[Mitgcm-support] Re: Bottom boundary Ekman layer in tank experiment

[mitgcm-support at dev.mitgcm.org]

Hi Allen,

 Thats right. What I meant to write was 
"u=0 at w(0)" ( but I wrote "u=0 and w(0)" for some reason! ).
u=0 at w(0) is what you describe i.e. u(0) is non-zero and
u(-1/2)=0.

Chris
Allen Kuo wrote:
> 
> Hi Chris: For all my output files, the u velocity is nonzero at the bottom-most level
> (which I have been assuming is u(0) rather than u(1) and I think this is a correct assumption).
> So I was thinking there was an assumption that u(-1/2)=w(0)=0 in the model, rather than u(0)=0.
> I think this changes the analysis- can you verify this is or is not the case ? -Allen
> 
> Allen:
> 
>  d2u/dz2 is written in a finite volume like
> form
> 
>   1/vol_of_u_cell dk(Area_z viscAz du/dz )
> 
> where dk(phi) is phi(k)-phi(k+1)
> 
> I think the the inner term ( du/dz ) is discretised with u=0
> and w(0) to give a noslip boundary condition.
> 
> Chris
> 
> >
> > yeah, I think the noslip is on- maybe you don't have to come by- do you know the finite difference
> > approximation for d^2u/dz^2 at the bottomost (u,v) level ?
> >
> >
> >
> > .
> > .
> > .
> > .
> >
> > -------- u(2)
> >
> > ------------ w(2)
> >
> > -------- u(1)
> >
> > ------------ w(1)
> >
> > -------- u(0)    <-------------- what is d^2u/dz^2 here assuming no slip ?
> >
> > ------------ w(0)
> 
> Allen C Kuo
> Massachusetts Institute of Technology
> Dept of Earth Atmospheric & Planetary Sciences
> Room 54-1721 , 77 Massachusetts Ave
> Cambridge, MA 02139
> 
> http://rossby.mit.edu/~akuo
> 
> (617) 253-5935 (office)
> (617) 253-3585 (lab)
> (617) 253 6208 (fax)