[Met_help] [rt.rap.ucar.edu #63193] History for SURFACE LEVEL
John Halley Gotway via RT
met_help at ucar.edu
Mon Oct 14 14:24:04 MDT 2013
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hello,
I would like to know what level to specify when I want to check accuracy
at surface level of metrics such as precipitation (APCP), pressure (PRES)
and humidity (SPFH).
Thanks!
Vladimir A. Malabanan
3F IBM Bldg J, UP Ayala Technohub
Applications Programmer
Advanced Analytics - Data Specialist
Diliman, Quezon City
IBM Solutions Delivery, Inc.
Philippines
Global Delivery Center
Phone:
+632-995-2IBM local 8639
e-mail:
malabava at ph.ibm.com
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
From: John Halley Gotway
Time: Thu Sep 26 10:30:05 2013
Vladimir,
I see from your earlier email that you're using little R point
observations in MET. I assume that you're using them for doing
verification with the Point-Stat tool. Support for little R point
observations was recently added in METv4.1, and we've only had a
handful of users try it out to my knowledge. So I'll be interested to
hear how it goes and what issues, if any, arise.
I see that you're asking about what "level" to use when doing
verification of surface variables. Surface fields are kind of a
special case within Point-Stat.
For upper-air variables, like 500mb temperature for example, you would
specify a range of observation pressure values that you'd like to
compare to that field. For example, you may want to match any
observations with a pressure value between 475 and 525mb to the 500mb
forecast level. However for surface variables, Point-Stat does not
really use the "level" value. Instead, the matching is
controlled by the message type. Observations with message types of
ADPSFC (over land) and SFCSHP (over water) are assumed to be at the
surface. When verifying surface fields, like 2-m temperature or
10-m winds, you should compare them to the APDSFC and/or SFCSHP
message types. Any observations with those message types will be
compared against the forecast at the surface.
Since you're using little R data, things may get a little messy.
Currently ascii2nc contains a mapping of little R message type to the
conventional message types used in the rest of MET (like ADPSFC
and SFCSHP). However, I suspect that mapping may be incomplete.
Please try running ASCII2NC and Point-Stat, and if you run into
problems, just let me know. It'll probably be easiest to just have
you send me some sample data. That way I can run your data here and
figure out what changes, if any, will be required to get it working
more smoothly in MET. You can post data to our anonymous ftp site by
following these instructions:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
Thanks,
John Halley Gotway
met_help at ucar.edu
On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>
> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
> Queue: met_help
> Subject: SURFACE LEVEL
> Owner: Nobody
> Requestors: vlad.malabanan at ph.ibm.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
>
> Hello,
>
> I would like to know what level to specify when I want to check
accuracy
> at surface level of metrics such as precipitation (APCP), pressure
(PRES)
> and humidity (SPFH).
>
> Thanks!
>
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
------------------------------------------------
Subject: SURFACE LEVEL
From: Vladimir A Malabanan
Time: Fri Sep 27 03:00:37 2013
Hello John,
Thanks for the response.
Yes, you are right. I am using little_r file format for my observation
file to match with my forecast file (in GRIB2 format, converted from
WRF
netcdf file using UPP).
As for my question, I would just like to know the correct level to
compute
for the accuracy of precipitation amount (APCP) (and other surface
measurements)
{
name = "APSP";
level = [ "__" ];
cat_thresh = [ >=0 ];
},
I have already uploaded my sample data (netcdf from converted little_r
for
observed values and GRIB2 file from converted netcdf for the forecast
file) in the ftp site. I am just not sure if I was able to upload it
successfully.
In any case, I am attaching the files here:
Thanks for your assistance! I am really new in MET (and in weather
forecasting!). I am just trying out some things for a project.
Thank you!
Vladimir A. Malabanan
3F IBM Bldg J, UP Ayala Technohub
Applications Programmer
Advanced Analytics - Data Specialist
Diliman, Quezon City
IBM Solutions Delivery, Inc.
Philippines
Global Delivery Center
Phone:
+632-995-2IBM local 8639
e-mail:
malabava at ph.ibm.com
From: "John Halley Gotway via RT" <met_help at ucar.edu>
To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
Date: 09/27/2013 12:28 AM
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
Vladimir,
I see from your earlier email that you're using little R point
observations in MET. I assume that you're using them for doing
verification with the Point-Stat tool. Support for little R point
observations was recently added in METv4.1, and we've only had a
handful
of users try it out to my knowledge. So I'll be interested to hear
how it
goes and what issues, if any, arise.
I see that you're asking about what "level" to use when doing
verification
of surface variables. Surface fields are kind of a special case
within
Point-Stat.
For upper-air variables, like 500mb temperature for example, you would
specify a range of observation pressure values that you'd like to
compare
to that field. For example, you may want to match any
observations with a pressure value between 475 and 525mb to the 500mb
forecast level. However for surface variables, Point-Stat does not
really
use the "level" value. Instead, the matching is
controlled by the message type. Observations with message types of
ADPSFC
(over land) and SFCSHP (over water) are assumed to be at the surface.
When
verifying surface fields, like 2-m temperature or
10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
types. Any observations with those message types will be compared
against
the forecast at the surface.
Since you're using little R data, things may get a little messy.
Currently
ascii2nc contains a mapping of little R message type to the
conventional
message types used in the rest of MET (like ADPSFC
and SFCSHP). However, I suspect that mapping may be incomplete.
Please try running ASCII2NC and Point-Stat, and if you run into
problems,
just let me know. It'll probably be easiest to just have you send me
some
sample data. That way I can run your data here and
figure out what changes, if any, will be required to get it working
more
smoothly in MET. You can post data to our anonymous ftp site by
following
these instructions:
http://www.dtcenter.org/met/users/support/met_help.php#ftp
Thanks,
John Halley Gotway
met_help at ucar.edu
On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>
> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
> Queue: met_help
> Subject: SURFACE LEVEL
> Owner: Nobody
> Requestors: vlad.malabanan at ph.ibm.com
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
>
> Hello,
>
> I would like to know what level to specify when I want to check
accuracy
> at surface level of metrics such as precipitation (APCP), pressure
(PRES)
> and humidity (SPFH).
>
> Thanks!
>
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
From: John Halley Gotway
Time: Fri Sep 27 10:52:22 2013
Vladimir,
I took a look at the example data you sent. For verifying surface
fields, you should use the ADPSFC message type. I see that that's
what you have in the NetCDF point observation file you sent.
When I run with your data, I get 711 matched pairs for 2-m temperature
for the "FULL" domain (which is just the full domain of the forecast
data). However, I get 0 matched pairs for the U and V
components of the wind. Running Point-Stat at verbosity level 3 (-v
3) reveals:
DEBUG 2: Processing UGRD/Z10 versus UGRD/Z10, for observation type
ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
DEBUG 3: Number of matched pairs = 0
DEBUG 3: Observations processed = 28295
DEBUG 3: Rejected: GRIB code = 28295
DEBUG 3: Rejected: valid time = 0
DEBUG 3: Rejected: bad obs value = 0
DEBUG 3: Rejected: off the grid = 0
DEBUG 3: Rejected: level mismatch = 0
DEBUG 3: Rejected: quality marker = 0
DEBUG 3: Rejected: message type = 0
DEBUG 3: Rejected: masking region = 0
DEBUG 3: Rejected: bad fcst value = 0
DEBUG 3: Rejected: duplicates = 0
So we're getting 0 matched pairs because none of the observations GRIB
code matches the UGRD GRIB code, which is 33. See GRIB table here:
http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html
I wanted to inspect your observations easily so I converted them back
to ASCII by running the following R script:
Rscript METv4.1/scripts/Rscripts/pntnc2ascii.R
SURFACE_OBS_20130401.nc > SURFACE_OBS_20130401.txt
Looking in SURFACE_OBS_20130401.txt, I see that you have GRIB code
observations of wind speed and direction (codes 31 and 32). You can
verify wind speed directly in Point-Stat, but for wind
direction, you'd need to U and V components. So I modified your
config file like this:
field = [
{
name = "TMP";
level = [ "Z2" ];
cat_thresh = [ <=273, >273 ];
},
{
name = "WIND";
level = [ "Z10" ];
cat_thresh = [ >=5 ];
}
];
Running Point-Stat again, I now get 711 matched pairs for both 2-m
temperature and 10-m wind speed.
Hope that helps.
Thanks,
John Halley Gotway
met_help at ucar.edu
On 09/27/2013 03:00 AM, Vladimir A Malabanan via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
> Hello John,
>
> Thanks for the response.
>
> Yes, you are right. I am using little_r file format for my
observation
> file to match with my forecast file (in GRIB2 format, converted from
WRF
> netcdf file using UPP).
>
> As for my question, I would just like to know the correct level to
compute
> for the accuracy of precipitation amount (APCP) (and other surface
> measurements)
>
> {
> name = "APSP";
> level = [ "__" ];
> cat_thresh = [ >=0 ];
> },
>
> I have already uploaded my sample data (netcdf from converted
little_r for
> observed values and GRIB2 file from converted netcdf for the
forecast
> file) in the ftp site. I am just not sure if I was able to upload it
> successfully.
>
> In any case, I am attaching the files here:
>
>
>
> Thanks for your assistance! I am really new in MET (and in weather
> forecasting!). I am just trying out some things for a project.
>
> Thank you!
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
>
>
>
> From: "John Halley Gotway via RT" <met_help at ucar.edu>
> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
> Date: 09/27/2013 12:28 AM
> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>
>
>
> Vladimir,
>
> I see from your earlier email that you're using little R point
> observations in MET. I assume that you're using them for doing
> verification with the Point-Stat tool. Support for little R point
> observations was recently added in METv4.1, and we've only had a
handful
> of users try it out to my knowledge. So I'll be interested to hear
how it
> goes and what issues, if any, arise.
>
> I see that you're asking about what "level" to use when doing
verification
> of surface variables. Surface fields are kind of a special case
within
> Point-Stat.
>
> For upper-air variables, like 500mb temperature for example, you
would
> specify a range of observation pressure values that you'd like to
compare
> to that field. For example, you may want to match any
> observations with a pressure value between 475 and 525mb to the
500mb
> forecast level. However for surface variables, Point-Stat does not
really
> use the "level" value. Instead, the matching is
> controlled by the message type. Observations with message types of
ADPSFC
> (over land) and SFCSHP (over water) are assumed to be at the
surface. When
> verifying surface fields, like 2-m temperature or
> 10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
> types. Any observations with those message types will be compared
against
> the forecast at the surface.
>
> Since you're using little R data, things may get a little messy.
Currently
> ascii2nc contains a mapping of little R message type to the
conventional
> message types used in the rest of MET (like ADPSFC
> and SFCSHP). However, I suspect that mapping may be incomplete.
>
> Please try running ASCII2NC and Point-Stat, and if you run into
problems,
> just let me know. It'll probably be easiest to just have you send
me some
> sample data. That way I can run your data here and
> figure out what changes, if any, will be required to get it working
more
> smoothly in MET. You can post data to our anonymous ftp site by
following
> these instructions:
> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>>
>> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
>> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
>> Queue: met_help
>> Subject: SURFACE LEVEL
>> Owner: Nobody
>> Requestors: vlad.malabanan at ph.ibm.com
>> Status: new
>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>
>>
>> Hello,
>>
>> I would like to know what level to specify when I want to check
accuracy
>> at surface level of metrics such as precipitation (APCP), pressure
> (PRES)
>> and humidity (SPFH).
>>
>> Thanks!
>>
>>
>>
>> Vladimir A. Malabanan
>> 3F IBM Bldg J, UP Ayala Technohub
>>
>> Applications Programmer
>> Advanced Analytics - Data Specialist
>> Diliman, Quezon City
>> IBM Solutions Delivery, Inc.
>> Philippines
>> Global Delivery Center
>>
>>
>> Phone:
>> +632-995-2IBM local 8639
>>
>>
>> e-mail:
>> malabava at ph.ibm.com
>>
>>
>>
>>
>>
>>
>
>
>
------------------------------------------------
Subject: SURFACE LEVEL
From: Vladimir A Malabanan
Time: Fri Sep 27 19:02:12 2013
Thanks John!
Actually, I also just figured it out that in order to compute for
accuracy
for WIND, I should add WIND in the forecast fields ( instead of the
individual U and V).
However, I still don't know the proper config in order to verify rain
value/precipitation amount forecasts. Do you know how to set the
configuration file (the levels,etc) to verify this measure?
Thanks again!
Vladimir A. Malabanan
3F IBM Bldg J, UP Ayala Technohub
Applications Programmer
Advanced Analytics - Data Specialist
Diliman, Quezon City
IBM Solutions Delivery, Inc.
Philippines
Global Delivery Center
Phone:
+632-995-2IBM local 8639
e-mail:
malabava at ph.ibm.com
From: "John Halley Gotway via RT" <met_help at ucar.edu>
To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
Date: 09/28/2013 12:51 AM
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
Vladimir,
I took a look at the example data you sent. For verifying surface
fields,
you should use the ADPSFC message type. I see that that's what you
have
in the NetCDF point observation file you sent.
When I run with your data, I get 711 matched pairs for 2-m temperature
for
the "FULL" domain (which is just the full domain of the forecast
data).
However, I get 0 matched pairs for the U and V
components of the wind. Running Point-Stat at verbosity level 3 (-v
3)
reveals:
DEBUG 2: Processing UGRD/Z10 versus UGRD/Z10, for observation type
ADPSFC,
over region FULL, for interpolation method UW_MEAN(1), using 0 pairs.
DEBUG 3: Number of matched pairs = 0
DEBUG 3: Observations processed = 28295
DEBUG 3: Rejected: GRIB code = 28295
DEBUG 3: Rejected: valid time = 0
DEBUG 3: Rejected: bad obs value = 0
DEBUG 3: Rejected: off the grid = 0
DEBUG 3: Rejected: level mismatch = 0
DEBUG 3: Rejected: quality marker = 0
DEBUG 3: Rejected: message type = 0
DEBUG 3: Rejected: masking region = 0
DEBUG 3: Rejected: bad fcst value = 0
DEBUG 3: Rejected: duplicates = 0
So we're getting 0 matched pairs because none of the observations GRIB
code matches the UGRD GRIB code, which is 33. See GRIB table here:
http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html
I wanted to inspect your observations easily so I converted them back
to
ASCII by running the following R script:
Rscript METv4.1/scripts/Rscripts/pntnc2ascii.R
SURFACE_OBS_20130401.nc
> SURFACE_OBS_20130401.txt
Looking in SURFACE_OBS_20130401.txt, I see that you have GRIB code
observations of wind speed and direction (codes 31 and 32). You can
verify wind speed directly in Point-Stat, but for wind
direction, you'd need to U and V components. So I modified your
config
file like this:
field = [
{
name = "TMP";
level = [ "Z2" ];
cat_thresh = [ <=273, >273 ];
},
{
name = "WIND";
level = [ "Z10" ];
cat_thresh = [ >=5 ];
}
];
Running Point-Stat again, I now get 711 matched pairs for both 2-m
temperature and 10-m wind speed.
Hope that helps.
Thanks,
John Halley Gotway
met_help at ucar.edu
On 09/27/2013 03:00 AM, Vladimir A Malabanan via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
> Hello John,
>
> Thanks for the response.
>
> Yes, you are right. I am using little_r file format for my
observation
> file to match with my forecast file (in GRIB2 format, converted from
WRF
> netcdf file using UPP).
>
> As for my question, I would just like to know the correct level to
compute
> for the accuracy of precipitation amount (APCP) (and other surface
> measurements)
>
> {
> name = "APSP";
> level = [ "__" ];
> cat_thresh = [ >=0 ];
> },
>
> I have already uploaded my sample data (netcdf from converted
little_r
for
> observed values and GRIB2 file from converted netcdf for the
forecast
> file) in the ftp site. I am just not sure if I was able to upload it
> successfully.
>
> In any case, I am attaching the files here:
>
>
>
> Thanks for your assistance! I am really new in MET (and in weather
> forecasting!). I am just trying out some things for a project.
>
> Thank you!
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
>
>
>
> From: "John Halley Gotway via RT" <met_help at ucar.edu>
> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
> Date: 09/27/2013 12:28 AM
> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>
>
>
> Vladimir,
>
> I see from your earlier email that you're using little R point
> observations in MET. I assume that you're using them for doing
> verification with the Point-Stat tool. Support for little R point
> observations was recently added in METv4.1, and we've only had a
handful
> of users try it out to my knowledge. So I'll be interested to hear
how
it
> goes and what issues, if any, arise.
>
> I see that you're asking about what "level" to use when doing
verification
> of surface variables. Surface fields are kind of a special case
within
> Point-Stat.
>
> For upper-air variables, like 500mb temperature for example, you
would
> specify a range of observation pressure values that you'd like to
compare
> to that field. For example, you may want to match any
> observations with a pressure value between 475 and 525mb to the
500mb
> forecast level. However for surface variables, Point-Stat does not
really
> use the "level" value. Instead, the matching is
> controlled by the message type. Observations with message types of
ADPSFC
> (over land) and SFCSHP (over water) are assumed to be at the
surface.
When
> verifying surface fields, like 2-m temperature or
> 10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
> types. Any observations with those message types will be compared
against
> the forecast at the surface.
>
> Since you're using little R data, things may get a little messy.
Currently
> ascii2nc contains a mapping of little R message type to the
conventional
> message types used in the rest of MET (like ADPSFC
> and SFCSHP). However, I suspect that mapping may be incomplete.
>
> Please try running ASCII2NC and Point-Stat, and if you run into
problems,
> just let me know. It'll probably be easiest to just have you send
me
some
> sample data. That way I can run your data here and
> figure out what changes, if any, will be required to get it working
more
> smoothly in MET. You can post data to our anonymous ftp site by
following
> these instructions:
> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>>
>> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
>> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
>> Queue: met_help
>> Subject: SURFACE LEVEL
>> Owner: Nobody
>> Requestors: vlad.malabanan at ph.ibm.com
>> Status: new
>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193
>
>>
>>
>> Hello,
>>
>> I would like to know what level to specify when I want to check
accuracy
>> at surface level of metrics such as precipitation (APCP), pressure
> (PRES)
>> and humidity (SPFH).
>>
>> Thanks!
>>
>>
>>
>> Vladimir A. Malabanan
>> 3F IBM Bldg J, UP Ayala Technohub
>>
>> Applications Programmer
>> Advanced Analytics - Data Specialist
>> Diliman, Quezon City
>> IBM Solutions Delivery, Inc.
>> Philippines
>> Global Delivery Center
>>
>>
>> Phone:
>> +632-995-2IBM local 8639
>>
>>
>> e-mail:
>> malabava at ph.ibm.com
>>
>>
>>
>>
>>
>>
>
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
From: John Halley Gotway
Time: Mon Sep 30 11:12:22 2013
Vladimir,
For verifying precipitation, the ADPSFC message type will work fine.
But you do need to specify a level value - and that level value is the
accumulation interval of the precip.
For example, suppose you have 6-hour precip in the GRIB output of UPP.
And you have 6-hour accumulations of precip in your point
observations. Here's how you'd set up the config file:
++++++++++++++++++++++++++++++++++++
fcst = {
wind_thresh = [ NA ];
message_type = [ "ADPSFC" ];
field = [
{
name = "APCP";
level = [ "A06" ];
cat_thresh = [ >0.0 ];
}
];
};
obs = fcst;
++++++++++++++++++++++++++++++++++++
Assuming you're using ASCII2NC, the ascii input to that tool would
look like this:
ADPSFC STATION_NAME 20111017_120000 34.0081 -118.4989 -9999.00 61
06 -9999.00 -9999.00 1.00
The GRIB code is 61 for accumulated precip, and the "level" value is
06 for 6-hours of accumulation. For precip, the level is given in
HH[MMSS] format, where the minutes and seconds are optional.
Please give that a shot and let me know if you have more questions.
Thanks,
John
On 09/27/2013 07:02 PM, Vladimir A Malabanan via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
> Thanks John!
>
> Actually, I also just figured it out that in order to compute for
accuracy
> for WIND, I should add WIND in the forecast fields ( instead of the
> individual U and V).
>
> However, I still don't know the proper config in order to verify
rain
> value/precipitation amount forecasts. Do you know how to set the
> configuration file (the levels,etc) to verify this measure?
>
> Thanks again!
>
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
>
>
>
> From: "John Halley Gotway via RT" <met_help at ucar.edu>
> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
> Date: 09/28/2013 12:51 AM
> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>
>
>
> Vladimir,
>
> I took a look at the example data you sent. For verifying surface
fields,
> you should use the ADPSFC message type. I see that that's what you
have
> in the NetCDF point observation file you sent.
>
> When I run with your data, I get 711 matched pairs for 2-m
temperature for
> the "FULL" domain (which is just the full domain of the forecast
data).
> However, I get 0 matched pairs for the U and V
> components of the wind. Running Point-Stat at verbosity level 3 (-v
3)
> reveals:
>
> DEBUG 2: Processing UGRD/Z10 versus UGRD/Z10, for observation type
ADPSFC,
> over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 28295
> DEBUG 3: Rejected: GRIB code = 28295
> DEBUG 3: Rejected: valid time = 0
> DEBUG 3: Rejected: bad obs value = 0
> DEBUG 3: Rejected: off the grid = 0
> DEBUG 3: Rejected: level mismatch = 0
> DEBUG 3: Rejected: quality marker = 0
> DEBUG 3: Rejected: message type = 0
> DEBUG 3: Rejected: masking region = 0
> DEBUG 3: Rejected: bad fcst value = 0
> DEBUG 3: Rejected: duplicates = 0
>
> So we're getting 0 matched pairs because none of the observations
GRIB
> code matches the UGRD GRIB code, which is 33. See GRIB table here:
> http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html
>
> I wanted to inspect your observations easily so I converted them
back to
> ASCII by running the following R script:
> Rscript METv4.1/scripts/Rscripts/pntnc2ascii.R
SURFACE_OBS_20130401.nc
>> SURFACE_OBS_20130401.txt
>
> Looking in SURFACE_OBS_20130401.txt, I see that you have GRIB code
> observations of wind speed and direction (codes 31 and 32). You can
> verify wind speed directly in Point-Stat, but for wind
> direction, you'd need to U and V components. So I modified your
config
> file like this:
> field = [
> {
> name = "TMP";
> level = [ "Z2" ];
> cat_thresh = [ <=273, >273 ];
> },
>
> {
> name = "WIND";
> level = [ "Z10" ];
> cat_thresh = [ >=5 ];
> }
> ];
>
> Running Point-Stat again, I now get 711 matched pairs for both 2-m
> temperature and 10-m wind speed.
>
> Hope that helps.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On 09/27/2013 03:00 AM, Vladimir A Malabanan via RT wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>
>> Hello John,
>>
>> Thanks for the response.
>>
>> Yes, you are right. I am using little_r file format for my
observation
>> file to match with my forecast file (in GRIB2 format, converted
from WRF
>> netcdf file using UPP).
>>
>> As for my question, I would just like to know the correct level to
> compute
>> for the accuracy of precipitation amount (APCP) (and other surface
>> measurements)
>>
>> {
>> name = "APSP";
>> level = [ "__" ];
>> cat_thresh = [ >=0 ];
>> },
>>
>> I have already uploaded my sample data (netcdf from converted
little_r
> for
>> observed values and GRIB2 file from converted netcdf for the
forecast
>> file) in the ftp site. I am just not sure if I was able to upload
it
>> successfully.
>>
>> In any case, I am attaching the files here:
>>
>>
>>
>> Thanks for your assistance! I am really new in MET (and in weather
>> forecasting!). I am just trying out some things for a project.
>>
>> Thank you!
>>
>>
>> Vladimir A. Malabanan
>> 3F IBM Bldg J, UP Ayala Technohub
>>
>> Applications Programmer
>> Advanced Analytics - Data Specialist
>> Diliman, Quezon City
>> IBM Solutions Delivery, Inc.
>> Philippines
>> Global Delivery Center
>>
>>
>> Phone:
>> +632-995-2IBM local 8639
>>
>>
>> e-mail:
>> malabava at ph.ibm.com
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> From: "John Halley Gotway via RT" <met_help at ucar.edu>
>> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
>> Date: 09/27/2013 12:28 AM
>> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>>
>>
>>
>> Vladimir,
>>
>> I see from your earlier email that you're using little R point
>> observations in MET. I assume that you're using them for doing
>> verification with the Point-Stat tool. Support for little R point
>> observations was recently added in METv4.1, and we've only had a
handful
>> of users try it out to my knowledge. So I'll be interested to hear
how
> it
>> goes and what issues, if any, arise.
>>
>> I see that you're asking about what "level" to use when doing
> verification
>> of surface variables. Surface fields are kind of a special case
within
>> Point-Stat.
>>
>> For upper-air variables, like 500mb temperature for example, you
would
>> specify a range of observation pressure values that you'd like to
> compare
>> to that field. For example, you may want to match any
>> observations with a pressure value between 475 and 525mb to the
500mb
>> forecast level. However for surface variables, Point-Stat does not
> really
>> use the "level" value. Instead, the matching is
>> controlled by the message type. Observations with message types of
> ADPSFC
>> (over land) and SFCSHP (over water) are assumed to be at the
surface.
> When
>> verifying surface fields, like 2-m temperature or
>> 10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
>> types. Any observations with those message types will be compared
> against
>> the forecast at the surface.
>>
>> Since you're using little R data, things may get a little messy.
> Currently
>> ascii2nc contains a mapping of little R message type to the
conventional
>> message types used in the rest of MET (like ADPSFC
>> and SFCSHP). However, I suspect that mapping may be incomplete.
>>
>> Please try running ASCII2NC and Point-Stat, and if you run into
> problems,
>> just let me know. It'll probably be easiest to just have you send
me
> some
>> sample data. That way I can run your data here and
>> figure out what changes, if any, will be required to get it working
more
>> smoothly in MET. You can post data to our anonymous ftp site by
> following
>> these instructions:
>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>
>> Thanks,
>> John Halley Gotway
>> met_help at ucar.edu
>>
>> On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>>>
>>> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
>>> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
>>> Queue: met_help
>>> Subject: SURFACE LEVEL
>>> Owner: Nobody
>>> Requestors: vlad.malabanan at ph.ibm.com
>>> Status: new
>>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193
>>
>>>
>>>
>>> Hello,
>>>
>>> I would like to know what level to specify when I want to check
> accuracy
>>> at surface level of metrics such as precipitation (APCP), pressure
>> (PRES)
>>> and humidity (SPFH).
>>>
>>> Thanks!
>>>
>>>
>>>
>>> Vladimir A. Malabanan
>>> 3F IBM Bldg J, UP Ayala Technohub
>>>
>>> Applications Programmer
>>> Advanced Analytics - Data Specialist
>>> Diliman, Quezon City
>>> IBM Solutions Delivery, Inc.
>>> Philippines
>>> Global Delivery Center
>>>
>>>
>>> Phone:
>>> +632-995-2IBM local 8639
>>>
>>>
>>> e-mail:
>>> malabava at ph.ibm.com
>>>
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>
>
>
------------------------------------------------
Subject: SURFACE LEVEL
From: Vladimir A Malabanan
Time: Tue Oct 01 03:30:06 2013
Hi John,
I have applied the configuration that you provided to be able to
compute
for accuracy for rain value/precipitation. However, I am getting this
warning:
DEBUG 2: Reading data for APCP/A01.
DEBUG 3: MetGrib1DataFile::data_plane_array() -> Found 0 GRIB records
matching VarInfo "APCP/A01" in GRIB file "../INPUT/WRFPRS_d01.00".
WARNING:
WARNING: process_fcst_climo_files() -> no fields matching APCP/A01
found
in file: ../INPUT/WRFPRS_d01.00
WARNING:
It seems like precipitation was not found in my forecast file. My
forecast
file is a grib2 file for one forecast time converted from WRF netcdf
file.
I have another question though. I tried working on other metrics
(pressure, humidity).
For Pressure (PRES), I am not sure at which level do I need to
specify. I
tried P500 but I was not able to find it in my forecast file. How can
I
know what are the valid levels for a specific measure? Is there a
reference table for that? I specifically would like to make an
accuracy
report for the forecast for pressure (PSFC in WRF netcdf file).
DEBUG 2: Reading data for PRES/P500.
DEBUG 3: MetGrib1DataFile::data_plane_array() -> Found 0 GRIB records
matching VarInfo "PRES/P500" in GRIB file "../INPUT/WRFPRS_d01.00".
WARNING:
WARNING: process_fcst_climo_files() -> no fields matching PRES/P500
found
in file: ../INPUT/WRFPRS_d01.00
I also tried checking on relative humidity. It was found in the
forecast
file but was not able to make any matches against the observed file. I
received a message that point were rejected due to GRIB Code.
DEBUG 2: Processing SPFH/P500 versus SPFH/P500, for observation type
ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
DEBUG 3: Number of matched pairs = 0
DEBUG 3: Observations processed = 28295
DEBUG 3: Rejected: GRIB code = 28295
I have another question on my plate regarding little_r support. I will
just send a separate ticket regarding that.
Thanks John! I really appreciate your help!
Vladimir A. Malabanan
3F IBM Bldg J, UP Ayala Technohub
Applications Programmer
Advanced Analytics - Data Specialist
Diliman, Quezon City
IBM Solutions Delivery, Inc.
Philippines
Global Delivery Center
Phone:
+632-995-2IBM local 8639
e-mail:
malabava at ph.ibm.com
From: "John Halley Gotway via RT" <met_help at ucar.edu>
To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
Date: 10/01/2013 02:13 AM
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
Vladimir,
For verifying precipitation, the ADPSFC message type will work fine.
But
you do need to specify a level value - and that level value is the
accumulation interval of the precip.
For example, suppose you have 6-hour precip in the GRIB output of UPP.
And
you have 6-hour accumulations of precip in your point observations.
Here's
how you'd set up the config file:
++++++++++++++++++++++++++++++++++++
fcst = {
wind_thresh = [ NA ];
message_type = [ "ADPSFC" ];
field = [
{
name = "APCP";
level = [ "A06" ];
cat_thresh = [ >0.0 ];
}
];
};
obs = fcst;
++++++++++++++++++++++++++++++++++++
Assuming you're using ASCII2NC, the ascii input to that tool would
look
like this:
ADPSFC STATION_NAME 20111017_120000 34.0081 -118.4989 -9999.00 61
06
-9999.00 -9999.00 1.00
The GRIB code is 61 for accumulated precip, and the "level" value is
06
for 6-hours of accumulation. For precip, the level is given in
HH[MMSS]
format, where the minutes and seconds are optional.
Please give that a shot and let me know if you have more questions.
Thanks,
John
On 09/27/2013 07:02 PM, Vladimir A Malabanan via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
> Thanks John!
>
> Actually, I also just figured it out that in order to compute for
accuracy
> for WIND, I should add WIND in the forecast fields ( instead of the
> individual U and V).
>
> However, I still don't know the proper config in order to verify
rain
> value/precipitation amount forecasts. Do you know how to set the
> configuration file (the levels,etc) to verify this measure?
>
> Thanks again!
>
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
>
>
>
> From: "John Halley Gotway via RT" <met_help at ucar.edu>
> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
> Date: 09/28/2013 12:51 AM
> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>
>
>
> Vladimir,
>
> I took a look at the example data you sent. For verifying surface
fields,
> you should use the ADPSFC message type. I see that that's what you
have
> in the NetCDF point observation file you sent.
>
> When I run with your data, I get 711 matched pairs for 2-m
temperature
for
> the "FULL" domain (which is just the full domain of the forecast
data).
> However, I get 0 matched pairs for the U and V
> components of the wind. Running Point-Stat at verbosity level 3 (-v
3)
> reveals:
>
> DEBUG 2: Processing UGRD/Z10 versus UGRD/Z10, for observation type
ADPSFC,
> over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 28295
> DEBUG 3: Rejected: GRIB code = 28295
> DEBUG 3: Rejected: valid time = 0
> DEBUG 3: Rejected: bad obs value = 0
> DEBUG 3: Rejected: off the grid = 0
> DEBUG 3: Rejected: level mismatch = 0
> DEBUG 3: Rejected: quality marker = 0
> DEBUG 3: Rejected: message type = 0
> DEBUG 3: Rejected: masking region = 0
> DEBUG 3: Rejected: bad fcst value = 0
> DEBUG 3: Rejected: duplicates = 0
>
> So we're getting 0 matched pairs because none of the observations
GRIB
> code matches the UGRD GRIB code, which is 33. See GRIB table here:
> http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html
>
> I wanted to inspect your observations easily so I converted them
back to
> ASCII by running the following R script:
> Rscript METv4.1/scripts/Rscripts/pntnc2ascii.R
SURFACE_OBS_20130401.nc
>> SURFACE_OBS_20130401.txt
>
> Looking in SURFACE_OBS_20130401.txt, I see that you have GRIB code
> observations of wind speed and direction (codes 31 and 32). You can
> verify wind speed directly in Point-Stat, but for wind
> direction, you'd need to U and V components. So I modified your
config
> file like this:
> field = [
> {
> name = "TMP";
> level = [ "Z2" ];
> cat_thresh = [ <=273, >273 ];
> },
>
> {
> name = "WIND";
> level = [ "Z10" ];
> cat_thresh = [ >=5 ];
> }
> ];
>
> Running Point-Stat again, I now get 711 matched pairs for both 2-m
> temperature and 10-m wind speed.
>
> Hope that helps.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> On 09/27/2013 03:00 AM, Vladimir A Malabanan via RT wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>
>> Hello John,
>>
>> Thanks for the response.
>>
>> Yes, you are right. I am using little_r file format for my
observation
>> file to match with my forecast file (in GRIB2 format, converted
from
WRF
>> netcdf file using UPP).
>>
>> As for my question, I would just like to know the correct level to
> compute
>> for the accuracy of precipitation amount (APCP) (and other surface
>> measurements)
>>
>> {
>> name = "APSP";
>> level = [ "__" ];
>> cat_thresh = [ >=0 ];
>> },
>>
>> I have already uploaded my sample data (netcdf from converted
little_r
> for
>> observed values and GRIB2 file from converted netcdf for the
forecast
>> file) in the ftp site. I am just not sure if I was able to upload
it
>> successfully.
>>
>> In any case, I am attaching the files here:
>>
>>
>>
>> Thanks for your assistance! I am really new in MET (and in weather
>> forecasting!). I am just trying out some things for a project.
>>
>> Thank you!
>>
>>
>> Vladimir A. Malabanan
>> 3F IBM Bldg J, UP Ayala Technohub
>>
>> Applications Programmer
>> Advanced Analytics - Data Specialist
>> Diliman, Quezon City
>> IBM Solutions Delivery, Inc.
>> Philippines
>> Global Delivery Center
>>
>>
>> Phone:
>> +632-995-2IBM local 8639
>>
>>
>> e-mail:
>> malabava at ph.ibm.com
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> From: "John Halley Gotway via RT" <met_help at ucar.edu>
>> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
>> Date: 09/27/2013 12:28 AM
>> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>>
>>
>>
>> Vladimir,
>>
>> I see from your earlier email that you're using little R point
>> observations in MET. I assume that you're using them for doing
>> verification with the Point-Stat tool. Support for little R point
>> observations was recently added in METv4.1, and we've only had a
handful
>> of users try it out to my knowledge. So I'll be interested to hear
how
> it
>> goes and what issues, if any, arise.
>>
>> I see that you're asking about what "level" to use when doing
> verification
>> of surface variables. Surface fields are kind of a special case
within
>> Point-Stat.
>>
>> For upper-air variables, like 500mb temperature for example, you
would
>> specify a range of observation pressure values that you'd like to
> compare
>> to that field. For example, you may want to match any
>> observations with a pressure value between 475 and 525mb to the
500mb
>> forecast level. However for surface variables, Point-Stat does not
> really
>> use the "level" value. Instead, the matching is
>> controlled by the message type. Observations with message types of
> ADPSFC
>> (over land) and SFCSHP (over water) are assumed to be at the
surface.
> When
>> verifying surface fields, like 2-m temperature or
>> 10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
>> types. Any observations with those message types will be compared
> against
>> the forecast at the surface.
>>
>> Since you're using little R data, things may get a little messy.
> Currently
>> ascii2nc contains a mapping of little R message type to the
conventional
>> message types used in the rest of MET (like ADPSFC
>> and SFCSHP). However, I suspect that mapping may be incomplete.
>>
>> Please try running ASCII2NC and Point-Stat, and if you run into
> problems,
>> just let me know. It'll probably be easiest to just have you send
me
> some
>> sample data. That way I can run your data here and
>> figure out what changes, if any, will be required to get it working
more
>> smoothly in MET. You can post data to our anonymous ftp site by
> following
>> these instructions:
>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>
>> Thanks,
>> John Halley Gotway
>> met_help at ucar.edu
>>
>> On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>>>
>>> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
>>> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
>>> Queue: met_help
>>> Subject: SURFACE LEVEL
>>> Owner: Nobody
>>> Requestors: vlad.malabanan at ph.ibm.com
>>> Status: new
>>> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193
>>
>>>
>>>
>>> Hello,
>>>
>>> I would like to know what level to specify when I want to check
> accuracy
>>> at surface level of metrics such as precipitation (APCP), pressure
>> (PRES)
>>> and humidity (SPFH).
>>>
>>> Thanks!
>>>
>>>
>>>
>>> Vladimir A. Malabanan
>>> 3F IBM Bldg J, UP Ayala Technohub
>>>
>>> Applications Programmer
>>> Advanced Analytics - Data Specialist
>>> Diliman, Quezon City
>>> IBM Solutions Delivery, Inc.
>>> Philippines
>>> Global Delivery Center
>>>
>>>
>>> Phone:
>>> +632-995-2IBM local 8639
>>>
>>>
>>> e-mail:
>>> malabava at ph.ibm.com
>>>
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
>
>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
From: John Halley Gotway
Time: Tue Oct 01 14:43:25 2013
Vladimir,
You're trying to retrieve an hour's worth of accumulated precipitation
from the file named "WRFPRS_d01.00". I assume that's an analysis file
- meaning that it's the 0-hour forecast. There cannot be
a 1-hour accumulation in the 0-hour forecast file because the model
has just begun. It's not possible to verify accumulated precipitation
at analysis time - but you could evaluate the state
variables, like temperature and winds, since those are instantaneous
values rather than accumulated ones.
I'd suggest using the wgrib utility to inventory your GRIB files to
see what they contain. For example:
wgrib WRFPRS_d01.00 | grep PRES
That will tell you the levels at which you have pressure available.
If you want data at other pressure levels, you can modify the
configuration you're using when running the Unified Post Processor
(UPP). There's a file called wrf_cntrl.parm that specifies which
variables/levels you'd like in the GRIB output.
Regarding you're point observations, you just need to take a look at
them. If you have observations of relative humidity, you can use them
in the verification step. If not, you can't. Since you're
using ASCII observations, it's up to you look at them and figure out
what comparisons you'd like to make.
Hope that helps.
John
On 10/01/2013 03:30 AM, Vladimir A Malabanan via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
> Hi John,
>
> I have applied the configuration that you provided to be able to
compute
> for accuracy for rain value/precipitation. However, I am getting
this
> warning:
>
> DEBUG 2: Reading data for APCP/A01.
> DEBUG 3: MetGrib1DataFile::data_plane_array() -> Found 0 GRIB
records
> matching VarInfo "APCP/A01" in GRIB file "../INPUT/WRFPRS_d01.00".
> WARNING:
> WARNING: process_fcst_climo_files() -> no fields matching APCP/A01
found
> in file: ../INPUT/WRFPRS_d01.00
> WARNING:
>
> It seems like precipitation was not found in my forecast file. My
forecast
> file is a grib2 file for one forecast time converted from WRF netcdf
file.
>
> I have another question though. I tried working on other metrics
> (pressure, humidity).
>
> For Pressure (PRES), I am not sure at which level do I need to
specify. I
> tried P500 but I was not able to find it in my forecast file. How
can I
> know what are the valid levels for a specific measure? Is there a
> reference table for that? I specifically would like to make an
accuracy
> report for the forecast for pressure (PSFC in WRF netcdf file).
>
> DEBUG 2: Reading data for PRES/P500.
> DEBUG 3: MetGrib1DataFile::data_plane_array() -> Found 0 GRIB
records
> matching VarInfo "PRES/P500" in GRIB file "../INPUT/WRFPRS_d01.00".
> WARNING:
> WARNING: process_fcst_climo_files() -> no fields matching PRES/P500
found
> in file: ../INPUT/WRFPRS_d01.00
>
> I also tried checking on relative humidity. It was found in the
forecast
> file but was not able to make any matches against the observed file.
I
> received a message that point were rejected due to GRIB Code.
>
> DEBUG 2: Processing SPFH/P500 versus SPFH/P500, for observation type
> ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using
0
> pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 28295
> DEBUG 3: Rejected: GRIB code = 28295
>
> I have another question on my plate regarding little_r support. I
will
> just send a separate ticket regarding that.
>
> Thanks John! I really appreciate your help!
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
>
>
>
> From: "John Halley Gotway via RT" <met_help at ucar.edu>
> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
> Date: 10/01/2013 02:13 AM
> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>
>
>
> Vladimir,
>
> For verifying precipitation, the ADPSFC message type will work fine.
But
> you do need to specify a level value - and that level value is the
> accumulation interval of the precip.
>
> For example, suppose you have 6-hour precip in the GRIB output of
UPP. And
> you have 6-hour accumulations of precip in your point observations.
Here's
> how you'd set up the config file:
>
> ++++++++++++++++++++++++++++++++++++
>
> fcst = {
> wind_thresh = [ NA ];
> message_type = [ "ADPSFC" ];
>
> field = [
> {
> name = "APCP";
> level = [ "A06" ];
> cat_thresh = [ >0.0 ];
> }
> ];
>
> };
> obs = fcst;
>
> ++++++++++++++++++++++++++++++++++++
>
> Assuming you're using ASCII2NC, the ascii input to that tool would
look
> like this:
>
> ADPSFC STATION_NAME 20111017_120000 34.0081 -118.4989 -9999.00
61 06
> -9999.00 -9999.00 1.00
>
> The GRIB code is 61 for accumulated precip, and the "level" value is
06
> for 6-hours of accumulation. For precip, the level is given in
HH[MMSS]
> format, where the minutes and seconds are optional.
>
> Please give that a shot and let me know if you have more questions.
>
> Thanks,
> John
>
> On 09/27/2013 07:02 PM, Vladimir A Malabanan via RT wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>
>> Thanks John!
>>
>> Actually, I also just figured it out that in order to compute for
> accuracy
>> for WIND, I should add WIND in the forecast fields ( instead of the
>> individual U and V).
>>
>> However, I still don't know the proper config in order to verify
rain
>> value/precipitation amount forecasts. Do you know how to set the
>> configuration file (the levels,etc) to verify this measure?
>>
>> Thanks again!
>>
>>
>>
>> Vladimir A. Malabanan
>> 3F IBM Bldg J, UP Ayala Technohub
>>
>> Applications Programmer
>> Advanced Analytics - Data Specialist
>> Diliman, Quezon City
>> IBM Solutions Delivery, Inc.
>> Philippines
>> Global Delivery Center
>>
>>
>> Phone:
>> +632-995-2IBM local 8639
>>
>>
>> e-mail:
>> malabava at ph.ibm.com
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> From: "John Halley Gotway via RT" <met_help at ucar.edu>
>> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
>> Date: 09/28/2013 12:51 AM
>> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>>
>>
>>
>> Vladimir,
>>
>> I took a look at the example data you sent. For verifying surface
> fields,
>> you should use the ADPSFC message type. I see that that's what you
have
>> in the NetCDF point observation file you sent.
>>
>> When I run with your data, I get 711 matched pairs for 2-m
temperature
> for
>> the "FULL" domain (which is just the full domain of the forecast
data).
>> However, I get 0 matched pairs for the U and V
>> components of the wind. Running Point-Stat at verbosity level 3 (-
v 3)
>> reveals:
>>
>> DEBUG 2: Processing UGRD/Z10 versus UGRD/Z10, for observation type
> ADPSFC,
>> over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
>> DEBUG 3: Number of matched pairs = 0
>> DEBUG 3: Observations processed = 28295
>> DEBUG 3: Rejected: GRIB code = 28295
>> DEBUG 3: Rejected: valid time = 0
>> DEBUG 3: Rejected: bad obs value = 0
>> DEBUG 3: Rejected: off the grid = 0
>> DEBUG 3: Rejected: level mismatch = 0
>> DEBUG 3: Rejected: quality marker = 0
>> DEBUG 3: Rejected: message type = 0
>> DEBUG 3: Rejected: masking region = 0
>> DEBUG 3: Rejected: bad fcst value = 0
>> DEBUG 3: Rejected: duplicates = 0
>>
>> So we're getting 0 matched pairs because none of the observations
GRIB
>> code matches the UGRD GRIB code, which is 33. See GRIB table here:
>> http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html
>>
>> I wanted to inspect your observations easily so I converted them
back to
>> ASCII by running the following R script:
>> Rscript METv4.1/scripts/Rscripts/pntnc2ascii.R
> SURFACE_OBS_20130401.nc
>>> SURFACE_OBS_20130401.txt
>>
>> Looking in SURFACE_OBS_20130401.txt, I see that you have GRIB code
>> observations of wind speed and direction (codes 31 and 32). You
can
>> verify wind speed directly in Point-Stat, but for wind
>> direction, you'd need to U and V components. So I modified your
config
>> file like this:
>> field = [
>> {
>> name = "TMP";
>> level = [ "Z2" ];
>> cat_thresh = [ <=273, >273 ];
>> },
>>
>> {
>> name = "WIND";
>> level = [ "Z10" ];
>> cat_thresh = [ >=5 ];
>> }
>> ];
>>
>> Running Point-Stat again, I now get 711 matched pairs for both 2-m
>> temperature and 10-m wind speed.
>>
>> Hope that helps.
>>
>> Thanks,
>> John Halley Gotway
>> met_help at ucar.edu
>>
>> On 09/27/2013 03:00 AM, Vladimir A Malabanan via RT wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>>
>>> Hello John,
>>>
>>> Thanks for the response.
>>>
>>> Yes, you are right. I am using little_r file format for my
observation
>>> file to match with my forecast file (in GRIB2 format, converted
from
> WRF
>>> netcdf file using UPP).
>>>
>>> As for my question, I would just like to know the correct level to
>> compute
>>> for the accuracy of precipitation amount (APCP) (and other surface
>>> measurements)
>>>
>>> {
>>> name = "APSP";
>>> level = [ "__" ];
>>> cat_thresh = [ >=0 ];
>>> },
>>>
>>> I have already uploaded my sample data (netcdf from converted
little_r
>> for
>>> observed values and GRIB2 file from converted netcdf for the
forecast
>>> file) in the ftp site. I am just not sure if I was able to upload
it
>>> successfully.
>>>
>>> In any case, I am attaching the files here:
>>>
>>>
>>>
>>> Thanks for your assistance! I am really new in MET (and in weather
>>> forecasting!). I am just trying out some things for a project.
>>>
>>> Thank you!
>>>
>>>
>>> Vladimir A. Malabanan
>>> 3F IBM Bldg J, UP Ayala Technohub
>>>
>>> Applications Programmer
>>> Advanced Analytics - Data Specialist
>>> Diliman, Quezon City
>>> IBM Solutions Delivery, Inc.
>>> Philippines
>>> Global Delivery Center
>>>
>>>
>>> Phone:
>>> +632-995-2IBM local 8639
>>>
>>>
>>> e-mail:
>>> malabava at ph.ibm.com
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> From: "John Halley Gotway via RT" <met_help at ucar.edu>
>>> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
>>> Date: 09/27/2013 12:28 AM
>>> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>>>
>>>
>>>
>>> Vladimir,
>>>
>>> I see from your earlier email that you're using little R point
>>> observations in MET. I assume that you're using them for doing
>>> verification with the Point-Stat tool. Support for little R point
>>> observations was recently added in METv4.1, and we've only had a
> handful
>>> of users try it out to my knowledge. So I'll be interested to
hear how
>> it
>>> goes and what issues, if any, arise.
>>>
>>> I see that you're asking about what "level" to use when doing
>> verification
>>> of surface variables. Surface fields are kind of a special case
within
>>> Point-Stat.
>>>
>>> For upper-air variables, like 500mb temperature for example, you
would
>>> specify a range of observation pressure values that you'd like to
>> compare
>>> to that field. For example, you may want to match any
>>> observations with a pressure value between 475 and 525mb to the
500mb
>>> forecast level. However for surface variables, Point-Stat does
not
>> really
>>> use the "level" value. Instead, the matching is
>>> controlled by the message type. Observations with message types
of
>> ADPSFC
>>> (over land) and SFCSHP (over water) are assumed to be at the
surface.
>> When
>>> verifying surface fields, like 2-m temperature or
>>> 10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
>>> types. Any observations with those message types will be compared
>> against
>>> the forecast at the surface.
>>>
>>> Since you're using little R data, things may get a little messy.
>> Currently
>>> ascii2nc contains a mapping of little R message type to the
> conventional
>>> message types used in the rest of MET (like ADPSFC
>>> and SFCSHP). However, I suspect that mapping may be incomplete.
>>>
>>> Please try running ASCII2NC and Point-Stat, and if you run into
>> problems,
>>> just let me know. It'll probably be easiest to just have you send
me
>> some
>>> sample data. That way I can run your data here and
>>> figure out what changes, if any, will be required to get it
working
> more
>>> smoothly in MET. You can post data to our anonymous ftp site by
>> following
>>> these instructions:
>>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>>
>>> Thanks,
>>> John Halley Gotway
>>> met_help at ucar.edu
>>>
>>> On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>>>>
>>>> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
>>>> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
>>>> Queue: met_help
>>>> Subject: SURFACE LEVEL
>>>> Owner: Nobody
>>>> Requestors: vlad.malabanan at ph.ibm.com
>>>> Status: new
>>>> Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193
>>>
>>>>
>>>>
>>>> Hello,
>>>>
>>>> I would like to know what level to specify when I want to check
>> accuracy
>>>> at surface level of metrics such as precipitation (APCP),
pressure
>>> (PRES)
>>>> and humidity (SPFH).
>>>>
>>>> Thanks!
>>>>
>>>>
>>>>
>>>> Vladimir A. Malabanan
>>>> 3F IBM Bldg J, UP Ayala Technohub
>>>>
>>>> Applications Programmer
>>>> Advanced Analytics - Data Specialist
>>>> Diliman, Quezon City
>>>> IBM Solutions Delivery, Inc.
>>>> Philippines
>>>> Global Delivery Center
>>>>
>>>>
>>>> Phone:
>>>> +632-995-2IBM local 8639
>>>>
>>>>
>>>> e-mail:
>>>> malabava at ph.ibm.com
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>
>>
>>
>
>
>
------------------------------------------------
Subject: SURFACE LEVEL
From: Vladimir A Malabanan
Time: Tue Oct 01 18:23:51 2013
Hi John,
Thanks for the tips! I will apply them and I will let you know of any
progress!
Maraming Salamat!
Vladimir A. Malabanan
3F IBM Bldg J, UP Ayala Technohub
Applications Programmer
Advanced Analytics - Data Specialist
Diliman, Quezon City
IBM Solutions Delivery, Inc.
Philippines
Global Delivery Center
Phone:
+632-995-2IBM local 8639
e-mail:
malabava at ph.ibm.com
From: "John Halley Gotway via RT" <met_help at ucar.edu>
To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
Date: 10/02/2013 04:42 AM
Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
Vladimir,
You're trying to retrieve an hour's worth of accumulated precipitation
from the file named "WRFPRS_d01.00". I assume that's an analysis file
-
meaning that it's the 0-hour forecast. There cannot be
a 1-hour accumulation in the 0-hour forecast file because the model
has
just begun. It's not possible to verify accumulated precipitation at
analysis time - but you could evaluate the state
variables, like temperature and winds, since those are instantaneous
values rather than accumulated ones.
I'd suggest using the wgrib utility to inventory your GRIB files to
see
what they contain. For example:
wgrib WRFPRS_d01.00 | grep PRES
That will tell you the levels at which you have pressure available.
If
you want data at other pressure levels, you can modify the
configuration
you're using when running the Unified Post Processor
(UPP). There's a file called wrf_cntrl.parm that specifies which
variables/levels you'd like in the GRIB output.
Regarding you're point observations, you just need to take a look at
them.
If you have observations of relative humidity, you can use them in
the
verification step. If not, you can't. Since you're
using ASCII observations, it's up to you look at them and figure out
what
comparisons you'd like to make.
Hope that helps.
John
On 10/01/2013 03:30 AM, Vladimir A Malabanan via RT wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>
> Hi John,
>
> I have applied the configuration that you provided to be able to
compute
> for accuracy for rain value/precipitation. However, I am getting
this
> warning:
>
> DEBUG 2: Reading data for APCP/A01.
> DEBUG 3: MetGrib1DataFile::data_plane_array() -> Found 0 GRIB
records
> matching VarInfo "APCP/A01" in GRIB file "../INPUT/WRFPRS_d01.00".
> WARNING:
> WARNING: process_fcst_climo_files() -> no fields matching APCP/A01
found
> in file: ../INPUT/WRFPRS_d01.00
> WARNING:
>
> It seems like precipitation was not found in my forecast file. My
forecast
> file is a grib2 file for one forecast time converted from WRF netcdf
file.
>
> I have another question though. I tried working on other metrics
> (pressure, humidity).
>
> For Pressure (PRES), I am not sure at which level do I need to
specify.
I
> tried P500 but I was not able to find it in my forecast file. How
can I
> know what are the valid levels for a specific measure? Is there a
> reference table for that? I specifically would like to make an
accuracy
> report for the forecast for pressure (PSFC in WRF netcdf file).
>
> DEBUG 2: Reading data for PRES/P500.
> DEBUG 3: MetGrib1DataFile::data_plane_array() -> Found 0 GRIB
records
> matching VarInfo "PRES/P500" in GRIB file "../INPUT/WRFPRS_d01.00".
> WARNING:
> WARNING: process_fcst_climo_files() -> no fields matching PRES/P500
found
> in file: ../INPUT/WRFPRS_d01.00
>
> I also tried checking on relative humidity. It was found in the
forecast
> file but was not able to make any matches against the observed file.
I
> received a message that point were rejected due to GRIB Code.
>
> DEBUG 2: Processing SPFH/P500 versus SPFH/P500, for observation type
> ADPSFC, over region FULL, for interpolation method UW_MEAN(1), using
0
> pairs.
> DEBUG 3: Number of matched pairs = 0
> DEBUG 3: Observations processed = 28295
> DEBUG 3: Rejected: GRIB code = 28295
>
> I have another question on my plate regarding little_r support. I
will
> just send a separate ticket regarding that.
>
> Thanks John! I really appreciate your help!
>
>
> Vladimir A. Malabanan
> 3F IBM Bldg J, UP Ayala Technohub
>
> Applications Programmer
> Advanced Analytics - Data Specialist
> Diliman, Quezon City
> IBM Solutions Delivery, Inc.
> Philippines
> Global Delivery Center
>
>
> Phone:
> +632-995-2IBM local 8639
>
>
> e-mail:
> malabava at ph.ibm.com
>
>
>
>
>
>
>
>
>
> From: "John Halley Gotway via RT" <met_help at ucar.edu>
> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
> Date: 10/01/2013 02:13 AM
> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>
>
>
> Vladimir,
>
> For verifying precipitation, the ADPSFC message type will work fine.
But
> you do need to specify a level value - and that level value is the
> accumulation interval of the precip.
>
> For example, suppose you have 6-hour precip in the GRIB output of
UPP.
And
> you have 6-hour accumulations of precip in your point observations.
Here's
> how you'd set up the config file:
>
> ++++++++++++++++++++++++++++++++++++
>
> fcst = {
> wind_thresh = [ NA ];
> message_type = [ "ADPSFC" ];
>
> field = [
> {
> name = "APCP";
> level = [ "A06" ];
> cat_thresh = [ >0.0 ];
> }
> ];
>
> };
> obs = fcst;
>
> ++++++++++++++++++++++++++++++++++++
>
> Assuming you're using ASCII2NC, the ascii input to that tool would
look
> like this:
>
> ADPSFC STATION_NAME 20111017_120000 34.0081 -118.4989 -9999.00
61
06
> -9999.00 -9999.00 1.00
>
> The GRIB code is 61 for accumulated precip, and the "level" value is
06
> for 6-hours of accumulation. For precip, the level is given in
HH[MMSS]
> format, where the minutes and seconds are optional.
>
> Please give that a shot and let me know if you have more questions.
>
> Thanks,
> John
>
> On 09/27/2013 07:02 PM, Vladimir A Malabanan via RT wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>
>> Thanks John!
>>
>> Actually, I also just figured it out that in order to compute for
> accuracy
>> for WIND, I should add WIND in the forecast fields ( instead of the
>> individual U and V).
>>
>> However, I still don't know the proper config in order to verify
rain
>> value/precipitation amount forecasts. Do you know how to set the
>> configuration file (the levels,etc) to verify this measure?
>>
>> Thanks again!
>>
>>
>>
>> Vladimir A. Malabanan
>> 3F IBM Bldg J, UP Ayala Technohub
>>
>> Applications Programmer
>> Advanced Analytics - Data Specialist
>> Diliman, Quezon City
>> IBM Solutions Delivery, Inc.
>> Philippines
>> Global Delivery Center
>>
>>
>> Phone:
>> +632-995-2IBM local 8639
>>
>>
>> e-mail:
>> malabava at ph.ibm.com
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> From: "John Halley Gotway via RT" <met_help at ucar.edu>
>> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
>> Date: 09/28/2013 12:51 AM
>> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>>
>>
>>
>> Vladimir,
>>
>> I took a look at the example data you sent. For verifying surface
> fields,
>> you should use the ADPSFC message type. I see that that's what you
have
>> in the NetCDF point observation file you sent.
>>
>> When I run with your data, I get 711 matched pairs for 2-m
temperature
> for
>> the "FULL" domain (which is just the full domain of the forecast
data).
>> However, I get 0 matched pairs for the U and V
>> components of the wind. Running Point-Stat at verbosity level 3 (-
v 3)
>> reveals:
>>
>> DEBUG 2: Processing UGRD/Z10 versus UGRD/Z10, for observation type
> ADPSFC,
>> over region FULL, for interpolation method UW_MEAN(1), using 0
pairs.
>> DEBUG 3: Number of matched pairs = 0
>> DEBUG 3: Observations processed = 28295
>> DEBUG 3: Rejected: GRIB code = 28295
>> DEBUG 3: Rejected: valid time = 0
>> DEBUG 3: Rejected: bad obs value = 0
>> DEBUG 3: Rejected: off the grid = 0
>> DEBUG 3: Rejected: level mismatch = 0
>> DEBUG 3: Rejected: quality marker = 0
>> DEBUG 3: Rejected: message type = 0
>> DEBUG 3: Rejected: masking region = 0
>> DEBUG 3: Rejected: bad fcst value = 0
>> DEBUG 3: Rejected: duplicates = 0
>>
>> So we're getting 0 matched pairs because none of the observations
GRIB
>> code matches the UGRD GRIB code, which is 33. See GRIB table here:
>> http://www.nco.ncep.noaa.gov/pmb/docs/on388/table2.html
>>
>> I wanted to inspect your observations easily so I converted them
back
to
>> ASCII by running the following R script:
>> Rscript METv4.1/scripts/Rscripts/pntnc2ascii.R
> SURFACE_OBS_20130401.nc
>>> SURFACE_OBS_20130401.txt
>>
>> Looking in SURFACE_OBS_20130401.txt, I see that you have GRIB code
>> observations of wind speed and direction (codes 31 and 32). You
can
>> verify wind speed directly in Point-Stat, but for wind
>> direction, you'd need to U and V components. So I modified your
config
>> file like this:
>> field = [
>> {
>> name = "TMP";
>> level = [ "Z2" ];
>> cat_thresh = [ <=273, >273 ];
>> },
>>
>> {
>> name = "WIND";
>> level = [ "Z10" ];
>> cat_thresh = [ >=5 ];
>> }
>> ];
>>
>> Running Point-Stat again, I now get 711 matched pairs for both 2-m
>> temperature and 10-m wind speed.
>>
>> Hope that helps.
>>
>> Thanks,
>> John Halley Gotway
>> met_help at ucar.edu
>>
>> On 09/27/2013 03:00 AM, Vladimir A Malabanan via RT wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193 >
>>>
>>> Hello John,
>>>
>>> Thanks for the response.
>>>
>>> Yes, you are right. I am using little_r file format for my
observation
>>> file to match with my forecast file (in GRIB2 format, converted
from
> WRF
>>> netcdf file using UPP).
>>>
>>> As for my question, I would just like to know the correct level to
>> compute
>>> for the accuracy of precipitation amount (APCP) (and other surface
>>> measurements)
>>>
>>> {
>>> name = "APSP";
>>> level = [ "__" ];
>>> cat_thresh = [ >=0 ];
>>> },
>>>
>>> I have already uploaded my sample data (netcdf from converted
little_r
>> for
>>> observed values and GRIB2 file from converted netcdf for the
forecast
>>> file) in the ftp site. I am just not sure if I was able to upload
it
>>> successfully.
>>>
>>> In any case, I am attaching the files here:
>>>
>>>
>>>
>>> Thanks for your assistance! I am really new in MET (and in weather
>>> forecasting!). I am just trying out some things for a project.
>>>
>>> Thank you!
>>>
>>>
>>> Vladimir A. Malabanan
>>> 3F IBM Bldg J, UP Ayala Technohub
>>>
>>> Applications Programmer
>>> Advanced Analytics - Data Specialist
>>> Diliman, Quezon City
>>> IBM Solutions Delivery, Inc.
>>> Philippines
>>> Global Delivery Center
>>>
>>>
>>> Phone:
>>> +632-995-2IBM local 8639
>>>
>>>
>>> e-mail:
>>> malabava at ph.ibm.com
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> From: "John Halley Gotway via RT" <met_help at ucar.edu>
>>> To: Vladimir A Malabanan/Philippines/IBM at IBMPH,
>>> Date: 09/27/2013 12:28 AM
>>> Subject: Re: [rt.rap.ucar.edu #63193] SURFACE LEVEL
>>>
>>>
>>>
>>> Vladimir,
>>>
>>> I see from your earlier email that you're using little R point
>>> observations in MET. I assume that you're using them for doing
>>> verification with the Point-Stat tool. Support for little R point
>>> observations was recently added in METv4.1, and we've only had a
> handful
>>> of users try it out to my knowledge. So I'll be interested to
hear
how
>> it
>>> goes and what issues, if any, arise.
>>>
>>> I see that you're asking about what "level" to use when doing
>> verification
>>> of surface variables. Surface fields are kind of a special case
within
>>> Point-Stat.
>>>
>>> For upper-air variables, like 500mb temperature for example, you
would
>>> specify a range of observation pressure values that you'd like to
>> compare
>>> to that field. For example, you may want to match any
>>> observations with a pressure value between 475 and 525mb to the
500mb
>>> forecast level. However for surface variables, Point-Stat does
not
>> really
>>> use the "level" value. Instead, the matching is
>>> controlled by the message type. Observations with message types
of
>> ADPSFC
>>> (over land) and SFCSHP (over water) are assumed to be at the
surface.
>> When
>>> verifying surface fields, like 2-m temperature or
>>> 10-m winds, you should compare them to the APDSFC and/or SFCSHP
message
>>> types. Any observations with those message types will be compared
>> against
>>> the forecast at the surface.
>>>
>>> Since you're using little R data, things may get a little messy.
>> Currently
>>> ascii2nc contains a mapping of little R message type to the
> conventional
>>> message types used in the rest of MET (like ADPSFC
>>> and SFCSHP). However, I suspect that mapping may be incomplete.
>>>
>>> Please try running ASCII2NC and Point-Stat, and if you run into
>> problems,
>>> just let me know. It'll probably be easiest to just have you send
me
>> some
>>> sample data. That way I can run your data here and
>>> figure out what changes, if any, will be required to get it
working
> more
>>> smoothly in MET. You can post data to our anonymous ftp site by
>> following
>>> these instructions:
>>> http://www.dtcenter.org/met/users/support/met_help.php#ftp
>>>
>>> Thanks,
>>> John Halley Gotway
>>> met_help at ucar.edu
>>>
>>> On 09/26/2013 04:25 AM, Vladimir A Malabanan via RT wrote:
>>>>
>>>> Thu Sep 26 04:25:01 2013: Request 63193 was acted upon.
>>>> Transaction: Ticket created by vlad.malabanan at ph.ibm.com
>>>> Queue: met_help
>>>> Subject: SURFACE LEVEL
>>>> Owner: Nobody
>>>> Requestors: vlad.malabanan at ph.ibm.com
>>>> Status: new
>>>> Ticket <URL:
> https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=63193
>>>
>>>>
>>>>
>>>> Hello,
>>>>
>>>> I would like to know what level to specify when I want to check
>> accuracy
>>>> at surface level of metrics such as precipitation (APCP),
pressure
>>> (PRES)
>>>> and humidity (SPFH).
>>>>
>>>> Thanks!
>>>>
>>>>
>>>>
>>>> Vladimir A. Malabanan
>>>> 3F IBM Bldg J, UP Ayala Technohub
>>>>
>>>> Applications Programmer
>>>> Advanced Analytics - Data Specialist
>>>> Diliman, Quezon City
>>>> IBM Solutions Delivery, Inc.
>>>> Philippines
>>>> Global Delivery Center
>>>>
>>>>
>>>> Phone:
>>>> +632-995-2IBM local 8639
>>>>
>>>>
>>>> e-mail:
>>>> malabava at ph.ibm.com
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>
>>
>>
>
>
>
------------------------------------------------
More information about the Met_help
mailing list