Bug Fix for Morrison Microphysics Option (posted 11/16/11)
Problem: A bug was introduced in V3.3.1 in Morrison microphysics scheme. The bug affects cloud ice fallspeed, and hence cirrus lifetime and radiative forcing. The effect of the bug is less in shorter simulations, but significant in terms of high cloud cover and LW radiative forcing in longer-term climate runs. This bug is not present in V3.3 and earlier.
Solution: Edit phys/module_mp_morr_two_moment.F, and change line 1289 to
AIN(K) = (RHOSU/RHO(K))**0.35*AI
Or download the routine here.
Compile V3.3 with ifort (posted 4/6/11)
Problem: With the use of ifort 10.1, the model may seg fault with CAM and RRTMG radiation options with default compiler option.
Solution: Reduce the optimization level from -O3 to -O2 appears to resolve the problem.
OpenMP And MPI For Some Physics in V3.3 (posted 4/6/11)
Problem: Certain physics options do not work correctly with OpenMP or MPI compile. These are
- mp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
- sf_urban_physics = 2 (BEM) (OpenMP)
- ishallow = 1
(OpenMP and MPI)
There is no known problem with these physics options if OpenMP or MPI is not used.
Solution: Do not attempt to use the above listed options with OpenMP or MPI compile.
"HANGING" PROBLEM FOR SOME MACHINES
For machines with relatively small per-processor memory, such as blue gene, this fix removes the problem of just "hanging" when trying to open netcdf files. For users who have no troubles with the current netcdf IO, this fix is not required.
Solution: Reduce the MaxTimes variable to ten thousand from nine hundred thousand. Please look at code external/io_netcdf/wrf_io.F90 (or external/io_pnetcdf/wrf_io.F90 if you run parallel netCDF). Change line 44 from:
integer , parameter :: MaxTimes = 900000
to:
integer , parameter :: MaxTimes = 10000
The code needs to be recompiled (no ./clean -a is necessary).
PGI dmpar compile not giving bit-for-bit results between two identical runs (posted 12/31/13)
Problem: When compiling with PGI in dmpar, the results will not be identical from one run to the next, using the exact same namelist configuration.
Solution: In the arch/configure_new.defaults file, you will see the following line:
FCOPTIM = -fastsse -Mvect=noaltcode -Msmartalloc -Mprefetch=distance:8 -Mfprelaxed # -Minfo=all =Mneginfo=all
Comment out '-fastsse' so that it reads:
FCOPTIM = -O3 #-fastsse -Mvect=noaltcode -Msmartalloc -Mprefetch=distance:8 -Mfprelaxed # -Minfo=all =Mneginfo=all
You will then need to do a './clean -a' and then reconfigure and recompile.