Replicating a Realistic Mesoscale WRF Simulation Using ERF

[Armin-Ha]

Dear ERF Development Team,

I am beginning to work with ERF and would like to replicate a realistic mesoscale simulation I previously conducted using WRF. I understand that ERF supports initialization from met_em or wrfinput-wrfbdy files.

To get started, I would like to initialize ERF using the met_em files generated by WPS. I’ve drafted an initial ERF input file, but there are still several aspects I need to clarify, particularly regarding the incorporation of the NoahMP land-surface and RTE-RRTMGP radiation models.

I would greatly appreciate it if you could:
- Provide any guidance or documentation on best practices for converting WRF-based input configurations to ERF-compatible setups.
- Review my current input file and let me know what needs to be modified or added.

To provide full context, I am sharing the input files I used for WPS and WRF, along with the initial ERF input file I’ve started preparing. I would also be happy to share the met_em files if that would help with troubleshooting or validation.

Thank you for your support and your continued work on the ERF model.

Best regards,
Armin

namelist_input.txt
namelist_wps.txt
inputs_mesoscale.txt

[asalmgren]

@akashdhruv / @AMLattanzi -- is the combination of NOAH-MP and RTE-RRTMGP ready for testing?

[akashdhruv]

@akashdhruv / @AMLattanzi -- is the combination of NOAH-MP and RTE-RRTMGP ready for testing?

It will be ready this week. Ironing out segmentation faults.

[asalmgren]

@Armin-Ha -- I think this is ready for trying -- can you do a git pull and give it a try and let us know what happens?

[akashdhruv]

@Armin-Ha Yes this is ready for testing. We are currently doing testing of our and will be pushing updates, but if you see a discrepancies in your simulation results can you please report those to us?

You can use this Tall Towers case as a starting point:

https://anl.app.box.com/folder/319728445405?s=aazocir503xom3rihxph78vpth41h0ea

[Armin-Ha]

@asalmgren, @akashdhruv -- Thank you for the updates. I will start with testing the new features as soon as possible and keep you posted on any discrepancies or feedback that may arise during the testing process.

[Armin-Ha]

Dear ERF Development Team,

Thank you again for enabling the integration of NoahMP and RTE-RRTMGP in ERF. I’ve started testing the setup and would like to share a few initial observations:

Memory Usage
The radiation model requires substantial memory during advancement. On a machine with 256 GB RAM, I was unable to run the Tall Towers case even with a single domain.

Simulation Stability
I successfully ran a two-domain simulation for my own case using the following configuration:
- erf.land_surface_model = "NOAHMP"
- erf.moisture_model = "Kessler"
- erf.radiation_model = "RRTMGP"
- zlo.type = "NoSlipWall"
- erf.pbl_type = "None"
This setup is similar to the Tall Towers configuration and resulted in a stable simulation. However, as expected, the results were not particularly meaningful without a PBL scheme.

PBL Compatibility Issues and Runtime Error
When switching to erf.pbl_type = MYNN25 and zlo.type = surface_layer (with the same land surface, moisture, and radiation models), I encountered the following:

1. Warning: "Moisture may not yet be compatible with PBL models, proceed with caution.
2. Runtime Error: Assertion `qvel(i,j,k) > 0.0' failed, file "/home/armin/ERF/Source/PBL/ERF_ComputeDiffusivityMYNN25.cpp", line 80, Msg: KE must have a positive value !!!

This was traced to RhoKE not being initialized when MYNN25 is selected, causing qvel to be zero and triggering the assertion. I resolved this by modifying ERF_Prob.cpp to initialize RhoKE with a vertical decay profile.

Unstable Buoyancy Length Scale Calculation
After resolving the initialization issue, I encountered another critical error during the calculation of the buoyancy length scale in ERF_ComputeDiffusivityMYNN25.cpp. Specifically, the velocity scale qc, computed as qc = CONST_GRAV / theta0 * surface_heat_flux * l_T frequently becomes negative, which leads to a floating-point exception when passed to std::pow(qc, 1.0/3.0). This occurs in stable boundary layers where the surface heat flux is negative.

I would appreciate any guidance on how to handle these cases.

inputs_erf.txt

Best regards,
Armin

[ewquon]

@Armin-Ha, would it be possible to share your wrfinput and wrfbdy?

[Armin-Ha]

@ewquon, yes, certainly! Could you please verify if you can download the data from the following link?

https://drive.google.com/drive/folders/1QyCeqCBM9_ByLHaj4P2vAjnz_RuwD5AS?usp=drive_link

Please note that I encounter the same error when calculating the buoyancy length scale in the Tall Towers case. This occurs when I set erf.pbl_type = MYNN25 and zlo.type = surface_layer, while using erf.radiation_model = None to prevent the out of memory error.