Urban scheme in nest domain affecting mother domain

[mesego]

Dear WRF community,

I am using WRF4.4 and is testing with the urban options.
I have two domains and intend to use urban only for the nested domain. My made three tests:
For test1: sf_urban_physics = 0, 0,
For test2: sf_urban_physics = 0, 1,
For test3: sf_urban_physics = 0, 2,

All the other options are set to be the same among the three tests. And feedback = 0,
So, I expected the difference should only occur on the nested domain, but it turns out that the result of mother domain from the three tests are also different (both temperature and precipitation). Is there any problem with it?

best regards
Liying

 

[kwerner]

Hi,
Unfortunately if you're using an urban option, you must have all domains to the same value. If you're using an earlier version of the code, you may not have gotten a message about this in your output log (e.g., rsl.out.0000), but the more recent code has a check that should print this out, and the code will modify your settings to ensure all domains are set to the same value, which likely explains the behavior you're seeing.

 

[mesego]

Hi,
Unfortunately if you're using an urban option, you must have all domains to the same value. If you're using an earlier version of the code, you may not have gotten a message about this in your output log (e.g., rsl.out.0000), but the more recent code has a check that should print this out, and the code will modify your settings to ensure all domains are set to the same value, which likely explains the behavior you're seeing.

oh... I see. I didn't realize this before. thank you very much for the information!

 

[DUN ZHU]

Hi,
Unfortunately if you're using an urban option, you must have all domains to the same value. If you're using an earlier version of the code, you may not have gotten a message about this in your output log (e.g., rsl.out.0000), but the more recent code has a check that should print this out, and the code will modify your settings to ensure all domains are set to the same value, which likely explains the behavior you're seeing.

Hi, Kwerner.
I also ran into this problem recently, in my four domains simulation, sf_urban_physics was set to (0,0,0,3) and then it was modified to (3,3,3,3) by real.exe.
This bothered me for a long time until I saw your reply. Can you provide me some documentation, or tell me which program I should look for evidence?
By the way, if WRF forces an urban canopy model on for each nested domain, does it become less sensible to run 4 domains simulations for BEP+BEM?
Thanks in advance.

 

[kwerner]

@DUN ZHU
The best documentation for the urban schemes can probably be found from the NCAR/RAL page on the Urban Canopy Model. As for your question about running 4 domains, I think that is more dependent on what resolution you are ultimately interested in simulating and how many domains that requires you to use. If you're planning to use BEP+BEM anyway, then you will just have to use it on all the domains, regardless. If you have your domain set-up the way that makes sense for your application, I would advise to keep all 4 domains.

 

[DUN ZHU]

Thank you for your response, Kwerner~ but I still have several questions.
I now find that the difference of specific humidity between the initial value and the boundary conditions of the most inner domain (domain4) is very large, and therefore this variable always changed sharply after this simulation began. Then it deviates more and more from the observed value.
I wonder if it's because my domain1 covers almost the whole Japan, but for such a huge range, it is diffiIcult to determine the urban canopy parameters for each urban catagories (as well as the domain 2 and 3)...The domain 4 is a 50km * 50km area with a spatial resolution of 500m * 500m.
I would appreciate it if you could check my namelist files~

 

[kwerner]

@DUN ZHU
Since your new questions are different than the original topic here, please post this as a new topic question. This helps with searchability and readability for other users in the future. Thanks!

 

[struggling_phd]

Hi all, please confirm that I understand this correctly:
If I run two different cases: 1) sf_urban_physics=0,0 & 2) sf_urban_physics=0,3
and for both of these I run separate real.exe and then wrf.exe. With all the namelist parameters the same, except sf_urb_phys, you are telling me there will be differences in the outer domain?

 

[kwerner]

@struggling_phd
When running ANY urban options, you must set all domains to the same setting. So you would need to set sf_urban_physics = 2, 2, 2 or = 3, 3, 3. That being said, I think your question becomes obsolete.

 

[struggling_phd]

@kwerner
Dear ms. Werner,

I have one more question regarding this topic of mother/child domains and urban schemes. Here is what I did, I ran real.exe for both mother/child domains with BEM (sf_urban_phys=3,3,). Then I run two separate wrf cases, one where I turn off the urban scheme in the outer domain, but leave it in the child domain (sf_urban_phys=0,3) , and one where urban is turned off in both cases (sf_urban_phys=0,0). From what I understand, the code will revoke my physics settings, and either switch to sf_urban_phys=3,3 or 0,0 .

It would override the setting of sf_urban_phys=0,3. That being the case, I must ask if I am doing something wrong, because from the snapshot I provide below, you can see that after about 42h of simulation there are significant differences regarding wind in the outer (mother domain). ANd not even at urban places but over the ocean? I am doubt-full in my run, should this be happening?

thanks!

 

[kwerner]

@struggling_phd
Per the code (assuming you're using WRF version 4.0+), if you have sf_urban_physics = 0, 3, the model corrects this by reassigning the value of domain 1 to the value given by domain 3, which means it would actually be running as sf_urban_physics = 3, 3. Does that help to explain the differences you're seeing? If not, will you clarify which settings the figures apply to? I'm not sure which case (e.g., sf_urban_physics = 0, 3, sf_urban_physics = 3, 3, etc.) they correspond to, or if you're comparing two different settings. Thanks!

 

[struggling_phd]

@kwerner

thanks for replying. In order to simplify the comparison, I completely remove the child domain. Now I run real.exe with sf_urban_physics = 3 . Then I run two separate cases with the same domain, IC and BC conditions.
First case I leave sf_urban_physics = 3 . Second case I put sf_urban_physics = 0 .
I expected the differences in the run to be only over the urban areas. However, the differences, after about 48h of simulations can be seen everywhere.




It feels like I am looking at two different simulations. Is this a normal behavior? The left most subplot is the difference between the middle plot (BEM case ) and the right most plot (No Urban case). I use WRF 4.4., and all the namelist setting are the same, except sf_urban_physics.

Thanks!

 

[kwerner]

That makes sense. Thanks for the explanation. Yes, we do expect a change over all areas. As in the real world, environment changes, such as urban characteristics can significantly modify the weather in surrounding areas. Over time, these differences spread throughout the environment.