[Met_help] building MET on Ubuntu 9.10
John Halley Gotway
johnhg at ucar.edu
Tue Feb 16 09:44:20 MST 2010
Joe,
Sorry for the delay in getting back to you. I was out of the office last week.
When I run STAT-Analysis with the data you sent, I actually see a different error:
WARNING: do_job_aggr_stat() -> no matching STAT lines found for job
This message states that it can't find any STAT lines that match for the job you're running.
I think I know what's going on here. In the STAT output file, you have 104 matched pairs for the 10-meter U-component of wind (UGRD/L10), the 10-meter V-component of wind (VGRD/L10), and 2-meter
temperature (TMP/Z2). It appears, you'd like to aggregate those matched pairs and recompute continuous statistics. However, you're trying to do so with a single job. Instead, you should be using 3
different jobs, 1 for UGRD, 1 for VGRD, and 1 for TMP.
Try running the following three commands:
./stat_analysis -lookin point_stat_wrf_mynn_480000L_20020503_000000V.stat -job aggregate_stat -line_type MPR -out_line_type CNT -vx_mask FULL -fcst_var UGRD -dump_row UGRD_lines.stat -v 2
./stat_analysis -lookin point_stat_wrf_mynn_480000L_20020503_000000V.stat -job aggregate_stat -line_type MPR -out_line_type CNT -vx_mask FULL -fcst_var VGRD -dump_row VGRD_lines.stat -v 2
./stat_analysis -lookin point_stat_wrf_mynn_480000L_20020503_000000V.stat -job aggregate_stat -line_type MPR -out_line_type CNT -vx_mask FULL -fcst_var TMP -dump_row TMP_lines.stat -v 2
When you run an "aggregate_stat" job, you tell it an input line type (-line_type) and an output line type (-out_line_type). You should also specify the forecast variable to be used for each job
(-fcst_var), because it wouldn't make sense to combine UGRD matched pairs with TMP matched pairs. Also notice that I've dumped out the lines used for these jobs by using the -dump_row option. When
first setting up a STAT-Analysis job, we strongly recommend the use of the -dump_row option. After the job runs, take a look at the lines that were used for the job and verify that they're the ones
you intended. Once you're confident that your job is doing what you expect, you don't need to use -dump_row anymore.
Also note that if you had used multiple interpolation methods, you should pick only one to use for each job. But it looks like you've only used one in your Point-Stat output.
Once you get the jobs working on the command line the way you want, you could move them to a configuration file.
Hope that helps. If you have any more questions, just let me know.
Thanks,
John
Joe Eastman wrote:
> Hi John,
>
> OK, I am doing something silly again. I created all the stat files using the allpoints.sh which points to pointstat. When I try to run an aggregate_stat I get the following message. I am probably going about this all in the wrong manner so I will describe what I would like to do. We typically run the model for 2 to 4 weeks (I am also developing a short term forecast system that will be analyzed similarly). I would like to tabulate linear statistics (correlation MAE ME BIAS etc) at hourly intervals using all stations in my domain. I would also like to see the results over the entire integration, and also for discrete time intervals such as day 1, day 2, etc. and perhaps over 3-7 day periods.
>
> Thanks for your time.
>
> Take care,
> Joe
>
>
> [06:20 PM grsci104]$stat_analysis -lookin ./ -config statanalysis
>
>
> ERROR: read_mpr_lines() -> the forecast variable type or observation variable type should remain the same!
>
>
>
> ERROR: main() -> encountered an error value of 1. Calling clean_up() and usage() before exiting.
>
>
> *** Model Evaluation Tools (METV2.0) ***
>
> Usage: stat_analysis
> -lookin path
> [-out filename]
> [-v level]
> [-config config_file] | [JOB COMMAND LINE]
>
> where "-lookin path" specifies a STAT file or top-level directory containing STAT files. It allows the use of wildcards (at least one required).
> "-out filename" specifies a file name to which output should written rather than the screen (optional).
> "-v level" overrides the default level of logging (1) (optional).
> "-config config_file" specifies a STATAnalysis config file containing STATAnalysis jobs to be run.
> "JOB COMMAND LINE" specifies all the arguments necessary to perform a single STATAnalysis job.
>
> NOTE: Refer to a STATAnalysis config file for details on how to specify the JOB COMMAND LINE.
>
>
> Take care,
> Joe
>
> Dr. Joseph L. Eastman
> Senior Atmospheric Scientist
> WindLogics Inc.
> 201 4th St NW
> Grand Rapids, MN
> 55744
>
> c - 410.279.9702
> w - 651.556.4297
> jeastman at windlogics.com
>
>
> -----Original Message-----
> From: met_help-bounces at mailman.ucar.edu [mailto:met_help-bounces at mailman.ucar.edu] On Behalf Of John Halley Gotway
> Sent: Monday, February 08, 2010 4:32 PM
> To: Cory Davis
> Cc: met_help
> Subject: Re: [Met_help] building MET on Ubuntu 9.10
>
> Cory,
>
> Thanks for letting me know. We're running version 4.3.2 of the GNU compilers here. So that's why we haven't seen/fixed the problems you found.
>
> If we're able to get a hold of a newer version of the compilers, we'll try to get these issues ironed out before the next release of MET in the spring of 2010.
>
> Thanks,
> John
>
> Cory Davis wrote:
>> gcc version 4.4.1 (Ubuntu 4.4.1-4ubuntu9)
>>
>> John Halley Gotway wrote:
>>> Cory,
>>>
>>> Thanks for sending these details. I'm sorry to hear you had such a
>>> difficult time building MET on Ubuntu. Can you please tell me the
>>> names and version numbers of the compilers you were using?
>>>
>>> Thanks,
>>> John Halley-Gotway
>>> johnhg at ucar.edu
>>>
>>> Cory Davis wrote:
>>>> Hi there,
>>>> I have had to make quite a few changes in order to compile MET v2.0
>>>> with all patches METv2.0_patches_20091223.tar.gz on my Ubuntu 9.10
>>>> machines.
>>>>
>>>> These changes include adding
>>>> #include <cstdio>
>>>> to
>>>> ./lib/vx_cal/time_strings.cc
>>>> ./lib/vx_econfig/machine.cc
>>>> ./lib/vx_met_util/stat_hdr_columns.cc
>>>> ./lib/vx_util/ascii_table.cc
>>>> ./lib/vx_util/concat_string.cc
>>>> ./lib/vx_util/string_fxns.cc
>>>> ./lib/vx_util/threshold.cc
>>>>
>>>> this is in order to have sprintf, fopen and fclose defined.
>>>>
>>>> Also, the use of the strrchr function in
>>>> ./lib/vx_met_util/file_type.cc, and
>>>> ./lib/vx_met_util/stat_columns.cc
>>>> is not quite right. In C++ you are apprently allowed
>>>>
>>>> const char * strrchr ( const char * str, int character );
>>>> char * strrchr ( char * str, int character );
>>>>
>>>> and NOT
>>>> char * strrchr ( const char * str, int character );
>>>>
>>>> so to get around this I have done this
>>>> ./lib/vx_met_util/file_type.cc:43
>>>> if((ptr = (char *) strrchr(filename, '.')) != NULL) {
>>>>
>>>> and
>>>> ./lib/vx_met_util/stat_columns.cc:
>>>>
>>>> ./lib/vx_met_util/file_type.cc
>>>> line 44: if((ptr = (char *) strrchr(filename, '.')) != NULL) {
>>>> const char * strrchr ( const char * str, int character );
>>>> char * strrchr ( char * str, int character );
>>>>
>>>> ./lib/vx_met_util/stat_columns.cc:411
>>>> if((ptr = (char *) strrchr(col_name, '_')) != NULL) i = atoi(++ptr);
>>>>
>>>> Cheers,
>>>> Cory
>>>>
>>>>
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help
>
> CONFIDENTIALITY NOTICE: This e-mail message is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution of any kind is strictly prohibited. If you are not the intended recipient, please contact the sender via reply e-mail and destroy all copies of the original message. Thank you.
More information about the Met_help
mailing list