[Met_help] [rt.rap.ucar.edu #94701] History for Some Questions regarding Met verif tool

John Halley Gotway via RT met_help at ucar.edu
Thu Mar 26 12:54:12 MDT 2020


----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

Hi everybody!

My name is Rachid Moustabchir. First of all, I would like to thank all the
stuff of met help for this mail database, and I really appreciate their
willingness to assist the community.

We developed a real-time NWP over North-Africa and Morocco using UEMS
system. Since verification is an integral part of weather modeling, it is
almost one year that I am learning and using Met verif tool (learning from
the official website and this mail database).

Until the present day I am able to use pointstat verification for: TMP,
DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR and ceiling. I
also use gridstat verification for precipitation using CMORPH as
observation data.

Now I want to verify cloud cover, could you, please, tell me where to find
the data for cloud cover (Satellite…)? I think there is no cloud cover in
GDAS PrepBUFR.

4 other questions:

   1.

   If I look for PBL and surface CAPE in PrepBUFR using ADPSFC, SFCSHP and
   MSONET I could not find any data. Do I need to use upper level data
   (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface obs?
   2.

     I use pointstat analysis for wind direction (for thresholds: '>=0.0,
   >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use the command:

aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var UGRD_VGRD
-fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR

I change -fcst_thresh value for other thresholds. is that correct?

   1.

   for visibility and ceiling, I use conditional verification (for
   thresholds: <8000, <5000, <1500, <1000, <500) I use the command in
   pointstat analysis:

aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS -line_type
MPR -column_thresh OBS <500 -out_line_type CNT

I change -column_thresh OBS value for other thresholds. is that correct?

   1.

   What is the difference between -fcst_thresh and column_thresh when using
   pointstat analysis?

Thanking you most sincerely for your time and consideration.



Cordially,



Rachid Moustabchir

Assistant Professor

Science Faculty, Agadir

Ibn Zohr Uiversity


----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: Some Questions regarding Met verif tool
From: Julie Prestopnik
Time: Tue Mar 24 14:37:25 2020

Hi Rachid.

Thank you for your interest in MET.  I have assigned this ticket to
John
Halley Gotway who is best suited to answer your questions.  Please
allow a
few days for a response.

Julie

On Tue, Mar 24, 2020 at 1:46 PM Moustabchir Rachid via RT
<met_help at ucar.edu>
wrote:

>
> Tue Mar 24 13:40:54 2020: Request 94701 was acted upon.
> Transaction: Ticket created by rachid.moustabchir at gmail.com
>        Queue: met_help
>      Subject: Some Questions regarding Met verif tool
>        Owner: Nobody
>   Requestors: rachid.moustabchir at gmail.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701 >
>
>
> Hi everybody!
>
> My name is Rachid Moustabchir. First of all, I would like to thank
all the
> stuff of met help for this mail database, and I really appreciate
their
> willingness to assist the community.
>
> We developed a real-time NWP over North-Africa and Morocco using
UEMS
> system. Since verification is an integral part of weather modeling,
it is
> almost one year that I am learning and using Met verif tool
(learning from
> the official website and this mail database).
>
> Until the present day I am able to use pointstat verification for:
TMP,
> DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR and
ceiling. I
> also use gridstat verification for precipitation using CMORPH as
> observation data.
>
> Now I want to verify cloud cover, could you, please, tell me where
to find
> the data for cloud cover (Satellite…)? I think there is no cloud
cover in
> GDAS PrepBUFR.
>
> 4 other questions:
>
>    1.
>
>    If I look for PBL and surface CAPE in PrepBUFR using ADPSFC,
SFCSHP and
>    MSONET I could not find any data. Do I need to use upper level
data
>    (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface obs?
>    2.
>
>      I use pointstat analysis for wind direction (for thresholds:
'>=0.0,
>    >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use the
command:
>
> aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var UGRD_VGRD
> -fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR
>
> I change -fcst_thresh value for other thresholds. is that correct?
>
>    1.
>
>    for visibility and ceiling, I use conditional verification (for
>    thresholds: <8000, <5000, <1500, <1000, <500) I use the command
in
>    pointstat analysis:
>
> aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
-line_type
> MPR -column_thresh OBS <500 -out_line_type CNT
>
> I change -column_thresh OBS value for other thresholds. is that
correct?
>
>    1.
>
>    What is the difference between -fcst_thresh and column_thresh
when using
>    pointstat analysis?
>
> Thanking you most sincerely for your time and consideration.
>
>
>
> Cordially,
>
>
>
> Rachid Moustabchir
>
> Assistant Professor
>
> Science Faculty, Agadir
>
> Ibn Zohr Uiversity
>
>

--
Julie Prestopnik
Software Engineer
National Center for Atmospheric Research
Research Applications Laboratory
Phone: 303.497.8399
Email: jpresto at ucar.edu

My working day may not be your working day.  Please do not feel
obliged to
reply to this email outside of your normal working hours.

------------------------------------------------
Subject: Some Questions regarding Met verif tool
From: Moustabchir Rachid
Time: Tue Mar 24 14:42:08 2020

Thank you very much Julie!

On Tue, 24 Mar 2020 at 21:37, Julie Prestopnik via RT
<met_help at ucar.edu>
wrote:

> Hi Rachid.
>
> Thank you for your interest in MET.  I have assigned this ticket to
John
> Halley Gotway who is best suited to answer your questions.  Please
allow a
> few days for a response.
>
> Julie
>
> On Tue, Mar 24, 2020 at 1:46 PM Moustabchir Rachid via RT <
> met_help at ucar.edu>
> wrote:
>
> >
> > Tue Mar 24 13:40:54 2020: Request 94701 was acted upon.
> > Transaction: Ticket created by rachid.moustabchir at gmail.com
> >        Queue: met_help
> >      Subject: Some Questions regarding Met verif tool
> >        Owner: Nobody
> >   Requestors: rachid.moustabchir at gmail.com
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701 >
> >
> >
> > Hi everybody!
> >
> > My name is Rachid Moustabchir. First of all, I would like to thank
all
> the
> > stuff of met help for this mail database, and I really appreciate
their
> > willingness to assist the community.
> >
> > We developed a real-time NWP over North-Africa and Morocco using
UEMS
> > system. Since verification is an integral part of weather
modeling, it is
> > almost one year that I am learning and using Met verif tool
(learning
> from
> > the official website and this mail database).
> >
> > Until the present day I am able to use pointstat verification for:
TMP,
> > DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR and
> ceiling. I
> > also use gridstat verification for precipitation using CMORPH as
> > observation data.
> >
> > Now I want to verify cloud cover, could you, please, tell me where
to
> find
> > the data for cloud cover (Satellite…)? I think there is no cloud
cover in
> > GDAS PrepBUFR.
> >
> > 4 other questions:
> >
> >    1.
> >
> >    If I look for PBL and surface CAPE in PrepBUFR using ADPSFC,
SFCSHP
> and
> >    MSONET I could not find any data. Do I need to use upper level
data
> >    (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface
obs?
> >    2.
> >
> >      I use pointstat analysis for wind direction (for thresholds:
'>=0.0,
> >    >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use the
command:
> >
> > aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var
UGRD_VGRD
> > -fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR
> >
> > I change -fcst_thresh value for other thresholds. is that correct?
> >
> >    1.
> >
> >    for visibility and ceiling, I use conditional verification (for
> >    thresholds: <8000, <5000, <1500, <1000, <500) I use the command
in
> >    pointstat analysis:
> >
> > aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
-line_type
> > MPR -column_thresh OBS <500 -out_line_type CNT
> >
> > I change -column_thresh OBS value for other thresholds. is that
correct?
> >
> >    1.
> >
> >    What is the difference between -fcst_thresh and column_thresh
when
> using
> >    pointstat analysis?
> >
> > Thanking you most sincerely for your time and consideration.
> >
> >
> >
> > Cordially,
> >
> >
> >
> > Rachid Moustabchir
> >
> > Assistant Professor
> >
> > Science Faculty, Agadir
> >
> > Ibn Zohr Uiversity
> >
> >
>
> --
> Julie Prestopnik
> Software Engineer
> National Center for Atmospheric Research
> Research Applications Laboratory
> Phone: 303.497.8399
> Email: jpresto at ucar.edu
>
> My working day may not be your working day.  Please do not feel
obliged to
> reply to this email outside of your normal working hours.
>
>

------------------------------------------------
Subject: Some Questions regarding Met verif tool
From: John Halley Gotway
Time: Tue Mar 24 15:10:49 2020

Hello Rachid,

I'm glad to hear that you've had success running the Point-Stat and
Grid-Stat tools.  Thank you for laying out your questions so clearly:

(1) Where to find cloud cover satellite observations?

Unfortunately, I'm not the best person to recommend data source for
satellite observations.  I'd refer you to NASA for that, perhaps cloud
fraction data would be useful:
https://earthobservatory.nasa.gov/global-maps/MODAL2_M_CLD_FR

(2) How do I get PBL and CAPE observations?

You're correct that PBL and CAPE observations do not exist directly in
the
PrepBUFR files.  However, the PB2NC tool has recently been enhanced to
derive them from the observations that do exist.  We coordinate with
NOAA/EMC staff to migrate their derivation logic over to PB2NC.
Please try
running pb2nc from met-9.0.  Here's the default PB2NC config file for
met-9.0:
https://github.com/NCAR/MET/blob/master_v9.0/met/data/config/PB2NCConfig_default

Notice the variable names which begin with "D_".  All of these
observation
types can be derived from the input obs:
{ key = "D_DPT"; val = "DPT"; },
{ key = "D_WDIR"; val = "WDIR"; },
{ key = "D_WIND"; val = "WIND"; },
{ key = "D_RH"; val = "RH"; },
{ key = "D_MIXR"; val = "MIXR"; },
{ key = "D_PRMSL"; val = "PRMSL"; },
{ key = "D_PBL"; val = "PBL"; },
{ key = "D_CAPE"; val = "CAPE"; }

But you do need to configure PB2NC to actually request that they be
derived
by including them in the "obs_bufr_var" list:
obs_bufr_var = [ "QOB", "TOB", "ZOB", "UOB", "VOB", "D_PBL", "D_CAPE"
];

By listing them in obs_bufr_var, you request that they be computed.
And
the obs_prepbufr_map dictionary indicates how they should be renamed
in the
output.  So you should see obs in the output named "PBL" and "CAPE".
And
yes, I believe they are only derived for the ADPUPA message type.

(3) Wind direction statistics in Stat-Analysis.

There are a few ways you can see wind direction statistics.  The VCNT
line
type from Point-Stat and Grid-Stat includes continuous vector
statistics.
I recommend looking at those.
When running Stat-Analysis, you're right there is an output type named
"WDIR" which can be computed either from the vector partial sums line
type
(VL1L2) or from the matched pair (MPR) line type for individual U and
V
components of winds.

Using MPR for U/V pairs directly:
-job aggregate_stat -fcst_lev Z10 \
-line_type MPR -out_line_type WDIR \
-out_wind_thresh >0 -out_wind_logic INTERSECTION

Using VL1L2 lines:
-job aggregate_stat -fcst_lev Z10 \
-line_type VL1L2 -out_line_type WDIR \
-fcst_thresh ge3.0

The job above will produce output for VL1L2 lines where the wind speed
>=3.0 m/s.  However I'd suggest using the "-by" option instead:
-job aggregate_stat -fcst_lev Z10 \
-line_type VL1L2 -out_line_type WDIR \
-by FCST_THRESH

That produces separate output for each unique value found in the
FCST_THRESH column.

(4) Take a look at your Point-Stat output.  If you did conditional vx
with
those thresholds, you should see separate CNT lines for each
threshold.
If you want to aggregate results across multiple cases, I'd recommend
the
following type of job:

-job aggregate_stat -line_type SL1L2 -out_line_type CNT -fcst_var VIS
-by
FCST_THRESH

That'll aggregate all the SL1L2 lines for VIS together across all of
your
output times, but do so separately for each unique FCST_THRESH value
in the
data.

If you did write out MPR data, yes, you can also run the type of job
you
listed.  And since you have the raw MPR values available, you can
filter
them however you'd like:
-job aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
-line_type MPR -column_thresh OBS <500 -out_line_type CNT

(5) The -fcst_thresh job command option filters the input data using
the
contents of the FCST_THRESH header column.  It just does string
matching.
The -column_thresh job command option takes two arguments: a column
name
and a threshold.  It is used to filter by numeric input columns.  You
last
example is a perfect use of it "-line_type MPR -column_thresh OBS
lt500".
This means, only use the input MPR lines where the values in the OBS
data
column are less than 500.

Thanks,
John

On Tue, Mar 24, 2020 at 1:46 PM Moustabchir Rachid via RT
<met_help at ucar.edu>
wrote:

>
> Tue Mar 24 13:40:54 2020: Request 94701 was acted upon.
> Transaction: Ticket created by rachid.moustabchir at gmail.com
>        Queue: met_help
>      Subject: Some Questions regarding Met verif tool
>        Owner: Nobody
>   Requestors: rachid.moustabchir at gmail.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701 >
>
>
> Hi everybody!
>
> My name is Rachid Moustabchir. First of all, I would like to thank
all the
> stuff of met help for this mail database, and I really appreciate
their
> willingness to assist the community.
>
> We developed a real-time NWP over North-Africa and Morocco using
UEMS
> system. Since verification is an integral part of weather modeling,
it is
> almost one year that I am learning and using Met verif tool
(learning from
> the official website and this mail database).
>
> Until the present day I am able to use pointstat verification for:
TMP,
> DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR and
ceiling. I
> also use gridstat verification for precipitation using CMORPH as
> observation data.
>
> Now I want to verify cloud cover, could you, please, tell me where
to find
> the data for cloud cover (Satellite…)? I think there is no cloud
cover in
> GDAS PrepBUFR.
>
> 4 other questions:
>
>    1.
>
>    If I look for PBL and surface CAPE in PrepBUFR using ADPSFC,
SFCSHP and
>    MSONET I could not find any data. Do I need to use upper level
data
>    (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface obs?
>    2.
>
>      I use pointstat analysis for wind direction (for thresholds:
'>=0.0,
>    >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use the
command:
>
> aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var UGRD_VGRD
> -fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR
>
> I change -fcst_thresh value for other thresholds. is that correct?
>
>    1.
>
>    for visibility and ceiling, I use conditional verification (for
>    thresholds: <8000, <5000, <1500, <1000, <500) I use the command
in
>    pointstat analysis:
>
> aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
-line_type
> MPR -column_thresh OBS <500 -out_line_type CNT
>
> I change -column_thresh OBS value for other thresholds. is that
correct?
>
>    1.
>
>    What is the difference between -fcst_thresh and column_thresh
when using
>    pointstat analysis?
>
> Thanking you most sincerely for your time and consideration.
>
>
>
> Cordially,
>
>
>
> Rachid Moustabchir
>
> Assistant Professor
>
> Science Faculty, Agadir
>
> Ibn Zohr Uiversity
>
>

------------------------------------------------
Subject: Some Questions regarding Met verif tool
From: Moustabchir Rachid
Time: Tue Mar 24 18:08:47 2020

Hello John,

Thanks as always for your fast and comprehensive answers.

I use the last version of MET (9.0) and config files. Tomorrow, I will
do a
run for the verification of PBL and CAPE observations.

I will follow your instructions about wind direction and visibility
and use
the “by” option and look the difference with the way I am doing the
job
right now.

John, the way I am doing wind direction is it correct? In this case I
have
an output file for every threshold for wind direction, visibility and
ceiling height.

For cloud fraction taking into account the link you give me, I will
make a
bibliographic search over the Internet to learn how and what type of
data
people use. Since I am a slow learner, perhaps I will have some
questions
about this.

By the way, The archive location of imerg data for precipitation is
https://gpm1.gesdisc.eosdis.nasa.gov/data/GPM_L3/GPM_3IMERGHH.06/.

Thanks,

Rachid

On Tue, 24 Mar 2020 at 22:10, John Halley Gotway via RT
<met_help at ucar.edu>
wrote:

> Hello Rachid,
>
> I'm glad to hear that you've had success running the Point-Stat and
> Grid-Stat tools.  Thank you for laying out your questions so
clearly:
>
> (1) Where to find cloud cover satellite observations?
>
> Unfortunately, I'm not the best person to recommend data source for
> satellite observations.  I'd refer you to NASA for that, perhaps
cloud
> fraction data would be useful:
> https://earthobservatory.nasa.gov/global-maps/MODAL2_M_CLD_FR
>
> (2) How do I get PBL and CAPE observations?
>
> You're correct that PBL and CAPE observations do not exist directly
in the
> PrepBUFR files.  However, the PB2NC tool has recently been enhanced
to
> derive them from the observations that do exist.  We coordinate with
> NOAA/EMC staff to migrate their derivation logic over to PB2NC.
Please try
> running pb2nc from met-9.0.  Here's the default PB2NC config file
for
> met-9.0:
>
>
https://github.com/NCAR/MET/blob/master_v9.0/met/data/config/PB2NCConfig_default
>
> Notice the variable names which begin with "D_".  All of these
observation
> types can be derived from the input obs:
> { key = "D_DPT"; val = "DPT"; },
> { key = "D_WDIR"; val = "WDIR"; },
> { key = "D_WIND"; val = "WIND"; },
> { key = "D_RH"; val = "RH"; },
> { key = "D_MIXR"; val = "MIXR"; },
> { key = "D_PRMSL"; val = "PRMSL"; },
> { key = "D_PBL"; val = "PBL"; },
> { key = "D_CAPE"; val = "CAPE"; }
>
> But you do need to configure PB2NC to actually request that they be
derived
> by including them in the "obs_bufr_var" list:
> obs_bufr_var = [ "QOB", "TOB", "ZOB", "UOB", "VOB", "D_PBL",
"D_CAPE" ];
>
> By listing them in obs_bufr_var, you request that they be computed.
And
> the obs_prepbufr_map dictionary indicates how they should be renamed
in the
> output.  So you should see obs in the output named "PBL" and "CAPE".
And
> yes, I believe they are only derived for the ADPUPA message type.
>
> (3) Wind direction statistics in Stat-Analysis.
>
> There are a few ways you can see wind direction statistics.  The
VCNT line
> type from Point-Stat and Grid-Stat includes continuous vector
statistics.
> I recommend looking at those.
> When running Stat-Analysis, you're right there is an output type
named
> "WDIR" which can be computed either from the vector partial sums
line type
> (VL1L2) or from the matched pair (MPR) line type for individual U
and V
> components of winds.
>
> Using MPR for U/V pairs directly:
> -job aggregate_stat -fcst_lev Z10 \
> -line_type MPR -out_line_type WDIR \
> -out_wind_thresh >0 -out_wind_logic INTERSECTION
>
> Using VL1L2 lines:
> -job aggregate_stat -fcst_lev Z10 \
> -line_type VL1L2 -out_line_type WDIR \
> -fcst_thresh ge3.0
>
> The job above will produce output for VL1L2 lines where the wind
speed
> >=3.0 m/s.  However I'd suggest using the "-by" option instead:
> -job aggregate_stat -fcst_lev Z10 \
> -line_type VL1L2 -out_line_type WDIR \
> -by FCST_THRESH
>
> That produces separate output for each unique value found in the
> FCST_THRESH column.
>
> (4) Take a look at your Point-Stat output.  If you did conditional
vx with
> those thresholds, you should see separate CNT lines for each
threshold.
> If you want to aggregate results across multiple cases, I'd
recommend the
> following type of job:
>
> -job aggregate_stat -line_type SL1L2 -out_line_type CNT -fcst_var
VIS -by
> FCST_THRESH
>
> That'll aggregate all the SL1L2 lines for VIS together across all of
your
> output times, but do so separately for each unique FCST_THRESH value
in the
> data.
>
> If you did write out MPR data, yes, you can also run the type of job
you
> listed.  And since you have the raw MPR values available, you can
filter
> them however you'd like:
> -job aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
> -line_type MPR -column_thresh OBS <500 -out_line_type CNT
>
> (5) The -fcst_thresh job command option filters the input data using
the
> contents of the FCST_THRESH header column.  It just does string
matching.
> The -column_thresh job command option takes two arguments: a column
name
> and a threshold.  It is used to filter by numeric input columns.
You last
> example is a perfect use of it "-line_type MPR -column_thresh OBS
lt500".
> This means, only use the input MPR lines where the values in the OBS
data
> column are less than 500.
>
> Thanks,
> John
>
> On Tue, Mar 24, 2020 at 1:46 PM Moustabchir Rachid via RT <
> met_help at ucar.edu>
> wrote:
>
> >
> > Tue Mar 24 13:40:54 2020: Request 94701 was acted upon.
> > Transaction: Ticket created by rachid.moustabchir at gmail.com
> >        Queue: met_help
> >      Subject: Some Questions regarding Met verif tool
> >        Owner: Nobody
> >   Requestors: rachid.moustabchir at gmail.com
> >       Status: new
> >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701 >
> >
> >
> > Hi everybody!
> >
> > My name is Rachid Moustabchir. First of all, I would like to thank
all
> the
> > stuff of met help for this mail database, and I really appreciate
their
> > willingness to assist the community.
> >
> > We developed a real-time NWP over North-Africa and Morocco using
UEMS
> > system. Since verification is an integral part of weather
modeling, it is
> > almost one year that I am learning and using Met verif tool
(learning
> from
> > the official website and this mail database).
> >
> > Until the present day I am able to use pointstat verification for:
TMP,
> > DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR and
> ceiling. I
> > also use gridstat verification for precipitation using CMORPH as
> > observation data.
> >
> > Now I want to verify cloud cover, could you, please, tell me where
to
> find
> > the data for cloud cover (Satellite…)? I think there is no cloud
cover in
> > GDAS PrepBUFR.
> >
> > 4 other questions:
> >
> >    1.
> >
> >    If I look for PBL and surface CAPE in PrepBUFR using ADPSFC,
SFCSHP
> and
> >    MSONET I could not find any data. Do I need to use upper level
data
> >    (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface
obs?
> >    2.
> >
> >      I use pointstat analysis for wind direction (for thresholds:
'>=0.0,
> >    >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use the
command:
> >
> > aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var
UGRD_VGRD
> > -fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR
> >
> > I change -fcst_thresh value for other thresholds. is that correct?
> >
> >    1.
> >
> >    for visibility and ceiling, I use conditional verification (for
> >    thresholds: <8000, <5000, <1500, <1000, <500) I use the command
in
> >    pointstat analysis:
> >
> > aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
-line_type
> > MPR -column_thresh OBS <500 -out_line_type CNT
> >
> > I change -column_thresh OBS value for other thresholds. is that
correct?
> >
> >    1.
> >
> >    What is the difference between -fcst_thresh and column_thresh
when
> using
> >    pointstat analysis?
> >
> > Thanking you most sincerely for your time and consideration.
> >
> >
> >
> > Cordially,
> >
> >
> >
> > Rachid Moustabchir
> >
> > Assistant Professor
> >
> > Science Faculty, Agadir
> >
> > Ibn Zohr Uiversity
> >
> >
>
>

------------------------------------------------
Subject: Some Questions regarding Met verif tool
From: John Halley Gotway
Time: Thu Mar 26 09:47:19 2020

Rachid,

Yes, the way you're doing wind direction verification is fine.

When you run the Grid-Stat and Point-Stat tools and generate VL1L2
vector
partial sums output and VCNT vector continuous statistics, you're
using the
"wind_thresh" configuration option.  And that option filters the U and
V
wind component matched pairs, only including the pairs which meet your
specified wind speed threshold criteria.

The motivation for this approach is that for light winds, the wind
direction is variable and not well-defined.  So the user applies one
or
more wind speed thresholds to see how the vector winds perform when
the
wind direction is well-defined.

And you can use Stat-Analysis (as you're already doing) to aggregate
those
results across multiple cases.

Thanks,
John

On Tue, Mar 24, 2020 at 6:09 PM Moustabchir Rachid via RT
<met_help at ucar.edu>
wrote:

>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701 >
>
> Hello John,
>
> Thanks as always for your fast and comprehensive answers.
>
> I use the last version of MET (9.0) and config files. Tomorrow, I
will do a
> run for the verification of PBL and CAPE observations.
>
> I will follow your instructions about wind direction and visibility
and use
> the “by” option and look the difference with the way I am doing the
job
> right now.
>
> John, the way I am doing wind direction is it correct? In this case
I have
> an output file for every threshold for wind direction, visibility
and
> ceiling height.
>
> For cloud fraction taking into account the link you give me, I will
make a
> bibliographic search over the Internet to learn how and what type of
data
> people use. Since I am a slow learner, perhaps I will have some
questions
> about this.
>
> By the way, The archive location of imerg data for precipitation is
> https://gpm1.gesdisc.eosdis.nasa.gov/data/GPM_L3/GPM_3IMERGHH.06/.
>
> Thanks,
>
> Rachid
>
> On Tue, 24 Mar 2020 at 22:10, John Halley Gotway via RT
<met_help at ucar.edu
> >
> wrote:
>
> > Hello Rachid,
> >
> > I'm glad to hear that you've had success running the Point-Stat
and
> > Grid-Stat tools.  Thank you for laying out your questions so
clearly:
> >
> > (1) Where to find cloud cover satellite observations?
> >
> > Unfortunately, I'm not the best person to recommend data source
for
> > satellite observations.  I'd refer you to NASA for that, perhaps
cloud
> > fraction data would be useful:
> > https://earthobservatory.nasa.gov/global-maps/MODAL2_M_CLD_FR
> >
> > (2) How do I get PBL and CAPE observations?
> >
> > You're correct that PBL and CAPE observations do not exist
directly in
> the
> > PrepBUFR files.  However, the PB2NC tool has recently been
enhanced to
> > derive them from the observations that do exist.  We coordinate
with
> > NOAA/EMC staff to migrate their derivation logic over to PB2NC.
Please
> try
> > running pb2nc from met-9.0.  Here's the default PB2NC config file
for
> > met-9.0:
> >
> >
>
https://github.com/NCAR/MET/blob/master_v9.0/met/data/config/PB2NCConfig_default
> >
> > Notice the variable names which begin with "D_".  All of these
> observation
> > types can be derived from the input obs:
> > { key = "D_DPT"; val = "DPT"; },
> > { key = "D_WDIR"; val = "WDIR"; },
> > { key = "D_WIND"; val = "WIND"; },
> > { key = "D_RH"; val = "RH"; },
> > { key = "D_MIXR"; val = "MIXR"; },
> > { key = "D_PRMSL"; val = "PRMSL"; },
> > { key = "D_PBL"; val = "PBL"; },
> > { key = "D_CAPE"; val = "CAPE"; }
> >
> > But you do need to configure PB2NC to actually request that they
be
> derived
> > by including them in the "obs_bufr_var" list:
> > obs_bufr_var = [ "QOB", "TOB", "ZOB", "UOB", "VOB", "D_PBL",
"D_CAPE" ];
> >
> > By listing them in obs_bufr_var, you request that they be
computed.  And
> > the obs_prepbufr_map dictionary indicates how they should be
renamed in
> the
> > output.  So you should see obs in the output named "PBL" and
"CAPE".  And
> > yes, I believe they are only derived for the ADPUPA message type.
> >
> > (3) Wind direction statistics in Stat-Analysis.
> >
> > There are a few ways you can see wind direction statistics.  The
VCNT
> line
> > type from Point-Stat and Grid-Stat includes continuous vector
statistics.
> > I recommend looking at those.
> > When running Stat-Analysis, you're right there is an output type
named
> > "WDIR" which can be computed either from the vector partial sums
line
> type
> > (VL1L2) or from the matched pair (MPR) line type for individual U
and V
> > components of winds.
> >
> > Using MPR for U/V pairs directly:
> > -job aggregate_stat -fcst_lev Z10 \
> > -line_type MPR -out_line_type WDIR \
> > -out_wind_thresh >0 -out_wind_logic INTERSECTION
> >
> > Using VL1L2 lines:
> > -job aggregate_stat -fcst_lev Z10 \
> > -line_type VL1L2 -out_line_type WDIR \
> > -fcst_thresh ge3.0
> >
> > The job above will produce output for VL1L2 lines where the wind
speed
> > >=3.0 m/s.  However I'd suggest using the "-by" option instead:
> > -job aggregate_stat -fcst_lev Z10 \
> > -line_type VL1L2 -out_line_type WDIR \
> > -by FCST_THRESH
> >
> > That produces separate output for each unique value found in the
> > FCST_THRESH column.
> >
> > (4) Take a look at your Point-Stat output.  If you did conditional
vx
> with
> > those thresholds, you should see separate CNT lines for each
threshold.
> > If you want to aggregate results across multiple cases, I'd
recommend the
> > following type of job:
> >
> > -job aggregate_stat -line_type SL1L2 -out_line_type CNT -fcst_var
VIS -by
> > FCST_THRESH
> >
> > That'll aggregate all the SL1L2 lines for VIS together across all
of your
> > output times, but do so separately for each unique FCST_THRESH
value in
> the
> > data.
> >
> > If you did write out MPR data, yes, you can also run the type of
job you
> > listed.  And since you have the raw MPR values available, you can
filter
> > them however you'd like:
> > -job aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var
VIS
> > -line_type MPR -column_thresh OBS <500 -out_line_type CNT
> >
> > (5) The -fcst_thresh job command option filters the input data
using the
> > contents of the FCST_THRESH header column.  It just does string
matching.
> > The -column_thresh job command option takes two arguments: a
column name
> > and a threshold.  It is used to filter by numeric input columns.
You
> last
> > example is a perfect use of it "-line_type MPR -column_thresh OBS
lt500".
> > This means, only use the input MPR lines where the values in the
OBS data
> > column are less than 500.
> >
> > Thanks,
> > John
> >
> > On Tue, Mar 24, 2020 at 1:46 PM Moustabchir Rachid via RT <
> > met_help at ucar.edu>
> > wrote:
> >
> > >
> > > Tue Mar 24 13:40:54 2020: Request 94701 was acted upon.
> > > Transaction: Ticket created by rachid.moustabchir at gmail.com
> > >        Queue: met_help
> > >      Subject: Some Questions regarding Met verif tool
> > >        Owner: Nobody
> > >   Requestors: rachid.moustabchir at gmail.com
> > >       Status: new
> > >  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701
> >
> > >
> > >
> > > Hi everybody!
> > >
> > > My name is Rachid Moustabchir. First of all, I would like to
thank all
> > the
> > > stuff of met help for this mail database, and I really
appreciate their
> > > willingness to assist the community.
> > >
> > > We developed a real-time NWP over North-Africa and Morocco using
UEMS
> > > system. Since verification is an integral part of weather
modeling, it
> is
> > > almost one year that I am learning and using Met verif tool
(learning
> > from
> > > the official website and this mail database).
> > >
> > > Until the present day I am able to use pointstat verification
for: TMP,
> > > DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR and
> > ceiling. I
> > > also use gridstat verification for precipitation using CMORPH as
> > > observation data.
> > >
> > > Now I want to verify cloud cover, could you, please, tell me
where to
> > find
> > > the data for cloud cover (Satellite…)? I think there is no cloud
cover
> in
> > > GDAS PrepBUFR.
> > >
> > > 4 other questions:
> > >
> > >    1.
> > >
> > >    If I look for PBL and surface CAPE in PrepBUFR using ADPSFC,
SFCSHP
> > and
> > >    MSONET I could not find any data. Do I need to use upper
level data
> > >    (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface
obs?
> > >    2.
> > >
> > >      I use pointstat analysis for wind direction (for
thresholds:
> '>=0.0,
> > >    >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use the
> command:
> > >
> > > aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var
UGRD_VGRD
> > > -fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR
> > >
> > > I change -fcst_thresh value for other thresholds. is that
correct?
> > >
> > >    1.
> > >
> > >    for visibility and ceiling, I use conditional verification
(for
> > >    thresholds: <8000, <5000, <1500, <1000, <500) I use the
command in
> > >    pointstat analysis:
> > >
> > > aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
> -line_type
> > > MPR -column_thresh OBS <500 -out_line_type CNT
> > >
> > > I change -column_thresh OBS value for other thresholds. is that
> correct?
> > >
> > >    1.
> > >
> > >    What is the difference between -fcst_thresh and column_thresh
when
> > using
> > >    pointstat analysis?
> > >
> > > Thanking you most sincerely for your time and consideration.
> > >
> > >
> > >
> > > Cordially,
> > >
> > >
> > >
> > > Rachid Moustabchir
> > >
> > > Assistant Professor
> > >
> > > Science Faculty, Agadir
> > >
> > > Ibn Zohr Uiversity
> > >
> > >
> >
> >
>
>

------------------------------------------------
Subject: Some Questions regarding Met verif tool
From: Moustabchir Rachid
Time: Thu Mar 26 11:44:38 2020

Dear John!

Thank you very much for clarifying this. By using upper observation
data
for PBL and surface CAPE as you said, I find data and we verification
work.

Have a good weekend.

Rachid

On Thu, 26 Mar 2020 at 16:47, John Halley Gotway via RT
<met_help at ucar.edu>
wrote:

> Rachid,
>
> Yes, the way you're doing wind direction verification is fine.
>
> When you run the Grid-Stat and Point-Stat tools and generate VL1L2
vector
> partial sums output and VCNT vector continuous statistics, you're
using the
> "wind_thresh" configuration option.  And that option filters the U
and V
> wind component matched pairs, only including the pairs which meet
your
> specified wind speed threshold criteria.
>
> The motivation for this approach is that for light winds, the wind
> direction is variable and not well-defined.  So the user applies one
or
> more wind speed thresholds to see how the vector winds perform when
the
> wind direction is well-defined.
>
> And you can use Stat-Analysis (as you're already doing) to aggregate
those
> results across multiple cases.
>
> Thanks,
> John
>
> On Tue, Mar 24, 2020 at 6:09 PM Moustabchir Rachid via RT <
> met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701 >
> >
> > Hello John,
> >
> > Thanks as always for your fast and comprehensive answers.
> >
> > I use the last version of MET (9.0) and config files. Tomorrow, I
will
> do a
> > run for the verification of PBL and CAPE observations.
> >
> > I will follow your instructions about wind direction and
visibility and
> use
> > the “by” option and look the difference with the way I am doing
the job
> > right now.
> >
> > John, the way I am doing wind direction is it correct? In this
case I
> have
> > an output file for every threshold for wind direction, visibility
and
> > ceiling height.
> >
> > For cloud fraction taking into account the link you give me, I
will make
> a
> > bibliographic search over the Internet to learn how and what type
of data
> > people use. Since I am a slow learner, perhaps I will have some
questions
> > about this.
> >
> > By the way, The archive location of imerg data for precipitation
is
> > https://gpm1.gesdisc.eosdis.nasa.gov/data/GPM_L3/GPM_3IMERGHH.06/.
> >
> > Thanks,
> >
> > Rachid
> >
> > On Tue, 24 Mar 2020 at 22:10, John Halley Gotway via RT <
> met_help at ucar.edu
> > >
> > wrote:
> >
> > > Hello Rachid,
> > >
> > > I'm glad to hear that you've had success running the Point-Stat
and
> > > Grid-Stat tools.  Thank you for laying out your questions so
clearly:
> > >
> > > (1) Where to find cloud cover satellite observations?
> > >
> > > Unfortunately, I'm not the best person to recommend data source
for
> > > satellite observations.  I'd refer you to NASA for that, perhaps
cloud
> > > fraction data would be useful:
> > > https://earthobservatory.nasa.gov/global-maps/MODAL2_M_CLD_FR
> > >
> > > (2) How do I get PBL and CAPE observations?
> > >
> > > You're correct that PBL and CAPE observations do not exist
directly in
> > the
> > > PrepBUFR files.  However, the PB2NC tool has recently been
enhanced to
> > > derive them from the observations that do exist.  We coordinate
with
> > > NOAA/EMC staff to migrate their derivation logic over to PB2NC.
Please
> > try
> > > running pb2nc from met-9.0.  Here's the default PB2NC config
file for
> > > met-9.0:
> > >
> > >
> >
>
https://github.com/NCAR/MET/blob/master_v9.0/met/data/config/PB2NCConfig_default
> > >
> > > Notice the variable names which begin with "D_".  All of these
> > observation
> > > types can be derived from the input obs:
> > > { key = "D_DPT"; val = "DPT"; },
> > > { key = "D_WDIR"; val = "WDIR"; },
> > > { key = "D_WIND"; val = "WIND"; },
> > > { key = "D_RH"; val = "RH"; },
> > > { key = "D_MIXR"; val = "MIXR"; },
> > > { key = "D_PRMSL"; val = "PRMSL"; },
> > > { key = "D_PBL"; val = "PBL"; },
> > > { key = "D_CAPE"; val = "CAPE"; }
> > >
> > > But you do need to configure PB2NC to actually request that they
be
> > derived
> > > by including them in the "obs_bufr_var" list:
> > > obs_bufr_var = [ "QOB", "TOB", "ZOB", "UOB", "VOB", "D_PBL",
"D_CAPE"
> ];
> > >
> > > By listing them in obs_bufr_var, you request that they be
computed.
> And
> > > the obs_prepbufr_map dictionary indicates how they should be
renamed in
> > the
> > > output.  So you should see obs in the output named "PBL" and
"CAPE".
> And
> > > yes, I believe they are only derived for the ADPUPA message
type.
> > >
> > > (3) Wind direction statistics in Stat-Analysis.
> > >
> > > There are a few ways you can see wind direction statistics.  The
VCNT
> > line
> > > type from Point-Stat and Grid-Stat includes continuous vector
> statistics.
> > > I recommend looking at those.
> > > When running Stat-Analysis, you're right there is an output type
named
> > > "WDIR" which can be computed either from the vector partial sums
line
> > type
> > > (VL1L2) or from the matched pair (MPR) line type for individual
U and V
> > > components of winds.
> > >
> > > Using MPR for U/V pairs directly:
> > > -job aggregate_stat -fcst_lev Z10 \
> > > -line_type MPR -out_line_type WDIR \
> > > -out_wind_thresh >0 -out_wind_logic INTERSECTION
> > >
> > > Using VL1L2 lines:
> > > -job aggregate_stat -fcst_lev Z10 \
> > > -line_type VL1L2 -out_line_type WDIR \
> > > -fcst_thresh ge3.0
> > >
> > > The job above will produce output for VL1L2 lines where the wind
speed
> > > >=3.0 m/s.  However I'd suggest using the "-by" option instead:
> > > -job aggregate_stat -fcst_lev Z10 \
> > > -line_type VL1L2 -out_line_type WDIR \
> > > -by FCST_THRESH
> > >
> > > That produces separate output for each unique value found in the
> > > FCST_THRESH column.
> > >
> > > (4) Take a look at your Point-Stat output.  If you did
conditional vx
> > with
> > > those thresholds, you should see separate CNT lines for each
threshold.
> > > If you want to aggregate results across multiple cases, I'd
recommend
> the
> > > following type of job:
> > >
> > > -job aggregate_stat -line_type SL1L2 -out_line_type CNT
-fcst_var VIS
> -by
> > > FCST_THRESH
> > >
> > > That'll aggregate all the SL1L2 lines for VIS together across
all of
> your
> > > output times, but do so separately for each unique FCST_THRESH
value in
> > the
> > > data.
> > >
> > > If you did write out MPR data, yes, you can also run the type of
job
> you
> > > listed.  And since you have the raw MPR values available, you
can
> filter
> > > them however you'd like:
> > > -job aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var
VIS
> > > -line_type MPR -column_thresh OBS <500 -out_line_type CNT
> > >
> > > (5) The -fcst_thresh job command option filters the input data
using
> the
> > > contents of the FCST_THRESH header column.  It just does string
> matching.
> > > The -column_thresh job command option takes two arguments: a
column
> name
> > > and a threshold.  It is used to filter by numeric input columns.
You
> > last
> > > example is a perfect use of it "-line_type MPR -column_thresh
OBS
> lt500".
> > > This means, only use the input MPR lines where the values in the
OBS
> data
> > > column are less than 500.
> > >
> > > Thanks,
> > > John
> > >
> > > On Tue, Mar 24, 2020 at 1:46 PM Moustabchir Rachid via RT <
> > > met_help at ucar.edu>
> > > wrote:
> > >
> > > >
> > > > Tue Mar 24 13:40:54 2020: Request 94701 was acted upon.
> > > > Transaction: Ticket created by rachid.moustabchir at gmail.com
> > > >        Queue: met_help
> > > >      Subject: Some Questions regarding Met verif tool
> > > >        Owner: Nobody
> > > >   Requestors: rachid.moustabchir at gmail.com
> > > >       Status: new
> > > >  Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=94701
> > >
> > > >
> > > >
> > > > Hi everybody!
> > > >
> > > > My name is Rachid Moustabchir. First of all, I would like to
thank
> all
> > > the
> > > > stuff of met help for this mail database, and I really
appreciate
> their
> > > > willingness to assist the community.
> > > >
> > > > We developed a real-time NWP over North-Africa and Morocco
using UEMS
> > > > system. Since verification is an integral part of weather
modeling,
> it
> > is
> > > > almost one year that I am learning and using Met verif tool
(learning
> > > from
> > > > the official website and this mail database).
> > > >
> > > > Until the present day I am able to use pointstat verification
for:
> TMP,
> > > > DPT, PRMSL, UGRD, VGRD, WIND, WDIR, VIS, RH, SPFH, HGT, MIXR
and
> > > ceiling. I
> > > > also use gridstat verification for precipitation using CMORPH
as
> > > > observation data.
> > > >
> > > > Now I want to verify cloud cover, could you, please, tell me
where to
> > > find
> > > > the data for cloud cover (Satellite…)? I think there is no
cloud
> cover
> > in
> > > > GDAS PrepBUFR.
> > > >
> > > > 4 other questions:
> > > >
> > > >    1.
> > > >
> > > >    If I look for PBL and surface CAPE in PrepBUFR using
ADPSFC,
> SFCSHP
> > > and
> > > >    MSONET I could not find any data. Do I need to use upper
level
> data
> > > >    (AIRCAR, PROFLR, ADPUPA). If yes, how can I extract surface
obs?
> > > >    2.
> > > >
> > > >      I use pointstat analysis for wind direction (for
thresholds:
> > '>=0.0,
> > > >    >=1.0, >=2.0, >=3.0, >=4.0, >=5.0, >=6.0, >=7.0'), I use
the
> > command:
> > > >
> > > > aggregate_stat -model wrf1_d01 -fcst_lead 010000 -fcst_var
UGRD_VGRD
> > > > -fcst_thresh >=4.0 -line_type VL1L2 -out_line_type WDIR
> > > >
> > > > I change -fcst_thresh value for other thresholds. is that
correct?
> > > >
> > > >    1.
> > > >
> > > >    for visibility and ceiling, I use conditional verification
(for
> > > >    thresholds: <8000, <5000, <1500, <1000, <500) I use the
command in
> > > >    pointstat analysis:
> > > >
> > > > aggregate_stat -model wrf2_d02 -fcst_lead 000000 -fcst_var VIS
> > -line_type
> > > > MPR -column_thresh OBS <500 -out_line_type CNT
> > > >
> > > > I change -column_thresh OBS value for other thresholds. is
that
> > correct?
> > > >
> > > >    1.
> > > >
> > > >    What is the difference between -fcst_thresh and
column_thresh when
> > > using
> > > >    pointstat analysis?
> > > >
> > > > Thanking you most sincerely for your time and consideration.
> > > >
> > > >
> > > >
> > > > Cordially,
> > > >
> > > >
> > > >
> > > > Rachid Moustabchir
> > > >
> > > > Assistant Professor
> > > >
> > > > Science Faculty, Agadir
> > > >
> > > > Ibn Zohr Uiversity
> > > >
> > > >
> > >
> > >
> >
> >
>
>

------------------------------------------------


More information about the Met_help mailing list