WRF compiled with double-precision support

[akila.gothandaraman@gmail]

Hello,

We compiled WRF version 4.6.1 with double-precision support using Intel compilers (using ./configure r8). While trying to use this version, real.exe completes successfully, however WRF fails with segmentation fault. Increasing the debug level in the WRF namelist shows the seg fault occurs right after the cumulus driver is called.

d01 2005-02-28_12:05:43+37/50 in MYJPBL
d01 2005-02-28_12:05:43+37/50 call cumulus_driver
d01 2005-02-28_12:05:43+37/50 calling inc/HALO_CUP_G3_IN_inline.inc
d01 2005-02-28_12:05:43+37/50 in kf_eta_cps
forrtl: severe (174): SIGSEGV, segmentation fault occurred

When the same WRF run is performed with our standard installation of version 4.6.1 with single-precision, the run completes without any issues. The stack size is set to unlimited in both cases. I haven't yet build a debug version.

Any help or guidance you can offer is much appreciated.

 

[kwerner]

Hi,
Can you please package all of your rsl* files into a single *.tar file and attach that, as well as your namelist.input? Thanks!

 

[akila.gothandaraman@gmail]

Hello,

Thanks for your reply and apologies for not posting sooner. After some testing and re-compiling WRF with debug flag, we determined that the issue was related to the use of double precision version with spectral nudging enabled. Turning off spectral nudging (grid_fdda=0) in the WRF namelist resolved our issue. For the purposes of our current simulation, we did not need spectral nudging enabled.