Here we explain the CANDIE A-gird, or more precisely, the CANDIE mixed A-grid and C-grid model configuration. This is perhaps best shown graphically in Fig. 3, which shows a horizontal cross section of the fluid grid cells. There are IO cells in the X-direction numbered left to right (west to east) I=1,2,..., I0, and J0 cells in the Y-direction numbered bottom to top (south to north) J=1,2,..., J0. The values IO and J0 are specified by the user.
The black dot at the center of each grid cell in Fig. 3 is a model A-grid point. The CANDIE array variables calculated at these points are represented by the pressure array P, and have array dimensions (IO, JO ) in the (X,Y) directions. The CANDIE scalar variables are calculated at these A-grid points. Some of these scalar variables are the pressure P(I0,J0,K0), surface pressure PS(I0,J0), temperatures T1(I0,J0,K0) and T2(I0,J0,K0), salinities S1(I0,J0,K0) and S2(I0,J0,K0), and density RHOP(I0,J0,K0), where K0 is the array dimension in the vertical Z-direction. The A-grid CANDIE vector variables calculated at these points are the X-direction velocities U1(I0,J0,K0) and U2(I0,J0,K0), and the Y-direction velocities V1(I0,J0,K0) and V2(I0,J0,K0). The vertical velocity W(I0,J0,K0) is also calculated at the A-grid points but at a different level in the vertical. The model forcings such as wind stresses TAUX(I0,J0) and TAUY(I0,J0), and the heat and salinity fluxes are also applied at the surface A-grid points.
The Fig. 3 also shows the C-grid advective velocities used for the conservation equations, used in calculating fluxes into and out of the cell volume. The X-direction U advective velocity point is marked by a cross, and the Y-direction V advective velocity point is marked by a square. These C-grid advective velocity arrays have different dimensions from the A-grid variable arrays. If we denote IP= I0+1 and JP= J0+1, then we have U(IP,J0,K0) and V(I0,JP,K0).
The CANDIE model uses an elliptic solver for the surface pressure, so that the model boundaries require special attention. The grid cells at the boundary must be either land cells, or if at an open boundary the open boundary conditions must be specified there. The one grid cell border all around the domain is referred to as the ``ghost zone", and is marked as the shadded region in Fig. 3. Of course, land points can also be in the interior of the domain, and the model can also accomodate islands within the domain. A discussion of how to set up the grid will be given later when discussing the prep.F file and model preparations.