Net Flux Through Open Boundaries

Message

rwdell · #1 Unread post by **rwdell** » Fri Aug 14, 2009 1:05 pm

Hi ROMS Community,

I'm a grad student getting started with ROMS, and I was wondering if anyone could help me with an idealized boundary-layer I'm trying to model. I'm looking at a thermal boundary layer over a sloping bottom, which causes a net flux up the slope. I would like the boundary layer flux to exit the domain through an open boundary, and be replaced by an equal net flux into the domain on the opposite side. However, I have tried many different configurations, and in all cases the boundary layer flux reaches the edge of the domain, then remains in the model domain, exits the boundary layer vertically, and forms a return flow in the interior, leaving zero net flux through any column. For a typical velocity profile snapshot, see:
http://web.mit.edu/~rwdell/Public/DDF2-04_vSnapshot.pdf

In order to force the boundary layer flux to leave the domain, I tried applying the option NORTH_M3CLAMPED (where the northern boundary is the upslope boundary), then specifying the following velocity profile on the boundary:

Code: Select all

# ifdef NORTH_M3OBC
        IF (NORTHERN_EDGE) THEN
	fac = 2*user(2)*user(4)*COS(user(1))/SIN(user(1))
         DO k=1,N(ng)
          DO i=IstrR,IendR
	BOUNDARY(ng)%v_north(i,k)= fac*EXP(-user(4)*                     &
        &			  (z_r(i, Jend, k)-z_r(i,Jend,1)))*  &
	&                         SIN(user(4)*(z_r(i, Jend, k)-z_r(i,Jend,1)))
          END DO
          DO i=Istr,IendR
      	    BOUNDARY(ng)%u_north(i,k)=0.0_r8
          END DO
         END DO
        END IF

The total flux in this velocity profile should be:
Flux = user(2)*cot(user(1))

The other boundary conditions that I used were:

Code: Select all

#define EW_PERIODIC
#define NORTH_FSCHAPMAN
#define NORTH_TRADIATION
#define SOUTH_M3RADIATION 
#define SOUTH_FSCHAPMAN
#define SOUTH_TRADIATION

When I examin the model output, the velocity profile in the northern-most column
of velocity grid points matches the above velocity for the first timestep. However, by timestep 2, all the velocities of the boundary condition has been shifted by the right amount to make no net flux through the boundary. This suggests to me that the problem might have something to do with the calculation of the barotropic mode (at the fast time step), but I really don't know. Why is ROMS so keen to prevent volume fluxes in and out of the domain? If anyone could help me with this problem, I would be very appreciative.

Thanks very much,
Rebecca