[Met_help] Output format MET Tools

[John Halley Gotway]

Thomas,

OK, I understand.  Unfortunately, we can't ensure alignment across files in a really general way.  However, MET should be doing a good job lining up data nicely into columns within any single file.

One reason why we can't choose fixed column widths is the "MODEL" column.  The user chooses a string to describe the model name so as to differentiate between models they're evaluating.  Since the
user can choose any string for that, we can't anticipate what column width should be used.

I tried running a "filter" job through the STAT-Analysis tool hoping that it'd line up the STAT data nicely into columns.  However, it didn't.  And I understand why.  If you ran that type of filter
job over a whole lot of STAT data, lining it up nicely would use up a lot of memory.

I looked through the code that writes that STAT data and there is no easy way of hard coded the column widths to be used.

One solution would be to add a new job to the STAT-Analysis tool, name is something like "align", that just reads the input STAT data and lines it up like you'd need.  Does that sound like it'd meet
your needs?  You'd need to run that STAT-Analysis job on the STAT output before reading it with your FORTRAN code.

John

Thomas Schwitalla wrote:
> John,
> 
> I attached two example files for wind verification. My point can be seen
> when looking at the positions of "VOBAR" and the corresponding values in
> both files. They are shifted by one position in this case.
> 
> Thomas
> 
> 
> John Halley Gotway schrieb:
>> Thomas,
>>
>> I'm not entirely clear on what you're asking.  I can say that it's
>> pretty easy to control the precision of the numbers in the output -
>> i.e. the number of decimal places used.  However, I can't think
>> of an easy way to output exactly the same number of digits every time.
>>
>> Could you send me a sample MET output file, and explain which numbers
>> would need to change to get it in the form you'd like?  That'd help me
>> understand the issue better.
>>
>> Thanks,
>> John
>>
>> Thomas Schwitalla wrote:
>>  
>>> Hi John,
>>>
>>> is it possible, to output variables always in the same format with the
>>> same number of digits (like F12.5 in Fortran)?
>>> Otherwise it may be difficult to read out the variables e.g. with
>>> FORTRAN as I am doing it...
>>> I saw this possible problem in the vl1l2 files for windspeed
>>> verification.
>>>
>>> Can you give me a quick solution for that?
>>>
>>> Thomas
>>>
>>>     
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help