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

UCM model Issue

C

climatewind

New member
I tried to run SL-UCM model by putting sf_urban physics value as 0,0,0,1 in a 4 domain model. The model ran successfully, but whe I ran the same model with urban physics as 0,0,0,0; and removing the use_wudapt lcz from the model, the result came out to be same as previous. Kindly, tell the possible error.
The model version of WRF is V4.4.
 

Attachments

  • namelist.input_non_UCM.txt
    5 KB · Views: 9
  • namelist.input_UCM.txt
    5.1 KB · Views: 10
What files did you compare with and without UCM? How many urban points are there in your domains?
When you say the results are same, did you mean they are bit-for-bit identical?

Another issue is that, it is recommended that all physics schemes (except cumulus) should be same for all domains. In your case, you set feedback = 1 and sf_urban_physics = 0, 0, 0, 1. Such settings can cause inconsistency between D03 and D04. I am perplexed that the model ran to the end successfully.
 
The output file for domain 4 (D04) are compared in both the cases. As the study area under the domain 4 is a city at 333.33 m resolution, many of the points are urban. The results are bit to bit same, which is creating the concern.

The sf_urban_physics was chosen as 0,0,0,1 to minimize the computational power and memory consumption.
 
can you upload your wrfout_d04 files (with and without urban physics) for me to take a look? Thanks.
 
Make sure your urban_physics option equals to 0 and 1 for your wrfinput (NO_UCM and UCM).
How identical are your results? Have you compared area-averaged? For example, can you generate a xy plot for 2-m temperature?
 
By saying "point-to-point same", do you mean the wrfout files from the two runs are identical?

Can you check whether you have the file 'diffwrf' existing in in WRF/external/io_netcdf? if so, please issue the command:

WRF/external/io_netcdf/diffwrf wrfout_urban wrfout_no_urban > log

(Note I use wrfout_urban and wrfout_no_urban to represent wrfout files form your two runs)

And upload the log file for me to take a look.
 
Ming Chen said:
By saying "point-to-point same", do you mean the wrfout files from the two runs are identical?

Can you check whether you have the file 'diffwrf' existing in in WRF/external/io_netcdf? if so, please issue the command:

WRF/external/io_netcdf/diffwrf wrfout_urban wrfout_no_urban > log

(Note I use wrfout_urban and wrfout_no_urban to represent wrfout files form your two runs)

And upload the log file for me to take a look.
Click to expand...
 

Attachments

  • log.txt
    8.2 KB · Views: 1
The log.txt file indicates that the wrfout files are bit-for-bit dental at times from 2022-05-21_12 to 2022-05-21_21.

However, starting from 2022-05-21_22, the wrfout files become different in many fields like LFMASS, T2V, T2B, etc.

It remains mysterious why the results are bit-for-bit identical in the first 9 hours of simulation. Sorry that I don't have an explanation right now. I will talk to our expert and get back to you if I have more information.
 
Thank You and in the case when simulation is after 2022-05-21_22 parameters like T2 shows similar results too. The output graphs for UCM and Non-UCM models are identical which is also a concern
 
You must log in or register to reply here.
Top