WRF 3.9 DFI broken?

[meteoadriatic]

Hello,

Had anyone tried WRF 3.9 with DFI filter? It seems to me it is broken in a sense that it can't continue after DFI finishes (might have something with broken restart capability described in "known problems" for the release)?

Question is not EMS related.

thx

[surgeon]

I use DFI and it's working fine like in 3.8.1.

[meteoadriatic]

OK, thanks, I should find a mistake then.

[meteoadriatic]

I'm getting segmentation fault right after DFI finishes and model tries to start to integrate normally from initial time:

...
Writing out initialized model state
mediation_integrate.G 1728 DATASET=HISTORY
mediation_integrate.G 1729 grid%id 1 grid%oid 1
Timing for Writing wrfout_d01_2016-06-18_00:00:00 for domain 1: 4.76986 elapsed seconds
Timing for processing lateral boundary for domain 1: 0.12908 elapsed seconds
forrtl: severe (174): SIGSEGV, segmentation fault occurred

With 3.8.1 this does not happen. Strange...

[meteoadriatic]

OK, new information. It seems I get seg fault only if I use physics_suite = 'conus'. However, it 'conus' works without DFI. Will try to narrow down the cause, probably one of physics settings from 'conus' make problem.

[meteoadriatic]

Well, it looks like it is not a particular setting within 'conus' suite, but actually physics_suite is the cause. Funny, but that my tests show.

If I set old way all physics from 'conus' and remove physics_suite from namelist.input, then everything works with DFI. If I use physics_suite and remove traditional entries defined in it, run segfaults after DFI phase. On the other hand as I wrote physics_suite = 'conus' works fine if no DFI is used.

I guess, case closed for me.