[Met_help] [rt.rap.ucar.edu #96952] History for Point_stat output for station list
John Halley Gotway via RT
met_help at ucar.edu
Thu Oct 8 21:05:25 MDT 2020
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
MET help team,
When I run point_stat and output anything but the MPR file I only get
output for two unique VX_masks ( see attached output). My goal is to
obtain output from point stat from each station in my station list. Is
there a way to configure point_stat so the output is station by station?
(i.e. for each station in sid-- config file is attached)
Thanks for any input you can provide!
Best,
Brooke Adams
Scientific Programmer, AceInfo Solutions
Supporting Meteorological Development Laboratory
Cell: (518)361-6900
Brooke.Karlee.Adams at noaa.gov
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Point_stat output for station list
From: John Halley Gotway
Time: Mon Oct 05 10:54:00 2020
Brooke,
I see you have a question about Point-Stat. You are interested in
performance at a particular set of stations and are wondering how to
create
output for each individual station.
It sounds like you already know about the MPR line type and are able
to
generate those output lines. Thanks for sending your Point-Stat
configuration file. I see that you're using the "mask.sid" option to
define
a list of stations. In the output you see 2 CNT lines, one for VX_MASK
=
FULL and one for VX_MASK = CWSU. Let me explain what's happening in
the
code:
(1) The default point_stat config file contains an entry for:
* mask = { grid = ["FULL"]; ... }*
"FULL" just indicates that all grid points in the input model domain
should
be included in that stats. When you run point_stat, the config file
you
pass on the command line is used to override the settings in the
default
config file.
In your config file, you commented out the "grid = " line instead of
just
emptying it out. By commenting it out, your config file entry for
"grid" no
longer overrides the default value. So that's why you're seeing output
for
VX_MASK = FULL. To disable that output, instead of commenting it out,
just
provide an empty list:
* mask = { grid = [ ]; ... }*
(2) For an explanation of the "mask.sid" entry, I've copied and pasted
a
selection from the user's guide about this:
https://dtcenter.github.io/MET/Users_Guide/data_io.html
When you set sid equal to a filename, the stations are processed as a
group. And the first entry in that file defines the name of that
group.
Stations names provided as a comma-separated list directly in the
config
file, with no name prefix, are processed as individual settings for
VX_MASK, rather than as one big group.
// - The "sid" entry is an array of strings which define groups of
// observation station ID's over which to compute statistics.
Each entry
// in the array is either a filename of a comma-separated list.
// - For a filename, the strings are whitespace-separated. The
first
// string is the mask "name" and the remaining strings are the
station
// ID's to be used.
// - For a comma-separated list, optionally use a colon to
specify a name.
// For "MY_LIST:SID1,SID2", name = MY_LIST and values = SID1
and SID2.
// - For a comma-separated list of length one with no name
specified, the
// mask "name" and value are both set to the single station ID
string.
// For "SID1", name = SID1 and value = SID1.
// - For a comma-separated list of length greater than one with
no name
// specified, the name is set to MASK_SID and the values are
the station
// ID's to be used.
// For "SID1,SID2", name = MASK_SID and values = SID1 and
SID2.
// - The "name" of the station ID mask is written to the VX_MASK
column
// of the MET output files.
While you can compute stats separately for each station, let me
suggest an
alternative approach.
Point-Stat is run once per model output time. As such, for each
variable
and level, you'll likely only have one matched pair at each station.
And
statistics computed using a single value are not meaningful. However,
that
is why we recommend writing the MPR line type. After running Point-
Stat to
create MPR output lines over many cases, you can run the STAT-Analysis
tool
to compute stats separately for each station.
Let's say you have an output directory named "out_point_stat" that
contains
many .stat files, each containing MPR lines for many stations. You can
run
the following type of stat_analysis job:
stat_analysis -lookin out_point_stat -job aggregate_stat -line_type
MPR
-out_line_type CNT \
-by FCST_VAR,FCST_LEV,FCST_LEAD,OBS_SID -out_stat
mpr_to_cnt_by_station.stat
That job will...
- search the out_point_stat directory recursively for files ending in
.stat
- read MPR lines from those files
- keep track of the unique combinations of the FCST_VAR, FCST_LEV,
FCST_LEAD, and OBS_SID columns
- for each unique combination of those, it'll use the MPR lines to
derive
CNT statistics
- write the output to a .stat file named mpr_to_cnt_by_station.stat
The "-by" options specify which columns should be included in the
grouping
of MPR lines. If you have output for multiple models, just include the
"MODEL" column in the "-by" options and your results will be broken
down by
model.
The example above derives continuous statistics, but many other output
line
types are also available by changing the "-out_line_type" option (e.g.
-out_line_type SL1L2 ... or -out_line_type CTC -out_thresh gt0).
Hope that helps.
Thanks,
John Halley Gotway
On Sun, Oct 4, 2020 at 1:47 PM Brooke Adams - NOAA Affiliate via RT <
met_help at ucar.edu> wrote:
>
> Sun Oct 04 13:46:48 2020: Request 96952 was acted upon.
> Transaction: Ticket created by brooke.karlee.adams at noaa.gov
> Queue: met_help
> Subject: Point_stat output for station list
> Owner: Nobody
> Requestors: brooke.karlee.adams at noaa.gov
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=96952 >
>
>
> MET help team,
>
> When I run point_stat and output anything but the MPR file I only
get
> output for two unique VX_masks ( see attached output). My goal is
to
> obtain output from point stat from each station in my station list.
Is
> there a way to configure point_stat so the output is station by
station?
> (i.e. for each station in sid-- config file is attached)
>
> Thanks for any input you can provide!
>
> Best,
> Brooke Adams
>
> Scientific Programmer, AceInfo Solutions
> Supporting Meteorological Development Laboratory
> Cell: (518)361-6900
> Brooke.Karlee.Adams at noaa.gov
>
>
------------------------------------------------
Subject: Point_stat output for station list
From: Brooke Adams - NOAA Affiliate
Time: Thu Oct 08 13:53:40 2020
John,
Thank you I was able to get successful runs for point stat! I
appreciate your help.
Best,
Brooke Adams
Scientific Programmer, AceInfo Solutions
Supporting Meteorological Development Laboratory
Cell: (518)361-6900
Brooke.Karlee.Adams at noaa.gov
On Mon, Oct 5, 2020 at 12:54 PM John Halley Gotway via RT
<met_help at ucar.edu>
wrote:
> Brooke,
>
> I see you have a question about Point-Stat. You are interested in
> performance at a particular set of stations and are wondering how to
create
> output for each individual station.
>
> It sounds like you already know about the MPR line type and are able
to
> generate those output lines. Thanks for sending your Point-Stat
> configuration file. I see that you're using the "mask.sid" option to
define
> a list of stations. In the output you see 2 CNT lines, one for
VX_MASK =
> FULL and one for VX_MASK = CWSU. Let me explain what's happening in
the
> code:
>
> (1) The default point_stat config file contains an entry for:
> * mask = { grid = ["FULL"]; ... }*
>
> "FULL" just indicates that all grid points in the input model domain
should
> be included in that stats. When you run point_stat, the config file
you
> pass on the command line is used to override the settings in the
default
> config file.
>
> In your config file, you commented out the "grid = " line instead of
just
> emptying it out. By commenting it out, your config file entry for
"grid" no
> longer overrides the default value. So that's why you're seeing
output for
> VX_MASK = FULL. To disable that output, instead of commenting it
out, just
> provide an empty list:
> * mask = { grid = [ ]; ... }*
>
> (2) For an explanation of the "mask.sid" entry, I've copied and
pasted a
> selection from the user's guide about this:
> https://dtcenter.github.io/MET/Users_Guide/data_io.html
>
> When you set sid equal to a filename, the stations are processed as
a
> group. And the first entry in that file defines the name of that
group.
> Stations names provided as a comma-separated list directly in the
config
> file, with no name prefix, are processed as individual settings for
> VX_MASK, rather than as one big group.
>
> // - The "sid" entry is an array of strings which define groups
of
> // observation station ID's over which to compute statistics.
Each
> entry
> // in the array is either a filename of a comma-separated list.
> // - For a filename, the strings are whitespace-separated. The
first
> // string is the mask "name" and the remaining strings are
the
> station
> // ID's to be used.
> // - For a comma-separated list, optionally use a colon to
specify a
> name.
> // For "MY_LIST:SID1,SID2", name = MY_LIST and values = SID1
and
> SID2.
> // - For a comma-separated list of length one with no name
> specified, the
> // mask "name" and value are both set to the single station
ID
> string.
> // For "SID1", name = SID1 and value = SID1.
> // - For a comma-separated list of length greater than one
with no
> name
> // specified, the name is set to MASK_SID and the values are
the
> station
> // ID's to be used.
> // For "SID1,SID2", name = MASK_SID and values = SID1 and
SID2.
> // - The "name" of the station ID mask is written to the
VX_MASK
> column
> // of the MET output files.
>
> While you can compute stats separately for each station, let me
suggest an
> alternative approach.
>
> Point-Stat is run once per model output time. As such, for each
variable
> and level, you'll likely only have one matched pair at each station.
And
> statistics computed using a single value are not meaningful.
However, that
> is why we recommend writing the MPR line type. After running Point-
Stat to
> create MPR output lines over many cases, you can run the STAT-
Analysis tool
> to compute stats separately for each station.
>
> Let's say you have an output directory named "out_point_stat" that
contains
> many .stat files, each containing MPR lines for many stations. You
can run
> the following type of stat_analysis job:
>
> stat_analysis -lookin out_point_stat -job aggregate_stat -line_type
MPR
> -out_line_type CNT \
> -by FCST_VAR,FCST_LEV,FCST_LEAD,OBS_SID -out_stat
> mpr_to_cnt_by_station.stat
>
> That job will...
> - search the out_point_stat directory recursively for files ending
in
> .stat
> - read MPR lines from those files
> - keep track of the unique combinations of the FCST_VAR, FCST_LEV,
> FCST_LEAD, and OBS_SID columns
> - for each unique combination of those, it'll use the MPR lines to
derive
> CNT statistics
> - write the output to a .stat file named mpr_to_cnt_by_station.stat
>
> The "-by" options specify which columns should be included in the
grouping
> of MPR lines. If you have output for multiple models, just include
the
> "MODEL" column in the "-by" options and your results will be broken
down by
> model.
>
> The example above derives continuous statistics, but many other
output line
> types are also available by changing the "-out_line_type" option
(e.g.
> -out_line_type SL1L2 ... or -out_line_type CTC -out_thresh gt0).
>
> Hope that helps.
>
> Thanks,
> John Halley Gotway
>
> On Sun, Oct 4, 2020 at 1:47 PM Brooke Adams - NOAA Affiliate via RT
<
> met_help at ucar.edu> wrote:
>
> >
> > Sun Oct 04 13:46:48 2020: Request 96952 was acted upon.
> > Transaction: Ticket created by brooke.karlee.adams at noaa.gov
> > Queue: met_help
> > Subject: Point_stat output for station list
> > Owner: Nobody
> > Requestors: brooke.karlee.adams at noaa.gov
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=96952 >
> >
> >
> > MET help team,
> >
> > When I run point_stat and output anything but the MPR file I only
get
> > output for two unique VX_masks ( see attached output). My goal is
to
> > obtain output from point stat from each station in my station
list. Is
> > there a way to configure point_stat so the output is station by
station?
> > (i.e. for each station in sid-- config file is attached)
> >
> > Thanks for any input you can provide!
> >
> > Best,
> > Brooke Adams
> >
> > Scientific Programmer, AceInfo Solutions
> > Supporting Meteorological Development Laboratory
> > Cell: (518)361-6900
> > Brooke.Karlee.Adams at noaa.gov
> >
> >
>
>
------------------------------------------------
Subject: Point_stat output for station list
From: John Halley Gotway
Time: Thu Oct 08 14:18:46 2020
Brooke,
Great, thanks for confirming. I’ll go ahead and resolve this ticket.
Thanks,
John
On Thu, Oct 8, 2020 at 1:54 PM Brooke Adams - NOAA Affiliate via RT <
met_help at ucar.edu> wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=96952 >
>
> John,
>
> Thank you I was able to get successful runs for point stat! I
> appreciate your help.
>
> Best,
> Brooke Adams
>
> Scientific Programmer, AceInfo Solutions
> Supporting Meteorological Development Laboratory
> Cell: (518)361-6900
> Brooke.Karlee.Adams at noaa.gov
>
>
> On Mon, Oct 5, 2020 at 12:54 PM John Halley Gotway via RT <
> met_help at ucar.edu>
> wrote:
>
> > Brooke,
> >
> > I see you have a question about Point-Stat. You are interested in
> > performance at a particular set of stations and are wondering how
to
> create
> > output for each individual station.
> >
> > It sounds like you already know about the MPR line type and are
able to
> > generate those output lines. Thanks for sending your Point-Stat
> > configuration file. I see that you're using the "mask.sid" option
to
> define
> > a list of stations. In the output you see 2 CNT lines, one for
VX_MASK =
> > FULL and one for VX_MASK = CWSU. Let me explain what's happening
in the
> > code:
> >
> > (1) The default point_stat config file contains an entry for:
> > * mask = { grid = ["FULL"]; ... }*
> >
> > "FULL" just indicates that all grid points in the input model
domain
> should
> > be included in that stats. When you run point_stat, the config
file you
> > pass on the command line is used to override the settings in the
default
> > config file.
> >
> > In your config file, you commented out the "grid = " line instead
of just
> > emptying it out. By commenting it out, your config file entry for
"grid"
> no
> > longer overrides the default value. So that's why you're seeing
output
> for
> > VX_MASK = FULL. To disable that output, instead of commenting it
out,
> just
> > provide an empty list:
> > * mask = { grid = [ ]; ... }*
> >
> > (2) For an explanation of the "mask.sid" entry, I've copied and
pasted a
> > selection from the user's guide about this:
> > https://dtcenter.github.io/MET/Users_Guide/data_io.html
> >
> > When you set sid equal to a filename, the stations are processed
as a
> > group. And the first entry in that file defines the name of that
group.
> > Stations names provided as a comma-separated list directly in the
config
> > file, with no name prefix, are processed as individual settings
for
> > VX_MASK, rather than as one big group.
> >
> > // - The "sid" entry is an array of strings which define groups
of
> > // observation station ID's over which to compute statistics.
Each
> > entry
> > // in the array is either a filename of a comma-separated
list.
> > // - For a filename, the strings are whitespace-separated.
The
> first
> > // string is the mask "name" and the remaining strings are
the
> > station
> > // ID's to be used.
> > // - For a comma-separated list, optionally use a colon to
specify
> a
> > name.
> > // For "MY_LIST:SID1,SID2", name = MY_LIST and values =
SID1 and
> > SID2.
> > // - For a comma-separated list of length one with no name
> > specified, the
> > // mask "name" and value are both set to the single
station ID
> > string.
> > // For "SID1", name = SID1 and value = SID1.
> > // - For a comma-separated list of length greater than one
with no
> > name
> > // specified, the name is set to MASK_SID and the values
are the
> > station
> > // ID's to be used.
> > // For "SID1,SID2", name = MASK_SID and values = SID1 and
SID2.
> > // - The "name" of the station ID mask is written to the
VX_MASK
> > column
> > // of the MET output files.
> >
> > While you can compute stats separately for each station, let me
suggest
> an
> > alternative approach.
> >
> > Point-Stat is run once per model output time. As such, for each
variable
> > and level, you'll likely only have one matched pair at each
station. And
> > statistics computed using a single value are not meaningful.
However,
> that
> > is why we recommend writing the MPR line type. After running
Point-Stat
> to
> > create MPR output lines over many cases, you can run the STAT-
Analysis
> tool
> > to compute stats separately for each station.
> >
> > Let's say you have an output directory named "out_point_stat" that
> contains
> > many .stat files, each containing MPR lines for many stations. You
can
> run
> > the following type of stat_analysis job:
> >
> > stat_analysis -lookin out_point_stat -job aggregate_stat
-line_type MPR
> > -out_line_type CNT \
> > -by FCST_VAR,FCST_LEV,FCST_LEAD,OBS_SID -out_stat
> > mpr_to_cnt_by_station.stat
> >
> > That job will...
> > - search the out_point_stat directory recursively for files
ending in
> > .stat
> > - read MPR lines from those files
> > - keep track of the unique combinations of the FCST_VAR,
FCST_LEV,
> > FCST_LEAD, and OBS_SID columns
> > - for each unique combination of those, it'll use the MPR lines
to
> derive
> > CNT statistics
> > - write the output to a .stat file named
mpr_to_cnt_by_station.stat
> >
> > The "-by" options specify which columns should be included in the
> grouping
> > of MPR lines. If you have output for multiple models, just include
the
> > "MODEL" column in the "-by" options and your results will be
broken down
> by
> > model.
> >
> > The example above derives continuous statistics, but many other
output
> line
> > types are also available by changing the "-out_line_type" option
(e.g.
> > -out_line_type SL1L2 ... or -out_line_type CTC -out_thresh gt0).
> >
> > Hope that helps.
> >
> > Thanks,
> > John Halley Gotway
> >
> > On Sun, Oct 4, 2020 at 1:47 PM Brooke Adams - NOAA Affiliate via
RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > Sun Oct 04 13:46:48 2020: Request 96952 was acted upon.
> > > Transaction: Ticket created by brooke.karlee.adams at noaa.gov
> > > Queue: met_help
> > > Subject: Point_stat output for station list
> > > Owner: Nobody
> > > Requestors: brooke.karlee.adams at noaa.gov
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=96952
> >
> > >
> > >
> > > MET help team,
> > >
> > > When I run point_stat and output anything but the MPR file I
only get
> > > output for two unique VX_masks ( see attached output). My goal
is to
> > > obtain output from point stat from each station in my station
list. Is
> > > there a way to configure point_stat so the output is station by
> station?
> > > (i.e. for each station in sid-- config file is attached)
> > >
> > > Thanks for any input you can provide!
> > >
> > > Best,
> > > Brooke Adams
> > >
> > > Scientific Programmer, AceInfo Solutions
> > > Supporting Meteorological Development Laboratory
> > > Cell: (518)361-6900
> > > Brooke.Karlee.Adams at noaa.gov
> > >
> > >
> >
> >
>
>
------------------------------------------------
More information about the Met_help
mailing list