Lateral boundary conditions

Usually, a land mask is defined on the horizontal numerical grid. This mask is denoted by $a^z$ for T-points, $a^u$ for U-points and $a^v$ for V-points with $a^z$, $a^u$, and $a^v$ being integer fields. A T-point is either a land point ($a^z=0$) or a water point ($a^z>0$). All U- and V-points surrounding a land point are defined as closed boundary and masked out: $a^u=0$ and $a^v=0$. The velocities on such closed boundaries are always set to 0.

Open boundaries are defined by $a^z>1$ for T-points. Forced boundary points are marked by $a^z=2$ and passive boundary points by $a^z=3$. All other T-points are characterised by $a^z=1$. For velocity points, three different types are defined at the open boundaries. U-points are classified by $a^u=3$ if both the T-points east and west are open boundary points and by $a^u=2$ if one adjacent T-point is an open boundary point and the other an open water point with $a^z=1$. The same is carried out for V-points: They are classified by $a^v=3$ if both the T-points south and north are open boundary points and by $a^v=2$ if one adjacent T-point is an open boundary point and the other an open water point with $a^z=1$. U-points which are adjacent to T-points with $a^z=2$ and which are not denoted by $a^u=2$ or $a^u=3$ are the external U-points and are denoted by $a^u=4$. The same holds for V-points: Those which are adjacent to T-points with $a^z=2$ and which are not denoted by $a^v=2$ or $a^v=3$ are the external V-points and are denoted by $a^v=4$. For a simple example of grid point classification, see figure 9.

Figure 9: Classification of grid points for a simple $5 \times 5$ configuration ( $i_{\max}=j_{\max}=5$). On the locations for T-, U- and V-points, the values of $a^z$, $a^u$, and $a^v$, respectively, are written. The northern and eastern boundaries are closed and the western and southern boundaries are open and forced.
\begin{figure}\begin{center}
\fbox{
\begin{picture}(6.0,6.0)(-0.5,-0.5)
\dottedl...
...t(4.5,5){\makebox(0,0)[c]{\large\bf 0}}
\end{picture}}\end{center}\end{figure}

When the barotropic boundary forcing is carried out by means of prescribed surface elevations only, then the surface elevation $\zeta$ is prescribed in all T-points with $a^z=2$. For passive boundary conditions ($a^z=3$), where the curvature of the surface elevation is zero normal to the boundary, the surface slope is simply extrapolated to the boundary points. For a boundary point $(i,j)$ at the western boundary this results e.g. in the following calculation for the boundary point:

$\displaystyle \zeta_{i,j}=\zeta_{i+1,j}+(\zeta_{i+1,j}-\zeta_{i+2,j})=
2\zeta_{i+1,j}-\zeta_{i+2,j}.$ (48)