Domain design: the ratio among domain-resolutions

I feel confused about the domains.
Here are two configurations listed below: the differences between these two are domain-ratios: the regular one is designed with the domain-ratio of 3 containing 5 domains, while another is designed with a domain-ratio of 9 including3 domains.

My questions are: is the ratio of 3 among domain-resolutions necessary for WRF design or the ratio of 9 is also fine (where can I find the information talking about the ratio among domains)? Are there some differences in model results between these two options listed below?

Regular options:
max_dom = 5,
e_we = 222, 334, 667, 502, 301,
e_sn = 222, 334, 667, 502, 301,
dx = 27000, 9000, 3000, 1000, 333.3,
dy = 27000, 9000, 3000, 1000, 333.3,
grid_id = 1, 2, 3, 4, 5,
parent_id = 1, 1, 2, 3, 4,

Another options:

max_dom = 3,
e_we = 222, 667, 301,
e_sn = 222, 667, 301,
dx = 27000, 3000, 333.3,
dy = 27000, 3000, 333.3,
grid_id = 1, 2, 3,
parent_id = 1, 1, 2,


Thanks for your attention.
All the best!
Nannan

Hi Nannan,
We recommend using odd ratios - typically 3:1 or 5:1, and we say that you should keep it at least less than 7:1, if possible. As the model is having to interpolate down from one domain to the next, and the ratio is a much larger "jump" with a 9:1 ratio, we do expect that the results would differ some. You're welcome to do a short test to see how much difference the results make for your case.

You can find some information on nesting recommendations from these 2 pages:
http://www2.mmm.ucar.edu/wrf/users/namelist_best_prac_wps.html
http://www2.mmm.ucar.edu/wrf/users/tutorial/201807/werner_nesting.pdf

kwerner said:

Hi Nannan,
We recommend using odd ratios - typically 3:1 or 5:1, and we say that you should keep it at least less than 7:1, if possible. As the model is having to interpolate down from one domain to the next, and the ratio is a much larger "jump" with a 9:1 ratio, we do expect that the results would differ some. You're welcome to do a short test to see how much difference the results make for your case.

You can find some information on nesting recommendations from these 2 pages:
http://www2.mmm.ucar.edu/wrf/users/namelist_best_prac_wps.html
http://www2.mmm.ucar.edu/wrf/users/tutorial/201807/werner_nesting.pdf

Click to expand...

Hi Kelly,

Thanks for your kind reply and useful links.
Yes, odd ratios (3, 5, 7, 9...) are recommended in WRF. I still have no idea about why radio of 9 is not expected to use if it can save a lot computing time. What does " the ratio is a much larger "jump" with a 9:1 ratio" exactly mean? Is it a conventional rule according to WRF-ARW core?
Thanks again.

All the best!
Nannan

Hi Nannan,
I apologize for being unclear. Saying that a larger ratio creates a larger "jump" I simply mean that there is a larger ratio to interpolate down to from the coarse parent grid, to the child. In other words, it is more difficult to get a reasonable high-resolution result when you're interpolating from much more coarse resolution data. As I mentioned before, if you would like to try to use a 9:1 ratio, then I would recommend doing a small test to see if the results look reasonable. If so, then it is probably okay to use.

kwerner said:

Hi Nannan,
I apologize for being unclear. Saying that a larger ratio creates a larger "jump" I simply mean that there is a larger ratio to interpolate down to from the coarse parent grid, to the child. In other words, it is more difficult to get a reasonable high-resolution result when you're interpolating from much more coarse resolution data. As I mentioned before, if you would like to try to use a 9:1 ratio, then I would recommend doing a small test to see if the results look reasonable. If so, then it is probably okay to use.

Click to expand...

Hi Kelly,
Thanks for your reply, and I will do tests on that and lets see what happens.
Best!
Nannan