question on SOUTH_M2FLATHER and NORTH_M2FLATHER

[longmtm]

Dear all

Can some one explain the following section of the code Nonlinear/u2dbc_im.F ?
Link: https://www.myroms.org/svn/src/trunk/RO ... u2dbc_im.F

Code: Select all

# elif defined SOUTH_M2FLATHER || defined SOUTH_M2REDUCED
!
!  Southern edge, Chapman boundary condition.
!
        DO i=IstrU,Iend
          cff=dt2d*0.5_r8*(GRID(ng)%pn(i-1,Jstr)+                       &
     &                     GRID(ng)%pn(i  ,Jstr))
          cff1=SQRT(g*0.5_r8*(GRID(ng)%h(i-1,Jstr)+                     &
     &                        zeta(i-1,Jstr,know)+                      &
     &                        GRID(ng)%h(i  ,Jstr)+                     &
     &                        zeta(i  ,Jstr,know)))
          Ce=cff*cff1
          cff2=1.0_r8/(1.0_r8+Ce)
          ubar(i,Jstr-1,kout)=cff2*(ubar(i,Jstr-1,know)+                &
     &                              Ce*ubar(i,Jstr,kout))
#  ifdef MASKING
          ubar(i,Jstr-1,kout)=ubar(i,Jstr-1,kout)*                      &
     &                        GRID(ng)%umask(i,Jstr-1)
#  endif
        END DO

# elif defined SOUTH_M2CLAMPED

My question is: the CPP clearly says SOUTH_M2FLATHER, however the comments say it is
Chapman boundary condition. Does that mean no FLATHER condition is implemented for the southern boundary for ubar ? Or does it mean FLATHER condition is equivalent to Chapman condition for the southern edge for ubar?

Same is happening for the NORTH_M2FLATHER condition.


Thanks.

Wen

[wilkin]

The "Chapman" option is for sea level (zeta), and "Flather" option is for depth-average velocity. The comment in the code is misleading but the code is correct.

There is no Chapman for ubar, nor is there Flather for zeta.

[arango]

Maybe the label is misleading but it is done on purpose. The Flather boundary conditions or reduced physics boundary conditions are applied only to the normal component of the velocity. Then, a Chapman boundary condition is applied to the tangential component even if the xxx_M2FLATHER and xxx_M2REDUCED options are defined. Otherwise, we will need to have a set of options for the normal and tangential components for each boundary edge which complicates the logic. I think that this is documented somewhere...

There are ways to do 2D-radiation conditions but they are not implemented for ubar and vbar.

By the way, I will eliminate all the CPP options for all open boundary conditions when the nesting version of ROMS is released. I am now heavily working on this. We going to have logical switches instead, so users can select mixed boundary conditions per variable and boundary edge. This probably with rise some issues with beginners, but it is inevitable.

[longmtm]

Thanks a lot! Since a 2D radiation boundary for (ubar,vbar) is not yet implemented and nesting version of ROMS is not yet released, for now do you think EAST_M2RADIATION + EAST_M2NUDGING is better than EAST_M2FLATHER for nesting a model inside of an external model?

The reason that I'm asking is that we are nesting a Chesapeake Bay-Delaware Bay model in Ruoying He's MABGOM (tidally averaged results) using M2FLATHER + UV_TIDES (from ADCIRC). And we are running LTRANS to do particle tracking. Here is what I observe.

The MABGOM (outside model) gives OBC tidally (12.4 hr) averaged (ubar,vbar) for the inside model with the pink and red arrows indicating M2 boundary conditions for the inside model interpolated from the outside MABGOM model (black arrows):

http://131.118.211.30/macroms/2005/obc/ ... _00003.png

However, at the north-east corner of OBC, the inside model predicts the following picture (instantaneous ubar,vbar, no-time-averaging, black arrows for overall prediction, colored arrows for same predictions at boundary), with tides, M2FLATHER on, ADD_M2OBC on, UV_TIDES on:

http://131.118.211.30/macroms/wenlong-t ... _00057.png

We see the model velocity and particles race along the corner like a wall, which doesn't seem to be right. The red and pink arrows are model outputs at the boundary grids. Wonder whether this is due to the lack of FLATHER condition for tangential velocity.

We'll do multiple tests and inspect the closeness between prediction and forcing, and your comments are very much welcomed.