[Met_help] Re: Data interface about met program?

John Halley Gotway johnhg at rap.ucar.edu
Tue Feb 26 08:58:13 MST 2008


The default version of the PGI compilers I was using was version 5.2-4.  However, I do see that we have PGI version 6.1-6 available on our machines.  So I'll try to build MET using that and let you
know if I'm able to reproduce the error you're seeing.

Thanks,
John

Coffee.Z wrote:
> Dear John:
> 
>         I have replaced the file, but it is still not working. I have tried two version of PGI.
>            pgCC 7.0-7 32-bit target on x86 Linux
>            pgCC 6.1-4 32-bit target on x86 Linux
>         All of them reported the same error as before.
>         
> 
> User: Coffee.Z
> 
> 
> 
> 
> ·¢¼þÈË£º John Halley Gotway
> ·¢ËÍʱ¼ä£º 2008-02-26 00:53:28
> ÊÕ¼þÈË£º Coffee.Z; met_help
> ³­ËÍ£º 
> Ö÷Ì⣺ Re: Data interface about met program?
> 
> That's interesting.  That's the first time I've seen that error message.
> 
> Can you tell me what version of the PGI compilers you're using? Just type "pgCC --version" (that's with two '-' in front of version) to get the version number.
> 
> I really don't know why the compiler would have a problem with that line of code.  But let's try the following.  I'd like you to do 2 things:
> (1) First, replace the file METv1.0/lib/vx_met_util/apply_mask.cc with the version of that file attached to this message.
> (2) Second, replace the file METv1.0/lib/vx_met_util/constants.h with the version of that file attached to this message.
> 
> Then try rebuilding MET, and please let me know if that get's you past this error message.  If so, we'll need to edit a 3rd file in a similar way to get the "vx_wrfmode" library to build.
> 
> Thanks and please do let me know what compiler version you're using.
> 
> John
> 
> 
> 
> Coffee.Z wrote:
>> Dear John:
>>
>>      I'm back to school now. I have tried the pgi compiler, and it is still failed. This time it seems syntax error, but i cannot figure out why. Here is the log. I have already updated all the file according to your "know issues".
>>      Thanks!  
>>
>> user: CoffeeZ
>>
>>
>>
>>
>> ÄúºÃ£¡
>>
>>
>>
>>
>>     ÖÂ
>> Àñ£¡
>>
>>
>> СÖì
>> 2008-02-24
>>
>>
>>
>> ·¢¼þÈË£º John Halley Gotway
>> ·¢ËÍʱ¼ä£º 2008-01-26 00:23:10
>> ÊÕ¼þÈË£º Coffee.Z
>> ³­ËÍ£º met_help
>> Ö÷Ì⣺ Re: Data interface about met program?
>>
>> Hello,
>>
>> Based on the files you sent, I see that the error you're encountering is due to NetCDF.  The linker isn't finding the NetCDF library archive files it's looking for.
>>
>> I'm hoping that if you make a simple fix in the Makefile and recompile, it'll get past this error.
>>
>> In the Makefile, look at how the variable NETCDF_LIBS is defined.  List out those directories:
>> ls /home/users/zhucoffee/wrfnew/netcdf/libs
>> ls /home/users/zhucoffee/wrfnew/netcdf/include
>>
>> Do they contain the files libnetcdf.a and libnetcdf_c++.a?
>>
>> If not, locate those two files in the NetCDF directory.  You can try something like:
>> cd /home/users/zhucoffee/wrfnew/netcdf
>> find ./ -name libnetcdf.a
>> find ./ -name libnetcdf_c++.a
>>
>> Then reset NETCDF_LIBS to the directory (or directories) that contain them.  And then rebuild MET with something like:
>> make clean
>> make   >& make_met.log&
>>
>> Please let me know if that did the trick.
>>
>> Thanks,
>> John Halley Gotway
>>
>> So I'd suggest checking whether the version of NetCDF in /home/users/zhucoffee/wrfnew/netcdf was compiled using the PGI or the GNU compilers.
>>
>> Coffee.Z wrote:
>>> Dear John:
>>>
>>>
>>>         Here is the log. Although i have changed the PGI compiler to the 32 bit and all other external lib to 32 bit. It is still failed. 
>>>   
>>>
>>>
>>> User: CoffeeZ
>>>
>>>
>>>
>>>
>>> ÄúºÃ£¡
>>>
>>>
>>>
>>>
>>>     ÖÂ
>>> Àñ£¡
>>>
>>>
>>> СÖì
>>> 2008-01-25
>>>
>>>
>>>
>>> ·¢¼þÈË£º John Halley Gotway
>>> ·¢ËÍʱ¼ä£º 2008-01-25 12:44:42
>>> ÊÕ¼þÈË£º Coffee.Z
>>> ³­ËÍ£º met_help at ucar.edu
>>> Ö÷Ì⣺ Re: Data interface about met program?
>>>
>>>
>>> Sorry to hear about the PGI compilers.  If you'd like to send the Makefile
>>> and compilation log file, I could take a look and see if there's anything
>>> obviously wrong.
>>>
>>> What I mean is, when you build MET using a command like...
>>> make -f Makefile_pgi    >& make_met.log
>>> ... please send me a copy of the "Makefile_pgi" that you're using and a
>>> copy of the "make_met.log" logfile.
>>>
>>> Regarding the issue of using point observations in ASCII format, several
>>> users are encountering the same obstacle.  For the next release of MET,
>>> I'd like to add a utility that will convert formatted ASCII point
>>> observations into the NetCDF format that MET is expecting.
>>>
>>> However, that release won't be ready for a couple of months I'd say.  In
>>> the meantime, I need to write up some instructions describing what the
>>> NetCDF point observations should look like for those users who would like
>>> to reformat their observations into NetCDF themselves.
>>>
>>> So I'll work on those basic instructions tomorrow, post that info to the
>>> MET website, and let you know when it's available for you to take a look
>>> at.
>>>
>>> Thanks,
>>> John
>>>
>>>
>>>
>>>
>>>> Dear John:
>>>>
>>>>      Thanks very much for your kind guidance! I have try the PGI compiler,
>>>> but still failed. I will try it again when i back to my school.
>>>>
>>>>      Right now, I have some question about the met data interface. For in
>>>> my country, most of the observational data are stored in ASCII
>>>> format. That mean i can't use the 'pb2nc' util. So I'm thinking to
>>>> transfer my observational data  to  the netcdf format, instead of to
>>>> the prebufr format.  My question is do i need to pay special
>>>> attention when i write the netcdf code?
>>>>
>>>>      Another question is about the gridded input of the observational
>>>> data. I'm wondering if i use the data like 'ncep fnl global 1 degree
>>>> * 1 degree', do i need to intepolate the data into the domain just as
>>>> the wrf output?
>>>>
>>>>      Thanks!
>>>>
>>>> users: CoffeeZ
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> ÄúºÃ£¡
>>>>
>>>>
>>>>
>>>>
>>>>     ÖÂ
>>>> Àñ£¡
>>>>
>>>>
>>>> СÖì
>>>> 2008-01-25
>>>>
>>>>
>>>>
>>>> ·¢¼þÈË£º John Halley Gotway
>>>> ·¢ËÍʱ¼ä£º 2008-01-23 09:35:59
>>>> ÊÕ¼þÈË£º Coffee.Z
>>>> ³­ËÍ£º met_help
>>>> Ö÷Ì⣺ Re: [Met_help] Error when running met with the test_all.csh?
>>>>
>>>> In regards to compiling MET using the PGI compilers, I wanted to point
>>>> you
>>>> to a couple of known issues posted on the MET website.
>>>>
>>>> The "Makefile_pgi" file needs to be modified prior to compiling with the
>>>> PGI compilers, especially version 7 of the PGI compilers:
>>>> http://www.dtcenter.org/met/users/support/known_issues/METv1.0/METv1.0_known_issues.php
>>>>
>>>> So if you'd like to try using the PGI compilers again please refer to the
>>>> fixes posted.
>>>>
>>>> Thanks,
>>>> John Halley Gotway
>>>>
>>>>> Hello,
>>>>>
>>>>> I'm guessing that the problems you're encountering are related to
>>>>> running
>>>>> MET on a 64-bit machine.  Currently, BUFRLIB a library on which MET
>>>>> depends, doesn't work well on a 64-bit machine.  That's
>>>>> the reason for the error message you're seeing from PB2NC:
>>>>>
>>>>>  **************BUFR ARCHIVE LIBRARY ABORT*****************
>>>>>  BUFRLIB: OPENBF - FIRST 4 BYTES READ FROM RECORD IN INPUT FILE
>>>>> CONNECTED
>>>>> TO UNIT  11 NOT 'BUFR', DOES NOT CONTAIN BUFR DATA
>>>>>  **************BUFR ARCHIVE LIBRARY ABORT*****************
>>>>>
>>>>> However, for the other error messages you're seeing about the NetCDF
>>>>> file
>>>>> containing illegal characters... that may also be related to running on
>>>>> a
>>>>> 64-bit machine.
>>>>>
>>>>> I'm in the process of trying to get MET to run on a 64-bit machine we
>>>>> have
>>>>> here at NCAR.  What I'm trying to do is compile MET (and the libraries
>>>>> on
>>>>> which it depends) using the GNU -m32 compiler flag
>>>>> which should build 32-bit code.  But I'm still working on it.
>>>>>
>>>>> In the top-level MET directory, edit the file "Makefile".
>>>>> Replace the line:
>>>>> FLAGS           = -g -Wall -Wshadow -static
>>>>> with
>>>>> FLAGS           = -g -Wall -Wshadow -static -m32
>>>>>
>>>>> But the trick is to get the libraries on which MET depends to compile
>>>>> as
>>>>> 32-bit as well.
>>>>>
>>>>> Feel free to give it a shot yourself.  I'll let you know when I'm able
>>>>> to
>>>>> compile it on a 64-bit machine myself.  If you have access to a 32-bit
>>>>> machine, I'd suggest trying it over there as well.
>>>>>
>>>>> Thanks,
>>>>> John Halley Gotway
>>>>>
>>>>> Coffee.Z wrote:
>>>>>    > Hello,met_help:
>>>>>    >
>>>>>    >     i have been trying to run the met program with the sample data,
>>>>> but
>>>>>    > encountered the following problem, for more detail, you can see
>>>>>    > further in test.log:
>>>>>    >
>>>>>    > *** Running GRID_STAT using netCDF input for both forecast and
>>>>>    > observation ***
>>>>>    > Forecast File: ../out/pcp_combine/sample_fcst_12L_2005080712V_12A.nc
>>>>>    > Observation File: ../out/pcp_combine/sample_obs_2005080712V_12A.nc
>>>>>    > Configuration File: config/GridStatConfig_APCP_12
>>>>>    > NetCDF Precipitation (min, max) = (0, 38.8841)
>>>>>    > NetCDF Precipitation (min, max) = (0, 104.1)
>>>>>    > NetCDF: Name contains illegal characters
>>>>>    > ....
>>>>>    > ERROR: process_obs_file() -    > can't open observation netCDF file:
>>>>>    > ../out/pb2nc/sample_pb.nc
>>>>>    >
>>>>>    >    I have been tried to build the source code with pgf77, but failed.
>>>>>    > The above is the result build with g++ and g77. Since the netcdf is
>>>>>    > build with the pgf90 and the system is 64 bit, and i'm wondering if i
>>>>>    > need to build the source code with pgi.
>>>>>    >
>>>>>    >     Thanks!
>>>>>    >
>>>>>    >  user: CoffeeZ
>>>>>    >
>>>>>    >
>>>>>    >
>>>>>    >
>>>>>    > ÄúºÃ£¡
>>>>>    >
>>>>>    >
>>>>>    >
>>>>>    >
>>>>>    >     ÖÂ
>>>>>    > Àñ£¡
>>>>>    >
>>>>>    >
>>>>>    > СÖì
>>>>>    > 2008-01-23
>>>>>    >
>>>>>    >
>>>>>    > ------------------------------------------------------------------------
>>>>>    >
>>>>>    > _______________________________________________
>>>>>    > Met_help mailing list
>>>>>    > Met_help at mailman.ucar.edu
>>>>>    > http://mailman.ucar.edu/mailman/listinfo/met_help
>>>>> _______________________________________________
>>>>> Met_help mailing list
>>>>> Met_help at mailman.ucar.edu
>>>>> http://mailman.ucar.edu/mailman/listinfo/met_help
>>>>>
>>> .


More information about the Met_help mailing list