[Met_help] pcp_combine

John Halley Gotway johnhg at ucar.edu
Fri Apr 9 15:47:15 MDT 2010


I ran the files you sent and got the following error:

pcp_combine -subtract test_d02_19900930.000 21 test_d02_19900929.000 0 test.nc
Reading input file: test_d02_19900930.000
ERROR: get_field() -> can't find grib code 61 with accumulation of 21 hours in grib file: test_d02_19900930.000

That file doesn't contain a 21-hour accumulation of precipitation.  It contains a 0-hour accumulation.  So then I ran the following command instead:

pcp_combine -subtract test_d02_19900930.000 0 test_d02_19900929.000 0 test.nc
Reading input file: test_d02_19900930.000
Reading input file: test_d02_19900929.000
Performing subtraction command.
Writing output file: test.nc

And it worked fine.  This of course is a very silly thing to do - all we're doing is subtracting 0 minus 0.  But it does run without the NetCDF error you mentioned:
"NetCDF: Numeric conversion not representable"

If the test script runs fine, but this data doesn't work, I'm really not sure what's going on.  I haven't seen that type of NetCDF error before.  You command line looks fine.  I suppose you could try
the "-add" option instead of "-subtract" but I'm guessing you'll see the same error.  I've run it here using the original and updated versions of METv2.0 and using the GNU and PGI compilers, but I
don't see the error you're seeing.  I'm not sure how else to debug this issue.


Jianting.Chu at uni-hohenheim.de wrote:
> Dear John,
>     Thank you very much!
>     I changed back to PGI10.0 compiler, and tested./test_pcp_combine.sh,
> it has no problem. I did not patch at present.
>     If I use:
>      ./pcp_combine -subtract test_d02_19900930.000 21
> test_d02_19900929.000 0 test.nc
>     It still does not work, even can not open the file. (for d02, the
> total precipitation is 3-hr accumulated)
>     Is my commend correct? I have upload these two files again to your
> ftp, would you please testing it again (because I use PGIv10 compiler,
> instend of gfortran this time)? What is your expression in commend line?
> We still do not need type "0" for the second file?
>     In addition, these two files are de-rotated by CDO and some warning
> message occured during this process,I am not very clear if they are
> totally correct for MET, would you please checking them?
>     Thank you very much!
> Bests,
> Jianting
> 引用 John Halley Gotway <johnhg at ucar.edu>:
>> Jianting,
>> Thanks for sending the sample data.  I retrieved it from the FTP site
>> and ran it through pcp_combine here.  I ran it through pcp_combine in
>> two ways: once using the pcp_combine code as originally
>> released, and once using the pcp_combine code after applying all the
>> latest patches posted on the MET website
>> (http://www.dtcenter.org/met/users/support/known_issues/METv2.0/index.php).
>> In both cases, pcp_combine ran fine and created an output file - no
>> NetCDF errors.  I've attached the resulting NetCDF file.  However, I
>> would like to point out that you really don't need to run
>> pcp_combine to subract 24-hour accumulation minus 0-hour
>> accumulation.  The 0-hour accumulation should contain zero precip
>> values anyway.  So you could just use the 0-24 hour GRIB file in the
>> other
>> MET tools directly.
>> Since you're seeing this NetCDF problem when you run PCP-Combine, I'm
>> wondering if you could take a step back and try running the test script:
>>   cd METv2.0/scripts
>>   ./test_pcp_combine.sh >& test_pcp_combine.log
>> That'll run the PCP-Combine test script and dump its output to a log
>> file.  I'm wondering if you'll see NetCDF errors there as well.
>> Thanks,
>> John
>> Jianting Chu wrote:
>>> Dear Sir,
>>>     I encountered a trouble with pcp_combine.
>>>     With the wrfout files, the precpitation is accumulated every hour
>>> (history_interval=60), so I wanna use pcp_combine to get daily
>>> precipitation.
>>>     If I use the command:
>>>             > ./pcp_combine -subtract test_WRFPRS_d01_19890102.000 24
>>> test_WRFPRS_d01_19890101.000 0 test_19890101.nc -v 1000
>>>     it will say:
>>>             NetCDF: Numeric conversion not representable
>>>       So, it seems that the problem is caused by NetCDF, but I am not
>>> sure.
>>>     I have uploaded test_WRFPRS_d01_19890101.000.gz,
>>> test_WRFPRS_d01_19890102.000.gz and pcp_combine to your ftp:
>>> /incoming/irap/wrfhelp. Would you please checking it?
>>>     I need to explain something before:
>>>         1) we use rotated lat-lon projection for our WRFv3.1, and use
>>> the lasted WPP to transform its format to GRIB;
>>>         2) Because MET does not recognize the rotated data, we use CDO
>>> to do the de-rotation (copygb is not possible for linux currently). I
>>> have used Xconv and wgrib to check the de-rotated results, it seems
>>> reasonable;
>>>         3) pcp_combine is produced by gfortran compiler.
>>>     What is the problem?
>>> Best wishes,
>>> Jianting
>>> _______________________________________________
>>> 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