully it at least makes sense
now.
Rerunning with these settings produces the following output variables,
where the threshold applied is included in the variable name:
*> ncdump -h seriesAnalysisOutWRF.nc | grep float float lat(lat) ;
float
lon(lon) ; float series_fho_F_RATE_gt0.0(lat, lon) ; float
series_fho_O_RATE_gt0.0(lat, lon) ; float series_ctc_FY_OY_gt0.0(lat,
lon)
; float series_ctc_FN_ON_gt0.0(lat, lon) ; float
series_cts_CSI_gt0.0(lat,
lon) ; float series_cts_GSS_gt0.0(lat, lon) ; float
series_fho_F_RATE_ge5.0(lat, lon) ; float series_fho_O_RATE_ge5.0(lat,
lon)
; float series_ctc_FY_OY_ge5.0(lat, lon) ; float
series_ctc_FN_ON_ge5.0(lat, lon) ; float series_cts_CSI_ge5.0(lat,
lon) ;
float series_cts_GSS_ge5.0(lat, lon) ; float
series_fho_F_RATE_ge7.0(lat,
lon) ; float series_fho_O_RATE_ge7.0(lat, lon) ; float
series_ctc_FY_OY_ge7.0(lat, lon) ; float series_ctc_FN_ON_ge7.0(lat,
lon) ;
float series_cts_CSI_ge7.0(lat, lon) ; float series_cts_GSS_ge7.0(lat,
lon)
; float series_fho_F_RATE_ge10.0(lat, lon) ; float
series_fho_O_RATE_ge10.0(lat, lon) ; float
series_ctc_FY_OY_ge10.0(lat,
lon) ; float series_ctc_FN_ON_ge10.0(lat, lon) ; float
series_cts_CSI_ge10.0(lat, lon) ; float series_cts_GSS_ge10.0(lat,
lon) ;
float series_cnt_TOTAL(lat, lon) ; float series_cnt_RMSE(lat, lon) ;
float
series_cnt_FBAR(lat, lon) ;*
*float series_cnt_OBAR(lat, lon) ;*
Also, note that increasing the "block_size" setting would make this
run
faster... as long as you have sufficient memory available.
Thanks,
John
On Thu, Jun 4, 2020 at 6:38 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Dear John,
> Again, I specified three thresholds of precipitation but I can't
find that
> in the netcdf output.
> //
>
> // Forecast and observation fields to be verified
>
> //
>
> fcst = {
>
>
>
>
> field = [
>
> {
>
> name = "pr";
>
> level = "(0,*,*)";
>
> cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
>
> }
>
> ];
>
>
>
>
> }
>
> Also, the min and max values of CSI, GSS and some other statistics
are the
> same. Please what could be the cause?
> Thank you.
> Best regards,Imoleayo
> On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo <
> gbodeimoleayo at yahoo.com> wrote:
>
> Dear John,
> Thanks for pointing out my mistakes and thanks for helping to fix
the
> script.
> Please I would like to know why you choose to change the below
portion in
> the configure script? //
> // Number of grid points to be processed concurrently. Set smaller
to use
> // less memory but increase the number of passes through the data.
> //
> block_size = 10000;
>
> //
> // Ratio of valid matched pairs to compute statistics for a grid
point
> //
> vld_thresh = 0.1;.Why did you use 10000 and 0.1?
> Thank you.
> Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM GMT+2,
John
> Halley Gotway via RT <met_help at ucar.edu> wrote:
>
> This is a problem with the script you wrote (runSeriesAnalysis.sh).
You
> have some trailing spaces after the line continuation marks, and
those
> wreak havoc in the script.
>
> I tested this manually by create fcst_file_list and obs_file_list
> containing those long lists of files and running:
>
> *cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
> */glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
> fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc
-config
> mySeriesAnalysisConfig -v 2*
>
> Note that in my local config file, I updated:
>
> //
> // Number of grid points to be processed concurrently. Set smaller
to use
> // less memory but increase the number of passes through the data.
> //
> block_size = 10000;
>
> //
> // Ratio of valid matched pairs to compute statistics for a grid
point
> //
> vld_thresh = 0.1;
>
> See attached screenshot showing ncview of the FBAR and OBAR fields.
>
> Thanks,
> John
>
> On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
> >
> > Dear John,
> > Please I would like to know whether you saw my message?
> > Thank you.
> > Best regards,Imoleayo
> > On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo <
> > gbodeimoleayo at yahoo.com> wrote:
> >
> > Dear John,Thanks for your suggestions.
> > I am trying to run the series_analysis on my Cheyenne environment
but I
> > get error messages (see attached log.series). Please also find
attached
> the
> > run script and configure files for examination.
> > Thank you.
> > Best regards,Imoleayo
> >
> > On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT
> > <met_help at ucar.edu> wrote:
> >
> > Hello Imoleayo,
> >
> > Sorry for the delay in getting back to you. I see that you're
running the
> > Grid-Stat tool and would like to know the results at a single
location,
> > namely 5 degrees latitude north and 7 degrees longitude east.
> >
> > Before answering that question, let me take a step back to
describe the
> > purpose of the Grid-Stat tool. It is run to compare one or more
fields of
> > gridded forecast data to one or more fields of gridded analysis
data.
> > Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
> > verifying 2-m temperature from a model run which was initialized
at 00Z
> > with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times
to
> verify
> > each of those output times: 00-hour, 06-hour, 12-hour, and so on.
For
> each
> > time, Grid-Stat computes spatially aggregated statistics. The user
can
> > configure whatever regions they'd like for the computation of
statistics.
> > For example, perhaps you want stats computed over the northern
hemisphere
> > land points. You could define that region (using the gen_vx_mask
tool)
> and
> > configure Grid-Stat to compute statistics over all the northern
> hemisphere
> > land grid points.
> >
> > To answer your question, technically, yes, you could define a very
tiny
> > masking region that only includes the grid point for the single
lat/lon
> > location. But that region will only contain 1 matched pair! And
MET can't
> > compute meaningful statistics from a single matched pair.
> >
> > I don't know what you're really trying to do here. Perhaps what
you'd
> > really like to do is compute stats at that single location over a
time
> > series of data? If that's the case, then you should run the
> Series-Analysis
> > tool instead, which is described here:
> > https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
> >
> > If you can give me a better idea of what you're trying to do, I
might
> have
> > better suggestions for how to do it with MET.
> >
> > Thanks,
> > John Halley Gotway
> >
> >
> >
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Fri Jun 05 14:50:36 2020
Hi John,Sorry, mean to write that I want to compute Fractional Skill
Score in Series_Analysis. Is this possible?
Thanks.
Best regards,Imoleayo
On Friday, June 5, 2020, 10:47:13 PM GMT+2, gbode Imoleayo
<gbodeimoleayo at yahoo.com> wrote:
Great! Thanks John.
I am happy to now find values for the CSI and GSS. Thanks for your
support.
Please another thing, I want to compute Fractional Skill Score in Met
Tool. Is this possible?
Thank you.
Best regards,Imoleayo
On Friday, June 5, 2020, 08:44:48 PM GMT+2, John Halley Gotway via
RT <met_help at ucar.edu> wrote:
Imoleayo,
OK, I see what's going on here, and I realize why it's confusing.
To make it work, move the cat_thresh setting out of the "field"
dictionary
entry and up to the top-level of context, as I've done in the attached
version of your config file.
Looking in the source code (
https://github.com/NCAR/MET/blob/master_v9.0/met/src/tools/core/series_analysis/series_analysis_conf_info.cc)
at line 304, I see that we're parsing cat_thresh from the "fcst" and
"obs"
dictionary level of context. That means, if it's defined at that
level, or
higher, it'll be applied. In the attached file, I defined it at the
global
level of context, so that'll work.
And now, why is it set up this way? Series-Analysis can only be run on
a
single series at a time? That's usually a time-series, where you
define a
single field and pull it from multiple input files for different
times.
However, you could also define the series as multiple vertical levels
from
the same input file... in which case you'd have multiple fields
listed.
However, you can only define a single threshold for the entire series.
The
tool doesn't support different thresholds for each field entry.
I realize that this is confusing since it differs slightly from the
logic
in Point-Stat and Grid-Stat, but hopefully it at least makes sense
now.
Rerunning with these settings produces the following output variables,
where the threshold applied is included in the variable name:
*> ncdump -h seriesAnalysisOutWRF.nc | grep float float lat(lat) ;
float
lon(lon) ; float series_fho_F_RATE_gt0.0(lat, lon) ; float
series_fho_O_RATE_gt0.0(lat, lon) ; float series_ctc_FY_OY_gt0.0(lat,
lon)
; float series_ctc_FN_ON_gt0.0(lat, lon) ; float
series_cts_CSI_gt0.0(lat,
lon) ; float series_cts_GSS_gt0.0(lat, lon) ; float
series_fho_F_RATE_ge5.0(lat, lon) ; float series_fho_O_RATE_ge5.0(lat,
lon)
; float series_ctc_FY_OY_ge5.0(lat, lon) ; float
series_ctc_FN_ON_ge5.0(lat, lon) ; float series_cts_CSI_ge5.0(lat,
lon) ;
float series_cts_GSS_ge5.0(lat, lon) ; float
series_fho_F_RATE_ge7.0(lat,
lon) ; float series_fho_O_RATE_ge7.0(lat, lon) ; float
series_ctc_FY_OY_ge7.0(lat, lon) ; float series_ctc_FN_ON_ge7.0(lat,
lon) ;
float series_cts_CSI_ge7.0(lat, lon) ; float series_cts_GSS_ge7.0(lat,
lon)
; float series_fho_F_RATE_ge10.0(lat, lon) ; float
series_fho_O_RATE_ge10.0(lat, lon) ; float
series_ctc_FY_OY_ge10.0(lat,
lon) ; float series_ctc_FN_ON_ge10.0(lat, lon) ; float
series_cts_CSI_ge10.0(lat, lon) ; float series_cts_GSS_ge10.0(lat,
lon) ;
float series_cnt_TOTAL(lat, lon) ; float series_cnt_RMSE(lat, lon) ;
float
series_cnt_FBAR(lat, lon) ;*
*float series_cnt_OBAR(lat, lon) ;*
Also, note that increasing the "block_size" setting would make this
run
faster... as long as you have sufficient memory available.
Thanks,
John
On Thu, Jun 4, 2020 at 6:38 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Dear John,
> Again, I specified three thresholds of precipitation but I can't
find that
> in the netcdf output.
> //
>
> // Forecast and observation fields to be verified
>
> //
>
> fcst = {
>
>
>
>
> field = [
>
> {
>
> name = "pr";
>
> level = "(0,*,*)";
>
> cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
>
> }
>
> ];
>
>
>
>
> }
>
> Also, the min and max values of CSI, GSS and some other statistics
are the
> same. Please what could be the cause?
> Thank you.
> Best regards,Imoleayo
> On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo <
> gbodeimoleayo at yahoo.com> wrote:
>
> Dear John,
> Thanks for pointing out my mistakes and thanks for helping to fix
the
> script.
> Please I would like to know why you choose to change the below
portion in
> the configure script? //
> // Number of grid points to be processed concurrently. Set smaller
to use
> // less memory but increase the number of passes through the data.
> //
> block_size = 10000;
>
> //
> // Ratio of valid matched pairs to compute statistics for a grid
point
> //
> vld_thresh = 0.1;.Why did you use 10000 and 0.1?
> Thank you.
> Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM GMT+2,
John
> Halley Gotway via RT <met_help at ucar.edu> wrote:
>
> This is a problem with the script you wrote (runSeriesAnalysis.sh).
You
> have some trailing spaces after the line continuation marks, and
those
> wreak havoc in the script.
>
> I tested this manually by create fcst_file_list and obs_file_list
> containing those long lists of files and running:
>
> *cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
> */glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
> fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc
-config
> mySeriesAnalysisConfig -v 2*
>
> Note that in my local config file, I updated:
>
> //
> // Number of grid points to be processed concurrently. Set smaller
to use
> // less memory but increase the number of passes through the data.
> //
> block_size = 10000;
>
> //
> // Ratio of valid matched pairs to compute statistics for a grid
point
> //
> vld_thresh = 0.1;
>
> See attached screenshot showing ncview of the FBAR and OBAR fields.
>
> Thanks,
> John
>
> On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
> >
> > Dear John,
> > Please I would like to know whether you saw my message?
> > Thank you.
> > Best regards,Imoleayo
> > On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo <
> > gbodeimoleayo at yahoo.com> wrote:
> >
> > Dear John,Thanks for your suggestions.
> > I am trying to run the series_analysis on my Cheyenne environment
but I
> > get error messages (see attached log.series). Please also find
attached
> the
> > run script and configure files for examination.
> > Thank you.
> > Best regards,Imoleayo
> >
> > On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley Gotway
via RT
> > <met_help at ucar.edu> wrote:
> >
> > Hello Imoleayo,
> >
> > Sorry for the delay in getting back to you. I see that you're
running the
> > Grid-Stat tool and would like to know the results at a single
location,
> > namely 5 degrees latitude north and 7 degrees longitude east.
> >
> > Before answering that question, let me take a step back to
describe the
> > purpose of the Grid-Stat tool. It is run to compare one or more
fields of
> > gridded forecast data to one or more fields of gridded analysis
data.
> > Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's say
you're
> > verifying 2-m temperature from a model run which was initialized
at 00Z
> > with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times
to
> verify
> > each of those output times: 00-hour, 06-hour, 12-hour, and so on.
For
> each
> > time, Grid-Stat computes spatially aggregated statistics. The user
can
> > configure whatever regions they'd like for the computation of
statistics.
> > For example, perhaps you want stats computed over the northern
hemisphere
> > land points. You could define that region (using the gen_vx_mask
tool)
> and
> > configure Grid-Stat to compute statistics over all the northern
> hemisphere
> > land grid points.
> >
> > To answer your question, technically, yes, you could define a very
tiny
> > masking region that only includes the grid point for the single
lat/lon
> > location. But that region will only contain 1 matched pair! And
MET can't
> > compute meaningful statistics from a single matched pair.
> >
> > I don't know what you're really trying to do here. Perhaps what
you'd
> > really like to do is compute stats at that single location over a
time
> > series of data? If that's the case, then you should run the
> Series-Analysis
> > tool instead, which is described here:
> > https://dtcenter.org/met-online-tutorial-metv8-0/series-analysis
> >
> > If you can give me a better idea of what you're trying to do, I
might
> have
> > better suggestions for how to do it with MET.
> >
> > Thanks,
> > John Halley Gotway
> >
> >
> >
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: John Halley Gotway
Time: Fri Jun 05 14:56:51 2020
Yes, FSS is computed by the Grid-Stat tool.
John
On Fri, Jun 5, 2020 at 2:47 PM gbode Imoleayo via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>
> Great! Thanks John.
> I am happy to now find values for the CSI and GSS. Thanks for your
support.
> Please another thing, I want to compute Fractional Skill Score in
Met
> Tool. Is this possible?
> Thank you.
> Best regards,Imoleayo
> On Friday, June 5, 2020, 08:44:48 PM GMT+2, John Halley Gotway
via RT <
> met_help at ucar.edu> wrote:
>
> Imoleayo,
>
> OK, I see what's going on here, and I realize why it's confusing.
>
> To make it work, move the cat_thresh setting out of the "field"
dictionary
> entry and up to the top-level of context, as I've done in the
attached
> version of your config file.
>
> Looking in the source code (
>
>
https://github.com/NCAR/MET/blob/master_v9.0/met/src/tools/core/series_analysis/series_analysis_conf_info.cc
> )
> at line 304, I see that we're parsing cat_thresh from the "fcst" and
"obs"
> dictionary level of context. That means, if it's defined at that
level, or
> higher, it'll be applied. In the attached file, I defined it at the
global
> level of context, so that'll work.
>
> And now, why is it set up this way? Series-Analysis can only be run
on a
> single series at a time? That's usually a time-series, where you
define a
> single field and pull it from multiple input files for different
times.
> However, you could also define the series as multiple vertical
levels from
> the same input file... in which case you'd have multiple fields
listed.
>
> However, you can only define a single threshold for the entire
series. The
> tool doesn't support different thresholds for each field entry.
>
> I realize that this is confusing since it differs slightly from the
logic
> in Point-Stat and Grid-Stat, but hopefully it at least makes sense
now.
>
> Rerunning with these settings produces the following output
variables,
> where the threshold applied is included in the variable name:
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> *> ncdump -h seriesAnalysisOutWRF.nc | grep float float lat(lat) ;
float
> lon(lon) ; float series_fho_F_RATE_gt0.0(lat, lon) ; float
> series_fho_O_RATE_gt0.0(lat, lon) ; float
series_ctc_FY_OY_gt0.0(lat, lon)
> ; float series_ctc_FN_ON_gt0.0(lat, lon) ; float
series_cts_CSI_gt0.0(lat,
> lon) ; float series_cts_GSS_gt0.0(lat, lon) ; float
> series_fho_F_RATE_ge5.0(lat, lon) ; float
series_fho_O_RATE_ge5.0(lat, lon)
> ; float series_ctc_FY_OY_ge5.0(lat, lon) ; float
> series_ctc_FN_ON_ge5.0(lat, lon) ; float series_cts_CSI_ge5.0(lat,
lon) ;
> float series_cts_GSS_ge5.0(lat, lon) ; float
series_fho_F_RATE_ge7.0(lat,
> lon) ; float series_fho_O_RATE_ge7.0(lat, lon) ; float
> series_ctc_FY_OY_ge7.0(lat, lon) ; float series_ctc_FN_ON_ge7.0(lat,
lon) ;
> float series_cts_CSI_ge7.0(lat, lon) ; float
series_cts_GSS_ge7.0(lat, lon)
> ; float series_fho_F_RATE_ge10.0(lat, lon) ; float
> series_fho_O_RATE_ge10.0(lat, lon) ; float
series_ctc_FY_OY_ge10.0(lat,
> lon) ; float series_ctc_FN_ON_ge10.0(lat, lon) ; float
> series_cts_CSI_ge10.0(lat, lon) ; float series_cts_GSS_ge10.0(lat,
lon) ;
> float series_cnt_TOTAL(lat, lon) ; float series_cnt_RMSE(lat, lon) ;
float
> series_cnt_FBAR(lat, lon) ;*
> *float series_cnt_OBAR(lat, lon) ;*
>
> Also, note that increasing the "block_size" setting would make this
run
> faster... as long as you have sufficient memory available.
>
> Thanks,
> John
>
>
> On Thu, Jun 4, 2020 at 6:38 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
> >
> > Dear John,
> > Again, I specified three thresholds of precipitation but I can't
find
> that
> > in the netcdf output.
> > //
> >
> > // Forecast and observation fields to be verified
> >
> > //
> >
> > fcst = {
> >
> >
> >
> >
> > field = [
> >
> > {
> >
> > name = "pr";
> >
> > level = "(0,*,*)";
> >
> > cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
> >
> > }
> >
> > ];
> >
> >
> >
> >
> > }
> >
> > Also, the min and max values of CSI, GSS and some other statistics
are
> the
> > same. Please what could be the cause?
> > Thank you.
> > Best regards,Imoleayo
> > On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo <
> > gbodeimoleayo at yahoo.com> wrote:
> >
> > Dear John,
> > Thanks for pointing out my mistakes and thanks for helping to fix
the
> > script.
> > Please I would like to know why you choose to change the below
portion in
> > the configure script? //
> > // Number of grid points to be processed concurrently. Set
smaller to
> use
> > // less memory but increase the number of passes through the data.
> > //
> > block_size = 10000;
> >
> > //
> > // Ratio of valid matched pairs to compute statistics for a grid
point
> > //
> > vld_thresh = 0.1;.Why did you use 10000 and 0.1?
> > Thank you.
> > Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM
GMT+2, John
> > Halley Gotway via RT <met_help at ucar.edu> wrote:
> >
> > This is a problem with the script you wrote
(runSeriesAnalysis.sh). You
> > have some trailing spaces after the line continuation marks, and
those
> > wreak havoc in the script.
> >
> > I tested this manually by create fcst_file_list and obs_file_list
> > containing those long lists of files and running:
> >
> > *cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
> > */glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
> > fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc
-config
> > mySeriesAnalysisConfig -v 2*
> >
> > Note that in my local config file, I updated:
> >
> > //
> > // Number of grid points to be processed concurrently. Set
smaller to
> use
> > // less memory but increase the number of passes through the data.
> > //
> > block_size = 10000;
> >
> > //
> > // Ratio of valid matched pairs to compute statistics for a grid
point
> > //
> > vld_thresh = 0.1;
> >
> > See attached screenshot showing ncview of the FBAR and OBAR
fields.
> >
> > Thanks,
> > John
> >
> > On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> > wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
> > >
> > > Dear John,
> > > Please I would like to know whether you saw my message?
> > > Thank you.
> > > Best regards,Imoleayo
> > > On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo
<
> > > gbodeimoleayo at yahoo.com> wrote:
> > >
> > > Dear John,Thanks for your suggestions.
> > > I am trying to run the series_analysis on my Cheyenne
environment but I
> > > get error messages (see attached log.series). Please also find
attached
> > the
> > > run script and configure files for examination.
> > > Thank you.
> > > Best regards,Imoleayo
> > >
> > > On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley
Gotway via
> RT
> > > <met_help at ucar.edu> wrote:
> > >
> > > Hello Imoleayo,
> > >
> > > Sorry for the delay in getting back to you. I see that you're
running
> the
> > > Grid-Stat tool and would like to know the results at a single
location,
> > > namely 5 degrees latitude north and 7 degrees longitude east.
> > >
> > > Before answering that question, let me take a step back to
describe the
> > > purpose of the Grid-Stat tool. It is run to compare one or more
fields
> of
> > > gridded forecast data to one or more fields of gridded analysis
data.
> > > Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's
say
> you're
> > > verifying 2-m temperature from a model run which was initialized
at 00Z
> > > with 6-hourly output out to 5 days. You'd run Grid-Stat 21 times
to
> > verify
> > > each of those output times: 00-hour, 06-hour, 12-hour, and so
on. For
> > each
> > > time, Grid-Stat computes spatially aggregated statistics. The
user can
> > > configure whatever regions they'd like for the computation of
> statistics.
> > > For example, perhaps you want stats computed over the northern
> hemisphere
> > > land points. You could define that region (using the gen_vx_mask
tool)
> > and
> > > configure Grid-Stat to compute statistics over all the northern
> > hemisphere
> > > land grid points.
> > >
> > > To answer your question, technically, yes, you could define a
very tiny
> > > masking region that only includes the grid point for the single
lat/lon
> > > location. But that region will only contain 1 matched pair! And
MET
> can't
> > > compute meaningful statistics from a single matched pair.
> > >
> > > I don't know what you're really trying to do here. Perhaps what
you'd
> > > really like to do is compute stats at that single location over
a time
> > > series of data? If that's the case, then you should run the
> > Series-Analysis
> > > tool instead, which is described here:
> > > https://dtcenter.org/met-online-tutorial-metv8-0/series-
analysis
> > >
> > > If you can give me a better idea of what you're trying to do, I
might
> > have
> > > better suggestions for how to do it with MET.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > >
> > >
> >
> >
> >
>
>
>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: John Halley Gotway
Time: Fri Jun 05 14:59:00 2020
FSS is currently only computed by the Grid-Stat tool and is not
available
in Series-Analysis. However, I do believe that in the coming year,
we'll
work with staff from the UK MetOffice to add the computation of FSS to
Series-Analysis.
It is a logical extension of the functionality, but we just haven't
had any
requests for it previously.
John
On Fri, Jun 5, 2020 at 2:56 PM John Halley Gotway <johnhg at ucar.edu>
wrote:
> Yes, FSS is computed by the Grid-Stat tool.
>
> John
>
> On Fri, Jun 5, 2020 at 2:47 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> wrote:
>
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>>
>> Great! Thanks John.
>> I am happy to now find values for the CSI and GSS. Thanks for your
>> support.
>> Please another thing, I want to compute Fractional Skill Score in
Met
>> Tool. Is this possible?
>> Thank you.
>> Best regards,Imoleayo
>> On Friday, June 5, 2020, 08:44:48 PM GMT+2, John Halley Gotway
via RT
>> <met_help at ucar.edu> wrote:
>>
>> Imoleayo,
>>
>> OK, I see what's going on here, and I realize why it's confusing.
>>
>> To make it work, move the cat_thresh setting out of the "field"
dictionary
>> entry and up to the top-level of context, as I've done in the
attached
>> version of your config file.
>>
>> Looking in the source code (
>>
>>
https://github.com/NCAR/MET/blob/master_v9.0/met/src/tools/core/series_analysis/series_analysis_conf_info.cc
>> )
>> at line 304, I see that we're parsing cat_thresh from the "fcst"
and "obs"
>> dictionary level of context. That means, if it's defined at that
level, or
>> higher, it'll be applied. In the attached file, I defined it at the
global
>> level of context, so that'll work.
>>
>> And now, why is it set up this way? Series-Analysis can only be run
on a
>> single series at a time? That's usually a time-series, where you
define a
>> single field and pull it from multiple input files for different
times.
>> However, you could also define the series as multiple vertical
levels from
>> the same input file... in which case you'd have multiple fields
listed.
>>
>> However, you can only define a single threshold for the entire
series. The
>> tool doesn't support different thresholds for each field entry.
>>
>> I realize that this is confusing since it differs slightly from the
logic
>> in Point-Stat and Grid-Stat, but hopefully it at least makes sense
now.
>>
>> Rerunning with these settings produces the following output
variables,
>> where the threshold applied is included in the variable name:
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> *> ncdump -h seriesAnalysisOutWRF.nc | grep float float lat(lat) ;
float
>> lon(lon) ; float series_fho_F_RATE_gt0.0(lat, lon) ; float
>> series_fho_O_RATE_gt0.0(lat, lon) ; float
series_ctc_FY_OY_gt0.0(lat, lon)
>> ; float series_ctc_FN_ON_gt0.0(lat, lon) ; float
series_cts_CSI_gt0.0(lat,
>> lon) ; float series_cts_GSS_gt0.0(lat, lon) ; float
>> series_fho_F_RATE_ge5.0(lat, lon) ; float
series_fho_O_RATE_ge5.0(lat,
>> lon)
>> ; float series_ctc_FY_OY_ge5.0(lat, lon) ; float
>> series_ctc_FN_ON_ge5.0(lat, lon) ; float series_cts_CSI_ge5.0(lat,
lon) ;
>> float series_cts_GSS_ge5.0(lat, lon) ; float
series_fho_F_RATE_ge7.0(lat,
>> lon) ; float series_fho_O_RATE_ge7.0(lat, lon) ; float
>> series_ctc_FY_OY_ge7.0(lat, lon) ; float
series_ctc_FN_ON_ge7.0(lat, lon)
>> ;
>> float series_cts_CSI_ge7.0(lat, lon) ; float
series_cts_GSS_ge7.0(lat,
>> lon)
>> ; float series_fho_F_RATE_ge10.0(lat, lon) ; float
>> series_fho_O_RATE_ge10.0(lat, lon) ; float
series_ctc_FY_OY_ge10.0(lat,
>> lon) ; float series_ctc_FN_ON_ge10.0(lat, lon) ; float
>> series_cts_CSI_ge10.0(lat, lon) ; float series_cts_GSS_ge10.0(lat,
lon) ;
>> float series_cnt_TOTAL(lat, lon) ; float series_cnt_RMSE(lat, lon)
; float
>> series_cnt_FBAR(lat, lon) ;*
>> *float series_cnt_OBAR(lat, lon) ;*
>>
>> Also, note that increasing the "block_size" setting would make this
run
>> faster... as long as you have sufficient memory available.
>>
>> Thanks,
>> John
>>
>>
>> On Thu, Jun 4, 2020 at 6:38 PM gbode Imoleayo via RT
<met_help at ucar.edu>
>> wrote:
>>
>> >
>> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>> >
>> > Dear John,
>> > Again, I specified three thresholds of precipitation but I can't
find
>> that
>> > in the netcdf output.
>> > //
>> >
>> > // Forecast and observation fields to be verified
>> >
>> > //
>> >
>> > fcst = {
>> >
>> >
>> >
>> >
>> > field = [
>> >
>> > {
>> >
>> > name = "pr";
>> >
>> > level = "(0,*,*)";
>> >
>> > cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
>> >
>> > }
>> >
>> > ];
>> >
>> >
>> >
>> >
>> > }
>> >
>> > Also, the min and max values of CSI, GSS and some other
statistics are
>> the
>> > same. Please what could be the cause?
>> > Thank you.
>> > Best regards,Imoleayo
>> > On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo <
>> > gbodeimoleayo at yahoo.com> wrote:
>> >
>> > Dear John,
>> > Thanks for pointing out my mistakes and thanks for helping to fix
the
>> > script.
>> > Please I would like to know why you choose to change the below
portion
>> in
>> > the configure script? //
>> > // Number of grid points to be processed concurrently. Set
smaller to
>> use
>> > // less memory but increase the number of passes through the
data.
>> > //
>> > block_size = 10000;
>> >
>> > //
>> > // Ratio of valid matched pairs to compute statistics for a grid
point
>> > //
>> > vld_thresh = 0.1;.Why did you use 10000 and 0.1?
>> > Thank you.
>> > Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM
GMT+2,
>> John
>> > Halley Gotway via RT <met_help at ucar.edu> wrote:
>> >
>> > This is a problem with the script you wrote
(runSeriesAnalysis.sh). You
>> > have some trailing spaces after the line continuation marks, and
those
>> > wreak havoc in the script.
>> >
>> > I tested this manually by create fcst_file_list and obs_file_list
>> > containing those long lists of files and running:
>> >
>> > *cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
>> > */glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
>> > fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc
-config
>> > mySeriesAnalysisConfig -v 2*
>> >
>> > Note that in my local config file, I updated:
>> >
>> > //
>> > // Number of grid points to be processed concurrently. Set
smaller to
>> use
>> > // less memory but increase the number of passes through the
data.
>> > //
>> > block_size = 10000;
>> >
>> > //
>> > // Ratio of valid matched pairs to compute statistics for a grid
point
>> > //
>> > vld_thresh = 0.1;
>> >
>> > See attached screenshot showing ncview of the FBAR and OBAR
fields.
>> >
>> > Thanks,
>> > John
>> >
>> > On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu
>> >
>> > wrote:
>> >
>> > >
>> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>> > >
>> > > Dear John,
>> > > Please I would like to know whether you saw my message?
>> > > Thank you.
>> > > Best regards,Imoleayo
>> > > On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo
<
>> > > gbodeimoleayo at yahoo.com> wrote:
>> > >
>> > > Dear John,Thanks for your suggestions.
>> > > I am trying to run the series_analysis on my Cheyenne
environment but
>> I
>> > > get error messages (see attached log.series). Please also find
>> attached
>> > the
>> > > run script and configure files for examination.
>> > > Thank you.
>> > > Best regards,Imoleayo
>> > >
>> > > On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley
Gotway
>> via RT
>> > > <met_help at ucar.edu> wrote:
>> > >
>> > > Hello Imoleayo,
>> > >
>> > > Sorry for the delay in getting back to you. I see that you're
running
>> the
>> > > Grid-Stat tool and would like to know the results at a single
>> location,
>> > > namely 5 degrees latitude north and 7 degrees longitude east.
>> > >
>> > > Before answering that question, let me take a step back to
describe
>> the
>> > > purpose of the Grid-Stat tool. It is run to compare one or more
>> fields of
>> > > gridded forecast data to one or more fields of gridded analysis
data.
>> > > Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's
say
>> you're
>> > > verifying 2-m temperature from a model run which was
initialized at
>> 00Z
>> > > with 6-hourly output out to 5 days. You'd run Grid-Stat 21
times to
>> > verify
>> > > each of those output times: 00-hour, 06-hour, 12-hour, and so
on. For
>> > each
>> > > time, Grid-Stat computes spatially aggregated statistics. The
user can
>> > > configure whatever regions they'd like for the computation of
>> statistics.
>> > > For example, perhaps you want stats computed over the northern
>> hemisphere
>> > > land points. You could define that region (using the
gen_vx_mask tool)
>> > and
>> > > configure Grid-Stat to compute statistics over all the northern
>> > hemisphere
>> > > land grid points.
>> > >
>> > > To answer your question, technically, yes, you could define a
very
>> tiny
>> > > masking region that only includes the grid point for the single
>> lat/lon
>> > > location. But that region will only contain 1 matched pair! And
MET
>> can't
>> > > compute meaningful statistics from a single matched pair.
>> > >
>> > > I don't know what you're really trying to do here. Perhaps what
you'd
>> > > really like to do is compute stats at that single location over
a time
>> > > series of data? If that's the case, then you should run the
>> > Series-Analysis
>> > > tool instead, which is described here:
>> > > https://dtcenter.org/met-online-tutorial-metv8-0/series-
analysis
>> > >
>> > > If you can give me a better idea of what you're trying to do, I
might
>> > have
>> > > better suggestions for how to do it with MET.
>> > >
>> > > Thanks,
>> > > John Halley Gotway
>> > >
>> > >
>> > >
>> >
>> >
>> >
>>
>>
>>
------------------------------------------------
Subject: Running Grid_Stat for a specific location
From: gbode Imoleayo
Time: Fri Jun 05 15:13:05 2020
Ok John,
Thanks for the clarification. I look forward to seeing FSS computed in
series_analysis.
Once again, thanks for these solutions.
Best regards,Imoleayo
On Friday, June 5, 2020, 11:06:43 PM GMT+2, John Halley Gotway via
RT <met_help at ucar.edu> wrote:
FSS is currently only computed by the Grid-Stat tool and is not
available
in Series-Analysis. However, I do believe that in the coming year,
we'll
work with staff from the UK MetOffice to add the computation of FSS to
Series-Analysis.
It is a logical extension of the functionality, but we just haven't
had any
requests for it previously.
John
On Fri, Jun 5, 2020 at 2:56 PM John Halley Gotway <johnhg at ucar.edu>
wrote:
> Yes, FSS is computed by the Grid-Stat tool.
>
> John
>
> On Fri, Jun 5, 2020 at 2:47 PM gbode Imoleayo via RT
<met_help at ucar.edu>
> wrote:
>
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>>
>> Great! Thanks John.
>> I am happy to now find values for the CSI and GSS. Thanks for your
>> support.
>> Please another thing, I want to compute Fractional Skill Score in
Met
>> Tool. Is this possible?
>> Thank you.
>> Best regards,Imoleayo
>> On Friday, June 5, 2020, 08:44:48 PM GMT+2, John Halley Gotway
via RT
>> <met_help at ucar.edu> wrote:
>>
>> Imoleayo,
>>
>> OK, I see what's going on here, and I realize why it's confusing.
>>
>> To make it work, move the cat_thresh setting out of the "field"
dictionary
>> entry and up to the top-level of context, as I've done in the
attached
>> version of your config file.
>>
>> Looking in the source code (
>>
>>
https://github.com/NCAR/MET/blob/master_v9.0/met/src/tools/core/series_analysis/series_analysis_conf_info.cc
>> )
>> at line 304, I see that we're parsing cat_thresh from the "fcst"
and "obs"
>> dictionary level of context. That means, if it's defined at that
level, or
>> higher, it'll be applied. In the attached file, I defined it at the
global
>> level of context, so that'll work.
>>
>> And now, why is it set up this way? Series-Analysis can only be run
on a
>> single series at a time? That's usually a time-series, where you
define a
>> single field and pull it from multiple input files for different
times.
>> However, you could also define the series as multiple vertical
levels from
>> the same input file... in which case you'd have multiple fields
listed.
>>
>> However, you can only define a single threshold for the entire
series. The
>> tool doesn't support different thresholds for each field entry.
>>
>> I realize that this is confusing since it differs slightly from the
logic
>> in Point-Stat and Grid-Stat, but hopefully it at least makes sense
now.
>>
>> Rerunning with these settings produces the following output
variables,
>> where the threshold applied is included in the variable name:
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> *> ncdump -h seriesAnalysisOutWRF.nc | grep float float lat(lat) ;
float
>> lon(lon) ; float series_fho_F_RATE_gt0.0(lat, lon) ; float
>> series_fho_O_RATE_gt0.0(lat, lon) ; float
series_ctc_FY_OY_gt0.0(lat, lon)
>> ; float series_ctc_FN_ON_gt0.0(lat, lon) ; float
series_cts_CSI_gt0.0(lat,
>> lon) ; float series_cts_GSS_gt0.0(lat, lon) ; float
>> series_fho_F_RATE_ge5.0(lat, lon) ; float
series_fho_O_RATE_ge5.0(lat,
>> lon)
>> ; float series_ctc_FY_OY_ge5.0(lat, lon) ; float
>> series_ctc_FN_ON_ge5.0(lat, lon) ; float series_cts_CSI_ge5.0(lat,
lon) ;
>> float series_cts_GSS_ge5.0(lat, lon) ; float
series_fho_F_RATE_ge7.0(lat,
>> lon) ; float series_fho_O_RATE_ge7.0(lat, lon) ; float
>> series_ctc_FY_OY_ge7.0(lat, lon) ; float
series_ctc_FN_ON_ge7.0(lat, lon)
>> ;
>> float series_cts_CSI_ge7.0(lat, lon) ; float
series_cts_GSS_ge7.0(lat,
>> lon)
>> ; float series_fho_F_RATE_ge10.0(lat, lon) ; float
>> series_fho_O_RATE_ge10.0(lat, lon) ; float
series_ctc_FY_OY_ge10.0(lat,
>> lon) ; float series_ctc_FN_ON_ge10.0(lat, lon) ; float
>> series_cts_CSI_ge10.0(lat, lon) ; float series_cts_GSS_ge10.0(lat,
lon) ;
>> float series_cnt_TOTAL(lat, lon) ; float series_cnt_RMSE(lat, lon)
; float
>> series_cnt_FBAR(lat, lon) ;*
>> *float series_cnt_OBAR(lat, lon) ;*
>>
>> Also, note that increasing the "block_size" setting would make this
run
>> faster... as long as you have sufficient memory available.
>>
>> Thanks,
>> John
>>
>>
>> On Thu, Jun 4, 2020 at 6:38 PM gbode Imoleayo via RT
<met_help at ucar.edu>
>> wrote:
>>
>> >
>> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>> >
>> > Dear John,
>> > Again, I specified three thresholds of precipitation but I can't
find
>> that
>> > in the netcdf output.
>> > //
>> >
>> > // Forecast and observation fields to be verified
>> >
>> > //
>> >
>> > fcst = {
>> >
>> >
>> >
>> >
>> > field = [
>> >
>> > {
>> >
>> > name = "pr";
>> >
>> > level = "(0,*,*)";
>> >
>> > cat_thresh = [ >0.0, >=5.0, >=7.0, >=10.0 ];
>> >
>> > }
>> >
>> > ];
>> >
>> >
>> >
>> >
>> > }
>> >
>> > Also, the min and max values of CSI, GSS and some other
statistics are
>> the
>> > same. Please what could be the cause?
>> > Thank you.
>> > Best regards,Imoleayo
>> > On Friday, June 5, 2020, 02:30:36 AM GMT+2, gbode Imoleayo <
>> > gbodeimoleayo at yahoo.com> wrote:
>> >
>> > Dear John,
>> > Thanks for pointing out my mistakes and thanks for helping to fix
the
>> > script.
>> > Please I would like to know why you choose to change the below
portion
>> in
>> > the configure script? //
>> > // Number of grid points to be processed concurrently. Set
smaller to
>> use
>> > // less memory but increase the number of passes through the
data.
>> > //
>> > block_size = 10000;
>> >
>> > //
>> > // Ratio of valid matched pairs to compute statistics for a grid
point
>> > //
>> > vld_thresh = 0.1;.Why did you use 10000 and 0.1?
>> > Thank you.
>> > Best regards,Imoleayo On Friday, June 5, 2020, 01:54:59 AM
GMT+2,
>> John
>> > Halley Gotway via RT <met_help at ucar.edu> wrote:
>> >
>> > This is a problem with the script you wrote
(runSeriesAnalysis.sh). You
>> > have some trailing spaces after the line continuation marks, and
those
>> > wreak havoc in the script.
>> >
>> > I tested this manually by create fcst_file_list and obs_file_list
>> > containing those long lists of files and running:
>> >
>> > *cd /glade/p/ral/jntp/MET/MET_Help/imoleayo_data_20200604*
>> > */glade/p/ral/jntp/MET/MET_releases/8.0/bin/series_analysis -fcst
>> > fcst_file_list -obs obs_file_list -out seriesAnalysisOutWRF.nc
-config
>> > mySeriesAnalysisConfig -v 2*
>> >
>> > Note that in my local config file, I updated:
>> >
>> > //
>> > // Number of grid points to be processed concurrently. Set
smaller to
>> use
>> > // less memory but increase the number of passes through the
data.
>> > //
>> > block_size = 10000;
>> >
>> > //
>> > // Ratio of valid matched pairs to compute statistics for a grid
point
>> > //
>> > vld_thresh = 0.1;
>> >
>> > See attached screenshot showing ncview of the FBAR and OBAR
fields.
>> >
>> > Thanks,
>> > John
>> >
>> > On Thu, Jun 4, 2020 at 4:18 PM gbode Imoleayo via RT
<met_help at ucar.edu
>> >
>> > wrote:
>> >
>> > >
>> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95398 >
>> > >
>> > > Dear John,
>> > > Please I would like to know whether you saw my message?
>> > > Thank you.
>> > > Best regards,Imoleayo
>> > > On Thursday, June 4, 2020, 02:29:44 AM GMT+2, gbode Imoleayo
<
>> > > gbodeimoleayo at yahoo.com> wrote:
>> > >
>> > > Dear John,Thanks for your suggestions.
>> > > I am trying to run the series_analysis on my Cheyenne
environment but
>> I
>> > > get error messages (see attached log.series). Please also find
>> attached
>> > the
>> > > run script and configure files for examination.
>> > > Thank you.
>> > > Best regards,Imoleayo
>> > >
>> > > On Tuesday, June 2, 2020, 06:41:21 AM GMT+2, John Halley
Gotway
>> via RT
>> > > <met_help at ucar.edu> wrote:
>> > >
>> > > Hello Imoleayo,
>> > >
>> > > Sorry for the delay in getting back to you. I see that you're
running
>> the
>> > > Grid-Stat tool and would like to know the results at a single
>> location,
>> > > namely 5 degrees latitude north and 7 degrees longitude east.
>> > >
>> > > Before answering that question, let me take a step back to
describe
>> the
>> > > purpose of the Grid-Stat tool. It is run to compare one or more
>> fields of
>> > > gridded forecast data to one or more fields of gridded analysis
data.
>> > > Grid-Stat is run once FOR EACH OUTPUT TIME. For example, let's
say
>> you're
>> > > verifying 2-m temperature from a model run which was
initialized at
>> 00Z
>> > > with 6-hourly output out to 5 days. You'd run Grid-Stat 21
times to
>> > verify
>> > > each of those output times: 00-hour, 06-hour, 12-hour, and so
on. For
>> > each
>> > > time, Grid-Stat computes spatially aggregated statistics. The
user can
>> > > configure whatever regions they'd like for the computation of
>> statistics.
>> > > For example, perhaps you want stats computed over the northern
>> hemisphere
>> > > land points. You could define that region (using the
gen_vx_mask tool)
>> > and
>> > > configure Grid-Stat to compute statistics over all the northern
>> > hemisphere
>> > > land grid points.
>> > >
>> > > To answer your question, technically, yes, you could define a
very
>> tiny
>> > > masking region that only includes the grid point for the single
>> lat/lon
>> > > location. But that region will only contain 1 matched pair! And
MET
>> can't
>> > > compute meaningful statistics from a single matched pair.
>> > >
>> > > I don't know what you're really trying to do here. Perhaps what
you'd
>> > > really like to do is compute stats at that single location over
a time
>> > > series of data? If that's the case, then you should run the
>> > Series-Analysis
>> > > tool instead, which is described here:
>> > > https://dtcenter.org/met-online-tutorial-metv8-0/series-
analysis
>> > >
>> > > If you can give me a better idea of what you're trying to do, I
might
>> > have
>> > > better suggestions for how to do it with MET.
>> > >
>> > > Thanks,
>> > > John Halley Gotway
>> > >
>> > >
>> > >
>> >
>> >
>> >
>>
>>
>>
------------------------------------------------
More information about the Met_help
mailing list