[Met_help] [rt.rap.ucar.edu #47148] History for question on the output of stat_analysis
RAL HelpDesk {for John Halley Gotway}
met_help at ucar.edu
Fri Jun 17 14:12:54 MDT 2011
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi,
We tried to use stat_analysis to calculate temperature and relative humidity's RMSE. After successfully using pb2nc and point_stat as below,
/discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-3.2.1/MET/bin/pb2nc_anil prepbufr.gdas.2008062312.wo40 ./2008062312_test_new.nc ./PB2NCConfig_G212_SFC_WRF_sjz
/discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-3.2.1/MET/bin/point_stat wrfout_d01_2008-06-23_12:
00:00_PLEV ./2008062312_test_new.nc PointStatConfig_SFC_WRF_sjz
we use stat_analysis as blow:
/discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-3.2.1/MET/bin/stat_analysis -lookin ./point_stat/point_stat_15000000L_20080623_120000V.stat -out test_stat_output -job summary -line_type cnt -fcst_var RH -dump_row temp_cnt_out -column RMSE
It works for RH as well as TT as fcst_var not TMP. In the outputs, we get the values for min,max, mean are same as shown below:
JOB_LIST: -job summary -fcst_var RH -line_type CNT -dump_row temp_cnt_out -column RMSE -out_alpha 0.050000
COL_NAME: TOTAL MEAN MEAN_NCL MEAN_NCU MEAN_BCL MEAN_BCU STDEV STDEV_BCL STDEV_BCU MIN P10 P25 P50 P75 P90 MAX
SUMMARY: 1 26.99475 NA NA NA NA NA NA NA 26.99475 26.99475 26.99475 26.99475 26.99475 26.99475 26.99475
Any suggestion?
Thanks,
Shujia
Anil
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47148] question on the output of stat_analysis
From: John Halley Gotway
Time: Thu May 26 10:52:05 2011
Shujia,
I see that you're not currently a registered user of the MET software.
Prior to providing direct support in the use of MET, we require that
users fill out the free online registration form. That
enables us to ensure that we remain in compliance with the
international export laws of the United States.
To fill out the registration, just go here:
http://www.dtcenter.org/met/users/downloads/index.php
Enter you email address and hit enter to be taken to the form.
Please write back to let me know when you've registered. Also please
send me the Point-Stat output file that you're running through STAT-
Analysis and the STAT-Analysis command line you're running
that's returning no data.
Thanks,
John Halley Gotway
met_help at ucar.edu
On 05/26/2011 10:06 AM, RAL HelpDesk {for Shujia Zhou} wrote:
> /discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-
3.2.1/MET/bin/stat_analysis -lookin
./point_stat/point_stat_15000000L_20080623_120000V.stat -out
test_stat_output -job summary -line_type cnt -fcst_var RH -dump_row
temp_cnt_out -column RMSE
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47148] question on the output of stat_analysis
From: Shujia Zhou
Time: Thu May 26 11:30:49 2011
Hi, John:
On May 26, 2011, at 12:52 PM, RAL HelpDesk {for John Halley Gotway}
wrote:
> Shujia,
>
> I see that you're not currently a registered user of the MET
software. Prior to providing direct support in the use of MET, we
require that users fill out the free online registration form. That
> enables us to ensure that we remain in compliance with the
international export laws of the United States.
>
> To fill out the registration, just go here:
> http://www.dtcenter.org/met/users/downloads/index.php
> Enter you email address and hit enter to be taken to the form.
>
> Please write back to let me know when you've registered.
Done.
> Also please send me the Point-Stat output file that you're running
through STAT-Analysis and the STAT-Analysis command line you're
running
> that's returning no data.
point_stat_15000000L_20080623_120000V.stat
is in the next email.
Thanks,
Shujia
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
>
>
> On 05/26/2011 10:06 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>> /discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-
3.2.1/MET/bin/stat_analysis -lookin
./point_stat/point_stat_15000000L_20080623_120000V.stat -out
test_stat_output -job summary -line_type cnt -fcst_var RH -dump_row
temp_cnt_out -column RMSE
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47148] question on the output of stat_analysis
From: John Halley Gotway
Time: Thu May 26 11:40:53 2011
Shujia,
Please take a look in the FCST_VAR column of the data file you just
sent me. You'll see that it only contains entries for "TT" and "RH".
When you run your STAT-Analysis job using the "-fcst_var RH"
option, you'll match those lines with RH in the FCST_VAR column, and
likewise for TT. If you run with "-fcst_var TMP", you'll match 0 STAT
lines because none of them have TMP in the FCST_VAR column.
Hope that helps.
John
On 05/26/2011 11:30 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47148 >
>
> Hi, John:
>
>
> On May 26, 2011, at 12:52 PM, RAL HelpDesk {for John Halley Gotway}
wrote:
>
>> Shujia,
>>
>> I see that you're not currently a registered user of the MET
software. Prior to providing direct support in the use of MET, we
require that users fill out the free online registration form. That
>> enables us to ensure that we remain in compliance with the
international export laws of the United States.
>>
>> To fill out the registration, just go here:
>> http://www.dtcenter.org/met/users/downloads/index.php
>> Enter you email address and hit enter to be taken to the form.
>>
>> Please write back to let me know when you've registered.
>
>
> Done.
>
>
>> Also please send me the Point-Stat output file that you're running
through STAT-Analysis and the STAT-Analysis command line you're
running
>> that's returning no data.
>
>
> point_stat_15000000L_20080623_120000V.stat
>
> is in the next email.
>
> Thanks,
>
> Shujia
>>
>> Thanks,
>> John Halley Gotway
>> met_help at ucar.edu
>>
>>
>>
>> On 05/26/2011 10:06 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>>> /discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-
3.2.1/MET/bin/stat_analysis -lookin
./point_stat/point_stat_15000000L_20080623_120000V.stat -out
test_stat_output -job summary -line_type cnt -fcst_var RH -dump_row
temp_cnt_out -column RMSE
>>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47148] question on the output of stat_analysis
From: Shujia Zhou
Time: Thu May 26 14:30:24 2011
John:
Thank for your prompt. Yes, I see it.
Can you tell why we can get min, max, mean with the same value after
using stat_analysis?
Shujia
On May 26, 2011, at 1:40 PM, RAL HelpDesk {for John Halley Gotway}
wrote:
> Shujia,
>
> Please take a look in the FCST_VAR column of the data file you just
sent me. You'll see that it only contains entries for "TT" and "RH".
When you run your STAT-Analysis job using the "-fcst_var RH"
> option, you'll match those lines with RH in the FCST_VAR column, and
likewise for TT. If you run with "-fcst_var TMP", you'll match 0 STAT
lines because none of them have TMP in the FCST_VAR column.
>
> Hope that helps.
>
> John
>
> On 05/26/2011 11:30 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47148 >
>>
>> Hi, John:
>>
>>
>> On May 26, 2011, at 12:52 PM, RAL HelpDesk {for John Halley Gotway}
wrote:
>>
>>> Shujia,
>>>
>>> I see that you're not currently a registered user of the MET
software. Prior to providing direct support in the use of MET, we
require that users fill out the free online registration form. That
>>> enables us to ensure that we remain in compliance with the
international export laws of the United States.
>>>
>>> To fill out the registration, just go here:
>>> http://www.dtcenter.org/met/users/downloads/index.php
>>> Enter you email address and hit enter to be taken to the form.
>>>
>>> Please write back to let me know when you've registered.
>>
>>
>> Done.
>>
>>
>>> Also please send me the Point-Stat output file that you're running
through STAT-Analysis and the STAT-Analysis command line you're
running
>>> that's returning no data.
>>
>>
>> point_stat_15000000L_20080623_120000V.stat
>>
>> is in the next email.
>>
>> Thanks,
>>
>> Shujia
>>>
>>> Thanks,
>>> John Halley Gotway
>>> met_help at ucar.edu
>>>
>>>
>>>
>>> On 05/26/2011 10:06 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>>>> /discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-
3.2.1/MET/bin/stat_analysis -lookin
./point_stat/point_stat_15000000L_20080623_120000V.stat -out
test_stat_output -job summary -line_type cnt -fcst_var RH -dump_row
temp_cnt_out -column RMSE
>>>
>>
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #47148] question on the output of stat_analysis
From: John Halley Gotway
Time: Thu May 26 15:55:21 2011
Shujia,
You're running a STAT-Analysis "summary" type job, telling it to look
at the continuous statistics (CNT) line type. The "summary" job in
STAT-Analysis is very simplistic. It just looks through the
stat files you pass it and applies the filtering criteria you
specified. In your case, it's looking for CNT lines where the
FCST_VAR column is RH. And since you set "-column RMSE", it's just
pulling
out the RMSE value from that line. From that distribution of RMSE
values it finds, it computes summary information like min, max, mean,
median, and standard deviation.
Since you only passed it a single STAT file, it's only finding 1 CNT
line and extracting a single RMSE value from that line. So it's
taking the min, max, mean, and so forth, of a single value.
That's why those numbers are the same.
One of the output columns for the summary job is "TOTAL" which tells
you the number of values that we're used to compute the summary
information. In your example, total is set to 1.
So I would not recommend running the summary job in the way you're
doing it. Instead, it'd be more appropriate to run it across a
season's worth of data, for example, where you have many, many RMSE
values to be summarized.
When you use the "-lookin" option for STAT-Analysis and pass it a
directory name instead of a STAT file, it'll search recursively
through that directory looking for STAT files to process. That should
make running a STAT-Analysis job over a large set of data pretty easy.
All you need to do is pass in the top-level directory and specify the
filtering criteria you'd like STAT-Analysis to apply when
running its job.
Hopefully that helps clarify.
Thanks,
John
On 05/26/2011 02:30 PM, RAL HelpDesk {for Shujia Zhou} wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47148 >
>
> John:
>
> Thank for your prompt. Yes, I see it.
>
> Can you tell why we can get min, max, mean with the same value after
using stat_analysis?
>
>
> Shujia
> On May 26, 2011, at 1:40 PM, RAL HelpDesk {for John Halley Gotway}
wrote:
>
>> Shujia,
>>
>> Please take a look in the FCST_VAR column of the data file you just
sent me. You'll see that it only contains entries for "TT" and "RH".
When you run your STAT-Analysis job using the "-fcst_var RH"
>> option, you'll match those lines with RH in the FCST_VAR column,
and likewise for TT. If you run with "-fcst_var TMP", you'll match 0
STAT lines because none of them have TMP in the FCST_VAR column.
>>
>> Hope that helps.
>>
>> John
>>
>> On 05/26/2011 11:30 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>>>
>>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=47148 >
>>>
>>> Hi, John:
>>>
>>>
>>> On May 26, 2011, at 12:52 PM, RAL HelpDesk {for John Halley
Gotway} wrote:
>>>
>>>> Shujia,
>>>>
>>>> I see that you're not currently a registered user of the MET
software. Prior to providing direct support in the use of MET, we
require that users fill out the free online registration form. That
>>>> enables us to ensure that we remain in compliance with the
international export laws of the United States.
>>>>
>>>> To fill out the registration, just go here:
>>>> http://www.dtcenter.org/met/users/downloads/index.php
>>>> Enter you email address and hit enter to be taken to the form.
>>>>
>>>> Please write back to let me know when you've registered.
>>>
>>>
>>> Done.
>>>
>>>
>>>> Also please send me the Point-Stat output file that you're
running through STAT-Analysis and the STAT-Analysis command line
you're running
>>>> that's returning no data.
>>>
>>>
>>> point_stat_15000000L_20080623_120000V.stat
>>>
>>> is in the next email.
>>>
>>> Thanks,
>>>
>>> Shujia
>>>>
>>>> Thanks,
>>>> John Halley Gotway
>>>> met_help at ucar.edu
>>>>
>>>>
>>>>
>>>> On 05/26/2011 10:06 AM, RAL HelpDesk {for Shujia Zhou} wrote:
>>>>> /discover/nobackup/szhou/nu-wrf-trunk/nu-wrf_v2beta2-
3.2.1/MET/bin/stat_analysis -lookin
./point_stat/point_stat_15000000L_20080623_120000V.stat -out
test_stat_output -job summary -line_type cnt -fcst_var RH -dump_row
temp_cnt_out -column RMSE
>>>>
>>>
>>
>
------------------------------------------------
More information about the Met_help
mailing list