Unexpected Oceanic Wind and Rainfall Sensitivity to Land Surface Models in WRF

[Arty]

Hello everyone,

I conducted several sensitivity tests on physics parameterizations and noticed significant discrepancies in rainfall, 10m wind, and GHI (Global Horizontal Irradiance) fields across my domain (primarily oceanic) when using identical configurations—except for the Land Surface Model (LSM): Noah LSM vs. Thermal Diffusion. Interestingly, the Thermal Diffusion scheme produced much better agreement with in-situ observations and ERA5 data.

This result puzzles me because, as described in Skamarock et al., 2008 (Chap. 8.4):

"The land-surface models (LSMs) use atmospheric information from the surface layer scheme, radiative forcing from the radiation scheme, and precipitation forcing from the microphysics and convective schemes, together with internal information on the land’s state variables and land-surface properties, to provide heat and moisture fluxes over land points and sea-ice points."

Given that more than 90% of my domain is ocean, I wouldn’t expect such large differences between the two LSMs. I reviewed Tewari et al., 2004 and Dudhia, 1996 (related to each scheme) and examined module_sf_noahlsm.F in the WRF source code, but I couldn’t pinpoint the cause.

I understand the Land Surface Model interacts with other components (e.g., Surface Layer, PBL, etc.). Could anyone help clarify how changes in the LSM could influence atmospheric conditions over the ocean in this context?

Thank you in advance for your insights!


Additional Context:

• Double-nested domain (finest d02 resolution: 2.333 km) over Tahiti, French Polynesia;
• WRF Version: 3.6 (I realize it’s old, but it’s what I’m working with);
• Attached files: _A means convection is activated on d02 while _D means convection is deactivated on d02.

 

[Arty]

I apologize for the inconvenience, but I would like to follow up on this post as I need your help in identifying the cause of the issue mentioned above. Thank you, and I wish you a wonderful holiday season.

 

[kwerner]

Hi,
Apologies for the delay in response. To determine what's causing the difference, we would need to do an in-depth study into the code. I know you are working with V3.6.1, but is it possible for you to try the latest version of the code to see if any updates make your results different?

 

[Arty]

Hi,
Apologies for the delay in response. To determine what's causing the difference, we would need to do an in-depth study into the code. I know you are working with V3.6.1, but is it possible for you to try the latest version of the code to see if any updates make your results different?

Dear Kelly,

Thank you for your response, and I also apologize for my delayed reply.

I truly appreciate your suggestion to test a more recent version of the model. While I will do my best to explore this, I must admit that my current schedule is quite constrained as I’m in the final year of my PhD. With multiple competing priorities, it’s a bit challenging to dedicate extensive focus to this specific issue at the moment.

That said, I find it both intriguing and puzzling how the model version could lead to such differences when the same physics modules are used. Would you mind sharing your perspective on what might be driving these discrepancies? Do the physics modules undergo updates alongside the model code? Any insights into how the LSM could influence results in an ocean-dominated domain would be invaluable in helping me better understand this phenomenon.

Thank you again for your time and assistance—I genuinely value your input and look forward to any further clarification you can provide.

Wishing you a Happy New Year!

 

[kwerner]

Yes, the physics modules often have modifications for each new release. There are always improvements and bug fixes happening for them. As I'm not as experienced as some of my colleagues when it comes to model physics, I reached out to one to ask their opinion. They said it's possible the roughness length is different. They suggested you may want to check the values for ZNT if you have that available in the output.

 

[Arty]

Yes, the physics modules often have modifications for each new release. There are always improvements and bug fixes happening for them. As I'm not as experienced as some of my colleagues when it comes to model physics, I reached out to one to ask their opinion. They said it's possible the roughness length is different. They suggested you may want to check the values for ZNT if you have that available in the output.

Thank you, Kelly, for reaching out to your colleagues—I really appreciate the effort! I checked my output files (and wrfinput) but couldn’t find the ZNT variable. That said, it’s not too surprising, as my LU_INDEX values are either 15 (mixed forest) or 16 (water bodies), which don’t change much over time.

I also reviewed the modules to better understand how ZNT is set up but couldn’t pinpoint where it’s initialized. If your colleagues would be open to discussing this further, I’d be very grateful to connect with them by email.

Thanks again for your help!

 

[kwerner]

Hi,
I'd like to apologize for the very long delay in response while we've had to tend to other obligations. Since it's been so long, I first want to ask if you're still experiencing this issue. If so, you should be able to get ZNT output in your wrfout* files by using the Runtime I/O option. Can you try that to see if you can obtain it and then check the values.

 

[Arty]

Hi,
I'd like to apologize for the very long delay in response while we've had to tend to other obligations. Since it's been so long, I first want to ask if you're still experiencing this issue. If so, you should be able to get ZNT output in your wrfout* files by using the Runtime I/O option. Can you try that to see if you can obtain it and then check the values.

Hello Kelly,

I'm still interested in understanding these discrepancies, but I have to set that aside for now, as I have only seven months left before submitting my PhD manuscript and have a lot to prioritize. Nevertheless, I'll run test simulations with the activated option you mentioned when I have more time and update you on that post.

No worries about the delay—I understand you're all busy, and I truly appreciate that WRF has such a dedicated support team. I hope the current U.S. politics won't interfere too much with the future work of NCAR and other meteorology/climate organizations.