Hi,
1) What is the reason you cannot change the domains? With your current set-up, NCL will not even plot the domains because d01 and d02 are too close together. It's not reasonable to have them so close and you will not obtain reasonable results with that set-up. You can make d01 (much) larger so that d02 will be positioned in a safe place within d01. As I previously mentioned, you're not saving very much computationally by having a small outer domain. The majority of computation cost is on the higher-resolution domains.
2) Please see my post from Oct 4, and please let me know if this is an idealized simulation and which one, if so?
3) Please see my post from Oct 4, and once you modify the domains, if you're still having the T2 issue, please send a plot showing constant temperature (T2) at a few locations.