[MITgcm-support] RBCS relaxation time scale limits

Martin Losch Martin.Losch at awi.de
Sun Mar 11 14:32:20 EDT 2018


Hi Dhruv,

if you ignore other sources for your tracer, you’ll have from discretizing d Tracer/ dt  = 1/tauRelax * ( Tracer_r - Tracer ) in time (it’s more complicated, but just conceptually):
Tracer(n+1) = Tracer(n) + dt/tauRelax * ( Tracer_r - Tracer(n) )
for tauRelax = deltaT, this means Tracer(n+1) = Tracer_r, ie you replace your tracer with the restoring value. A shorter restoring time scale doesn’t make too much sense to me. And if you use the above logic for tauRelax << deltaT then (deltaT/tauRelax>>1) and I am not quite sure what you would be getting. Maybe your advection scheme limits the effect of the restoring after you get funny values from deltaT/tauRelax >> 1

I suggest that you don’t use tauRelax<deltaT.

Martin

> On 10. Mar 2018, at 21:27, Dhruv Balwada <db194 at nyu.edu> wrote:
> 
> Hello, 
> 
> I am trying to do some surface tracer restoration experiments with relaxation time scales that are very fast (1s), extreme limit. This number is smaller than the time step in my experiment. I use the RBCS package for this. I am noticing that if the restoration time scale is smaller than the time step, then no tracer enters the domain. However, things work fine/as expected if the restoration time scale is larger than or equal to the time step. Is there some reason why this is the case? The restoration time scale seems to enter as a constant, not sure why it is tied to the numerics.
> 
> Thanks,
> Dhruv
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support



More information about the MITgcm-support mailing list