Urban Hydrology Using Urban Canyon Model is likely not Realistic - Could be causing Goofy Interpretations

[GAaronAlexander]

Bug Description

I may be missing some documentation about how this error is treated but during simulations I was doing for an urban space using the Urban Canyon Model during a rainstorm, I found that at least the runoff & subsurface runoff are not realistic given the the what is being used to describe the urban region.

To Reproduce

Steps to reproduce the behavior:

1. Run a WRF model over an urban area and use the Urban Canyon model. Ensure that there is a rainstorm. I was using Noah-MP as my LSM, but it could be done with Noah, Noah-MP, CLM, or any other LSM that is coupled with the Urban canyon model
2. Examine outputs of the rainfall and the surface runoff. Note the surface runoff is significantly lower than the precipitation.
3. I have provided an example of the Milwaukee Area, where we would expect that the runoff would be at least .5 percent in most places within the darker squares (with some 0.8 and 0.95s in more urbanized locations), but instead it is closer to 0.25.

Description of current Behavior and what would be expected given current documentation

Right now, when employing the urban canyon model, a fraction (either the urban ones from the NUAPT project, local climate zones, or the hardcoded NCLD ones) is assigned. For example, let's say that the split is 95% urban 5% other (as would be assigned in a default Urban Canyon Model run with a Heavy industrialized downtown area using NLCD). One would expect that the amount of runoff would be at least 95% that of incoming precipitation if there was no urban ponding turned on. When I have done experiments, the runoff ratio is ranges from 0 - 80% of rainfall. This is similar for the underground runoff and potentially other hydrologic variables like soil moisture.

Potential Fix

I believe that one would need to add specific averaging like is already done for variables like temperature, humidity, surface roughness, as shown from the Noah LSM surface driver in the screenshot below. This would need to be done in all locations that the urban canyon model is called and subsequently averaged (noahmp driver, CLM driver, etc).

[weiwangncar]

@GAaronAlexander Is your urban canyon model one of the urban options in the WRF model as assigned by namelist sf_urban_physics?

[GAaronAlexander]

@weiwangncar : The urban canyon model that I am referring to in this issue is the standard urban canyon model (sf_urban_physics = 1). The behavior that I am describing is using this model without any changes.

[weiwangncar]

@cenlinhe If you have a moment, can you comment on this post?

[cenlinhe]

Hi Aaron, thank you for pointing out this issue. You are right that there is currently no diagnostic calculation of the grid-mean runoff by merging urban runoff and non-urban runoff in the driver module. Also, current SLUCM does not include a comprehensive treatment of urban hydrology, for example, it only includes the runoff treatment for the green roof part. I would treat this as a current model deficiency instead of a bug. This part needs future model enhancement.

[GAaronAlexander]

Hi @cenlinhe! You are welcome. I am happy to help contribute to a working group to help get this, and other issues with urban hydrology if there is one going at NCAR. I know that there are some for the Noah-MP group, and if that is the best place to start then count me in!

[cenlinhe]

@GAaronAlexander Sure, I will count you in if there is any relevant activity from our side (currently we do not have any on-going work specifically on urban hydrology). If you have any plans from your side, we are also happy to help and collaborate. Thank you for offering your help!