[Wrf-users] Bug Fix for 3DVAR_OBSPROC

[Brent L Shaw]

For those of you out there using the 3DVAR obs pre-processor with WRF
Var, there are a couple of minor bug fixes you need to make to correct
some MM5 heritage (backwards definitions of I/J).  Without these fixes,
the obs pre-processor is correcting incorrect (I,J) grid locations from
the latitude and longitude when checking to see if observations are
within the domain.  

 

In 3DVAR_OBSPROC/src/module_namelist.F90, in the "WRF" section near line
362, you need to change the computation of the center X/Y of the grid.
The original code has:

 

      xxc = real(jjx)/2

      yyc = real(iix)/2

 

The corrected code is:

 

    xxc = real(iix+1)*0.5

    yyc = real(jjx+1)*0.5

 

(Note that in addition to the +1 and use of multiplication instead of
division, xxc is now associated with iix instead of jjx, and yyc is now
a function of jjx instead of iix.  It's not so easy to notice that
change with my font!).  

 

Then, in module_inside.F90, right after "call latlon_to_ij", comment out
(or remove):

 

xjd = xjd + 0.5

xid = yid + 0.5

 

In the same file, immediately following the lines above, change the IF
statement from:

 

      IF ((yid .GE. 1.) .AND. (yid .LE. ins  )  .AND. & 

          (xjd .GE. 1.) .AND. (xjd .LE. jew  )) THEN

 

To:

 

   IF ((yid .GE. 1.) .AND. (yid .LE. jew  )  .AND. & 

         (xjd .GE. 1.) .AND. (xjd .LE. ins  )) THEN

 

Best regards,

 

Brent

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/wrf-users/attachments/20070927/8c088be7/attachment.html