Nudging values and OBCFAC clarification

[lcbernardo]

As the nudging parameters seem to be important for my current modeling runs, I've been looking at this more closely.

!------------------------------------------------------------------------------
! Nudging/relaxation time scales, inverse scales will be computed internally.
!------------------------------------------------------------------------------
!
! When passive/active open boundary conditions are activated, these nudging
! values correspond to the passive (outflow) nudging time scales.
!
! TNUDG Nudging time scale (days) for active tracer variables.
! (1:NAT+NPT,1:Ngrids) values are expected.
!
! ZNUDG Nudging time scale (days) for free-surface.
!
! M2NUDG Nudging time scale (days) for 2D momentum.
!
! M3NUDG Nudging time scale (days) for 3D momentum.
!
! OBCFAC Factor between passive (outflow) and active (inflow) open
! boundary conditions. The nudging time scales for the
! active (inflow) conditions are obtained by multiplying
! the passive values by OBCFAC. If OBCFAC > 1, nudging on
! inflow is stronger than on outflow (recommended).

So according to this, the values I input for the NUDG parameters correspond to the outflow conditions. I may just truly be lacking in common sense, but will I not get the active (inflow) conditions by dividing by rather than multiplying the passive (outflow) values by OBCFAC?

[kate]

Yes. But in actual use, there's an inverse, so maybe the code multiplies down in the guts.

[lcbernardo]

Thank you for your reply. I see. I guess that's what "inverse scales will be computed internally" implies.

Anyway, even so, I'd like to verify using a concrete example.

If I have TNUDG == 30.0d0 30.0d0, and my OBCFAC = 30.0d0, does this mean that my outflow (tracer) nudging time scale for is 30 days, and my inflow nudging time scale is 1 day? And larger values of OBCFAC would then correspond to a nudging time scale of less than a day?

[kate]

Yes, exactly.

[lcbernardo]

Thanks for the clarification Kate! Nevertheless, determining the most appropriate values for my application is turning out to be quite a challenge.

[kate]

Well, it is a kludge and you might need stronger measures than simply RadNud with the optimal values to get good behavior. One thing that I have found to help is more frequent boundary values to nudge to, such as 5-daily SODA or even daily HYCOM. Monthly never worked well.

[lcbernardo]

Thanks again for your thoughts. Since things haven't been really working well for me, I've been trying to nudge to a climatology now, in addition to using daily HYCOM data as lateral boundary conditions. I've had the best results so far when I set my *NUDG==1.0d0 and OBCFAC=100.0d0. And since I don't use a separate nudging coefficient file, I defining ANA_NUDGCOEF results in the same values being used for the climatological nudging. Might this be considered a relatively "strong" measure? I'm concerned though, after looking at your ARCTIC case where you used larger values for the outgoing nudging coefficients, that my approach is too "forced" and that maybe there are better ways to get my model to behave properly.