[Met_help] [rt.rap.ucar.edu #77523] History for grid_stat questions
John Halley Gotway via RT
met_help at ucar.edu
Fri Aug 12 11:38:13 MDT 2016
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi,
I'm attempting to use MET seriously for the first time to verify some
probabilistic precipitation forecasts. The tutorial slides were very
helpful in getting me going in the right direction, and grid_stat seems to
be running without complaint if I do things a certain way, but only the *
pairs.nc is populated with actual data - the .stat and .txt files just have
the header line.
First question is GRIB2 related - is the Product Definition Template Number
figured out internally by the code? In the grib2_vars_flat.txt file I
didn't see a way to distinguish deterministic APCP from probabilistic
APCP. It seems like it might be treating it as a normal APCP, at least in
the *pairs.nc file the forecast field gets labeled as being in kg/m^2.
Second is that I seem to have to provide as many observational cat_thresh
values as there are categories in the forecast block. This contradicts the
tutorial slide I'm following, so maybe provides a clue as to what is going
wrong for me. Doing it this way:
field = [
{
name = "APCP";
level = [ "A24" ]; *# <------ Note...doing as "(*,*)" as in
the example I was following created failures. Maybe specifying as A24 is
problematic??*
prob = TRUE;
cat_thresh = [ >=0.0, >=0.1, >=0.2, >=0.3, >=0.4, >=0.5, >=0.6,
>=0.7, >=0.8, >=0.9, >=1.0 ];
}
];
obs = {
field = [
{
name = "APCP";
level = [ "A24" ];
cat_thresh = [ >25.4 ];
}
];
}
Gives me this error message:
ERROR : GridStatConfInfo::process_config() -> The number of thresholds for
each field in "fcst.cat_thresh" must match the number of thresholds for
each field in "obs.cat_thresh".
Any hints on where I'm going wrong?
Thanks,
Matt
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: grid_stat questions
From: John Halley Gotway
Time: Thu Aug 11 13:08:12 2016
Hi Matt,
Getting started with configuration files can be tricky, particularly
for
probabilistic data. Perhaps it'd be simplest to work through an
example.
Please either point me to your gridded data files (i.e. on theia
perhaps)
or post them to our anonymous ftp site following these instructions:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
I need a sample probabilistic forecast file as well as the verifying
observations.
And also send me a description of the what you're trying to verify.
Presumably, you want to verify probability of precip exceeding a
certain
amount.
I'll work up a suggested Grid-Stat configuration file and send it back
to
you.
Thanks,
John Halley Gotway
On Thu, Aug 11, 2016 at 9:37 AM, matthew.pyle at noaa.gov via RT <
met_help at ucar.edu> wrote:
>
> Thu Aug 11 09:37:16 2016: Request 77523 was acted upon.
> Transaction: Ticket created by Matthew.Pyle at noaa.gov
> Queue: met_help
> Subject: grid_stat questions
> Owner: Nobody
> Requestors: Matthew.Pyle at noaa.gov
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77523 >
>
>
> Hi,
>
> I'm attempting to use MET seriously for the first time to verify
some
> probabilistic precipitation forecasts. The tutorial slides were
very
> helpful in getting me going in the right direction, and grid_stat
seems to
> be running without complaint if I do things a certain way, but only
the *
> pairs.nc is populated with actual data - the .stat and .txt files
just
> have
> the header line.
>
> First question is GRIB2 related - is the Product Definition Template
Number
> figured out internally by the code? In the grib2_vars_flat.txt file
I
> didn't see a way to distinguish deterministic APCP from
probabilistic
> APCP. It seems like it might be treating it as a normal APCP, at
least in
> the *pairs.nc file the forecast field gets labeled as being in
kg/m^2.
>
> Second is that I seem to have to provide as many observational
cat_thresh
> values as there are categories in the forecast block. This
contradicts the
> tutorial slide I'm following, so maybe provides a clue as to what is
going
> wrong for me. Doing it this way:
>
>
> field = [
> {
> name = "APCP";
> level = [ "A24" ]; *# <------ Note...doing as "(*,*)"
as in
> the example I was following created failures. Maybe specifying as
A24 is
> problematic??*
> prob = TRUE;
> cat_thresh = [ >=0.0, >=0.1, >=0.2, >=0.3, >=0.4, >=0.5,
>=0.6,
> >=0.7, >=0.8, >=0.9, >=1.0 ];
> }
> ];
>
> obs = {
>
> field = [
> {
> name = "APCP";
> level = [ "A24" ];
> cat_thresh = [ >25.4 ];
> }
> ];
> }
>
> Gives me this error message:
>
> ERROR : GridStatConfInfo::process_config() -> The number of
thresholds
> for
> each field in "fcst.cat_thresh" must match the number of thresholds
for
> each field in "obs.cat_thresh".
>
> Any hints on where I'm going wrong?
>
> Thanks,
>
> Matt
>
>
------------------------------------------------
Subject: grid_stat questions
From: "matthew.pyle at noaa.gov"
Time: Thu Aug 11 13:31:07 2016
Glad to know I wasn't overlooking something obvious, and thanks for
you
offer to help! I've uploaded stuff onto theia in
/scratch4/NCEPDEV/meso/save/Matthew.Pyle/dtc_stuff/met_case/
fcst.t12z.grib2.218 (ensemble probability of 24 h QPF > 25.4 mm)
ccpa.2016080612.24h.218 (24 h precip analysis file)
Ultimately I'm trying to plot some reliability diagrams for various
experiments and various thresholds. I'm using the 5.1_bugfix version
if
that makes a difference.
-Matt
On Thu, Aug 11, 2016 at 3:08 PM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:
> Hi Matt,
>
> Getting started with configuration files can be tricky, particularly
for
> probabilistic data. Perhaps it'd be simplest to work through an
example.
> Please either point me to your gridded data files (i.e. on theia
perhaps)
> or post them to our anonymous ftp site following these instructions:
> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>
> I need a sample probabilistic forecast file as well as the verifying
> observations.
>
> And also send me a description of the what you're trying to verify.
> Presumably, you want to verify probability of precip exceeding a
certain
> amount.
>
> I'll work up a suggested Grid-Stat configuration file and send it
back to
> you.
>
> Thanks,
> John Halley Gotway
>
> On Thu, Aug 11, 2016 at 9:37 AM, matthew.pyle at noaa.gov via RT <
> met_help at ucar.edu> wrote:
>
> >
> > Thu Aug 11 09:37:16 2016: Request 77523 was acted upon.
> > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> > Queue: met_help
> > Subject: grid_stat questions
> > Owner: Nobody
> > Requestors: Matthew.Pyle at noaa.gov
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77523 >
> >
> >
> > Hi,
> >
> > I'm attempting to use MET seriously for the first time to verify
some
> > probabilistic precipitation forecasts. The tutorial slides were
very
> > helpful in getting me going in the right direction, and grid_stat
seems
> to
> > be running without complaint if I do things a certain way, but
only the *
> > pairs.nc is populated with actual data - the .stat and .txt files
just
> > have
> > the header line.
> >
> > First question is GRIB2 related - is the Product Definition
Template
> Number
> > figured out internally by the code? In the grib2_vars_flat.txt
file I
> > didn't see a way to distinguish deterministic APCP from
probabilistic
> > APCP. It seems like it might be treating it as a normal APCP, at
least
> in
> > the *pairs.nc file the forecast field gets labeled as being in
kg/m^2.
> >
> > Second is that I seem to have to provide as many observational
cat_thresh
> > values as there are categories in the forecast block. This
contradicts
> the
> > tutorial slide I'm following, so maybe provides a clue as to what
is
> going
> > wrong for me. Doing it this way:
> >
> >
> > field = [
> > {
> > name = "APCP";
> > level = [ "A24" ]; *# <------ Note...doing as
"(*,*)" as in
> > the example I was following created failures. Maybe specifying as
A24 is
> > problematic??*
> > prob = TRUE;
> > cat_thresh = [ >=0.0, >=0.1, >=0.2, >=0.3, >=0.4, >=0.5,
>=0.6,
> > >=0.7, >=0.8, >=0.9, >=1.0 ];
> > }
> > ];
> >
> > obs = {
> >
> > field = [
> > {
> > name = "APCP";
> > level = [ "A24" ];
> > cat_thresh = [ >25.4 ];
> > }
> > ];
> > }
> >
> > Gives me this error message:
> >
> > ERROR : GridStatConfInfo::process_config() -> The number of
thresholds
> > for
> > each field in "fcst.cat_thresh" must match the number of
thresholds for
> > each field in "obs.cat_thresh".
> >
> > Any hints on where I'm going wrong?
> >
> > Thanks,
> >
> > Matt
> >
> >
>
>
------------------------------------------------
Subject: grid_stat questions
From: John Halley Gotway
Time: Thu Aug 11 14:12:01 2016
Matt,
OK, please try using this configuration file:
/home/John.H.Gotway/MET_Help/pyle_data_20160811/GridStatConfig_APCP_24_prob
The key here is the "prob" setting. It may be set to a boolean value,
i.e.
"TRUE" or to a dictionary which define the probability information to
be
found.
So here's what works for your data:
name = "PROB";
level = [ "A24" ];
prob = { name = "APCP"; thresh_lo = 25.4; }
However, looking on page 47 of the MET-5.1 user's guide, I see that
the
description is lacking. We'll update that info for the next release.
Hopefully that'll do the trick.
Thanks,
John
On Thu, Aug 11, 2016 at 1:31 PM, matthew.pyle at noaa.gov via RT <
met_help at ucar.edu> wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77523 >
>
> Glad to know I wasn't overlooking something obvious, and thanks for
you
> offer to help! I've uploaded stuff onto theia in
> /scratch4/NCEPDEV/meso/save/Matthew.Pyle/dtc_stuff/met_case/
>
> fcst.t12z.grib2.218 (ensemble probability of 24 h QPF > 25.4 mm)
> ccpa.2016080612.24h.218 (24 h precip analysis file)
>
> Ultimately I'm trying to plot some reliability diagrams for various
> experiments and various thresholds. I'm using the 5.1_bugfix
version if
> that makes a difference.
>
> -Matt
>
>
>
> On Thu, Aug 11, 2016 at 3:08 PM, John Halley Gotway via RT <
> met_help at ucar.edu> wrote:
>
> > Hi Matt,
> >
> > Getting started with configuration files can be tricky,
particularly for
> > probabilistic data. Perhaps it'd be simplest to work through an
> example.
> > Please either point me to your gridded data files (i.e. on theia
perhaps)
> > or post them to our anonymous ftp site following these
instructions:
> > http://www.dtcenter.org/met/users/support/met_help.php#ftp
> >
> > I need a sample probabilistic forecast file as well as the
verifying
> > observations.
> >
> > And also send me a description of the what you're trying to
verify.
> > Presumably, you want to verify probability of precip exceeding a
certain
> > amount.
> >
> > I'll work up a suggested Grid-Stat configuration file and send it
back to
> > you.
> >
> > Thanks,
> > John Halley Gotway
> >
> > On Thu, Aug 11, 2016 at 9:37 AM, matthew.pyle at noaa.gov via RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > Thu Aug 11 09:37:16 2016: Request 77523 was acted upon.
> > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> > > Queue: met_help
> > > Subject: grid_stat questions
> > > Owner: Nobody
> > > Requestors: Matthew.Pyle at noaa.gov
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77523
> >
> > >
> > >
> > > Hi,
> > >
> > > I'm attempting to use MET seriously for the first time to verify
some
> > > probabilistic precipitation forecasts. The tutorial slides were
very
> > > helpful in getting me going in the right direction, and
grid_stat seems
> > to
> > > be running without complaint if I do things a certain way, but
only
> the *
> > > pairs.nc is populated with actual data - the .stat and .txt
files just
> > > have
> > > the header line.
> > >
> > > First question is GRIB2 related - is the Product Definition
Template
> > Number
> > > figured out internally by the code? In the grib2_vars_flat.txt
file I
> > > didn't see a way to distinguish deterministic APCP from
probabilistic
> > > APCP. It seems like it might be treating it as a normal APCP,
at least
> > in
> > > the *pairs.nc file the forecast field gets labeled as being in
kg/m^2.
> > >
> > > Second is that I seem to have to provide as many observational
> cat_thresh
> > > values as there are categories in the forecast block. This
contradicts
> > the
> > > tutorial slide I'm following, so maybe provides a clue as to
what is
> > going
> > > wrong for me. Doing it this way:
> > >
> > >
> > > field = [
> > > {
> > > name = "APCP";
> > > level = [ "A24" ]; *# <------ Note...doing as
"(*,*)" as
> in
> > > the example I was following created failures. Maybe specifying
as A24
> is
> > > problematic??*
> > > prob = TRUE;
> > > cat_thresh = [ >=0.0, >=0.1, >=0.2, >=0.3, >=0.4, >=0.5,
>=0.6,
> > > >=0.7, >=0.8, >=0.9, >=1.0 ];
> > > }
> > > ];
> > >
> > > obs = {
> > >
> > > field = [
> > > {
> > > name = "APCP";
> > > level = [ "A24" ];
> > > cat_thresh = [ >25.4 ];
> > > }
> > > ];
> > > }
> > >
> > > Gives me this error message:
> > >
> > > ERROR : GridStatConfInfo::process_config() -> The number of
> thresholds
> > > for
> > > each field in "fcst.cat_thresh" must match the number of
thresholds for
> > > each field in "obs.cat_thresh".
> > >
> > > Any hints on where I'm going wrong?
> > >
> > > Thanks,
> > >
> > > Matt
> > >
> > >
> >
> >
>
>
------------------------------------------------
Subject: grid_stat questions
From: "matthew.pyle at noaa.gov"
Time: Thu Aug 11 17:11:44 2016
Thanks for the super quick response - worked like a charm on the first
try. Not sure when if ever I would have figured that out on my own,
so
thanks again!
-Matt
On Thu, Aug 11, 2016 at 4:12 PM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:
> Matt,
>
> OK, please try using this configuration file:
>
> /home/John.H.Gotway/MET_Help/pyle_data_20160811/
> GridStatConfig_APCP_24_prob
>
> The key here is the "prob" setting. It may be set to a boolean
value, i.e.
> "TRUE" or to a dictionary which define the probability information
to be
> found.
>
> So here's what works for your data:
> name = "PROB";
> level = [ "A24" ];
> prob = { name = "APCP"; thresh_lo = 25.4; }
>
> However, looking on page 47 of the MET-5.1 user's guide, I see that
the
> description is lacking. We'll update that info for the next
release.
>
> Hopefully that'll do the trick.
>
> Thanks,
> John
>
>
>
> On Thu, Aug 11, 2016 at 1:31 PM, matthew.pyle at noaa.gov via RT <
> met_help at ucar.edu> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77523 >
> >
> > Glad to know I wasn't overlooking something obvious, and thanks
for you
> > offer to help! I've uploaded stuff onto theia in
> > /scratch4/NCEPDEV/meso/save/Matthew.Pyle/dtc_stuff/met_case/
> >
> > fcst.t12z.grib2.218 (ensemble probability of 24 h QPF > 25.4 mm)
> > ccpa.2016080612.24h.218 (24 h precip analysis file)
> >
> > Ultimately I'm trying to plot some reliability diagrams for
various
> > experiments and various thresholds. I'm using the 5.1_bugfix
version if
> > that makes a difference.
> >
> > -Matt
> >
> >
> >
> > On Thu, Aug 11, 2016 at 3:08 PM, John Halley Gotway via RT <
> > met_help at ucar.edu> wrote:
> >
> > > Hi Matt,
> > >
> > > Getting started with configuration files can be tricky,
particularly
> for
> > > probabilistic data. Perhaps it'd be simplest to work through
an
> > example.
> > > Please either point me to your gridded data files (i.e. on theia
> perhaps)
> > > or post them to our anonymous ftp site following these
instructions:
> > > http://www.dtcenter.org/met/users/support/met_help.php#ftp
> > >
> > > I need a sample probabilistic forecast file as well as the
verifying
> > > observations.
> > >
> > > And also send me a description of the what you're trying to
verify.
> > > Presumably, you want to verify probability of precip exceeding a
> certain
> > > amount.
> > >
> > > I'll work up a suggested Grid-Stat configuration file and send
it back
> to
> > > you.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > > On Thu, Aug 11, 2016 at 9:37 AM, matthew.pyle at noaa.gov via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > >
> > > > Thu Aug 11 09:37:16 2016: Request 77523 was acted upon.
> > > > Transaction: Ticket created by Matthew.Pyle at noaa.gov
> > > > Queue: met_help
> > > > Subject: grid_stat questions
> > > > Owner: Nobody
> > > > Requestors: Matthew.Pyle at noaa.gov
> > > > Status: new
> > > > Ticket <URL: https://rt.rap.ucar.edu/rt/
> Ticket/Display.html?id=77523
> > >
> > > >
> > > >
> > > > Hi,
> > > >
> > > > I'm attempting to use MET seriously for the first time to
verify some
> > > > probabilistic precipitation forecasts. The tutorial slides
were very
> > > > helpful in getting me going in the right direction, and
grid_stat
> seems
> > > to
> > > > be running without complaint if I do things a certain way, but
only
> > the *
> > > > pairs.nc is populated with actual data - the .stat and .txt
files
> just
> > > > have
> > > > the header line.
> > > >
> > > > First question is GRIB2 related - is the Product Definition
Template
> > > Number
> > > > figured out internally by the code? In the
grib2_vars_flat.txt file
> I
> > > > didn't see a way to distinguish deterministic APCP from
probabilistic
> > > > APCP. It seems like it might be treating it as a normal APCP,
at
> least
> > > in
> > > > the *pairs.nc file the forecast field gets labeled as being in
> kg/m^2.
> > > >
> > > > Second is that I seem to have to provide as many observational
> > cat_thresh
> > > > values as there are categories in the forecast block. This
> contradicts
> > > the
> > > > tutorial slide I'm following, so maybe provides a clue as to
what is
> > > going
> > > > wrong for me. Doing it this way:
> > > >
> > > >
> > > > field = [
> > > > {
> > > > name = "APCP";
> > > > level = [ "A24" ]; *# <------ Note...doing as
"(*,*)"
> as
> > in
> > > > the example I was following created failures. Maybe
specifying as
> A24
> > is
> > > > problematic??*
> > > > prob = TRUE;
> > > > cat_thresh = [ >=0.0, >=0.1, >=0.2, >=0.3, >=0.4,
>=0.5,
> >=0.6,
> > > > >=0.7, >=0.8, >=0.9, >=1.0 ];
> > > > }
> > > > ];
> > > >
> > > > obs = {
> > > >
> > > > field = [
> > > > {
> > > > name = "APCP";
> > > > level = [ "A24" ];
> > > > cat_thresh = [ >25.4 ];
> > > > }
> > > > ];
> > > > }
> > > >
> > > > Gives me this error message:
> > > >
> > > > ERROR : GridStatConfInfo::process_config() -> The number of
> > thresholds
> > > > for
> > > > each field in "fcst.cat_thresh" must match the number of
thresholds
> for
> > > > each field in "obs.cat_thresh".
> > > >
> > > > Any hints on where I'm going wrong?
> > > >
> > > > Thanks,
> > > >
> > > > Matt
> > > >
> > > >
> > >
> > >
> >
> >
>
>
------------------------------------------------
More information about the Met_help
mailing list