Bug in LAI initialization for LCZ urban gridcells

[jesusff]

I copy here a problem reported by @LluisFB and explained in several wrf-forum and github entries:

Recently I found a bug in WRF, which has already been fixed 2 weeks ago. It is related to LAI initialization when using LCZ data...
wrf-model/WRF#1990
It is strange, because I only encountered it whilst trying to solve another issue when I compiled WRF (out git_FPS version) in debug mode.

I am trying to run some urban simulations for Buenos Aires, but I failed encountering a RRTMG seg fault. See my post and how it follows in the WRF-MPAS forum.

https://forum.mmm.ucar.edu/threads/segmentation-fault-with-rrtmg-rtmg_lw_taumoltaumol_mp_taugb3_-module_ra_rrtmg_lw-f90-0.15612/*post-38953 Hope it might help someone, or give some clues to some execution errors...

[jesusff]

There's a proposed fix for this: NCAR/noahmp@6f2bf7c , which we could apply to our WRF v4.5.1-devel branch and use it for the FPS-URB-RCC runs. Not sure why this bug did not affect the existing runs using LCZ for some groups. There are also some others reporting no failure NCAR/noahmp#104 (comment) It would be nice to find out in which scenario this issue leads to model crashes