nesting blow up

[simion1232006]

I have experienced blowing up problem with ROMS nesting scheme for my application. Since the real case is complicated and hard to tell , I cleaned up all unnecessary issues by turning off the surface heat flux and wind, making the refined grid exactly the same as the master grid (the same masking and bathymetry, only 3 times finer). I set all velocity components to zero for both initial and boundary conditions. After around 50 time step, the model blow up on the fine grid boundary. I attached the surface velocity results at final time step .
I have also tested individual simulations for both grid, and they are fine to run with no blowing up problems. So both grids should be smooth enough (maximal rx0 =0.2).I just don't understand why I have no problems running individual simulation for each grids, while have problems when doing nesting.

I know the nesting scheme is still in debugging, and it might be a potential issue in the code.I don't know what I missed. Is there anyone having the similar experiences? Put the question more specific, what is the boundary condition (Nes) mean in the input file?

LBC(isFsur) == Cha Clo Cha Cha \ ! free-surface
Nes Nes Nes Nes
LBC(isUbar) == Fla Clo Fla Fla \ ! 2D U-momentum
Nes Nes Nes Nes
LBC(isVbar) == Fla Clo Fla Fla \ ! 2D V-momentum
Nes Nes Nes Nes
LBC(isUvel) == Rad Clo Rad Rad \ ! 3D U-momentum
Nes Nes Nes Nes
LBC(isVvel) == Rad Clo Rad Rad \ ! 3D V-momentum
Nes Nes Nes Nes
LBC(isMtke) == Gra Clo Gra Gra \ ! mixing TKE
Nes Nes Nes Nes

LBC(isTvar) == RadNud Clo RadNud RadNud \ ! temperature
RadNud Clo RadNud RadNud \ ! salinity
Nes Nes Nes Nes \ ! temperature
Nes Nes Nes Nes ! salinity

[simion1232006]

I just found the radiation_2D flag should be switched on. Another test with this flag is running fine for 3000 time steps already. It could be the problem.

[simion1232006]

For the interests of anyone who working on ROMS nesting scheme, I feel the one_way nesting is more stable than the two_way nesting. I will dig into the codes to see their differences . while if someone know the reason, I really appreciate you can post them here.

[fdaryabor]

Did you check time step of the baroclinic and barotropic mode for the both domains. And the time intervals to save data in hisfile and avgfile? You should be careful apply the refinement coefficient for those.

Good Luck!

[simion1232006]

I have checked everything. Turning on the one_way nesting is so far the only solution.

[simion1232006]

On behalf of ROMS users, I need to say my poster here might be misleading. Prof Wilkin sent me a message about the potential volume conservative risk of one way nesting. So it might not be the best choice. The nesting scheme is still under active development, so will wait for the newest progress.

[Guotong_Deng]

I have met the same problem, have you solved this problem?