Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

WRF V4.3.2 NOAH_MP BEP/BEM very low nighttime 2m air temperatures


New member
Dear all,

I perform some ultra-high resolution simulations during a heat-wave period with WRF coupled to BEP/BEM urban canopy model using the NOAH_MP land surface model. Attached you can find the namelist.input I use. Initially, I used version 4.3, but recently I applied also version 4.3.2, as it contains the following bug fix, according to developers:

"An error in longwave radiation calculation in the BEP-BEM urban module is fixed. This error affects the calculation of TSK, which now is correctly calculated."

Indeed, the use of initial version led to very high TSK values (~ 330 K in night and ~ 347 in day), which were not correspond well to satellite LST observations. The application of the updated/corrected versions, lead to better simulation of TSK, but to worse results concerning near surface atmospheric variables and especially for 2m air temperature.

In the provided figures, the first one corresponds to version 4.3 simulations, while the second corresponds to version 4.3.2 simulation for the first six days (25/07 00:00 to 30/07 00:00). A significant drop of the simulated 2m air temperature by 2-3 degrees Celsius is evident during the night with the updated/corrected version. During the day the results are the same.

Maybe the longwave radiation calculation in BEP/BEM corrects the TSK calculation, but affects negatively the computation of 2m air temperature and other near surface variables?

Any ideas on how to resolve this issue would be very much appreciated.

Best regards,



  • namelist.input
    7.6 KB · Views: 48
  • August_T_all_FINAL_ALL_ERA5_v3.png
    974.7 KB · Views: 381
  • August_T_all_FINAL_ALL_ERA5_v2.png
    975.8 KB · Views: 381
Hi Christos,
You are right that the fix of TSK in WRFV4.3.2 introduces 'large' bias in T2. This is because T2 is diagnosed based on similarity theory. TSK and the lowest model level T are used as inout to calculate T2. With lower TSK, T2 will reduce correspondingly. This explains why you see lower T2 in WRFV4.3.2.
Surface and PBL processes in daytime are better described than in nighttime when the PBL becomes stable. This could be a possible reason that bias of daytime T2 is less in daytime than that in nighttime.
Last edited:
Dear Ming,

Thank you for the provided information. I changed the model physics configuration, so now I get better simulation results with the updated/corrected version (v4.3.2).

Best regards,

Hi Christos,

Would you please let me know which physics options you have changed? Thanks.

Hi Ming,

Sorry for the late reply.

I have changed YSU to BouLac PBL. This has provided a major improvement not only for 2m air temperature but also for 10m wind. I think that this is related to the fact that BouLac was specifically developed for use with BEP. Minor improvements were also provided by using SL option 2 instead of 1 and by using RRTM and Dudhia radiation schemes instead of RRTMG.

Thank you for the kind information, --- it is interesting that RRTM and Dudhia radiation schemes perform better than RRTMG.