[Met_help] [rt.rap.ucar.edu #89077] History for stat_analysis question
John Halley Gotway via RT
met_help at ucar.edu
Thu Apr 25 14:51:15 MDT 2019
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Good afternoon,
(John, if this is assigned to you, apologies for not articulating this as
well on the call today - this is a brief followup to that).
We had a couple of questions regarding the output of stat_analysis. We're
having some issues getting that station IDs out of the .stat files.
We would like the process to be:
1. Run point_stat on each validation hour and produce a .stat and an .mpr
file for each valid hour, listing the verification for each station out in
the output files.
2. Run stat_anlaysis monthly on each of the .stat or .mpr files to produce
our aggregated scores for each station individually (aggregating by site).
Unfortunately, our processing at the moment has some snags that are causing
some headaches.
When we run point_stat, writing both the .mpr and .stat files, the station
identifier column doesn't have a column name in the .stat file (not a big
deal since we can get that information from the .mpr file). But when run
stat_analysis over the month to aggregate the station data, when we
aggregate over OBS_SID, the stat_analysis output doesn't contain the unique
identifier for the station causing us not to know what station that
verification belongs to.
Are we missing something in our stat_analysis configuration for it to
output OBS_SID column in the .stat output? This would solve all of our
problems. :)
Thanks much for your help,
Dana
--
Dana Strom
Verification and Visualization Task Lead
Meteorological Development Lab
NOAA/National Weather Service
301-427-9451
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: stat_analysis question
From: John Halley Gotway
Time: Mon Feb 25 15:30:31 2019
Dana,
Is the OBS_SID column totally blank or is it set to some constant
string.
If it's totally blank, that'll likely mess up the parsing logic of
Stat-Analysis. If it's not blank, just constant, then the parsing
logic
should be OK.
Instead of using "-by OBS_SID", you could try using "-by
OBS_LAT,OBS_LON".
That'll concatenate the lat/lon of each station into a string and
report
results for the unique strings it finds.
Does that help? Feel free to send a sample file if you have
additional
questions.
Thanks,
John
On Mon, Feb 25, 2019 at 1:36 PM dana.strom at noaa.gov via RT <
met_help at ucar.edu> wrote:
>
> Mon Feb 25 13:35:47 2019: Request 89077 was acted upon.
> Transaction: Ticket created by dana.strom at noaa.gov
> Queue: met_help
> Subject: stat_analysis question
> Owner: Nobody
> Requestors: dana.strom at noaa.gov
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
>
>
> Good afternoon,
>
> (John, if this is assigned to you, apologies for not articulating
this as
> well on the call today - this is a brief followup to that).
>
> We had a couple of questions regarding the output of stat_analysis.
We're
> having some issues getting that station IDs out of the .stat files.
>
> We would like the process to be:
>
> 1. Run point_stat on each validation hour and produce a .stat and an
.mpr
> file for each valid hour, listing the verification for each station
out in
> the output files.
> 2. Run stat_anlaysis monthly on each of the .stat or .mpr files to
produce
> our aggregated scores for each station individually (aggregating by
site).
>
> Unfortunately, our processing at the moment has some snags that are
causing
> some headaches.
>
> When we run point_stat, writing both the .mpr and .stat files, the
station
> identifier column doesn't have a column name in the .stat file (not
a big
> deal since we can get that information from the .mpr file). But when
run
> stat_analysis over the month to aggregate the station data, when we
> aggregate over OBS_SID, the stat_analysis output doesn't contain the
unique
> identifier for the station causing us not to know what station that
> verification belongs to.
>
> Are we missing something in our stat_analysis configuration for it
to
> output OBS_SID column in the .stat output? This would solve all of
our
> problems. :)
>
> Thanks much for your help,
> Dana
>
> --
> Dana Strom
> Verification and Visualization Task Lead
> Meteorological Development Lab
> NOAA/National Weather Service
>
> 301-427-9451
>
>
------------------------------------------------
Subject: stat_analysis question
From: dana.strom at noaa.gov
Time: Tue Feb 26 08:36:16 2019
Hey John,
* Is the OBS_SID column totally blank or is it set to some constant
string.
*
*point_stat *contains the OBS_SID column with proper names of the
stations
we provided
*stat_analysis* does NOT contain a OBS_SID column in the output, even
when
specifying that as the stratifier. It DOES output all of the stations,
we
just don't know which one is which because the OBS_SID column does not
exist in the stat_analysis output.
We have not attempted to "-by OBS_LAT,OBS_LON". However, we did notice
that
OBS_LAT and OBS_LON are not present in the stat_analysis output, so
I'm not
sure we could use that either.
I hope this helps explain what issues we're having.
Thanks as always for your help,
Dana
On Mon, Feb 25, 2019 at 5:30 PM John Halley Gotway via RT
<met_help at ucar.edu>
wrote:
> Dana,
>
> Is the OBS_SID column totally blank or is it set to some constant
string.
> If it's totally blank, that'll likely mess up the parsing logic of
> Stat-Analysis. If it's not blank, just constant, then the parsing
logic
> should be OK.
>
> Instead of using "-by OBS_SID", you could try using "-by
OBS_LAT,OBS_LON".
> That'll concatenate the lat/lon of each station into a string and
report
> results for the unique strings it finds.
>
> Does that help? Feel free to send a sample file if you have
additional
> questions.
>
> Thanks,
> John
>
> On Mon, Feb 25, 2019 at 1:36 PM dana.strom at noaa.gov via RT <
> met_help at ucar.edu> wrote:
>
> >
> > Mon Feb 25 13:35:47 2019: Request 89077 was acted upon.
> > Transaction: Ticket created by dana.strom at noaa.gov
> > Queue: met_help
> > Subject: stat_analysis question
> > Owner: Nobody
> > Requestors: dana.strom at noaa.gov
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
> >
> >
> > Good afternoon,
> >
> > (John, if this is assigned to you, apologies for not articulating
this as
> > well on the call today - this is a brief followup to that).
> >
> > We had a couple of questions regarding the output of
stat_analysis. We're
> > having some issues getting that station IDs out of the .stat
files.
> >
> > We would like the process to be:
> >
> > 1. Run point_stat on each validation hour and produce a .stat and
an .mpr
> > file for each valid hour, listing the verification for each
station out
> in
> > the output files.
> > 2. Run stat_anlaysis monthly on each of the .stat or .mpr files to
> produce
> > our aggregated scores for each station individually (aggregating
by
> site).
> >
> > Unfortunately, our processing at the moment has some snags that
are
> causing
> > some headaches.
> >
> > When we run point_stat, writing both the .mpr and .stat files, the
> station
> > identifier column doesn't have a column name in the .stat file
(not a big
> > deal since we can get that information from the .mpr file). But
when run
> > stat_analysis over the month to aggregate the station data, when
we
> > aggregate over OBS_SID, the stat_analysis output doesn't contain
the
> unique
> > identifier for the station causing us not to know what station
that
> > verification belongs to.
> >
> > Are we missing something in our stat_analysis configuration for it
to
> > output OBS_SID column in the .stat output? This would solve all of
our
> > problems. :)
> >
> > Thanks much for your help,
> > Dana
> >
> > --
> > Dana Strom
> > Verification and Visualization Task Lead
> > Meteorological Development Lab
> > NOAA/National Weather Service
> >
> > 301-427-9451
> >
> >
>
>
--
Dana Strom
Verification and Visualization Task Lead
Meteorological Development Lab
NOAA/National Weather Service
301-427-9451
------------------------------------------------
Subject: stat_analysis question
From: John Halley Gotway
Time: Tue Feb 26 11:31:13 2019
Dana,
Ah, OK, I understand. Please try re-running your job using both the
"-out"
and the "-out_stat" options.
You'll find that the output file created with "-out" does actually
contain
the OBS_SID column strings.
However the output created with "-out_stat" does not.
For example, using a sample output MPR file create by "make test", I
ran
this job:
*stat_analysis -lookin
out/point_stat/point_stat_360000L_20070331_120000V_mpr.txt \*
*-job aggregate_stat -line_type MPR -out_line_type CNT -by
FCST_VAR,OBS_SID
\*
*-out mpr_to_cnt.out -out_stat mpr_to_cnt.stat*
The *mpr_to_cnt.out* file does contain the OBS_SID output column...
but
lacks the full 22 STAT header columns.
The *mpr_to_cnt.stat* file does have the full 22 STAT header
columns... but
lacks the OBS_SID info.
I won't go into the background as to why these 2 output file types
exist.
There is a job command option named "-set_hdr" which allows you to
specify
a value for the header columns written to the "-out_stat" output
column.
However, that's a constant string for all of the output for that job.
For
example, settings "-set_hdr VX_MASK TEST" would put the string "TEST"
into
the VX_MASK output column.
However, what we really need here is a way of writing the case-
specific
info to the -out_stat output file in some way. In this example, it'd
be
most useful to write the OBS_SID value to the VX_MASK output column.
So
supporting something like:
* -set_hdr VX_MASK OBS_SID*
Would do the trick.
Do you agree that would be a useful enhancement?
Thanks,
John
On Tue, Feb 26, 2019 at 8:47 AM dana.strom at noaa.gov via RT <
met_help at ucar.edu> wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
>
> Hey John,
>
>
> * Is the OBS_SID column totally blank or is it set to some constant
string.
> *
>
> *point_stat *contains the OBS_SID column with proper names of the
stations
> we provided
> *stat_analysis* does NOT contain a OBS_SID column in the output,
even when
> specifying that as the stratifier. It DOES output all of the
stations, we
> just don't know which one is which because the OBS_SID column does
not
> exist in the stat_analysis output.
>
> We have not attempted to "-by OBS_LAT,OBS_LON". However, we did
notice that
> OBS_LAT and OBS_LON are not present in the stat_analysis output, so
I'm not
> sure we could use that either.
>
> I hope this helps explain what issues we're having.
>
> Thanks as always for your help,
> Dana
>
> On Mon, Feb 25, 2019 at 5:30 PM John Halley Gotway via RT <
> met_help at ucar.edu>
> wrote:
>
> > Dana,
> >
> > Is the OBS_SID column totally blank or is it set to some constant
string.
> > If it's totally blank, that'll likely mess up the parsing logic of
> > Stat-Analysis. If it's not blank, just constant, then the parsing
logic
> > should be OK.
> >
> > Instead of using "-by OBS_SID", you could try using "-by
> OBS_LAT,OBS_LON".
> > That'll concatenate the lat/lon of each station into a string and
report
> > results for the unique strings it finds.
> >
> > Does that help? Feel free to send a sample file if you have
additional
> > questions.
> >
> > Thanks,
> > John
> >
> > On Mon, Feb 25, 2019 at 1:36 PM dana.strom at noaa.gov via RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > Mon Feb 25 13:35:47 2019: Request 89077 was acted upon.
> > > Transaction: Ticket created by dana.strom at noaa.gov
> > > Queue: met_help
> > > Subject: stat_analysis question
> > > Owner: Nobody
> > > Requestors: dana.strom at noaa.gov
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077
> >
> > >
> > >
> > > Good afternoon,
> > >
> > > (John, if this is assigned to you, apologies for not
articulating this
> as
> > > well on the call today - this is a brief followup to that).
> > >
> > > We had a couple of questions regarding the output of
stat_analysis.
> We're
> > > having some issues getting that station IDs out of the .stat
files.
> > >
> > > We would like the process to be:
> > >
> > > 1. Run point_stat on each validation hour and produce a .stat
and an
> .mpr
> > > file for each valid hour, listing the verification for each
station out
> > in
> > > the output files.
> > > 2. Run stat_anlaysis monthly on each of the .stat or .mpr files
to
> > produce
> > > our aggregated scores for each station individually (aggregating
by
> > site).
> > >
> > > Unfortunately, our processing at the moment has some snags that
are
> > causing
> > > some headaches.
> > >
> > > When we run point_stat, writing both the .mpr and .stat files,
the
> > station
> > > identifier column doesn't have a column name in the .stat file
(not a
> big
> > > deal since we can get that information from the .mpr file). But
when
> run
> > > stat_analysis over the month to aggregate the station data, when
we
> > > aggregate over OBS_SID, the stat_analysis output doesn't contain
the
> > unique
> > > identifier for the station causing us not to know what station
that
> > > verification belongs to.
> > >
> > > Are we missing something in our stat_analysis configuration for
it to
> > > output OBS_SID column in the .stat output? This would solve all
of our
> > > problems. :)
> > >
> > > Thanks much for your help,
> > > Dana
> > >
> > > --
> > > Dana Strom
> > > Verification and Visualization Task Lead
> > > Meteorological Development Lab
> > > NOAA/National Weather Service
> > >
> > > 301-427-9451
> > >
> > >
> >
> >
>
> --
> Dana Strom
> Verification and Visualization Task Lead
> Meteorological Development Lab
> NOAA/National Weather Service
>
> 301-427-9451
>
>
------------------------------------------------
Subject: stat_analysis question
From: John Halley Gotway
Time: Wed Feb 27 14:42:44 2019
Dana,
Good news.
I was able to update the development version of MET to support the
changes
you need. In met-8.1, Stat-Analysis will support:
-job aggregate_stat -line_type MPR -out_line_type CNT -out_stat
out.stat
-by FCST_VAR,OBS_SID -set_hdr VX_MASK OBS_SID -set_hdr DESC CASE
Where the "-set_hdr" options control the output written to "out.stat":
- VX_MASK in the out.stat file will contain the current observation
station
ID name, such as "KDEN".
- DECS in the out.stat file will contain the full case information
string,
such as "TMP:KDEN".
Hopefully that'll solve this issue.
Thanks,
John
On Tue, Feb 26, 2019 at 11:30 AM John Halley Gotway <johnhg at ucar.edu>
wrote:
> Dana,
>
> Ah, OK, I understand. Please try re-running your job using both the
> "-out" and the "-out_stat" options.
>
> You'll find that the output file created with "-out" does actually
contain
> the OBS_SID column strings.
> However the output created with "-out_stat" does not.
>
> For example, using a sample output MPR file create by "make test", I
ran
> this job:
>
> *stat_analysis -lookin
> out/point_stat/point_stat_360000L_20070331_120000V_mpr.txt \*
> *-job aggregate_stat -line_type MPR -out_line_type CNT -by
> FCST_VAR,OBS_SID \*
> *-out mpr_to_cnt.out -out_stat mpr_to_cnt.stat*
>
> The *mpr_to_cnt.out* file does contain the OBS_SID output column...
but
> lacks the full 22 STAT header columns.
> The *mpr_to_cnt.stat* file does have the full 22 STAT header
columns...
> but lacks the OBS_SID info.
>
> I won't go into the background as to why these 2 output file types
exist.
> There is a job command option named "-set_hdr" which allows you to
specify
> a value for the header columns written to the "-out_stat" output
column.
> However, that's a constant string for all of the output for that
job. For
> example, settings "-set_hdr VX_MASK TEST" would put the string
"TEST" into
> the VX_MASK output column.
>
> However, what we really need here is a way of writing the case-
specific
> info to the -out_stat output file in some way. In this example,
it'd be
> most useful to write the OBS_SID value to the VX_MASK output column.
So
> supporting something like:
> * -set_hdr VX_MASK OBS_SID*
> Would do the trick.
>
> Do you agree that would be a useful enhancement?
>
> Thanks,
> John
>
>
> On Tue, Feb 26, 2019 at 8:47 AM dana.strom at noaa.gov via RT <
> met_help at ucar.edu> wrote:
>
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
>>
>> Hey John,
>>
>>
>> * Is the OBS_SID column totally blank or is it set to some constant
>> string.
>> *
>>
>> *point_stat *contains the OBS_SID column with proper names of the
stations
>> we provided
>> *stat_analysis* does NOT contain a OBS_SID column in the output,
even when
>> specifying that as the stratifier. It DOES output all of the
stations, we
>> just don't know which one is which because the OBS_SID column does
not
>> exist in the stat_analysis output.
>>
>> We have not attempted to "-by OBS_LAT,OBS_LON". However, we did
notice
>> that
>> OBS_LAT and OBS_LON are not present in the stat_analysis output, so
I'm
>> not
>> sure we could use that either.
>>
>> I hope this helps explain what issues we're having.
>>
>> Thanks as always for your help,
>> Dana
>>
>> On Mon, Feb 25, 2019 at 5:30 PM John Halley Gotway via RT <
>> met_help at ucar.edu>
>> wrote:
>>
>> > Dana,
>> >
>> > Is the OBS_SID column totally blank or is it set to some constant
>> string.
>> > If it's totally blank, that'll likely mess up the parsing logic
of
>> > Stat-Analysis. If it's not blank, just constant, then the
parsing logic
>> > should be OK.
>> >
>> > Instead of using "-by OBS_SID", you could try using "-by
>> OBS_LAT,OBS_LON".
>> > That'll concatenate the lat/lon of each station into a string and
report
>> > results for the unique strings it finds.
>> >
>> > Does that help? Feel free to send a sample file if you have
additional
>> > questions.
>> >
>> > Thanks,
>> > John
>> >
>> > On Mon, Feb 25, 2019 at 1:36 PM dana.strom at noaa.gov via RT <
>> > met_help at ucar.edu> wrote:
>> >
>> > >
>> > > Mon Feb 25 13:35:47 2019: Request 89077 was acted upon.
>> > > Transaction: Ticket created by dana.strom at noaa.gov
>> > > Queue: met_help
>> > > Subject: stat_analysis question
>> > > Owner: Nobody
>> > > Requestors: dana.strom at noaa.gov
>> > > Status: new
>> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077
>> >
>> > >
>> > >
>> > > Good afternoon,
>> > >
>> > > (John, if this is assigned to you, apologies for not
articulating
>> this as
>> > > well on the call today - this is a brief followup to that).
>> > >
>> > > We had a couple of questions regarding the output of
stat_analysis.
>> We're
>> > > having some issues getting that station IDs out of the .stat
files.
>> > >
>> > > We would like the process to be:
>> > >
>> > > 1. Run point_stat on each validation hour and produce a .stat
and an
>> .mpr
>> > > file for each valid hour, listing the verification for each
station
>> out
>> > in
>> > > the output files.
>> > > 2. Run stat_anlaysis monthly on each of the .stat or .mpr files
to
>> > produce
>> > > our aggregated scores for each station individually
(aggregating by
>> > site).
>> > >
>> > > Unfortunately, our processing at the moment has some snags that
are
>> > causing
>> > > some headaches.
>> > >
>> > > When we run point_stat, writing both the .mpr and .stat files,
the
>> > station
>> > > identifier column doesn't have a column name in the .stat file
(not a
>> big
>> > > deal since we can get that information from the .mpr file). But
when
>> run
>> > > stat_analysis over the month to aggregate the station data,
when we
>> > > aggregate over OBS_SID, the stat_analysis output doesn't
contain the
>> > unique
>> > > identifier for the station causing us not to know what station
that
>> > > verification belongs to.
>> > >
>> > > Are we missing something in our stat_analysis configuration for
it to
>> > > output OBS_SID column in the .stat output? This would solve all
of our
>> > > problems. :)
>> > >
>> > > Thanks much for your help,
>> > > Dana
>> > >
>> > > --
>> > > Dana Strom
>> > > Verification and Visualization Task Lead
>> > > Meteorological Development Lab
>> > > NOAA/National Weather Service
>> > >
>> > > 301-427-9451
>> > >
>> > >
>> >
>> >
>>
>> --
>> Dana Strom
>> Verification and Visualization Task Lead
>> Meteorological Development Lab
>> NOAA/National Weather Service
>>
>> 301-427-9451
>>
>>
------------------------------------------------
Subject: stat_analysis question
From: dana.strom at noaa.gov
Time: Wed Feb 27 14:48:39 2019
Hey John,
That's fantastic news! We're more than willing to test those
functionalities out for you if you're able to roll it up into a docker
container.
FYI - I'll be on leave Thursday and Friday, so it won't be until next
week
that we could test.
Thanks as always for your help!
Dana
On Wed, Feb 27, 2019 at 4:42 PM John Halley Gotway via RT
<met_help at ucar.edu>
wrote:
> Dana,
>
> Good news.
>
> I was able to update the development version of MET to support the
changes
> you need. In met-8.1, Stat-Analysis will support:
>
> -job aggregate_stat -line_type MPR -out_line_type CNT -out_stat
out.stat
> -by FCST_VAR,OBS_SID -set_hdr VX_MASK OBS_SID -set_hdr DESC CASE
>
> Where the "-set_hdr" options control the output written to
"out.stat":
> - VX_MASK in the out.stat file will contain the current observation
station
> ID name, such as "KDEN".
> - DECS in the out.stat file will contain the full case information
string,
> such as "TMP:KDEN".
>
> Hopefully that'll solve this issue.
>
> Thanks,
> John
>
> On Tue, Feb 26, 2019 at 11:30 AM John Halley Gotway
<johnhg at ucar.edu>
> wrote:
>
> > Dana,
> >
> > Ah, OK, I understand. Please try re-running your job using both
the
> > "-out" and the "-out_stat" options.
> >
> > You'll find that the output file created with "-out" does actually
> contain
> > the OBS_SID column strings.
> > However the output created with "-out_stat" does not.
> >
> > For example, using a sample output MPR file create by "make test",
I ran
> > this job:
> >
> > *stat_analysis -lookin
> > out/point_stat/point_stat_360000L_20070331_120000V_mpr.txt \*
> > *-job aggregate_stat -line_type MPR -out_line_type CNT -by
> > FCST_VAR,OBS_SID \*
> > *-out mpr_to_cnt.out -out_stat mpr_to_cnt.stat*
> >
> > The *mpr_to_cnt.out* file does contain the OBS_SID output
column... but
> > lacks the full 22 STAT header columns.
> > The *mpr_to_cnt.stat* file does have the full 22 STAT header
columns...
> > but lacks the OBS_SID info.
> >
> > I won't go into the background as to why these 2 output file types
exist.
> > There is a job command option named "-set_hdr" which allows you to
> specify
> > a value for the header columns written to the "-out_stat" output
column.
> > However, that's a constant string for all of the output for that
job.
> For
> > example, settings "-set_hdr VX_MASK TEST" would put the string
"TEST"
> into
> > the VX_MASK output column.
> >
> > However, what we really need here is a way of writing the case-
specific
> > info to the -out_stat output file in some way. In this example,
it'd be
> > most useful to write the OBS_SID value to the VX_MASK output
column. So
> > supporting something like:
> > * -set_hdr VX_MASK OBS_SID*
> > Would do the trick.
> >
> > Do you agree that would be a useful enhancement?
> >
> > Thanks,
> > John
> >
> >
> > On Tue, Feb 26, 2019 at 8:47 AM dana.strom at noaa.gov via RT <
> > met_help at ucar.edu> wrote:
> >
> >>
> >> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
> >>
> >> Hey John,
> >>
> >>
> >> * Is the OBS_SID column totally blank or is it set to some
constant
> >> string.
> >> *
> >>
> >> *point_stat *contains the OBS_SID column with proper names of the
> stations
> >> we provided
> >> *stat_analysis* does NOT contain a OBS_SID column in the output,
even
> when
> >> specifying that as the stratifier. It DOES output all of the
stations,
> we
> >> just don't know which one is which because the OBS_SID column
does not
> >> exist in the stat_analysis output.
> >>
> >> We have not attempted to "-by OBS_LAT,OBS_LON". However, we did
notice
> >> that
> >> OBS_LAT and OBS_LON are not present in the stat_analysis output,
so I'm
> >> not
> >> sure we could use that either.
> >>
> >> I hope this helps explain what issues we're having.
> >>
> >> Thanks as always for your help,
> >> Dana
> >>
> >> On Mon, Feb 25, 2019 at 5:30 PM John Halley Gotway via RT <
> >> met_help at ucar.edu>
> >> wrote:
> >>
> >> > Dana,
> >> >
> >> > Is the OBS_SID column totally blank or is it set to some
constant
> >> string.
> >> > If it's totally blank, that'll likely mess up the parsing logic
of
> >> > Stat-Analysis. If it's not blank, just constant, then the
parsing
> logic
> >> > should be OK.
> >> >
> >> > Instead of using "-by OBS_SID", you could try using "-by
> >> OBS_LAT,OBS_LON".
> >> > That'll concatenate the lat/lon of each station into a string
and
> report
> >> > results for the unique strings it finds.
> >> >
> >> > Does that help? Feel free to send a sample file if you have
> additional
> >> > questions.
> >> >
> >> > Thanks,
> >> > John
> >> >
> >> > On Mon, Feb 25, 2019 at 1:36 PM dana.strom at noaa.gov via RT <
> >> > met_help at ucar.edu> wrote:
> >> >
> >> > >
> >> > > Mon Feb 25 13:35:47 2019: Request 89077 was acted upon.
> >> > > Transaction: Ticket created by dana.strom at noaa.gov
> >> > > Queue: met_help
> >> > > Subject: stat_analysis question
> >> > > Owner: Nobody
> >> > > Requestors: dana.strom at noaa.gov
> >> > > Status: new
> >> > > Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077
> >> >
> >> > >
> >> > >
> >> > > Good afternoon,
> >> > >
> >> > > (John, if this is assigned to you, apologies for not
articulating
> >> this as
> >> > > well on the call today - this is a brief followup to that).
> >> > >
> >> > > We had a couple of questions regarding the output of
stat_analysis.
> >> We're
> >> > > having some issues getting that station IDs out of the .stat
files.
> >> > >
> >> > > We would like the process to be:
> >> > >
> >> > > 1. Run point_stat on each validation hour and produce a .stat
and an
> >> .mpr
> >> > > file for each valid hour, listing the verification for each
station
> >> out
> >> > in
> >> > > the output files.
> >> > > 2. Run stat_anlaysis monthly on each of the .stat or .mpr
files to
> >> > produce
> >> > > our aggregated scores for each station individually
(aggregating by
> >> > site).
> >> > >
> >> > > Unfortunately, our processing at the moment has some snags
that are
> >> > causing
> >> > > some headaches.
> >> > >
> >> > > When we run point_stat, writing both the .mpr and .stat
files, the
> >> > station
> >> > > identifier column doesn't have a column name in the .stat
file (not
> a
> >> big
> >> > > deal since we can get that information from the .mpr file).
But when
> >> run
> >> > > stat_analysis over the month to aggregate the station data,
when we
> >> > > aggregate over OBS_SID, the stat_analysis output doesn't
contain the
> >> > unique
> >> > > identifier for the station causing us not to know what
station that
> >> > > verification belongs to.
> >> > >
> >> > > Are we missing something in our stat_analysis configuration
for it
> to
> >> > > output OBS_SID column in the .stat output? This would solve
all of
> our
> >> > > problems. :)
> >> > >
> >> > > Thanks much for your help,
> >> > > Dana
> >> > >
> >> > > --
> >> > > Dana Strom
> >> > > Verification and Visualization Task Lead
> >> > > Meteorological Development Lab
> >> > > NOAA/National Weather Service
> >> > >
> >> > > 301-427-9451
> >> > >
> >> > >
> >> >
> >> >
> >>
> >> --
> >> Dana Strom
> >> Verification and Visualization Task Lead
> >> Meteorological Development Lab
> >> NOAA/National Weather Service
> >>
> >> 301-427-9451
> >>
> >>
>
>
--
Dana Strom
Verification and Visualization Task Lead
Meteorological Development Lab
NOAA/National Weather Service
301-427-9451
------------------------------------------------
Subject: stat_analysis question
From: John Halley Gotway
Time: Wed Feb 27 16:12:37 2019
Dana,
OK, sounds good. These changes will be included in the next beta
release,
met-8.1_beta2. Unfortunately, we're not in a good place to cut a
release
right now. We're making changes to the config file code to make MET
Fortify-compliant and have a few more wrinkles to iron out to get our
nightly build working again.
Good to know you'd like to test it via a Docker container.
Thanks,
John
On Wed, Feb 27, 2019 at 2:48 PM dana.strom at noaa.gov via RT <
met_help at ucar.edu> wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
>
> Hey John,
>
> That's fantastic news! We're more than willing to test those
> functionalities out for you if you're able to roll it up into a
docker
> container.
>
> FYI - I'll be on leave Thursday and Friday, so it won't be until
next week
> that we could test.
>
> Thanks as always for your help!
> Dana
>
> On Wed, Feb 27, 2019 at 4:42 PM John Halley Gotway via RT <
> met_help at ucar.edu>
> wrote:
>
> > Dana,
> >
> > Good news.
> >
> > I was able to update the development version of MET to support the
> changes
> > you need. In met-8.1, Stat-Analysis will support:
> >
> > -job aggregate_stat -line_type MPR -out_line_type CNT -out_stat
out.stat
> > -by FCST_VAR,OBS_SID -set_hdr VX_MASK OBS_SID -set_hdr DESC CASE
> >
> > Where the "-set_hdr" options control the output written to
"out.stat":
> > - VX_MASK in the out.stat file will contain the current
observation
> station
> > ID name, such as "KDEN".
> > - DECS in the out.stat file will contain the full case information
> string,
> > such as "TMP:KDEN".
> >
> > Hopefully that'll solve this issue.
> >
> > Thanks,
> > John
> >
> > On Tue, Feb 26, 2019 at 11:30 AM John Halley Gotway
<johnhg at ucar.edu>
> > wrote:
> >
> > > Dana,
> > >
> > > Ah, OK, I understand. Please try re-running your job using both
the
> > > "-out" and the "-out_stat" options.
> > >
> > > You'll find that the output file created with "-out" does
actually
> > contain
> > > the OBS_SID column strings.
> > > However the output created with "-out_stat" does not.
> > >
> > > For example, using a sample output MPR file create by "make
test", I
> ran
> > > this job:
> > >
> > > *stat_analysis -lookin
> > > out/point_stat/point_stat_360000L_20070331_120000V_mpr.txt \*
> > > *-job aggregate_stat -line_type MPR -out_line_type CNT -by
> > > FCST_VAR,OBS_SID \*
> > > *-out mpr_to_cnt.out -out_stat mpr_to_cnt.stat*
> > >
> > > The *mpr_to_cnt.out* file does contain the OBS_SID output
column... but
> > > lacks the full 22 STAT header columns.
> > > The *mpr_to_cnt.stat* file does have the full 22 STAT header
columns...
> > > but lacks the OBS_SID info.
> > >
> > > I won't go into the background as to why these 2 output file
types
> exist.
> > > There is a job command option named "-set_hdr" which allows you
to
> > specify
> > > a value for the header columns written to the "-out_stat" output
> column.
> > > However, that's a constant string for all of the output for that
job.
> > For
> > > example, settings "-set_hdr VX_MASK TEST" would put the string
"TEST"
> > into
> > > the VX_MASK output column.
> > >
> > > However, what we really need here is a way of writing the case-
specific
> > > info to the -out_stat output file in some way. In this example,
it'd
> be
> > > most useful to write the OBS_SID value to the VX_MASK output
column.
> So
> > > supporting something like:
> > > * -set_hdr VX_MASK OBS_SID*
> > > Would do the trick.
> > >
> > > Do you agree that would be a useful enhancement?
> > >
> > > Thanks,
> > > John
> > >
> > >
> > > On Tue, Feb 26, 2019 at 8:47 AM dana.strom at noaa.gov via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > >>
> > >> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077 >
> > >>
> > >> Hey John,
> > >>
> > >>
> > >> * Is the OBS_SID column totally blank or is it set to some
constant
> > >> string.
> > >> *
> > >>
> > >> *point_stat *contains the OBS_SID column with proper names of
the
> > stations
> > >> we provided
> > >> *stat_analysis* does NOT contain a OBS_SID column in the
output, even
> > when
> > >> specifying that as the stratifier. It DOES output all of the
stations,
> > we
> > >> just don't know which one is which because the OBS_SID column
does not
> > >> exist in the stat_analysis output.
> > >>
> > >> We have not attempted to "-by OBS_LAT,OBS_LON". However, we did
notice
> > >> that
> > >> OBS_LAT and OBS_LON are not present in the stat_analysis
output, so
> I'm
> > >> not
> > >> sure we could use that either.
> > >>
> > >> I hope this helps explain what issues we're having.
> > >>
> > >> Thanks as always for your help,
> > >> Dana
> > >>
> > >> On Mon, Feb 25, 2019 at 5:30 PM John Halley Gotway via RT <
> > >> met_help at ucar.edu>
> > >> wrote:
> > >>
> > >> > Dana,
> > >> >
> > >> > Is the OBS_SID column totally blank or is it set to some
constant
> > >> string.
> > >> > If it's totally blank, that'll likely mess up the parsing
logic of
> > >> > Stat-Analysis. If it's not blank, just constant, then the
parsing
> > logic
> > >> > should be OK.
> > >> >
> > >> > Instead of using "-by OBS_SID", you could try using "-by
> > >> OBS_LAT,OBS_LON".
> > >> > That'll concatenate the lat/lon of each station into a string
and
> > report
> > >> > results for the unique strings it finds.
> > >> >
> > >> > Does that help? Feel free to send a sample file if you have
> > additional
> > >> > questions.
> > >> >
> > >> > Thanks,
> > >> > John
> > >> >
> > >> > On Mon, Feb 25, 2019 at 1:36 PM dana.strom at noaa.gov via RT <
> > >> > met_help at ucar.edu> wrote:
> > >> >
> > >> > >
> > >> > > Mon Feb 25 13:35:47 2019: Request 89077 was acted upon.
> > >> > > Transaction: Ticket created by dana.strom at noaa.gov
> > >> > > Queue: met_help
> > >> > > Subject: stat_analysis question
> > >> > > Owner: Nobody
> > >> > > Requestors: dana.strom at noaa.gov
> > >> > > Status: new
> > >> > > Ticket <URL:
> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=89077
> > >> >
> > >> > >
> > >> > >
> > >> > > Good afternoon,
> > >> > >
> > >> > > (John, if this is assigned to you, apologies for not
articulating
> > >> this as
> > >> > > well on the call today - this is a brief followup to that).
> > >> > >
> > >> > > We had a couple of questions regarding the output of
> stat_analysis.
> > >> We're
> > >> > > having some issues getting that station IDs out of the
.stat
> files.
> > >> > >
> > >> > > We would like the process to be:
> > >> > >
> > >> > > 1. Run point_stat on each validation hour and produce a
.stat and
> an
> > >> .mpr
> > >> > > file for each valid hour, listing the verification for each
> station
> > >> out
> > >> > in
> > >> > > the output files.
> > >> > > 2. Run stat_anlaysis monthly on each of the .stat or .mpr
files to
> > >> > produce
> > >> > > our aggregated scores for each station individually
(aggregating
> by
> > >> > site).
> > >> > >
> > >> > > Unfortunately, our processing at the moment has some snags
that
> are
> > >> > causing
> > >> > > some headaches.
> > >> > >
> > >> > > When we run point_stat, writing both the .mpr and .stat
files, the
> > >> > station
> > >> > > identifier column doesn't have a column name in the .stat
file
> (not
> > a
> > >> big
> > >> > > deal since we can get that information from the .mpr file).
But
> when
> > >> run
> > >> > > stat_analysis over the month to aggregate the station data,
when
> we
> > >> > > aggregate over OBS_SID, the stat_analysis output doesn't
contain
> the
> > >> > unique
> > >> > > identifier for the station causing us not to know what
station
> that
> > >> > > verification belongs to.
> > >> > >
> > >> > > Are we missing something in our stat_analysis configuration
for it
> > to
> > >> > > output OBS_SID column in the .stat output? This would solve
all of
> > our
> > >> > > problems. :)
> > >> > >
> > >> > > Thanks much for your help,
> > >> > > Dana
> > >> > >
> > >> > > --
> > >> > > Dana Strom
> > >> > > Verification and Visualization Task Lead
> > >> > > Meteorological Development Lab
> > >> > > NOAA/National Weather Service
> > >> > >
> > >> > > 301-427-9451
> > >> > >
> > >> > >
> > >> >
> > >> >
> > >>
> > >> --
> > >> Dana Strom
> > >> Verification and Visualization Task Lead
> > >> Meteorological Development Lab
> > >> NOAA/National Weather Service
> > >>
> > >> 301-427-9451
> > >>
> > >>
> >
> >
>
> --
> Dana Strom
> Verification and Visualization Task Lead
> Meteorological Development Lab
> NOAA/National Weather Service
>
> 301-427-9451
>
>
------------------------------------------------
More information about the Met_help
mailing list