Two-way and one-way nesting runs

[JiaWang]

WRF has two-way and one-way nesting.
My understanding is that, for D01, the two-way nesting with "feedback=0" should be equivalent to the one-way nesting.
But when I run two experiments, two-way nesting with feedback=0 and one-way nesting, using the same fixed time step and same other things, the results for D01 are different.
Anyone has any thoughts?

 

[kwerner]

I'm a bit confused by the question. Two-way nesting is when feedback=1, and one-way nesting is when feedback=0. It may just be that you mis-typed in the question, but can you clarify in a bit more detail the tests you're comparing? Thanks!

 

[JiaWang]

Sorry for the confusion.

For my experiments, I have 2 domains, D01 (parent) and D02 (nest). And I have done two experiments:
1. Set "max_dom=2", and "feedback=0", and run D01 and D02 concurrently
2. Set "max_dom=1", and run D01 alone.
Both experiments use the same parameterizations, and the same fixed time step.

I expect, for D01, the experiment 1 and 2 would yield the exactly same results. But it turns out that the results are different. I don't understand.

 

[kwerner]

Thank you. I understand. Can you attach the namelist(s) you use to run these two tests?

 

[JiaWang]

See the attached for the namelist. The only difference in the two experiments is the max_dom.

 

[kwerner]

Hi,
I ran a test and see the same results as you. I thought I remembered something about this in the past, so I dug through my notes and found that while, in theory, we would expect the results to be identical, but when running a nested simulation (even with feedback off) the process of the model building the inner domain causes noise that overwrites the variables in a relatively small way. To keep a pristine d01, you would probably need to use the ndown program to run d02, which may not be beneficial for your particular testing purpose.