[Met_help] [rt.rap.ucar.edu #99819] History for Having an issue running a PointStat job - Guidance welcome please
John Opatz via RT
met_help at ucar.edu
Tue Jun 15 11:05:23 MDT 2021
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi,
Have probably just missed something off somewhere, or put a typo in somewhere, but was hoping someone would be able to assist in telling me what I am doing wrong. I am able to replicate the issue when I run the PointStat job through METplus and also through a MET standalone setup which is suggesting it is the data.
The error I am seeing is:
DEBUG 1: Forecast File: /scratch/cfrd/NGVER/ncdf_march/prods_op_gl-mn_20210302_00_012.nc
DEBUG 1: Observation File: /data/users/cfrd/METplus_Out/output/ascii2nc/lndsyn_2021030212.nc
WARNING: MetNcCFDataFile::collect_time_offsets(VarInfo &) -> Not found time out of 1.
ERROR :
ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) -> variable "air_temperature" has unknown error (999).
ERROR :
It suggests it is something to do with a time setting somewhere, but not sure what and where. I've been looking in src/libcode/vx_data2d_nccf/data2d_nccf.cc and it seems to be with he setting up of time information (using info within the netcdf file?), but my expertise with netcdf files is minimal.
Context:
Trying to run a PointStat job using a Netcdf file of observations created through ascii2nc and a Netcdf file of MetOffice UM data. The MetOffice UM data does work when running it through a GridStat job (which suggests the codebase does different things between GridStat and PointStat). The filenames show the initialisation time and forecast range, so in this example I am trying to run a 12hr forecast valid at 12Z on the 2nd March (note our day and month values maybe the other way around to what you are used to).
I have previously got this PointStat job to run using a GRIB1 forecast file and an observation file generated in the same way as just mentioned.
I have successfully run plot_point_obs and plot_data_plane for the two files in question (so have tested that out)
I have attached the METplus output from one of my attempts which was run with verbosity set to 5. I have also attached the offending netcdf file in case you can spot what is wrong with it. It seems as though it contains all of the information it requires, but I did find it interesting that when I examined it (using one of our office viewers) comparing it to other files which did appear to run OK in PointStat, I couldn't see the time information (although it suggests it is there with ncdump).
Have also included the file of observations I was trying to run with.
Any guidance welcome.
Regards
Rob
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Having an issue running a PointStat job - Guidance welcome please
From: John Opatz
Time: Fri May 07 10:15:51 2021
Hi Rob,
Before we dig into this a bit more, would it be possible to share the
METplus PointStat configuration file you're working with? It'll help
make
sure we consider all angles of where the error may be coming from.
When you've replied, I'll see if I can replicate the error you're
experiencing and have the METplus wrapper folks take a look at it as
well.
Thanks,
-John O.
On Fri, May 7, 2021 at 8:05 AM Darvell, Rob via RT <met_help at ucar.edu>
wrote:
>
> Fri May 07 08:05:12 2021: Request 99819 was acted upon.
> Transaction: Ticket created by rob.darvell at metoffice.gov.uk
> Queue: met_help
> Subject: Having an issue running a PointStat job - Guidance
welcome
> please
> Owner: Nobody
> Requestors: rob.darvell at metoffice.gov.uk
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
>
>
> Hi,
>
> Have probably just missed something off somewhere, or put a typo in
> somewhere, but was hoping someone would be able to assist in telling
me
> what I am doing wrong. I am able to replicate the issue when I run
the
> PointStat job through METplus and also through a MET standalone
setup which
> is suggesting it is the data.
>
> The error I am seeing is:
> DEBUG 1: Forecast File: /scratch/cfrd/NGVER/ncdf_march/
> prods_op_gl-mn_20210302_00_012.nc
> DEBUG 1: Observation File:
/data/users/cfrd/METplus_Out/output/ascii2nc/
> lndsyn_2021030212.nc
> WARNING: MetNcCFDataFile::collect_time_offsets(VarInfo &) -> Not
found
> time out of 1.
> ERROR :
> ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
variable
> "air_temperature" has unknown error (999).
> ERROR :
>
> It suggests it is something to do with a time setting somewhere, but
not
> sure what and where. I've been looking in
> src/libcode/vx_data2d_nccf/data2d_nccf.cc and it seems to be with he
> setting up of time information (using info within the netcdf file?),
but my
> expertise with netcdf files is minimal.
>
> Context:
> Trying to run a PointStat job using a Netcdf file of observations
created
> through ascii2nc and a Netcdf file of MetOffice UM data. The
MetOffice UM
> data does work when running it through a GridStat job (which
suggests the
> codebase does different things between GridStat and PointStat). The
> filenames show the initialisation time and forecast range, so in
this
> example I am trying to run a 12hr forecast valid at 12Z on the 2nd
March
> (note our day and month values maybe the other way around to what
you are
> used to).
> I have previously got this PointStat job to run using a GRIB1
forecast
> file and an observation file generated in the same way as just
mentioned.
>
> I have successfully run plot_point_obs and plot_data_plane for the
two
> files in question (so have tested that out)
>
> I have attached the METplus output from one of my attempts which was
run
> with verbosity set to 5. I have also attached the offending netcdf
file in
> case you can spot what is wrong with it. It seems as though it
contains all
> of the information it requires, but I did find it interesting that
when I
> examined it (using one of our office viewers) comparing it to other
files
> which did appear to run OK in PointStat, I couldn't see the time
> information (although it suggests it is there with ncdump).
> Have also included the file of observations I was trying to run
with.
>
> Any guidance welcome.
>
> Regards
> Rob
>
>
--
John Opatz
Associate Scientist III
NCAR RAL and DTC
Boulder, Colorado
+1 303-497-8305
------------------------------------------------
Subject: Having an issue running a PointStat job - Guidance welcome please
From: John Halley Gotway
Time: Fri May 07 10:54:16 2021
John O and Rob,
I ran the following point_stat command using the input data that Rob
sent
and was able to replicate the error:
*point_stat prods_op_gl-mn_20210302_00_012.nc
<http://prods_op_gl-mn_20210302_00_012.nc> lndsyn_2021030212.nc
<http://lndsyn_2021030212.nc> PointStatConfig -outdir out*
*ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) -> variable
"air_temperature" has unknown error (999).*
I used the attached Point-Stat config file. The behavior is the same
for
both met-9.1 and the development version of met-10.0.0.
This is disconcerting because plot_data_plane does in fact plot this
data
without throwing any errors:
*plot_data_plane prods_op_gl-mn_20210302_00_012.nc
<http://prods_op_gl-mn_20210302_00_012.nc>
prods_op_gl-mn_20210302_00_012.ps
<http://prods_op_gl-mn_20210302_00_012.ps> 'name="air_temperature";
level="(*,*)";' -v 4*
This is a CF-compliant NetCDF file. Here's a selection of ncdump -h
output:
float air_temperature(latitude, longitude) ;
...
double time ;
It's most often the case that data variables are actually indexed by a
time
dimension... like air_temperature(time,latitude,longitude); But I do
recall
asking Howard to support this flavor of CF-compliant files as well...
as
evidenced by plot_data_plane. I strongly suspect that this is the
detail
causing trouble in this case. I'll work through the library code and
see if
I can fix it.
Thanks,
John HG
On Fri, May 7, 2021 at 10:16 AM John Opatz via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
>
> Hi Rob,
>
> Before we dig into this a bit more, would it be possible to share
the
> METplus PointStat configuration file you're working with? It'll help
make
> sure we consider all angles of where the error may be coming from.
>
> When you've replied, I'll see if I can replicate the error you're
> experiencing and have the METplus wrapper folks take a look at it as
well.
>
> Thanks,
>
> -John O.
>
> On Fri, May 7, 2021 at 8:05 AM Darvell, Rob via RT
<met_help at ucar.edu>
> wrote:
>
> >
> > Fri May 07 08:05:12 2021: Request 99819 was acted upon.
> > Transaction: Ticket created by rob.darvell at metoffice.gov.uk
> > Queue: met_help
> > Subject: Having an issue running a PointStat job - Guidance
welcome
> > please
> > Owner: Nobody
> > Requestors: rob.darvell at metoffice.gov.uk
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
> >
> >
> > Hi,
> >
> > Have probably just missed something off somewhere, or put a typo
in
> > somewhere, but was hoping someone would be able to assist in
telling me
> > what I am doing wrong. I am able to replicate the issue when I run
the
> > PointStat job through METplus and also through a MET standalone
setup
> which
> > is suggesting it is the data.
> >
> > The error I am seeing is:
> > DEBUG 1: Forecast File: /scratch/cfrd/NGVER/ncdf_march/
> > prods_op_gl-mn_20210302_00_012.nc
> > DEBUG 1: Observation File:
/data/users/cfrd/METplus_Out/output/ascii2nc/
> > lndsyn_2021030212.nc
> > WARNING: MetNcCFDataFile::collect_time_offsets(VarInfo &) -> Not
found
> > time out of 1.
> > ERROR :
> > ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
variable
> > "air_temperature" has unknown error (999).
> > ERROR :
> >
> > It suggests it is something to do with a time setting somewhere,
but not
> > sure what and where. I've been looking in
> > src/libcode/vx_data2d_nccf/data2d_nccf.cc and it seems to be with
he
> > setting up of time information (using info within the netcdf
file?), but
> my
> > expertise with netcdf files is minimal.
> >
> > Context:
> > Trying to run a PointStat job using a Netcdf file of observations
created
> > through ascii2nc and a Netcdf file of MetOffice UM data. The
MetOffice UM
> > data does work when running it through a GridStat job (which
suggests the
> > codebase does different things between GridStat and PointStat).
The
> > filenames show the initialisation time and forecast range, so in
this
> > example I am trying to run a 12hr forecast valid at 12Z on the 2nd
March
> > (note our day and month values maybe the other way around to what
you are
> > used to).
> > I have previously got this PointStat job to run using a GRIB1
forecast
> > file and an observation file generated in the same way as just
mentioned.
> >
> > I have successfully run plot_point_obs and plot_data_plane for the
two
> > files in question (so have tested that out)
> >
> > I have attached the METplus output from one of my attempts which
was run
> > with verbosity set to 5. I have also attached the offending netcdf
file
> in
> > case you can spot what is wrong with it. It seems as though it
contains
> all
> > of the information it requires, but I did find it interesting that
when I
> > examined it (using one of our office viewers) comparing it to
other files
> > which did appear to run OK in PointStat, I couldn't see the time
> > information (although it suggests it is there with ncdump).
> > Have also included the file of observations I was trying to run
with.
> >
> > Any guidance welcome.
> >
> > Regards
> > Rob
> >
> >
>
> --
> John Opatz
> Associate Scientist III
> NCAR RAL and DTC
> Boulder, Colorado
> +1 303-497-8305
>
>
------------------------------------------------
Subject: Having an issue running a PointStat job - Guidance welcome please
From: John Halley Gotway
Time: Fri May 07 11:53:53 2021
FYI, here's a GitHub issue to fix up this logic:
https://github.com/dtcenter/MET/issues/1782
Fingers crossed that we can get it fixed quickly and included in next
week's met-10.0.0 release.
John HG
On Fri, May 7, 2021 at 10:54 AM John Halley Gotway <johnhg at ucar.edu>
wrote:
> John O and Rob,
>
> I ran the following point_stat command using the input data that Rob
sent
> and was able to replicate the error:
>
> *point_stat prods_op_gl-mn_20210302_00_012.nc
> <http://prods_op_gl-mn_20210302_00_012.nc> lndsyn_2021030212.nc
> <http://lndsyn_2021030212.nc> PointStatConfig -outdir out*
>
> *ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
variable
> "air_temperature" has unknown error (999).*
>
> I used the attached Point-Stat config file. The behavior is the same
for
> both met-9.1 and the development version of met-10.0.0.
>
> This is disconcerting because plot_data_plane does in fact plot this
data
> without throwing any errors:
>
> *plot_data_plane prods_op_gl-mn_20210302_00_012.nc
> <http://prods_op_gl-mn_20210302_00_012.nc>
> prods_op_gl-mn_20210302_00_012.ps
> <http://prods_op_gl-mn_20210302_00_012.ps> 'name="air_temperature";
> level="(*,*)";' -v 4*
>
> This is a CF-compliant NetCDF file. Here's a selection of ncdump -h
output:
>
> float air_temperature(latitude, longitude) ;
> ...
> double time ;
>
> It's most often the case that data variables are actually indexed by
a
> time dimension... like air_temperature(time,latitude,longitude); But
I do
> recall asking Howard to support this flavor of CF-compliant files as
> well... as evidenced by plot_data_plane. I strongly suspect that
this is
> the detail causing trouble in this case. I'll work through the
library code
> and see if I can fix it.
>
> Thanks,
> John HG
>
> On Fri, May 7, 2021 at 10:16 AM John Opatz via RT
<met_help at ucar.edu>
> wrote:
>
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
>>
>> Hi Rob,
>>
>> Before we dig into this a bit more, would it be possible to share
the
>> METplus PointStat configuration file you're working with? It'll
help make
>> sure we consider all angles of where the error may be coming from.
>>
>> When you've replied, I'll see if I can replicate the error you're
>> experiencing and have the METplus wrapper folks take a look at it
as well.
>>
>> Thanks,
>>
>> -John O.
>>
>> On Fri, May 7, 2021 at 8:05 AM Darvell, Rob via RT
<met_help at ucar.edu>
>> wrote:
>>
>> >
>> > Fri May 07 08:05:12 2021: Request 99819 was acted upon.
>> > Transaction: Ticket created by rob.darvell at metoffice.gov.uk
>> > Queue: met_help
>> > Subject: Having an issue running a PointStat job - Guidance
welcome
>> > please
>> > Owner: Nobody
>> > Requestors: rob.darvell at metoffice.gov.uk
>> > Status: new
>> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
>> >
>> >
>> > Hi,
>> >
>> > Have probably just missed something off somewhere, or put a typo
in
>> > somewhere, but was hoping someone would be able to assist in
telling me
>> > what I am doing wrong. I am able to replicate the issue when I
run the
>> > PointStat job through METplus and also through a MET standalone
setup
>> which
>> > is suggesting it is the data.
>> >
>> > The error I am seeing is:
>> > DEBUG 1: Forecast File: /scratch/cfrd/NGVER/ncdf_march/
>> > prods_op_gl-mn_20210302_00_012.nc
>> > DEBUG 1: Observation File:
/data/users/cfrd/METplus_Out/output/ascii2nc/
>> > lndsyn_2021030212.nc
>> > WARNING: MetNcCFDataFile::collect_time_offsets(VarInfo &) -> Not
found
>> > time out of 1.
>> > ERROR :
>> > ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
variable
>> > "air_temperature" has unknown error (999).
>> > ERROR :
>> >
>> > It suggests it is something to do with a time setting somewhere,
but not
>> > sure what and where. I've been looking in
>> > src/libcode/vx_data2d_nccf/data2d_nccf.cc and it seems to be with
he
>> > setting up of time information (using info within the netcdf
file?),
>> but my
>> > expertise with netcdf files is minimal.
>> >
>> > Context:
>> > Trying to run a PointStat job using a Netcdf file of observations
>> created
>> > through ascii2nc and a Netcdf file of MetOffice UM data. The
MetOffice
>> UM
>> > data does work when running it through a GridStat job (which
suggests
>> the
>> > codebase does different things between GridStat and PointStat).
The
>> > filenames show the initialisation time and forecast range, so in
this
>> > example I am trying to run a 12hr forecast valid at 12Z on the
2nd March
>> > (note our day and month values maybe the other way around to what
you
>> are
>> > used to).
>> > I have previously got this PointStat job to run using a GRIB1
forecast
>> > file and an observation file generated in the same way as just
>> mentioned.
>> >
>> > I have successfully run plot_point_obs and plot_data_plane for
the two
>> > files in question (so have tested that out)
>> >
>> > I have attached the METplus output from one of my attempts which
was run
>> > with verbosity set to 5. I have also attached the offending
netcdf file
>> in
>> > case you can spot what is wrong with it. It seems as though it
contains
>> all
>> > of the information it requires, but I did find it interesting
that when
>> I
>> > examined it (using one of our office viewers) comparing it to
other
>> files
>> > which did appear to run OK in PointStat, I couldn't see the time
>> > information (although it suggests it is there with ncdump).
>> > Have also included the file of observations I was trying to run
with.
>> >
>> > Any guidance welcome.
>> >
>> > Regards
>> > Rob
>> >
>> >
>>
>> --
>> John Opatz
>> Associate Scientist III
>> NCAR RAL and DTC
>> Boulder, Colorado
>> +1 303-497-8305
>>
>>
------------------------------------------------
Subject: Having an issue running a PointStat job - Guidance welcome please
From: John Opatz
Time: Fri Jun 11 13:52:31 2021
Hi Rob,
I thought I'd return to this ticket, since the Github issue John HG
created
and shared was resolved with a Pull Request on May 7th that is now
included
in the MET v10.0 main branch.
Would you be able to test this latest update against the issue you
were having and let us know if the issue is resolved?
Thanks,
-John O.
On Fri, May 7, 2021 at 11:54 AM John Halley Gotway via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
>
> FYI, here's a GitHub issue to fix up this logic:
> https://github.com/dtcenter/MET/issues/1782
>
> Fingers crossed that we can get it fixed quickly and included in
next
> week's met-10.0.0 release.
>
> John HG
>
> On Fri, May 7, 2021 at 10:54 AM John Halley Gotway <johnhg at ucar.edu>
> wrote:
>
> > John O and Rob,
> >
> > I ran the following point_stat command using the input data that
Rob sent
> > and was able to replicate the error:
> >
> > *point_stat prods_op_gl-mn_20210302_00_012.nc
> > <http://prods_op_gl-mn_20210302_00_012.nc> lndsyn_2021030212.nc
> > <http://lndsyn_2021030212.nc> PointStatConfig -outdir out*
> >
> > *ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
variable
> > "air_temperature" has unknown error (999).*
> >
> > I used the attached Point-Stat config file. The behavior is the
same for
> > both met-9.1 and the development version of met-10.0.0.
> >
> > This is disconcerting because plot_data_plane does in fact plot
this data
> > without throwing any errors:
> >
> > *plot_data_plane prods_op_gl-mn_20210302_00_012.nc
> > <http://prods_op_gl-mn_20210302_00_012.nc>
> > prods_op_gl-mn_20210302_00_012.ps
> > <http://prods_op_gl-mn_20210302_00_012.ps>
'name="air_temperature";
> > level="(*,*)";' -v 4*
> >
> > This is a CF-compliant NetCDF file. Here's a selection of ncdump
-h
> output:
> >
> > float air_temperature(latitude, longitude) ;
> > ...
> > double time ;
> >
> > It's most often the case that data variables are actually indexed
by a
> > time dimension... like air_temperature(time,latitude,longitude);
But I do
> > recall asking Howard to support this flavor of CF-compliant files
as
> > well... as evidenced by plot_data_plane. I strongly suspect that
this is
> > the detail causing trouble in this case. I'll work through the
library
> code
> > and see if I can fix it.
> >
> > Thanks,
> > John HG
> >
> > On Fri, May 7, 2021 at 10:16 AM John Opatz via RT
<met_help at ucar.edu>
> > wrote:
> >
> >>
> >> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
> >>
> >> Hi Rob,
> >>
> >> Before we dig into this a bit more, would it be possible to share
the
> >> METplus PointStat configuration file you're working with? It'll
help
> make
> >> sure we consider all angles of where the error may be coming
from.
> >>
> >> When you've replied, I'll see if I can replicate the error you're
> >> experiencing and have the METplus wrapper folks take a look at it
as
> well.
> >>
> >> Thanks,
> >>
> >> -John O.
> >>
> >> On Fri, May 7, 2021 at 8:05 AM Darvell, Rob via RT
<met_help at ucar.edu>
> >> wrote:
> >>
> >> >
> >> > Fri May 07 08:05:12 2021: Request 99819 was acted upon.
> >> > Transaction: Ticket created by rob.darvell at metoffice.gov.uk
> >> > Queue: met_help
> >> > Subject: Having an issue running a PointStat job -
Guidance
> welcome
> >> > please
> >> > Owner: Nobody
> >> > Requestors: rob.darvell at metoffice.gov.uk
> >> > Status: new
> >> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819
> >
> >> >
> >> >
> >> > Hi,
> >> >
> >> > Have probably just missed something off somewhere, or put a
typo in
> >> > somewhere, but was hoping someone would be able to assist in
telling
> me
> >> > what I am doing wrong. I am able to replicate the issue when I
run the
> >> > PointStat job through METplus and also through a MET standalone
setup
> >> which
> >> > is suggesting it is the data.
> >> >
> >> > The error I am seeing is:
> >> > DEBUG 1: Forecast File: /scratch/cfrd/NGVER/ncdf_march/
> >> > prods_op_gl-mn_20210302_00_012.nc
> >> > DEBUG 1: Observation File:
> /data/users/cfrd/METplus_Out/output/ascii2nc/
> >> > lndsyn_2021030212.nc
> >> > WARNING: MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
Not found
> >> > time out of 1.
> >> > ERROR :
> >> > ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
variable
> >> > "air_temperature" has unknown error (999).
> >> > ERROR :
> >> >
> >> > It suggests it is something to do with a time setting
somewhere, but
> not
> >> > sure what and where. I've been looking in
> >> > src/libcode/vx_data2d_nccf/data2d_nccf.cc and it seems to be
with he
> >> > setting up of time information (using info within the netcdf
file?),
> >> but my
> >> > expertise with netcdf files is minimal.
> >> >
> >> > Context:
> >> > Trying to run a PointStat job using a Netcdf file of
observations
> >> created
> >> > through ascii2nc and a Netcdf file of MetOffice UM data. The
MetOffice
> >> UM
> >> > data does work when running it through a GridStat job (which
suggests
> >> the
> >> > codebase does different things between GridStat and PointStat).
The
> >> > filenames show the initialisation time and forecast range, so
in this
> >> > example I am trying to run a 12hr forecast valid at 12Z on the
2nd
> March
> >> > (note our day and month values maybe the other way around to
what you
> >> are
> >> > used to).
> >> > I have previously got this PointStat job to run using a GRIB1
forecast
> >> > file and an observation file generated in the same way as just
> >> mentioned.
> >> >
> >> > I have successfully run plot_point_obs and plot_data_plane for
the two
> >> > files in question (so have tested that out)
> >> >
> >> > I have attached the METplus output from one of my attempts
which was
> run
> >> > with verbosity set to 5. I have also attached the offending
netcdf
> file
> >> in
> >> > case you can spot what is wrong with it. It seems as though it
> contains
> >> all
> >> > of the information it requires, but I did find it interesting
that
> when
> >> I
> >> > examined it (using one of our office viewers) comparing it to
other
> >> files
> >> > which did appear to run OK in PointStat, I couldn't see the
time
> >> > information (although it suggests it is there with ncdump).
> >> > Have also included the file of observations I was trying to run
with.
> >> >
> >> > Any guidance welcome.
> >> >
> >> > Regards
> >> > Rob
> >> >
> >> >
> >>
> >> --
> >> John Opatz
> >> Associate Scientist III
> >> NCAR RAL and DTC
> >> Boulder, Colorado
> >> +1 303-497-8305
> >>
> >>
>
>
--
John Opatz
Associate Scientist III
NCAR RAL and DTC
Boulder, Colorado
+1 303-497-8305
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #99819] Having an issue running a PointStat job - Guidance welcome please
From: Darvell, Rob
Time: Mon Jun 14 01:35:05 2021
Hi,
The fix which went into MET10.0.0 has corrected the issue which I was
seeing, and things are running fine now.
Regards
Rob
-----Original Message-----
From: John Opatz via RT <met_help at ucar.edu>
Sent: 11 June 2021 20:53
To: Darvell, Rob <rob.darvell at metoffice.gov.uk>
Subject: Re: [rt.rap.ucar.edu #99819] Having an issue running a
PointStat job - Guidance welcome please
This email was received from an external source. Always check sender
details, links & attachments.
Hi Rob,
I thought I'd return to this ticket, since the Github issue John HG
created and shared was resolved with a Pull Request on May 7th that is
now included in the MET v10.0 main branch.
Would you be able to test this latest update against the issue you
were having and let us know if the issue is resolved?
Thanks,
-John O.
On Fri, May 7, 2021 at 11:54 AM John Halley Gotway via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
>
> FYI, here's a GitHub issue to fix up this logic:
> https://github.com/dtcenter/MET/issues/1782
>
> Fingers crossed that we can get it fixed quickly and included in
next
> week's met-10.0.0 release.
>
> John HG
>
> On Fri, May 7, 2021 at 10:54 AM John Halley Gotway <johnhg at ucar.edu>
> wrote:
>
> > John O and Rob,
> >
> > I ran the following point_stat command using the input data that
Rob
> > sent and was able to replicate the error:
> >
> > *point_stat prods_op_gl-mn_20210302_00_012.nc
> > <http://prods_op_gl-mn_20210302_00_012.nc> lndsyn_2021030212.nc
> > <http://lndsyn_2021030212.nc> PointStatConfig -outdir out*
> >
> > *ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
> > variable "air_temperature" has unknown error (999).*
> >
> > I used the attached Point-Stat config file. The behavior is the
same
> > for both met-9.1 and the development version of met-10.0.0.
> >
> > This is disconcerting because plot_data_plane does in fact plot
this
> > data without throwing any errors:
> >
> > *plot_data_plane prods_op_gl-mn_20210302_00_012.nc
> > <http://prods_op_gl-mn_20210302_00_012.nc>
> > prods_op_gl-mn_20210302_00_012.ps
> > <http://prods_op_gl-mn_20210302_00_012.ps>
'name="air_temperature";
> > level="(*,*)";' -v 4*
> >
> > This is a CF-compliant NetCDF file. Here's a selection of ncdump
-h
> output:
> >
> > float air_temperature(latitude, longitude) ; ...
> > double time ;
> >
> > It's most often the case that data variables are actually indexed
by
> > a time dimension... like air_temperature(time,latitude,longitude);
> > But I do recall asking Howard to support this flavor of CF-
compliant
> > files as well... as evidenced by plot_data_plane. I strongly
suspect
> > that this is the detail causing trouble in this case. I'll work
> > through the library
> code
> > and see if I can fix it.
> >
> > Thanks,
> > John HG
> >
> > On Fri, May 7, 2021 at 10:16 AM John Opatz via RT
> > <met_help at ucar.edu>
> > wrote:
> >
> >>
> >> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819 >
> >>
> >> Hi Rob,
> >>
> >> Before we dig into this a bit more, would it be possible to share
> >> the METplus PointStat configuration file you're working with?
It'll
> >> help
> make
> >> sure we consider all angles of where the error may be coming
from.
> >>
> >> When you've replied, I'll see if I can replicate the error you're
> >> experiencing and have the METplus wrapper folks take a look at it
> >> as
> well.
> >>
> >> Thanks,
> >>
> >> -John O.
> >>
> >> On Fri, May 7, 2021 at 8:05 AM Darvell, Rob via RT
> >> <met_help at ucar.edu>
> >> wrote:
> >>
> >> >
> >> > Fri May 07 08:05:12 2021: Request 99819 was acted upon.
> >> > Transaction: Ticket created by rob.darvell at metoffice.gov.uk
> >> > Queue: met_help
> >> > Subject: Having an issue running a PointStat job -
Guidance
> welcome
> >> > please
> >> > Owner: Nobody
> >> > Requestors: rob.darvell at metoffice.gov.uk
> >> > Status: new
> >> > Ticket <URL:
> >> > https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=99819
> >
> >> >
> >> >
> >> > Hi,
> >> >
> >> > Have probably just missed something off somewhere, or put a
typo
> >> > in somewhere, but was hoping someone would be able to assist in
> >> > telling
> me
> >> > what I am doing wrong. I am able to replicate the issue when I
> >> > run the PointStat job through METplus and also through a MET
> >> > standalone setup
> >> which
> >> > is suggesting it is the data.
> >> >
> >> > The error I am seeing is:
> >> > DEBUG 1: Forecast File: /scratch/cfrd/NGVER/ncdf_march/
> >> > prods_op_gl-mn_20210302_00_012.nc DEBUG 1: Observation File:
> /data/users/cfrd/METplus_Out/output/ascii2nc/
> >> > lndsyn_2021030212.nc
> >> > WARNING: MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
Not
> >> > found time out of 1.
> >> > ERROR :
> >> > ERROR : MetNcCFDataFile::collect_time_offsets(VarInfo &) ->
> >> > variable "air_temperature" has unknown error (999).
> >> > ERROR :
> >> >
> >> > It suggests it is something to do with a time setting
somewhere,
> >> > but
> not
> >> > sure what and where. I've been looking in
> >> > src/libcode/vx_data2d_nccf/data2d_nccf.cc and it seems to be
with
> >> > he setting up of time information (using info within the netcdf
> >> > file?),
> >> but my
> >> > expertise with netcdf files is minimal.
> >> >
> >> > Context:
> >> > Trying to run a PointStat job using a Netcdf file of
observations
> >> created
> >> > through ascii2nc and a Netcdf file of MetOffice UM data. The
> >> > MetOffice
> >> UM
> >> > data does work when running it through a GridStat job (which
> >> > suggests
> >> the
> >> > codebase does different things between GridStat and PointStat).
> >> > The filenames show the initialisation time and forecast range,
so
> >> > in this example I am trying to run a 12hr forecast valid at 12Z
> >> > on the 2nd
> March
> >> > (note our day and month values maybe the other way around to
what
> >> > you
> >> are
> >> > used to).
> >> > I have previously got this PointStat job to run using a GRIB1
> >> > forecast file and an observation file generated in the same way
> >> > as just
> >> mentioned.
> >> >
> >> > I have successfully run plot_point_obs and plot_data_plane for
> >> > the two files in question (so have tested that out)
> >> >
> >> > I have attached the METplus output from one of my attempts
which
> >> > was
> run
> >> > with verbosity set to 5. I have also attached the offending
> >> > netcdf
> file
> >> in
> >> > case you can spot what is wrong with it. It seems as though it
> contains
> >> all
> >> > of the information it requires, but I did find it interesting
> >> > that
> when
> >> I
> >> > examined it (using one of our office viewers) comparing it to
> >> > other
> >> files
> >> > which did appear to run OK in PointStat, I couldn't see the
time
> >> > information (although it suggests it is there with ncdump).
> >> > Have also included the file of observations I was trying to run
with.
> >> >
> >> > Any guidance welcome.
> >> >
> >> > Regards
> >> > Rob
> >> >
> >> >
> >>
> >> --
> >> John Opatz
> >> Associate Scientist III
> >> NCAR RAL and DTC
> >> Boulder, Colorado
> >> +1 303-497-8305
> >>
> >>
>
>
--
John Opatz
Associate Scientist III
NCAR RAL and DTC
Boulder, Colorado
+1 303-497-8305
------------------------------------------------
More information about the Met_help
mailing list