[Met_help] Re: mode_analysis scripting question

[John Halley Gotway]

Bill,

Good news.  You can use the -lookin path to specify a MODE output filename ***OR*** directory that contains MODE output somewhere down in it.

This is from the MODE-Analysis usage statement:
"-lookin path" specifies a MODE file or a top-level directory containing MODE files to be used (at least one required).

So all you'll need to do is pass in the top-level directory that contains all of your MODE output somewhere down beneath it.  MODE-Analysis should recursively search all of the subdirectories and read 
  all of the files ending in "_obj.txt" (that's the output file naming convention for the MODE files).  So MODE-Analysis will read all of the data and perform the job you specify.

However, let me caution you that you'll need to make sure your filtering parameters are set up properly.  Since you'll be reading ALL of the output data at once, you'll probably want to specify the 
model name for the job using the "-model" command line option - I assume that you'll want to only look at objects from one model at a time.  Think about how else you'd like to filter the data.

Sorry that wasn't more clear about how "-lookin" path works.

If you find that MODE-Analysis is running too slow because it's busy reading all of that data in and then throwing most of it away anyway, you could try using the MODE-Analysis configuration file 
instead.  In the configuration file you'd define many jobs to be performed for that data set rather than just one, like on the command line.  And it would make it so that the data is only read in once 
- which would make it faster.

Let me know if you have any questions.  And do be sure to keep using the "-dump_row" option and keep verifying that it's doing what you expect it to be doing.

Thanks,
John

wgallus at iastate.edu wrote:
> John,
> 
> I'm figuring out some stuff about mode_analysis and think I see a way to 
> do stuff I need (like using -summary to determine the average AREA from 
> all forecasted objects in the entire dataset).
> 
> But.... I'm runinng into a problem with creating the script to make this 
> easy.  I was following scripts I had used with mode.  Basically, I need 
> to have mode_analysis do the -lookin for all of my many cases and I 
> didn't want to have to list all 72 filenames in the script.  I tried to 
> use the following but it didn't like the syntax...
> 
> #! /bin/csh
> set cc = 1
> foreach date ( `cat dates2.txt` )
> echo ${date}
> /d1/johnhg/MET_releases/METv1.1beta5/bin/mode_analysis \
>    -lookin ${date} \
> @ cc = $cc + 1
> end
>    -summary -fcst -single -simple -matched \
>    -column AREA \
>    -dump_row area_summary_bycase.txt \
>    -out      area_summary_bycase.out
> 
> 
> 
> It seems to me one problem is that I have to get all filenames to show 
> up within the mode_analysis command/option lines, and it seems the 
> foreach loop and extra stuff needed messes up mode_analysis.  Any ideas 
> of how to do this without having to write the lookin line 72 times with 
> the 72 different filenames?
> 
> Bill
> 
> ------------------------------------------------
> |   William A. Gallus, Jr.                       |
> |   Professor of Meteorology                     |
> |   Professor-in-charge, Meteorology Program     |
> |   3025 Agronomy                                |
> |   Iowa State University                        |
> |   515-294-2270                                 |
>  ------------------------------------------------