
[Science] [BAS home] [Met home] [Beowulf home]	Antarctic Meteorology

Misc

Config files

These might be generally useful. Let me know if you think others should be up here... if you think anything is wrong, then even more let me know!

For the fujitsu compiler (old version):

config_4.5.cache for 32-bit model
config_4.5.cache for 64-bit model
compile_vars for 32-bit
compile_vars for 64-bit

For the portland compiler (unknown version; note: I haven't used portland much; I don't guarantee these to work):

config_4.5.cache for 32-bit model
config_4.5.cache for 64-bit model
compile_vars for 32-bit
compile_vars for 64-bit

Troublesome diagnostics

16,204 (RELATIVE HUMIDITY ON PRESSURE LEVELS) crashes my version of the model. Odd.

Filtering and Timestep

[nb: this applies to runs with the *wrong* level-by-level constants and a 20-min timestep, which (if you have any sense) means not your runs.]

In setfil1a.f are the lines:

! These diagnostic write statments allow the user to see how
! much of the globe is being filtered.

      WRITE(6,*) 'SETFIL : Updating filtering area to ',
     &           '2 - > ',NORTHERN_FILTERED_P_ROW,' and ',
     &           SOUTHERN_FILTERED_P_ROW,' -> ',glsize(2)-1
      WRITE(6,*) 'SETFIL: Filtering ',
     &  REAL((glsize(2)-2)-
     &  (SOUTHERN_FILTERED_P_ROW-NORTHERN_FILTERED_P_ROW-1))/
     &  REAL(glsize(2)-2),' % of grid (not counting polar rows)'

(by default commented out) which can be uncommented to show how much filtering is done. It prints whenever the filtering changes.

Over about 10 years of run, with a 20 min timestep (yabba), I get the following pattern of fraction filtered (nb the x-axis is not time, but just number-of-times-message-printed):

I presume this is an annual cycle.

But with a 30-min timestep and corrections to the lbl constants (see yabcc) the constraint changes to the adjustment courant number and the filtering is more, but changes less often: from a 15-year AMIP run (yabcd, to be precise):

 SETFIL : Updating filtering area to 2 - >  21  and  53  ->  72
 SETFIL: Filtering  0.563380301  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  23  and  53  ->  72
 SETFIL: Filtering  0.591549277  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  24  and  53  ->  72
 SETFIL: Filtering  0.605633795  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  23  and  53  ->  72
 SETFIL: Filtering  0.591549277  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  21  and  53  ->  72
 SETFIL: Filtering  0.563380301  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  22  and  53  ->  72
 SETFIL: Filtering  0.577464759  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  21  and  53  ->  72
 SETFIL: Filtering  0.563380301  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  22  and  53  ->  72
 SETFIL: Filtering  0.577464759  frac of grid (not counting polar rows)
 SETFIL : Updating filtering area to 2 - >  21  and  53  ->  72
 SETFIL: Filtering  0.563380301  frac of grid (not counting polar rows)

nb: those are %'s of the *rows* filtered: because the area shrinks towards the poles, 56% of the rows is about 37% of the area.

Past last modified: 13/12/2004 / wmc@bas.ac.uk