Why does the drag actually decrease and wind speed increase after enabling the urban canopy model?

[Kecheng Zhao]

First, these are the results I obtained from running the single-layer urban canopy model. I'm confused as to why enabling the urban canopy causes CD, Z0, and UST to decrease, which leads to an increase in 10-meter wind speed.
View attachment Multi_vars_diff_SLUCM_noUCM_fast.png

In addition, I also ran a simulation with BEP. The wind speed decreased as I expected, but this occurred despite the fact that CD, Z0, and UST in BEP are smaller compared to noUCM. I don't know how to explain this phenomenon.
View attachment Multi_vars_diff_BEP_noUCM_fast.png

I believe my simulation setup should be correct, but the results I obtained seem strange. I hope someone can give me some advice.

 

[Ming Chen]

My understanding is that, without urban canopy, urban areas are treated as rough terrain with large roughness length and large exchange coefficient. The strong shear between the surface and upper levels leads to large friction velocity. When UCM is activated, surface roughness length is for the surface below the urban canopy, and the buildings in urban areas are treated separately. U10 is less affected by drags inside the urban canopy.

 

[Kecheng Zhao]

My understanding is that, without urban canopy, urban areas are treated as rough terrain with large roughness length and large exchange coefficient. The strong shear between the surface and upper levels leads to large friction velocity. When UCM is activated, surface roughness length is for the surface below the urban canopy, and the buildings in urban areas are treated separately. U10 is less affected by drags inside the urban canopy.Thank you for your explanation. I'd like to follow up with a few questions to better understand this mechanism:

Thank you for your explanation. I'd like to follow up with a few questions to better understand this mechanism:

First, regarding the surface roughness length decrease when UCM is enabled:

Is this reduction related to the reason mentioned in this thread: Latent Heat Flux Is Zero in Urban Land? If I understand correctly, when UCM is activated, the surface roughness length represents only the ground surface beneath the urban canopy (rather than the effective roughness of the entire urban area including buildings). This smaller Z0 then leads to reduced UST and CD, which means U10 is less affected by drags inside the urban canopy, ultimately resulting in higher wind speeds in SLUCM compared to noUCM over urban areas. Is this interpretation correct?

Second, regarding the BEP results:
Following this logic, why does BEP show decreased wind speeds compared to noUCM, despite also having smaller CD, Z0, and UST values? Does BEP handle the momentum exchange or building drag effects differently than SLUCM? Is there perhaps an explicit treatment of building-induced drag in BEP that counteracts the effect of smaller surface parameters?

Third, regarding comparison with published literature:
I'm confused because when reviewing some literature, I found that SLUCM typically produces reduced wind speeds in their simulations (e.g., Liao et al., "Impacts of different urban canopy schemes in WRF/Chem on regional climate and air quality in Yangtze River Delta, China"; Wang et al., "Impact of different urban canopy models on air quality simulation in Chengdu, southwestern China"). This makes me wonder whether I've made an error in my simulation setup. However, my results show good agreement with observational data, which suggests the simulations may be reasonable.

Could there be regional differences, or differences in model configuration that might explain why my SLUCM results differ from these published studies?

I would appreciate any insights you can provide on these questions.

 

[Ming Chen]

Thank you for the follow-up. I guess our LSM expert is in a better position to explain these issues you raised. I have forwarded your question and hopefully he can get back top you soon.

 

[cenlinhe]

Hi, to answer your questions:
1. the surface roughness length decrease is not related to LH~0 in urban. The near-zero LH in urban is because there is almost no urban hydrology in urban physics (it is a dry urban traditionally).
2. the surface roughness length in urban is not always decreasing when activate urban, depending on which urban physics you are using and the urban parameters prescribed for your study domain. The surface roughness length can be tuned in parameter lookup table: URBPARM.TBL or URBPARM_LCZ.TBL (if you use LCZ). Also, the building height is also tunable in these parameter tables.
3. urban model uses a different formulation to compute drag coefficient that affects U10 than natural vegetated region when no urban physics is activated. you can take a look at those original urban physics papers for those formulations.
4. BEP and SLUCM use different building treatment and drag coefficient formulations as well as urban parameter values which lead to different results in wind speed changes.
5. Your results' difference from the literature may be due to the different urban parameters used in these studies. People may tune their urban parameters in their simulations.

 

[Kecheng Zhao]

Hi, to answer your questions:
1. the surface roughness length decrease is not related to LH~0 in urban. The near-zero LH in urban is because there is almost no urban hydrology in urban physics (it is a dry urban traditionally).
2. the surface roughness length in urban is not always decreasing when activate urban, depending on which urban physics you are using and the urban parameters prescribed for your study domain. The surface roughness length can be tuned in parameter lookup table: URBPARM.TBL or URBPARM_LCZ.TBL (if you use LCZ). Also, the building height is also tunable in these parameter tables.
3. urban model uses a different formulation to compute drag coefficient that affects U10 than natural vegetated region when no urban physics is activated. you can take a look at those original urban physics papers for those formulations.
4. BEP and SLUCM use different building treatment and drag coefficient formulations as well as urban parameter values which lead to different results in wind speed changes.
5. Your results' difference from the literature may be due to the different urban parameters used in these studies. People may tune their urban parameters in their simulations.

Thank you Ming Chen and cenlinhe for your responses. I've conducted additional analysis but still have some questions:

Observed phenomena：
After enabling UCM, compared to the noUCM case:
SLUCM: Z0↓, UST↓, CD↓, WS10↑, HFX↓, LH↑
BEP: Z0↓, UST↓, CD↓, WS10↓, HFX↓, LH↑

My questions：
1. Regarding roughness output:
What exactly does the ZNT/Z0 variable in wrfout represent after enabling UCM? The post I referenced mentioned "from which I guess the surface roughness for the urban portion is calculated in the urban module, but not combined with the rural counterparts in the NOAH module. So ZNT and Z0 in wrfout uses 'Natural' values instead of a combined value." I would like to know if this is the reason? Or as Ming Chen said, the output Z0 is just the surface roughness length for the surface below the urban canopy, and the buildings in urban areas are treated separately?

2. After enabling UCM, is the decrease in UST and CD in the wrfout output caused by the decrease in Z0?

3. When UST, CD, and Z0 in wrfout output are all reduced compared to noUCM, BEP reduces the 10-meter wind speed while SLUCM increases it. After reading some literature, I found that the contribution of vertical surfaces (walls) to momentum sink >> horizontal surfaces. I think this is because BEP has a multi-layer structure and performs more accurate building drag calculations, thus correctly reducing WS10.

4. In addition, I noticed that after enabling UCM, LH increases and HFX decreases in urban areas. I think this is partly due to FRC_URB2D, which will be read from URBPARM.TBL after enabling UCM, while the default noUCM directly uses 0.9; on the other hand, it may also be related to the explanation in the post I referenced, because after enabling UCM, it will first calculate according to NATURAL.

For the first and second points, I hope to get your further explanation.
In addition, do you think my understanding of the third and fourth points is correct?

Thank you again!

 

[cenlinhe]

To answer your questions:

1. The wrfout ZNT/Z0 variables are the values from land model calculations assuming "nature" vegetation type when UCM is activated. However, the ZNT/Z0 values are updated to be Z0C (e.g., from urban parameter table) within urban physics modules for urban physics calculation. These updated urban ZNT/Z0 is not passed to other wrf physics or wrfout modules, which appears to be a bug. At least, there should be a weighted average grid-level mean ZNT/Z0 as a diagnostic output variable to be used by other WRF physics (e.g., surface_clay_physics). These does not affect the urban physics calculations though.

2. Because of the above issue, you cannot use Z0/ZNT from wrfout to directly analyze how urban Z0 changes.

3. Because of the above issue, the change in U10, V10 from noUCM to UCM is also controlled by urban indirect impact on near-surface meteorology rather than direct impact from changing Z0, since the WRF surface physics uses Z0/ZNT from land model not urban model.

4. Your interpretation of BEP and SLUCM differences is also correct.

5. Regarding the LH and HFX changes from noUCM to UCM, your interpretation is also correct. Both FRC_URB2D (+ dry urban physics) and NATURAL (+general land surface model) play roles.

I think the first point I mentioned above is a bug and I will create a WRF GitHub issue for this to enable further discussions.