[Wrf-users] Re: Running ndown problem

[Steven G Decker]

> Message: 1
> Date: Tue, 12 Apr 2005 13:42:51 -0500
> From: crystal_paulsen at und.nodak.edu
> Subject: [Wrf-users] Running ndown problem
> To: wrf-users at ucar.edu
> Message-ID: <acbd7abf16.abf16acbd7 at und.nodak.edu>
> Content-Type: text/plain; charset=windows-1252
> 
> Hi,
> 
> I’m trying to run ndown for a one-way nest and am getting a floating 
> point error while running ndown.
> 
> For initialization of my outer domain, I used NCEP Reanalysis data.  
> My domains are configured as follows:
> 
> Domain 1
> 220x180  at 15 km grid spacing
> 51 vertical levels
> 
> Domain 2
> 601x601 at 1 km grid spacing
> 51 vertical levels
> 
> For the outer domain I generate hourly output files which I use to run 
> real.exe and wrf.exe for that domain.  Then renaming real_input_d02 to 
> real_input_d01 and using the output file from the 1st domain I run 
> real.exe for domain 2.  Then renaming wrfinput_d02 to wrfinput_d01 and 
> real_input_d01 back to real_input_d02, I try to run ndown, but it 
> fails with a floating point exception.
> 
> NDOWN_EM V2.0.3.1 
> PREPROCESSOR                                                  
>  DYNAMICS OPTION: Eulerian Mass Coordinate 
>  -------->>>  Processing data: loop= 1   date/time = 1992-03-
> 08_06:00:00
>  loopmax =  31    ending date = 1992-03-09_12:00:00
>  file date/time = 1992-03-08_06:00:00     desired date = 1992-03-
> 08_06:00:00     status =  0
>  NDOWN_EM V2.0.3.1 
> PREPROCESSOR                                                  
>  DYNAMICS OPTION: Eulerian Mass Coordinate 
>  DYNAMICS OPTION: Eulerian Mass Coordinate 
>  -------------- FATAL CALLED ---------------
>  frame/module_domain.f: Failed to allocate grid%rqiblten
> (sm31:em31,sm32:em32,sm33:em33). 
>  -------------------------------------------
>  
> If anyone has an ideas on solving this problem it would much 
> appreciated.
> 
> Crystal

Crystal,

I've noticed when running the WRF that I'll sometimes get floating point 
exceptions if a nested grid has a much smaller grid spacing than its 
parent. In the old MM5 you needed to have a 3:1 ratio. You're using a 
15:1 ratio, which seems quite high and might be the source of the 
problem. Perhaps using an intermediate inner nest will solve the 
problem. The WRF documentation does recommend the ratio be 4:1 or less.

Hope this helps,
Steve