Error when running Multi-scale Kain-Fritsch

Topics related to current and future physics in the WRF as well as any problems you may have.

Error when running Multi-scale Kain-Fritsch

Postby ronmcdo » Mon Mar 06, 2017 10:36 pm

G'day!

I'm getting this error when I run wrf.exe with the Multi-scale Kain-Fritsch (cu_physics = 11):

Timing for main: time 2017-01-06_10:43:20 on domain 2: 1.35804 elapsed seconds
Timing for main: time 2017-01-06_10:43:40 on domain 2: 1.33109 elapsed seconds
Timing for main: time 2017-01-06_10:44:00 on domain 2: 1.33436 elapsed seconds
Timing for main: time 2017-01-06_10:44:00 on domain 1: 6.31737 elapsed seconds
Timing for main: time 2017-01-06_10:44:20 on domain 2: 1.35746 elapsed seconds
Timing for main: time 2017-01-06_10:44:40 on domain 2: 1.32991 elapsed seconds
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2063
QG, QG(NK).LT.0
-------------------------------------------

Any ideas what is causing this or has anyone experience and solved the same thing?

Many thanks!
ronmcdo
 
Posts: 1
Joined: Mon Mar 06, 2017 10:30 pm

Re: Error when running Multi-scale Kain-Fritsch

Postby jjohnson » Tue Nov 28, 2017 8:38 pm

I ran into this same issue with MSKF. I had to modify the code to get more information since the error message is not helpful.

I found that the model was crashing for model grid cell columns over the Atlantic Ocean (presumably where extremely weak convection was being simulated) where the convective time scale was very long (more than 3.5 days).

The fix for me was to cap the convective time scale (TIMEC) at 24 hours (86400 seconds). The MSKF subroutine has not been updated from WRF v3.8 to v3.9.1.1, so this issue goes back at least that far (MSKF was introduced in WRF v3.7).

See my diff command and output for the two files (original and modified) below. You need to recompile WRF, but you shouldn't need to do a 'clean' first -- the build process will recompile the updated MSKF subroutine and create new executables.

I've emailed wrfhelp about this issue and hopefully a fix is implemented before the next release.

diff -u phys/module_cu_mskf.F.orig phys/module_cu_mskf.F
--- phys/module_cu_mskf.F.orig 2017-11-28 15:42:48.596887685 -0800
+++ phys/module_cu_mskf.F 2017-11-28 15:43:31.466887666 -0800
@@ -1702,6 +1702,7 @@

NIC=NINT(TIMEC/DT)
TIMEC=FLOAT(NIC)*DT
+ TIMEC=AMIN1(TIMEC,86400) !JRJ Ramboll: cap convective time scale at 24 hrs
!
!...COMPUTE WIND SHEAR AND PRECIPITATION EFFICIENCY.
!
jjohnson
 
Posts: 1
Joined: Tue Nov 28, 2017 8:15 pm

Re: Error when running Multi-scale Kain-Fritsch

Postby kwthomas » Wed Nov 29, 2017 6:23 pm

Your solution should be reported to "wrfhelp@ucar.edu" so that it can make it into future
releases.
Kevin W. Thomas
Center for Analysis and Prediction of Storms
University of Oklahoma
kwthomas
 
Posts: 208
Joined: Thu Aug 07, 2008 6:53 pm


Return to Physics

Who is online

Users browsing this forum: No registered users and 4 guests