Rules of thumb for rimwidth and halo for LBCs?

Atmosphere Unified Model
,
1

Just wondering if rim widths and halo regions are something that should be fiddled with as resolution changes, and if so, if there are any rules of thumb?

For context: @Paul.Gregory and I have started off with a very small 50x50 grid point test domain. With rim widths of 9 and halos of 7 (the default values inherited from the Lismore example’s first nested grid) mean that over a third of the domain is in the rim region, and we can only run on a max of 2x2 PEs because of the large halo + rim vs domain size

2

Hi @ashjbarnes, the size of the rim widths and halo regions are scientific details. If you go through the science papers they will describe the settings. The only “rule of thumb” one is when making an ancillary that is larger than your domain to cut-out the ERA5 data before processing into lateral boundary conditions.

I can’t remember the exact paper reference but let me know if you need one and I will get back to you.

3

You may want to talk to people who are more across the running of the UM in idealised mode.

4

Thanks Chermelle! Yes if you could send me a relevant paper that would be handy

What do you mean by “in idealised mode”? Does that apply to a GAL setup of rAM3?

5

I have seen talks in the UK about running the UM in an idealised mode. If you are interested in that @bethanwhite might be able to help you or put you in touch with people who are in the know.

6

@ashjbarnes The reference for the scientific definition of the RIM settings is in the RAL1 paper section 2.3.

7

Sorry I still don’t quite understand how idealised mode comes into this?

8

If you desperately want to run with a non-standard rim+halo you might be able to get away with it in a different mode – a.k.a. maybe use something idealised.

That is the extent of my knowledge you will have to reach out to the scientists to investigate.

9

ok that partly answers my question: if deviating from the settings defined already in rAM3 is a non-standard thing to do then we’d best not mess with it! I wasn’t sure if it was a setting that users were encouraged to mess with themselves or not.

@MartinDix indicated that since our test domain was so tiny we could maybe get away with changing rim widths, but if they’re set as they are for a reason and it’s not normal practice to change this with resolution then we won’t pursue it

10

I will leave this request up for a bit just to see if anyone else wants to respond, otherwise are you happy for me to close it?

12

I think this is the solution.

I’m currently trying to find out exactly who at the Met Office is running idealised configurations (if any) and what type, but I highly suspect they will be using non standard inputs etc. Even if changing these settings was something that was done in idealised setups, I suspect it wouldn’t be directly applicable here.

13

Yes thanks, sorry for poorly phrased question!