Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

4.1.

MODEL COORDINATES 77

uniform rectangular non−uniform rectangular

Figure 4.1: Example of a uniform (left) and non-uniform rectangular grid


(right).

In practice ∆ξ1 , ∆ξ2 are both normalised to 1, so that the metric coeffi-
cients h1 , h2 now become the grid spacings along the curvilinear coordinate
lines.
The following types of coordinate transformations are considered in CO-
HERENS:
• “Fully” curvilinear: h1 = h1 (ξ1 , ξ2 ) and h2 = h2 (ξ1 , ξ2 )
• Non-uniform rectangular: h1 = h1 (ξ1 ) and h2 = h2 (ξ2 ), i.e. ∂h1 /∂ξ2 =
∂h2 /∂ξ1 = 0
• Uniform rectangular: h1 and h2 are independent of ξ1 and ξ2 .
A computational model grid in the horizontal is constructed at the “grid
nodes” which are located at the orthogonal intersections of a series of coor-
dinate lines along which ξ1 is constant with coordinate lines along which ξ2
is constant. The boxes bounded by four neighbouring grid nodes are called
model “grid cells”. Figures 4.1 and 4.2 show examples of a uniform and
non-uniform rectangular grid, repectively a curvilinear grid. Although not
recommended, COHERENS offers the possibility to define model grids with
“ragged” boundaries, obtained by removing grid cells from the physical do-
main where the actual calculations are performed. An example is given in
Figure 4.3.
A model grid is defined in practice by supplying the coordinates of all
grid nodes. In case of a fully curvilinear grid, these coordinates need to be
provided by the user. For a rectangular grid, they are given by
x1 = xr
i−1
X
xi = xr + ∆x(k) for i = 2, . . . , Nx+1
k=1

You might also like