(RESOLVED) Domain setup for overlapping inner domains over large area

[userwrfch]

Hi,

I would like to use WRF for a coastal mountain range in western Canada/Alaska (red frame in the screenshot). I experimented with setting up WRF using multiple inner domains (black frames). They have to be as small as possible because I will do an expensive long-time run. The input data is on 30 km resolution and i am using a 1:3 ratio with lambert projection. It is important that I set up the inner domains such that I have data for all areas of the mountain range.

Can you please help me how to best set up the domains such that I have only the necessary overlap? Is it enough to discard 5 grid cells in each direction for the inner domains (and hence, the overlapping region should be 10 grid cells)?

Here is an example: I calculated: 24 (j_start d02) + (82-1)/3 - 10 = 41 (j_start d03)

parent_id = 1, 1, 1,
parent_grid_ratio = 1, 3, 3,
i_parent_start = 1, 52, 45,
j_parent_start = 1, 24, 41,
e_we = 111, 151, 94,
e_sn = 111, 82, 121,

Can you please let me know if this is the preferred way to set up the domains (I briefly looked into rotated lat-lon as well)? Thank you very much!

 

[kwerner]

Hi,
This is probably not the way you will want to set-up the domain. Since your highest resolution is only 3.33 km, this is not a very expensive set-up. You should simply just have a single inner-most domain to cover the entire area you're interested in. It really does not save that much computation to try to break the domains into smaller segments. You'll want to make sure to keep the boundaries of that domain away from the mountain ranges, and allow enough space between what you're interested in, and the edge of that domain. I would suggest looking at this page for suggestions on setting up a reasonable domain. You can also take a look at this presentation on nesting, which may be useful. That is the full recorded presentation, but if you only want to view the slides, you can find them here.

 

[userwrfch]

Okay, thank you very much!