WRF stops with `dzbot=30 and 50` but runs with `dzbot=100` for 1 km coastal Mumbai SLUCM + LCZ simulation

[UV Rohith]

Hello,

I am running a 4-domain nested WRF-UCM simulation for the coastal city of Mumbai (27 km, 9 km, 3 km, 1 km) using SLUCM with MODIS-CGLC-LCZ, with time_step = 54, e_vert = 48, dzstretch_s = 1.11, and dzstretch_u = 1.10.

Some 1 km urban pixels include high-rise LCZ classes, and the maximum building height is 37.5 m as specified in URBANPARAM_LCZ.TBL.

Earlier I used dzbot = 30, but the run stopped without clear errors in the rsl files. Then I presumed it's stopped because of the building height and I then tried dzbot = 50 (above maximum building height), but it still crashed.

When I increased to dzbot = 100, the model ran successfully and produced outputs. Could this be related to coastal dynamics + nested 1 km domains + SLUCM urban geometry interacting with first-layer thickness? For Mumbai urban simulations, is dzbot = 100 scientifically acceptable, or would another compromise value be better?

Also i have tried dzbot = 75, and it ran for 24hrs of simulation time (24 timesteps) and it stopped in between

 

[cenlinhe]

When the dzbot is smaller than building height, the SLUCM will not run. This is a hard requirement for SLUCM.
When the dzbot is larger than building height but close to building height, the model will become unstable which also depends on environmental conditions. Unfortunately, there is no good way to determine a best dzbot value right now and you will have to test a few different values in your specific case.