WRF 4.1.1 crashes line 1680 of file module_sf_noahdrv.f90 after changed physics options

Hi all.

I have been able to successfully run WRF (4.1.1) on the Amazon cloud using GFS input data on a region covering NW South America and a nest zoomed on on a mountainous part of Colombia. It ran fine using the "tropical" physics suite, however... when I change a few physics options, it always crashes in the land surface scheme.
See attached rsl and namelist. I copied a namelist that works on the same input files, and only changed these options, which crashes it:

mp_physics = -1, -1, 8,
cu_physics = 3, 3, 0,
lightning_option = 0, 0, 3,

(27-9-3 km domain grid spacing)

I wanted to experiment because the convective rainfall seems over-predicted near the mountains. On satellite there is no sign of convective clouds. With nwp_diagnostics =1 it shows even widespread signal in HAIL_MAX2D which is not simply cumulus to me.
I also got some meridional patterns in the cumulus rain which seem artifacts, maybe from interpolation of land surface properties?
Your suggestions are welcome.

best regards,
Oscar

I experimented further and it turns out I get no more errors after I eliminated the "lightning_option" line.

Thanks for the uoatde. Note that all physics in the parent and child domains should be the same except the cumulus scheme, which can be off for high resolution child domains.