How to start a WRF-LES simulation?

M

Michael Zhang

New member
I am a graduate student currently studying WRF. I would like to perform LES simulations, but I’ve run into some difficulties, so I’m here to ask for your help. I would like to know the specific steps involved in running an LES simulation. Do I simply need to adjust the parameters corresponding to the LES domain in the namelist and follow the standard WRF run procedure, or is a more complex process required (such as using ndown.exe and following the instructions in the official documentation “WRF-ARW V4: User’s Guide 5-15”)? I am eagerly seeking guidance on this matter and would greatly appreciate your insights. Thank you very much!!!
 
Hi Michael,
WRF provides ideal LES run in its officially released codes. is this what you are looking for? Or are you trying to run real-data LES simulation?
 
Yes, I'm trying to run real-data LES simulation, and I' m confused about what the exact procedures to start a real-data LES, “WRF-ARW V4: User’s Guide 5-15”--does the content mentioned in this section outline feasible steps for running LES?(Because I came across a document that outlined the steps for running LES, which matched the content in that section of the official documentation; however, the official documentation does not explicitly state that these are the steps for running LES.) And I asked several AI, their answers did not mention that 'ndown.exe' is required to launch LES, yet the course slides refer to the use of 'ndown.exe' throughout. So I get into trouble and I want to make clear it to continue next research work.
 

Attachments

Real-data LES simulation is tricky. This is because LES run requires high resolution of ~100m or less. No large-scale forcing data is available to drive a WRF run at such a high resolution. In addition, eddy structure is permanently missing along the lateral buffer zone, leading to incorrect results inside the model domain. To overcome these issues, ndown or multi-domain nesting is recommended for real-data LES simulation.
 
Thank you very much, Mr.Chen! I understand quite a few about the issue. The domains what i used include 5 nesting domains(27km, 9km, 3km 1km, 200m), the last domain is LES region i plan to use. As you say, LES run requires high resolution of ~100m or less, my simulation is not really true LES, i think it can be called 'VLES'(I saw from a literature). So if a i want to start a LES in 200m resolution(VLES), it is recommended to follow LES running procedures(using ndown.exe), or common wrf running steps is also advisable?
 
Michael Zhang said:
Thank you very much, Mr.Chen! I understand quite a few about the issue. The domains what i used include 5 nesting domains(27km, 9km, 3km 1km, 200m), the last domain is LES region i plan to use. As you say, LES run requires high resolution of ~100m or less, my simulation is not really true LES, i think it can be called 'VLES'(I saw from a literature). So if a i want to start a LES in 200m resolution(VLES), it is recommended to follow LES running procedures(using ndown.exe), or common wrf running steps is also advisable?
Click to expand...
If I may interject with a quick thought: Have you considered adding one more sub-kilometer nest in between 1 km and 200 m as a transitional LES run? This paper does something similar and might be a good reference.
 
One way to think about ndown: it's not really a requirement for LES run—it's more of a practical workaround for when running all nests at once gets too heavy. If a single coupled run fits within your resources, there's probably no need to complicate things with ndown.
 
Hans said:
One way to think about ndown: it's not really a requirement for LES run—it's more of a practical workaround for when running all nests at once gets too heavy. If a single coupled run fits within your resources, there's probably no need to complicate things with ndown.
Click to expand...
Am I correct in understanding that, if computational resources are sufficient, it is not necessary to use `ndown` when running LES, and that `ndown` is essentially a compromise made due to computational constraints? In that case, if I set up the physical and dynamical parameterization schemes (for example, by selecting `bl_pbl_physics = 0` for the LES domain) as shown in my attached `namelist.input` file, and then follow the steps for running the standard WRF mode (i.e., './real.exe' and then './wrf.exe'), can the results obtained from this run be considered an LES simulation?
 

Attachments

Michael Zhang said:
Am I correct in understanding that, if computational resources are sufficient, it is not necessary to use `ndown` when running LES, and that `ndown` is essentially a compromise made due to computational constraints? In that case, if I set up the physical and dynamical parameterization schemes (for example, by selecting `bl_pbl_physics = 0` for the LES domain) as shown in my attached `namelist.input` file, and then follow the steps for running the standard WRF mode (i.e., './real.exe' and then './wrf.exe'), can the results obtained from this run be considered an LES simulation?
Click to expand...
My understanding is indeed the same. Furthermore, LES run should not be used as a substitute for PBL parameterization until the grid spacing reaches the mesoscale limit. Here is an introduction to Additional PBL Options.
 
Last edited:
You must log in or register to reply here.
Back
Top