Date: Fri, 23 Apr 1999 17:06:48 -0600 (GMT)

>After I choose CCM
>radiation for MPP MM5, I'd like to output some variables in
>relation to radiation, but I don't know how to modify the
>code (should modify radout.F), is there any example you can
>give me how to add output variables? Thanks a lot.

How you do this will depend on what variables you want to see and how
you want to see them.  One complexity is that radiation is called
within a j-loop and so that from within the radiation package you can
only ever output data for one j slice at a time.  Furthermore, the data
is, of course, decomposed so you can't just WRITE or PRINT it and have
it come out in a sensible order.    There are a couple of ways I handle
this.

If I the values for a single point or column will suffice, I put things
like this in the code:

      if ( i .eq. IDEBUG .and. j .eq. JDEBUG .and. k .eq. KDEBUG ) then
        write(0,*)' at someplace ',i-idif_x0,j-jdif_x0, k
        write(0,*)'  VAR(I,J,K) ', VAR(I,J,K)
         ...
      endif

Assume that IDEBUG, JDEBUG, and KDEBUG are set to coordinates of the
point you're interested in.

You have to be careful about what loops you're in.  The above would be
put inside a triple nested loop (the j loop is outside the routine).
If you were in the routine but outside any i or k loop, you would do something
like this instead:

      if ( j .eq. JDEBUG ) then
	i = IDEBUG
	j = JDEBUG
        write(0,*)' at someplace ',i-idif_x0,j-jdif_x0, k
        write(0,*)'  VAR(I,J,K) ', VAR(I,J,K)
         ...
      endif

Note that you would use the name of j that's passed into the routine
(not necessarily j; might be jslc or something).  The i-idif_x0
converts the local index i to it's global value.  Note that FLIC will
handle most of the rest of this correctly -- if you're in doubt,
though, and are getting weird results that don't make sense, have a
look at the FLIC-expanded code in the little-f file that's generated in
MPP/build .  Note also that to see these, you will need to uncomment
the line #RM=ECHO in MPP/RSL/Makefile.RSL.

If you want to look at whole fields or whole layers, here's another
trick.  This will output to ascii file(s) all the values for a
horizontal slice of an array, which can then be sorted an looked at
post mortem.  (assume we're back in a triply nested loop again):

      if ( k .eq. KDEBUG ) then
	write(80+rsl_myproc,*)ktau,i-idif_x0,j-jdif_x0,VAL(i,j,k)
      endif

When the model runs, it will will generate files fort.80 on processor
zero, fort.81, on processor one, and so on.  The command:

     sort +0n -1 +1n -2 +2n -3 fort.8* > foo 

Will put the data into the file foo sorted correctly (i will be major
and j minor -- you can fiddle with this to change the ordering).  If
there are multiple time steps, it will handle this correctly too.

You can then inspect the values manually, or process them further to
use your favorite plotter.

I hope this helps.

John