WRF 4.5.2 LCZ simulation – suburban daytime temperature bias

[nanyin]

Hello everyone,


I am running WRF v4.5.2 with the LCZ urban parameterization (sf_urban_physics = 3, use_wudapt_lcz = 1). The model is generally working fine:

• Urban areas: near-surface temperature simulation looks very good.
• Suburban/rural areas: night-time temperature matches observations well, but during midday the simulated temperature is about 2–3°C too low compared to observations.

This systematic cold bias at noon in suburban regions makes me wonder if it is related to the choice of radiation or cloud microphysics schemes. I would really appreciate any suggestions on which physical parameterizations might be most sensitive for this issue (e.g., radiation, land surface model, PBL, etc.), or if additional namelist adjustments are typically required for LCZ runs.


Here is the relevant portion of my namelist.input:



&physics
physics_suite = 'CONUS'
mp_physics = 6, 6, 6,
cu_physics = 1, 1, 0,
ra_lw_physics = 4, 4, 4,
ra_sw_physics = 1, 1, 1,
bl_pbl_physics = 4, 4, 4,
sf_sfclay_physics = 2, 2, 2,
sf_surface_physics = 2, 2, 2,
radt = 2, 2, 2,
num_land_cat = 61,
sf_urban_physics = 0, 0, 3,
use_wudapt_lcz = 1,
/


Any advice or shared experience would be very helpful. Thank you!

 

[Ming Chen]

First if all, please turn on urban physics in all your three domains. Different physics options in child domains may lead to feedback errors.

Second, what is your simulation period? Which data did you use to drive your case?

Please run with RRTMG radiation, which is better than RRTM.