urban net longwave radiation error: no convergence. Critical = NaN > 0.001 !

[ahsanpmd]

Hi Graziano. I am using the latest version of the RegCM5 cloned from github. I am trying to run a simulation with the specifications mentioned in the attached namelist. But the models stops with the mentioned error.

test.i.txt

urban net longwave radiation error: no convergence
Critical = NaN > 0.001 !
-------------- FATAL CALLED ---------------
Fatal in file: mod_clm_urban.F90 at line: 2612
clm now stopping

urban net longwave radiation error: no convergence
Critical = NaN > 0.001 !
-------------- FATAL CALLED ---------------
Fatal in file: mod_clm_urban.F90 at line: 2612
clm now stopping

Abort called by computing node 15 at 2025-09-25 06:02:18.705 +0000

[DesmetQuentin]

Hi @ahsanpmd

I'm simply pointing out that this is the same error as previous issue #73
Still have no solution personally...
Could you please rerun the simulation with DEBUG enabled, to see if the simulation stop is triggered by some safety checks before reaching CLM?

Cheers,
Quentin

[graziano-giuliani]

Guys, can you give a try to the latest code? During the weekend I have found some race conditions introduced in GPU porting the CLM code which were popping up in the optimized intel build (the DEBUG was running just fine, the gfortran build was also running fine). They were related to implicit loop inside DO CONCURRENT which apparently the intel runs on multiple threads or vector, and gets lost on -O3. I should have fixed'em all, but I could not spot the error with ifort but only with ifx, and there may still be some lurking around because parts of the code are active only on conditional (glacier, wetland, lake, urban, etc). If error persists, can you post the namelist?