Problem with streaks in idealized WRF simulations

[jpaulsen]

Hi everyone,

After working a lot with the real WRF configurations, I am now in the process of setting up idealized WRF simulations.

I am using WRF v4.6.1 and compiled for the em_les case.

When running the simulation with two nested domains, I observe strong streaks in many of the 4D-output Variables (U, V, W, QKE, ...) propagating from the outflow towards the the inflow. The strength of the streaks decreases with increasing heights until they are gone by the height where the damping starts. Also, close to the ground, I observe that the streaks are exactly as large as the resolution of the inner domain.

I've been trying to mitigate this behaviour by changing boundary conditions, damping heights and coefficients, domain sizes,... but I couldnt get to the root of the issue. Attached, you can find the namelist I used for the current run with the input_sounding also included.

Thanks for any advice in advance.

Best regards
Johannes

 

[kwerner]

Hi Johannes,
I can't say for sure, but the issue may be related to the coarse grid resolution being used for this LES case. LES cases should use grid spacing up to about 100 m, and since the code for this idealized case is written with that in mind, I believe it could cause issues to use more coarse resolution.

 

[jpaulsen]

Hi,
I am not sure it stems from the coarse resolution, since with wind directions not aligned with the grid direction (i.e. not exactly westerly winds) these streaks appear to be significantly lower - roughly one order of magnitude. Also when simulating a single domain, this does not occur. I fear it might me an issue with the transport of information in and out of the inner domain.

Do you have any thoughts on this? Online I find different sources with conflicting information on what boundary conditions are best for simulation such an idealized case.

Best regards
Johannes

 

[kwerner]

Hi Johannes,
Apologies for the delay while I reached out to someone more expert on this subject. I mentioned to them my concern about grid spacing, but they didn't seem concerned about that. This was their response regarding the streaks:

The color scale doesn't give any indication of how large these perturbations are, but the LES case starts with low-level random theta perturbations, which are likely causing the streaks, if they aren't strong.

 

[jpaulsen]

Hi,

The streaks show ampltiudes of more than 0.5m/s with an average wind speed in u-direction of roughly 6m/s. For larger wind speeds, the amplitude gets smaller.

I will have a look in the potential temperature if I find the time. Thanks for the time you took.

In the meantime I compiled WRF 4.3.3 for the idealised case and here the streaks are not showing. Thus for now I plan on using this.

Best regards
Johannes