/[MITgcm]/manual/s_algorithm/text/spatial-discrete.tex

Diff of /manual/s_algorithm/text/spatial-discrete.tex

Parent Directory | Revision Log | View Revision Graph Revision Graph | View Patch Patch

-revision 1.1 by adcroft,
Wed Aug  8 16:15:21 2001 UTC
+revision 1.6 by adcroft,
Tue Sep 25 20:13:42 2001 UTC
 Line 3
  \section{Spatial discretization of the dynamical equations}
+ Spatial discretization is carried out using the finite volume
+ method. This amounts to a grid-point method (namely second-order
+ centered finite difference) in the fluid interior but allows
+ boundaries to intersect a regular grid allowing a more accurate
+ representation of the position of the boundary. We treat the
+ horizontal and veritical directions as seperable and differently.
+ \input{part2/notation}
+ \subsection{The finite volume method: finite volumes versus finite difference}
+ The finite volume method is used to discretize the equations in
+ space. The expression ``finite volume'' actually has two meanings; one
+ is the method of cut or instecting boundaries (shaved or lopped cells
+ in our terminology) and the other is non-linear interpolation methods
+ that can deal with non-smooth solutions such as shocks (i.e. flux
+ limiters for advection). Both make use of the integral form of the
+ conservation laws to which the {\it weak solution} is a solution on
+ each finite volume of (sub-domain). The weak solution can be
+ constructed outof piece-wise constant elements or be
+ differentiable. The differentiable equations can not be satisfied by
+ piece-wise constant functions.
+ As an example, the 1-D constant coefficient advection-diffusion
+ equation:
+ \begin{displaymath}
+ \partial_t \theta + \partial_x ( u \theta - \kappa \partial_x \theta ) = 0
+ \end{displaymath}
+ can be discretized by integrating over finite sub-domains, i.e.
+ the lengths $\Delta x_i$:
+ \begin{displaymath}
+ \Delta x \partial_t \theta + \delta_i ( F ) = 0
+ \end{displaymath}
+ is exact if $\theta(x)$ is peice-wise constant over the interval
+ $\Delta x_i$ or more generally if $\theta_i$ is defined as the average
+ over the interval $\Delta x_i$.
+ The flux, $F_{i-1/2}$, must be approximated:
+ \begin{displaymath}
+ F = u \overline{\theta} - \frac{\kappa}{\Delta x_c} \partial_i \theta
+ \end{displaymath}
+ and this is where truncation errors can enter the solution. The
+ method for obtaining $\overline{\theta}$ is unspecified and a wide
+ range of possibilities exist including centered and upwind
+ interpolation, polynomial fits based on the the volume average
+ definitions of quantities and non-linear interpolation such as
+ flux-limiters.
+ Choosing simple centered second-order interpolation and differencing
+ recovers the same ODE's resulting from finite differencing for the
+ interior of a fluid. Differences arise at boundaries where a boundary
+ is not positioned on a regular or smoothly varying grid. This method
+ is used to represent the topography using lopped cell, see
+ \cite{Adcroft98}. Subtle difference also appear in more than one
+ dimension away from boundaries. This happens because the each
+ direction is discretized independantly in the finite difference method
+ while the integrating over finite volume implicitly treats all
+ directions simultaneously. Illustration of this is given in
+ \cite{Adcroft02}.
  \subsection{C grid staggering of variables}
  \begin{figure}
  \centerline{ \resizebox{!}{2in}{ \includegraphics{part2/cgrid3d.eps}} }
- \label{fig-cgrid3d}
  \caption{Three dimensional staggering of velocity components. This
  facilitates the natural discretization of the continuity and tracer
  equations. }
+ \label{fig:cgrid3d}
  \end{figure}
+ The basic algorithm employed for stepping forward the momentum
+ equations is based on retaining non-divergence of the flow at all
+ times. This is most naturally done if the components of flow are
+ staggered in space in the form of an Arakawa C grid \cite{Arakawa70}.
+ Fig. \ref{fig:cgrid3d} shows the components of flow ($u$,$v$,$w$)
+ staggered in space such that the zonal component falls on the
+ interface between continiuty cells in the zonal direction. Similarly
+ for the meridional and vertical directions.  The continiuty cell is
+ synonymous with tracer cells (they are one and the same).
+ \subsection{Grid initialization and data}
+ Initialization of grid data is controlled by subroutine {\em
+ INI\_GRID} which in calls {\em INI\_VERTICAL\_GRID} to initialize the
+ vertical grid, and then either of {\em INI\_CARTESIAN\_GRID}, {\em
+ INI\_SPHERICAL\_POLAR\_GRID} or {\em INI\_CURV\-ILINEAR\_GRID} to
+ initialize the horizontal grid for cartesian, spherical-polar or
+ curvilinear coordinates respectively.
+ The reciprocals of all grid quantities are pre-calculated and this is
+ done in subroutine {\em INI\_MASKS\_ETC} which is called later by
+ subroutine {\em INITIALIZE\_FIXED}.
+ All grid descriptors are global arrays and stored in common blocks in
+ {\em GRID.h} and a generally declared as {\em \_RS}.
+ \fbox{ \begin{minipage}{4.75in}
+ {\em S/R INI\_GRID} ({\em model/src/ini\_grid.F})
+ {\em S/R INI\_MASKS\_ETC} ({\em model/src/ini\_masks\_etc.F})
+ grid data: ({\em model/inc/GRID.h})
+ \end{minipage} }
  \subsection{Horizontal grid}
  \begin{figure}
  \centerline{ \begin{tabular}{cc}
-   \resizebox{!}{2in}{ \includegraphics{part2/hgrid-Ac.eps}}
+   \raisebox{1.5in}{a)}\resizebox{!}{2in}{ \includegraphics{part2/hgrid-Ac.eps}}
- & \resizebox{!}{2in}{ \includegraphics{part2/hgrid-Az.eps}}
+ & \raisebox{1.5in}{b)}\resizebox{!}{2in}{ \includegraphics{part2/hgrid-Az.eps}}
  \\
-   \resizebox{!}{2in}{ \includegraphics{part2/hgrid-Au.eps}}
+   \raisebox{1.5in}{c)}\resizebox{!}{2in}{ \includegraphics{part2/hgrid-Au.eps}}
- & \resizebox{!}{2in}{ \includegraphics{part2/hgrid-Av.eps}}
+ & \raisebox{1.5in}{d)}\resizebox{!}{2in}{ \includegraphics{part2/hgrid-Av.eps}}
  \end{tabular} }
- \label{fig-hgrid}
+ \caption{
- \caption{Three dimensional staggering of velocity components. This
+ Staggering of horizontal grid descriptors (lengths and areas). The
- facilitates the natural discretization of the continuity and tracer
+ grid lines indicate the tracer cell boundaries and are the reference
- equations. }
+ grid for all panels. a) The area of a tracer cell, $A_c$, is bordered
+ by the lengths $\Delta x_g$ and $\Delta y_g$. b) The area of a
+ vorticity cell, $A_\zeta$, is bordered by the lengths $\Delta x_c$ and
+ $\Delta y_c$. c) The area of a u cell, $A_c$, is bordered by the
+ lengths $\Delta x_v$ and $\Delta y_f$. d) The area of a v cell, $A_c$,
+ is bordered by the lengths $\Delta x_f$ and $\Delta y_u$.}
+ \label{fig:hgrid}
  \end{figure}
+ The model domain is decomposed into tiles and within each tile a
+ quasi-regular grid is used. A tile is the basic unit of domain
+ decomposition for parallelization but may be used whether parallized
+ or not; see section \ref{sect:tiles} for more details. Although the
+ tiles may be patched together in an unstructured manner
+ (i.e. irregular or non-tessilating pattern), the interior of tiles is
+ a structered grid of quadrilateral cells. The horizontal coordinate
+ system is orthogonal curvilinear meaning we can not necessarily treat
+ the two horizontal directions as seperable. Instead, each cell in the
+ horizontal grid is described by the length of it's sides and it's
+ area.
+ The grid information is quite general and describes any of the
+ available coordinates systems, cartesian, spherical-polar or
+ curvilinear. All that is necessary to distinguish between the
+ coordinate systems is to initialize the grid data (discriptors)
+ appropriately.
+ In the following, we refer to the orientation of quantities on the
+ computational grid using geographic terminology such as points of the
+ compass.
+ \marginpar{Caution!}
+ This is purely for convenience but should note be confused
+ with the actual geographic orientation of model quantities.
+ Fig.~\ref{fig:hgrid}a shows the tracer cell (synonymous with the
+ continuity cell). The length of the southern edge, $\Delta x_g$,
+ western edge, $\Delta y_g$ and surface area, $A_c$, presented in the
+ vertical are stored in arrays {\bf DXg}, {\bf DYg} and {\bf rAc}.
+ \marginpar{$A_c$: {\bf rAc}}
+ \marginpar{$\Delta x_g$: {\bf DXg}}
+ \marginpar{$\Delta y_g$: {\bf DYg}}
+ The ``g'' suffix indicates that the lengths are along the defining
+ grid boundaries. The ``c'' suffix associates the quantity with the
+ cell centers. The quantities are staggered in space and the indexing
+ is such that {\bf DXg(i,j)} is positioned to the south of {\bf
+ rAc(i,j)} and {\bf DYg(i,j)} positioned to the west.
+ Fig.~\ref{fig:hgrid}b shows the vorticity cell. The length of the
+ southern edge, $\Delta x_c$, western edge, $\Delta y_c$ and surface
+ area, $A_\zeta$, presented in the vertical are stored in arrays {\bf
+ DXg}, {\bf DYg} and {\bf rAz}.
+ \marginpar{$A_\zeta$: {\bf rAz}}
+ \marginpar{$\Delta x_c$: {\bf DXc}}
+ \marginpar{$\Delta y_c$: {\bf DYc}}
+ The ``z'' suffix indicates that the lengths are measured between the
+ cell centers and the ``$\zeta$'' suffix associates points with the
+ vorticity points. The quantities are staggered in space and the
+ indexing is such that {\bf DXc(i,j)} is positioned to the north of
+ {\bf rAc(i,j)} and {\bf DYc(i,j)} positioned to the east.
+ Fig.~\ref{fig:hgrid}c shows the ``u'' or western (w) cell. The length of
+ the southern edge, $\Delta x_v$, eastern edge, $\Delta y_f$ and
+ surface area, $A_w$, presented in the vertical are stored in arrays
+ {\bf DXv}, {\bf DYf} and {\bf rAw}.
+ \marginpar{$A_w$: {\bf rAw}}
+ \marginpar{$\Delta x_v$: {\bf DXv}}
+ \marginpar{$\Delta y_f$: {\bf DYf}}
+ The ``v'' suffix indicates that the length is measured between the
+ v-points, the ``f'' suffix indicates that the length is measured
+ between the (tracer) cell faces and the ``w'' suffix associates points
+ with the u-points (w stands for west). The quantities are staggered in
+ space and the indexing is such that {\bf DXv(i,j)} is positioned to
+ the south of {\bf rAw(i,j)} and {\bf DYf(i,j)} positioned to the east.
+ Fig.~\ref{fig:hgrid}d shows the ``v'' or southern (s) cell. The length of
+ the northern edge, $\Delta x_f$, western edge, $\Delta y_u$ and
+ surface area, $A_s$, presented in the vertical are stored in arrays
+ {\bf DXf}, {\bf DYu} and {\bf rAs}.
+ \marginpar{$A_s$: {\bf rAs}}
+ \marginpar{$\Delta x_f$: {\bf DXf}}
+ \marginpar{$\Delta y_u$: {\bf DYu}}
+ The ``u'' suffix indicates that the length is measured between the
+ u-points, the ``f'' suffix indicates that the length is measured
+ between the (tracer) cell faces and the ``s'' suffix associates points
+ with the v-points (s stands for south). The quantities are staggered
+ in space and the indexing is such that {\bf DXf(i,j)} is positioned to
+ the north of {\bf rAs(i,j)} and {\bf DYu(i,j)} positioned to the west.
+ \fbox{ \begin{minipage}{4.75in}
+ {\em S/R INI\_CARTESIAN\_GRID} ({\em
+ model/src/ini\_cartesian\_grid.F})
+ {\em S/R INI\_SPHERICAL\_POLAR\_GRID} ({\em
+ model/src/ini\_spherical\_polar\_grid.F})
+ {\em S/R INI\_CURVILINEAR\_GRID} ({\em
+ model/src/ini\_curvilinear\_grid.F})
+ $A_c$, $A_\zeta$, $A_w$, $A_s$: {\bf rAc}, {\bf rAz}, {\bf rAw}, {\bf rAs}
+ ({\em GRID.h})
+ $\Delta x_g$, $\Delta y_g$: {\bf DXg}, {\bf DYg} ({\em GRID.h})
+ $\Delta x_c$, $\Delta y_c$: {\bf DXc}, {\bf DYc} ({\em GRID.h})
+ $\Delta x_f$, $\Delta y_f$: {\bf DXf}, {\bf DYf} ({\em GRID.h})
+ $\Delta x_v$, $\Delta y_u$: {\bf DXv}, {\bf DYu} ({\em GRID.h})
+ \end{minipage} }
+ \subsubsection{Reciprocals of horizontal grid descriptors}
+ %\marginpar{$A_c^{-1}$: {\bf RECIP\_rAc}}
+ %\marginpar{$A_\zeta^{-1}$: {\bf RECIP\_rAz}}
+ %\marginpar{$A_w^{-1}$: {\bf RECIP\_rAw}}
+ %\marginpar{$A_s^{-1}$: {\bf RECIP\_rAs}}
+ Lengths and areas appear in the denominator of expressions as much as
+ in the numerator. For efficiency and portability, we pre-calculate the
+ reciprocal of the horizontal grid quantities so that in-line divisions
+ can be avoided.
+ %\marginpar{$\Delta x_g^{-1}$: {\bf RECIP\_DXg}}
+ %\marginpar{$\Delta y_g^{-1}$: {\bf RECIP\_DYg}}
+ %\marginpar{$\Delta x_c^{-1}$: {\bf RECIP\_DXc}}
+ %\marginpar{$\Delta y_c^{-1}$: {\bf RECIP\_DYc}}
+ %\marginpar{$\Delta x_f^{-1}$: {\bf RECIP\_DXf}}
+ %\marginpar{$\Delta y_f^{-1}$: {\bf RECIP\_DYf}}
+ %\marginpar{$\Delta x_v^{-1}$: {\bf RECIP\_DXv}}
+ %\marginpar{$\Delta y_u^{-1}$: {\bf RECIP\_DYu}}
+ For each grid descriptor (array) there is a reciprocal named using the
+ prefix {\bf RECIP\_}. This doubles the amount of storage in {\em
+ GRID.h} but they are all only 2-D descriptors.
+ \fbox{ \begin{minipage}{4.75in}
+ {\em S/R INI\_MASKS\_ETC} ({\em
+ model/src/ini\_masks\_etc.F})
+ $A_c^{-1}$: {\bf RECIP\_Ac} ({\em GRID.h})
+ $A_\zeta^{-1}$: {\bf RECIP\_Az} ({\em GRID.h})
+ $A_w^{-1}$: {\bf RECIP\_Aw} ({\em GRID.h})
+ $A_s^{-1}$: {\bf RECIP\_As} ({\em GRID.h})
+ $\Delta x_g^{-1}$, $\Delta y_g^{-1}$: {\bf RECIP\_DXg}, {\bf RECIP\_DYg} ({\em GRID.h})
+ $\Delta x_c^{-1}$, $\Delta y_c^{-1}$: {\bf RECIP\_DXc}, {\bf RECIP\_DYc} ({\em GRID.h})
+ $\Delta x_f^{-1}$, $\Delta y_f^{-1}$: {\bf RECIP\_DXf}, {\bf RECIP\_DYf} ({\em GRID.h})
+ $\Delta x_v^{-1}$, $\Delta y_u^{-1}$: {\bf RECIP\_DXv}, {\bf RECIP\_DYu} ({\em GRID.h})
+ \end{minipage} }
+ \subsubsection{Cartesian coordinates}
+ Cartesian coordinates are selected when the logical flag {\bf
+ using\-Cartes\-ianGrid} in namelist {\em PARM04} is set to true. The grid
+ spacing can be set to uniform via scalars {\bf dXspacing} and {\bf
+ dYspacing} in namelist {\em PARM04} or to variable resolution by the
+ vectors {\bf DELX} and {\bf DELY}. Units are normally
+ meters. Non-dimensional coordinates can be used by interpretting the
+ gravitational constant as the Rayleigh number.
+ \subsubsection{Spherical-polar coordinates}
+ Spherical coordinates are selected when the logical flag {\bf
+ using\-Spherical\-PolarGrid} in namelist {\em PARM04} is set to true. The
+ grid spacing can be set to uniform via scalars {\bf dXspacing} and
+ {\bf dYspacing} in namelist {\em PARM04} or to variable resolution by
+ the vectors {\bf DELX} and {\bf DELY}. Units of these namelist
+ variables are alway degrees. The horizontal grid descriptors are
+ calculated from these namelist variables have units of meters.
+ \subsubsection{Curvilinear coordinates}
+ Curvilinear coordinates are selected when the logical flag {\bf
+ using\-Curvil\-inear\-Grid} in namelist {\em PARM04} is set to true. The
+ grid spacing can not be set via the namelist. Instead, the grid
+ descriptors are read from data files, one for each descriptor. As for
+ other grids, the horizontal grid descriptors have units of meters.
  \subsection{Vertical grid}
  \begin{figure}
  \centerline{ \begin{tabular}{cc}
-   \raisebox{4in}{a)}
+   \raisebox{4in}{a)} \resizebox{!}{4in}{
-   \resizebox{!}{4in}{ \includegraphics{part2/vgrid-cellcentered.eps}}
+   \includegraphics{part2/vgrid-cellcentered.eps}} & \raisebox{4in}{b)}
- &
+   \resizebox{!}{4in}{ \includegraphics{part2/vgrid-accurate.eps}}
-  \raisebox{4in}{b)}
-  \resizebox{!}{4in}{ \includegraphics{part2/vgrid-accurate.eps}}
  \end{tabular} }
- \label{fig-vgrid}
  \caption{Two versions of the vertical grid. a) The cell centered
  approach where the interface depths are specified and the tracer
  points centered in between the interfaces. b) The interface centered
  approach where tracer levels are specified and the w-interfaces are
  centered in between.}
+ \label{fig:vgrid}
  \end{figure}
+ As for the horizontal grid, we use the suffixes ``c'' and ``f'' to
+ indicates faces and centers. Fig.~\ref{fig:vgrid}a shows the default
+ vertical grid used by the model.
+ \marginpar{$\Delta r_f$: {\bf DRf}}
+ \marginpar{$\Delta r_c$: {\bf DRc}}
+ $\Delta r_f$ is the difference in $r$
+ (vertical coordinate) between the faces (i.e. $\Delta r_f \equiv -
+ \delta_k r$ where the minus sign appears due to the convention that the
+ surface layer has index $k=1$.).
+ The vertical grid is calculated in subroutine {\em
+ INI\_VERTICAL\_GRID} and specified via the vector {\bf DELR} in
+ namelist {\em PARM04}. The units of ``r'' are either meters or Pascals
+ depending on the isomorphism being used which in turn is dependent
+ only on the choise of equation of state.
+ There are alternative namelist vectors {\bf DELZ} and {\bf DELP} which
+ dictate whether z- or
+ \marginpar{Caution!}
+ p- coordinates are to be used but we intend to
+ phase this out since they are redundant.
+ The reciprocals $\Delta r_f^{-1}$ and $\Delta r_c^{-1}$ are
+ pre-calculated (also in subroutine {\em INI\_VERTICAL\_GRID}). All
+ vertical grid descriptors are stored in common blocks in {\em GRID.h}.
+ The above grid (Fig.~\ref{fig:vgrid}a) is known as the cell centered
+ approach because the tracer points are at cell centers; the cell
+ centers are mid-way between the cell interfaces. An alternative, the
+ vertex or interface centered approach, is shown in
+ Fig.~\ref{fig:vgrid}b. Here, the interior interfaces are positioned
+ mid-way between the tracer nodes (no longer cell centers). This
+ approach is formally more accurate for evaluation of hydrostatic
+ pressure and vertical advection but historically the cell centered
+ approach has been used. An alternative form of subroutine {\em
+ INI\_VERTICAL\_GRID} is used to select the interface centered approach
+ but no run time option is currently available.
+ \fbox{ \begin{minipage}{4.75in}
+ {\em S/R INI\_VERTICAL\_GRID} ({\em
+ model/src/ini\_vertical\_grid.F})
+ $\Delta r_f$: {\bf DRf} ({\em GRID.h})
+ $\Delta r_c$: {\bf DRc} ({\em GRID.h})
+ $\Delta r_f^{-1}$: {\bf RECIP\_DRf} ({\em GRID.h})
+ $\Delta r_c^{-1}$: {\bf RECIP\_DRc} ({\em GRID.h})
+ \end{minipage} }
+ \subsection{Topography: partially filled cells}
+ \begin{figure}
+ \centerline{
+ \resizebox{4.5in}{!}{\includegraphics{part2/vgrid-xz.eps}}
+ }
+ \caption{
+ A schematic of the x-r plane showing the location of the
+ non-dimensional fractions $h_c$ and $h_w$. The physical thickness of a
+ tracer cell is given by $h_c(i,j,k) \Delta r_f(k)$ and the physical
+ thickness of the open side is given by $h_w(i,j,k) \Delta r_f(k)$.}
+ \label{fig:hfacs}
+ \end{figure}
+ \cite{Adcroft97} presented two alternatives to the step-wise finite
+ difference representation of topography. The method is known to the
+ engineering community as {\em intersecting boundary method}. It
+ involves allowing the boundary to intersect a grid of cells thereby
+ modifying the shape of those cells intersected. We suggested allowing
+ the topgoraphy to take on a peice-wise linear representation (shaved
+ cells) or a simpler piecewise constant representaion (partial step).
+ Both show dramatic improvements in solution compared to the
+ traditional full step representation, the piece-wise linear being the
+ best. However, the storage requirements are excessive so the simpler
+ piece-wise constant or partial-step method is all that is currently
+ supported.
+ Fig.~\ref{fig:hfacs} shows a schematic of the x-r plane indicating how
+ the thickness of a level is determined at tracer and u points.
+ \marginpar{$h_c$: {\bf hFacC}}
+ \marginpar{$h_w$: {\bf hFacW}}
+ \marginpar{$h_s$: {\bf hFacS}}
+ The physical thickness of a tracer cell is given by $h_c(i,j,k) \Delta
+ r_f(k)$ and the physical thickness of the open side is given by
+ $h_w(i,j,k) \Delta r_f(k)$. Three 3-D discriptors $h_c$, $h_w$ and
+ $h_s$ are used to describe the geometry: {\bf hFacC}, {\bf hFacW} and
+ {\bf hFacS} respectively. These are calculated in subroutine {\em
+ INI\_MASKS\_ETC} along with there reciprocals {\bf RECIP\_hFacC}, {\bf
+ RECIP\_hFacW} and {\bf RECIP\_hFacS}.
+ The non-dimensional fractions (or h-facs as we call them) are
+ calculated from the model depth array and then processed to avoid tiny
+ volumes. The rule is that if a fraction is less than {\bf hFacMin}
+ then it is rounded to the nearer of $0$ or {\bf hFacMin} or if the
+ physical thickness is less than {\bf hFacMinDr} then it is similarly
+ rounded. The larger of the two methods is used when there is a
+ conflict. By setting {\bf hFacMinDr} equal to or larger than the
+ thinnest nominal layers, $\min{(\Delta z_f)}$, but setting {\bf
+ hFacMin} to some small fraction then the model will only lop thick
+ layers but retain stability based on the thinnest unlopped thickness;
+ $\min{(\Delta z_f,\mbox{\bf hFacMinDr})}$.
+ \fbox{ \begin{minipage}{4.75in}
+ {\em S/R INI\_MASKS\_ETC} ({\em model/src/ini\_masks\_etc.F})
+ $h_c$: {\bf hFacC} ({\em GRID.h})
+ $h_w$: {\bf hFacW} ({\em GRID.h})
- \subsection{Continuity and horizontal pressure gradient terms}
+ $h_s$: {\bf hFacS} ({\em GRID.h})
+ $h_c^{-1}$: {\bf RECIP\_hFacC} ({\em GRID.h})
+ $h_w^{-1}$: {\bf RECIP\_hFacW} ({\em GRID.h})
+ $h_s^{-1}$: {\bf RECIP\_hFacS} ({\em GRID.h})
+ \end{minipage} }
+ \section{Continuity and horizontal pressure gradient terms}
  The core algorithm is based on the ``C grid'' discretization of the
  continuity equation which can be summarized as:
  \begin{eqnarray}
- \partial_t u + \frac{1}{\Delta x_c} \delta_i \left. \frac{ \partial \Phi}{\partial r}\right|_{s} \eta + \frac{\epsilon_{nh}}{\Delta x_c} \delta_i \Phi_{nh}' & = & G_u - \frac{1}{\Delta x_c} \delta_i \Phi_h' \\
+ \partial_t u + \frac{1}{\Delta x_c} \delta_i \left. \frac{ \partial \Phi}{\partial r}\right|_{s} \eta + \frac{\epsilon_{nh}}{\Delta x_c} \delta_i \Phi_{nh}' & = & G_u - \frac{1}{\Delta x_c} \delta_i \Phi_h' \label{eq:discrete-momu} \\
- \partial_t v + \frac{1}{\Delta y_c} \delta_j \left. \frac{ \partial \Phi}{\partial r}\right|_{s} \eta + \frac{\epsilon_{nh}}{\Delta y_c} \delta_j \Phi_{nh}' & = & G_v - \frac{1}{\Delta y_c} \delta_j \Phi_h' \\
+ \partial_t v + \frac{1}{\Delta y_c} \delta_j \left. \frac{ \partial \Phi}{\partial r}\right|_{s} \eta + \frac{\epsilon_{nh}}{\Delta y_c} \delta_j \Phi_{nh}' & = & G_v - \frac{1}{\Delta y_c} \delta_j \Phi_h' \label{eq:discrete-momv} \\
- \epsilon_{nh} \left( \partial_t w + \frac{1}{\Delta r_c} \delta_k \Phi_{nh}' \right) & = & \epsilon_{nh} G_w + \overline{b}^k - \frac{1}{\Delta r_c} \delta_k \Phi_{h}' \\
+ \epsilon_{nh} \left( \partial_t w + \frac{1}{\Delta r_c} \delta_k \Phi_{nh}' \right) & = & \epsilon_{nh} G_w + \overline{b}^k - \frac{1}{\Delta r_c} \delta_k \Phi_{h}' \label{eq:discrete-momw} \\
  \delta_i \Delta y_g \Delta r_f h_w u +
  \delta_j \Delta x_g \Delta r_f h_s v +
  \delta_k {\cal A}_c w & = & {\cal A}_c \delta_k (P-E)_{r=0}
+ \label{eq:discrete-continuity}
  \end{eqnarray}
  where the continuity equation has been most naturally discretized by
  staggering the three components of velocity as shown in
-Line 77 
 A}_c$.  The factors $h_w$ and $h_s$ are
+Line 479 
 A}_c$.  The factors $h_w$ and $h_s$ are
  The last equation, the discrete continuity equation, can be summed in
  the vertical to yeild the free-surface equation:
  \begin{equation}
- {\cal A}_c \partial_t \eta + \delta_i \sum_k \Delta y_g \Delta r_f h_w u + \delta_j \sum_k \Delta x_g \Delta r_f h_s v =
+ {\cal A}_c \partial_t \eta + \delta_i \sum_k \Delta y_g \Delta r_f h_w
- {\cal A}_c(P-E)_{r=0}
+ u + \delta_j \sum_k \Delta x_g \Delta r_f h_s v = {\cal
+ A}_c(P-E)_{r=0} \label{eq:discrete-freesurface}
  \end{equation}
  The source term $P-E$ on the rhs of continuity accounts for the local
  addition of volume due to excess precipitation and run-off over
  evaporation and only enters the top-level of the {\em ocean} model.
- \subsection{Hydrostatic balance}
+ \section{Hydrostatic balance}
  The vertical momentum equation has the hydrostatic or
  quasi-hydrostatic balance on the right hand side. This discretization
-Line 98 
 discretized:
+Line 501 
 discretized:
  \begin{equation}
  \epsilon_{nh} \partial_t w
  + g \overline{\rho'}^k + \frac{1}{\Delta z} \delta_k \Phi_h' = \ldots
+ \label{eq:discrete_hydro_ocean}
  \end{equation}
  In the atmosphere, using p-coordinates, hydrostatic balance is
  discretized:
  \begin{equation}
  \overline{\theta'}^k + \frac{1}{\Delta \Pi} \delta_k \Phi_h' = 0
+ \label{eq:discrete_hydro_atmos}
  \end{equation}
  where $\Delta \Pi$ is the difference in Exner function between the
  pressure points. The non-hydrostatic equations are not available in
-Line 127 
 CALC\_PHI\_HYD}. Inside this routine, on
+Line 532 
 CALC\_PHI\_HYD}. Inside this routine, on
  atmospheric/oceanic form is selected based on the string variable {\bf
  buoyancyRelation}.
- \subsection{Flux-form momentum equations}
- The original finite volume model was based on the Eulerian flux form
- momentum equations. This is the default though the vector invariant
- form is optionally available (and recommended in some cases).
- The ``G's'' (our colloquial name for all terms on rhs!) are broken
- into the various advective, Coriolis, horizontal dissipation, vertical
- dissipation and metric forces:
- \marginpar{$G_u$: {\bf Gu} }
- \marginpar{$G_v$: {\bf Gv} }
- \marginpar{$G_w$: {\bf Gw} }
- \begin{eqnarray}
- G_u & = & G_u^{adv} + G_u^{cor} + G_u^{h-diss} + G_u^{v-diss} +
- G_u^{metric} + G_u^{nh-metric} \\
- G_v & = & G_v^{adv} + G_v^{cor} + G_v^{h-diss} + G_v^{v-diss} +
- G_v^{metric} + G_v^{nh-metric} \\
- G_w & = & G_w^{adv} + G_w^{cor} + G_w^{h-diss} + G_w^{v-diss} +
- G_w^{metric} + G_w^{nh-metric}
- \end{eqnarray}
- In the hydrostatic limit, $G_w=0$ and $\epsilon_{nh}=0$, reducing the
- vertical momentum to hydrostatic balance.
- These terms are calculated in routines called from subroutine {\em
- CALC\_MOM\_RHS} a collected into the global arrays {\bf Gu}, {\bf Gv},
- and {\bf Gw}.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R CALC\_MOM\_RHS} ({\em pkg/mom\_fluxform/calc\_mom\_rhs.F})
- $G_u$: {\bf Gu} ({\em DYNVARS.h})
- $G_v$: {\bf Gv} ({\em DYNVARS.h})
- $G_w$: {\bf Gw} ({\em DYNVARS.h})
- \end{minipage} }
- \subsubsection{Advection of momentum}
- The advective operator is second order accurate in space:
- \begin{eqnarray}
- {\cal A}_w \Delta r_f h_w G_u^{adv} & = &
-   \delta_i \overline{ U }^i \overline{ u }^i
- + \delta_j \overline{ V }^i \overline{ u }^j
- + \delta_k \overline{ W }^i \overline{ u }^k \\
- {\cal A}_s \Delta r_f h_s G_v^{adv} & = &
-   \delta_i \overline{ U }^j \overline{ v }^i
- + \delta_j \overline{ V }^j \overline{ v }^j
- + \delta_k \overline{ W }^j \overline{ v }^k \\
- {\cal A}_c \Delta r_c G_w^{adv} & = &
-   \delta_i \overline{ U }^k \overline{ w }^i
- + \delta_j \overline{ V }^k \overline{ w }^j
- + \delta_k \overline{ W }^k \overline{ w }^k \\
- \end{eqnarray}
- and because of the flux form does not contribute to the global budget
- of linear momentum. The quantities $U$, $V$ and $W$ are volume fluxes
- defined:
- \marginpar{$U$: {\bf uTrans} }
- \marginpar{$V$: {\bf vTrans} }
- \marginpar{$W$: {\bf rTrans} }
- \begin{eqnarray}
- U & = & \Delta y_g \Delta r_f h_w u \\
- V & = & \Delta x_g \Delta r_f h_s v \\
- W & = & {\cal A}_c w
- \end{eqnarray}
- The advection of momentum takes the same form as the advection of
- tracers but by a translated advective flow. Consequently, the
- conservation of second moments, derived for tracers later, applies to
- $u^2$ and $v^2$ and $w^2$ so that advection of momentum correctly
- conserves kinetic energy.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_ADV\_UU} ({\em mom\_u\_adv\_uu.F})
- {\em S/R MOM\_U\_ADV\_VU} ({\em mom\_u\_adv\_vu.F})
- {\em S/R MOM\_U\_ADV\_WU} ({\em mom\_u\_adv\_wu.F})
- {\em S/R MOM\_U\_ADV\_UV} ({\em mom\_u\_adv\_uv.F})
- {\em S/R MOM\_U\_ADV\_VV} ({\em mom\_u\_adv\_vv.F})
- {\em S/R MOM\_U\_ADV\_WV} ({\em mom\_u\_adv\_wv.F})
- $uu$, $uv$, $vu$, $vv$: {\bf aF} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Coriolis terms}
- The ``pure C grid'' Coriolis terms (i.e. in absence of C-D scheme) are
- discretized:
- \begin{eqnarray}
- {\cal A}_w \Delta r_f h_w G_u^{Cor} & = &
-   \overline{ f {\cal A}_c \Delta r_f h_c \overline{ v }^j }^i
- - \epsilon_{nh} \overline{ f' {\cal A}_c \Delta r_f h_c \overline{ w }^k }^i \\
- {\cal A}_s \Delta r_f h_s G_v^{Cor} & = &
- - \overline{ f {\cal A}_c \Delta r_f h_c \overline{ u }^i }^j \\
- {\cal A}_c \Delta r_c G_w^{Cor} & = &
-  \epsilon_{nh} \overline{ f' {\cal A}_c \Delta r_f h_c \overline{ u }^i }^k
- \end{eqnarray}
- where the Coriolis parameters $f$ and $f'$ are defined:
- \begin{eqnarray}
- f & = & 2 \Omega \sin{\phi} \\
- f' & = & 2 \Omega \cos{\phi}
- \end{eqnarray}
- when using spherical geometry, otherwise the $\beta$-plane definition is used:
- \begin{eqnarray}
- f & = & f_o + \beta y \\
- f' & = & 0
- \end{eqnarray}
- This discretization globally conserves kinetic energy. It should be
- noted that despite the use of this discretization in former
- publications, all calculations to date have used the following
- different discretization:
- \begin{eqnarray}
- G_u^{Cor} & = &
-   f_u \overline{ v }^{ji}
- - \epsilon_{nh} f_u' \overline{ w }^{ik} \\
- G_v^{Cor} & = &
- - f_v \overline{ u }^{ij} \\
- G_w^{Cor} & = &
-  \epsilon_{nh} f_w' \overline{ u }^{ik}
- \end{eqnarray}
- \marginpar{Need to change the default in code to match this}
- where the subscripts on $f$ and $f'$ indicate evaluation of the
- Coriolis parameters at the appropriate points in space. The above
- discretization does {\em not} conserve anything, especially energy. An
- option to recover this discretization has been retained for backward
- compatibility testing (set run-time logical {\bf
- useNonconservingCoriolis} to {\em true} which otherwise defaults to
- {\em false}).
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_CDSCHEME} ({\em mom\_cdscheme.F})
- {\em S/R MOM\_U\_CORIOLIS} ({\em mom\_u\_coriolis.F})
- {\em S/R MOM\_V\_CORIOLIS} ({\em mom\_v\_coriolis.F})
- $G_u^{Cor}$, $G_v^{Cor}$: {\bf cF} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Curvature metric terms}
- The most commonly used coordinate system on the sphere is the
- geographic system $(\lambda,\phi)$. The curvilinear nature of these
- coordinates on the sphere lead to some ``metric'' terms in the
- component momentum equations. Under the thin-atmosphere and
- hydrostatic approximations these terms are discretized:
- \begin{eqnarray}
- {\cal A}_w \Delta r_f h_w G_u^{metric} & = &
-   \overline{ \frac{ \overline{u}^i }{a} \tan{\phi} {\cal A}_c \Delta r_f h_c \overline{ v }^j }^i \\
- {\cal A}_s \Delta r_f h_s G_v^{metric} & = &
- - \overline{ \frac{ \overline{u}^i }{a} \tan{\phi} {\cal A}_c \Delta r_f h_c \overline{ u }^i }^j \\
- G_w^{metric} & = & 0
- \end{eqnarray}
- where $a$ is the radius of the planet (sphericity is assumed) or the
- radial distance of the particle (i.e. a function of height).  It is
- easy to see that this discretization satisfies all the properties of
- the discrete Coriolis terms since the metric factor $\frac{u}{a}
- \tan{\phi}$ can be viewed as a modification of the vertical Coriolis
- parameter: $f \rightarrow f+\frac{u}{a} \tan{\phi}$.
- However, as for the Coriolis terms, a non-energy conserving form has
- exclusively been used to date:
- \begin{eqnarray}
- G_u^{metric} & = & \frac{u \overline{v}^{ij} }{a} \tan{\phi} \\
- G_v^{metric} & = & \frac{ \overline{u}^{ij} \overline{u}^{ij}}{a} \tan{\phi}
- \end{eqnarray}
- where $\tan{\phi}$ is evaluated at the $u$ and $v$ points
- respectively.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_METRIC\_SPHERE} ({\em mom\_u\_metric\_sphere.F})
- {\em S/R MOM\_V\_METRIC\_SPHERE} ({\em mom\_v\_metric\_sphere.F})
- $G_u^{metric}$, $G_v^{metric}$: {\bf mT} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Non-hydrostatic metric terms}
- For the non-hydrostatic equations, dropping the thin-atmosphere
- approximation re-introduces metric terms involving $w$ and are
- required to conserve anglular momentum:
- \begin{eqnarray}
- {\cal A}_w \Delta r_f h_w G_u^{metric} & = &
- - \overline{ \frac{ \overline{u}^i \overline{w}^k }{a} {\cal A}_c \Delta r_f h_c }^i \\
- {\cal A}_s \Delta r_f h_s G_v^{metric} & = &
- - \overline{ \frac{ \overline{v}^j \overline{w}^k }{a} {\cal A}_c \Delta r_f h_c}^j \\
- {\cal A}_c \Delta r_c G_w^{metric} & = &
-   \overline{ \frac{ {\overline{u}^i}^2 + {\overline{v}^j}^2}{a} {\cal A}_c \Delta r_f h_c }^k
- \end{eqnarray}
- Because we are always consistent, even if consistently wrong, we have,
- in the past, used a different discretization in the model which is:
- \begin{eqnarray}
- G_u^{metric} & = &
- - \frac{u}{a} \overline{w}^{ik} \\
- G_v^{metric} & = &
- - \frac{v}{a} \overline{w}^{jk} \\
- G_w^{metric} & = &
-   \frac{1}{a} ( {\overline{u}^{ik}}^2 + {\overline{v}^{jk}}^2 )
- \end{eqnarray}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_METRIC\_NH} ({\em mom\_u\_metric\_nh.F})
- {\em S/R MOM\_V\_METRIC\_NH} ({\em mom\_v\_metric\_nh.F})
- $G_u^{metric}$, $G_v^{metric}$: {\bf mT} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Lateral dissipation}
- Historically, we have represented the SGS Reynolds stresses as simply
- down gradient momentum fluxes, ignoring constraints on the stress
- tensor such as symmetry.
- \begin{eqnarray}
- {\cal A}_w \Delta r_f h_w G_u^{h-diss} & = &
-   \delta_i  \Delta y_f \Delta r_f h_c \tau_{11}
- + \delta_j  \Delta x_v \Delta r_f h_\zeta \tau_{12} \\
- {\cal A}_s \Delta r_f h_s G_v^{h-diss} & = &
-   \delta_i  \Delta y_u \Delta r_f h_\zeta \tau_{21}
- + \delta_j  \Delta x_f \Delta r_f h_c \tau_{22}
- \end{eqnarray}
- \marginpar{Check signs of stress definitions}
- The lateral viscous stresses are discretized:
- \begin{eqnarray}
- \tau_{11} & = & A_h c_{11\Delta}(\phi) \frac{1}{\Delta x_f} \delta_i u
-                -A_4 c_{11\Delta^2}(\phi) \frac{1}{\Delta x_f} \delta_i \nabla^2 u \\
- \tau_{12} & = & A_h c_{12\Delta}(\phi) \frac{1}{\Delta y_u} \delta_j u
-                -A_4 c_{12\Delta^2}(\phi)\frac{1}{\Delta y_u} \delta_j \nabla^2 u \\
- \tau_{21} & = & A_h c_{21\Delta}(\phi) \frac{1}{\Delta x_v} \delta_i v
-                -A_4 c_{21\Delta^2}(\phi) \frac{1}{\Delta x_v} \delta_i \nabla^2 v \\
- \tau_{22} & = & A_h c_{22\Delta}(\phi) \frac{1}{\Delta y_f} \delta_j v
-                -A_4 c_{22\Delta^2}(\phi) \frac{1}{\Delta y_f} \delta_j \nabla^2 v
- \end{eqnarray}
- where the non-dimensional factors $c_{lm\Delta^n}(\phi), \{l,m,n\} \in
- \{1,2\}$ define the ``cosine'' scaling with latitude which can be
- applied in various ad-hoc ways. For instance, $c_{11\Delta} =
- c_{21\Delta} = (\cos{\phi})^{3/2}$, $c_{12\Delta}=c_{22\Delta}=0$ would
- represent the an-isotropic cosine scaling typically used on the
- ``lat-lon'' grid for Laplacian viscosity.
- \marginpar{Need to tidy up method for controlling this in code}
- It should be noted that dispite the ad-hoc nature of the scaling, some
- scaling must be done since on a lat-lon grid the converging meridians
- make it very unlikely that a stable viscosity parameter exists across
- the entire model domain.
- The Laplacian viscosity coefficient, $A_h$ ({\bf viscAh}), has units
- of $m^2 s^{-1}$. The bi-harmonic viscosity coefficient, $A_4$ ({\bf
- viscA4}), has units of $m^4 s^{-1}$.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_XVISCFLUX} ({\em mom\_u\_xviscflux.F})
- {\em S/R MOM\_U\_YVISCFLUX} ({\em mom\_u\_yviscflux.F})
- {\em S/R MOM\_V\_XVISCFLUX} ({\em mom\_v\_xviscflux.F})
- {\em S/R MOM\_V\_YVISCFLUX} ({\em mom\_v\_yviscflux.F})
- $\tau_{11}$, $\tau_{12}$, $\tau_{22}$, $\tau_{22}$: {\bf vF}, {\bf
- v4F} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- Two types of lateral boundary condition exist for the lateral viscous
- terms, no-slip and free-slip.
- The free-slip condition is most convenient to code since it is
- equivalent to zero-stress on boundaries. Simple masking of the stress
- components sets them to zero. The fractional open stress is properly
- handled using the lopped cells.
- The no-slip condition defines the normal gradient of a tangential flow
- such that the flow is zero on the boundary. Rather than modify the
- stresses by using complicated functions of the masks and ``ghost''
- points (see \cite{Adcroft+Marshall98}) we add the boundary stresses as
- an additional source term in cells next to solid boundaries. This has
- the advantage of being able to cope with ``thin walls'' and also makes
- the interior stress calculation (code) independent of the boundary
- conditions. The ``body'' force takes the form:
- \begin{eqnarray}
- G_u^{side-drag} & = &
- \frac{4}{\Delta z_f} \overline{ (1-h_\zeta) \frac{\Delta x_v}{\Delta y_u} }^j
- \left( A_h c_{12\Delta}(\phi) u - A_4 c_{12\Delta^2}(\phi) \nabla^2 u \right)
- \\
- G_v^{side-drag} & = &
- \frac{4}{\Delta z_f} \overline{ (1-h_\zeta) \frac{\Delta y_u}{\Delta x_v} }^i
- \left( A_h c_{21\Delta}(\phi) v - A_4 c_{21\Delta^2}(\phi) \nabla^2 v \right)
- \end{eqnarray}
- In fact, the above discretization is not quite complete because it
- assumes that the bathymetry at velocity points is deeper than at
- neighbouring vorticity points, e.g. $1-h_w < 1-h_\zeta$
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_SIDEDRAG} ({\em mom\_u\_sidedrag.F})
- {\em S/R MOM\_V\_SIDEDRAG} ({\em mom\_v\_sidedrag.F})
- $G_u^{side-drag}$, $G_v^{side-drag}$: {\bf vF} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Vertical dissipation}
- Vertical viscosity terms are discretized with only partial adherence
- to the variable grid lengths introduced by the finite volume
- formulation. This reduces the formal accuracy of these terms to just
- first order but only next to boundaries; exactly where other terms
- appear such as linar and quadratic bottom drag.
- \begin{eqnarray}
- G_u^{v-diss} & = &
- \frac{1}{\Delta r_f h_w} \delta_k \tau_{13} \\
- G_v^{v-diss} & = &
- \frac{1}{\Delta r_f h_s} \delta_k \tau_{23} \\
- G_w^{v-diss} & = & \epsilon_{nh}
- \frac{1}{\Delta r_f h_d} \delta_k \tau_{33}
- \end{eqnarray}
- represents the general discrete form of the vertical dissipation terms.
- In the interior the vertical stresses are discretized:
- \begin{eqnarray}
- \tau_{13} & = & A_v \frac{1}{\Delta r_c} \delta_k u \\
- \tau_{23} & = & A_v \frac{1}{\Delta r_c} \delta_k v \\
- \tau_{33} & = & A_v \frac{1}{\Delta r_f} \delta_k w
- \end{eqnarray}
- It should be noted that in the non-hydrostatic form, the stress tensor
- is even less consistent than for the hydrostatic (see Wazjowicz
- \cite{Waojz}). It is well known how to do this properly (see Griffies
- \cite{Griffies}) and is on the list of to-do's.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_RVISCLFUX} ({\em mom\_u\_rviscflux.F})
- {\em S/R MOM\_V\_RVISCLFUX} ({\em mom\_v\_rviscflux.F})
- $\tau_{13}$: {\bf urf} (local to {\em calc\_mom\_rhs.F})
- $\tau_{23}$: {\bf vrf} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- As for the lateral viscous terms, the free-slip condition is
- equivalent to simply setting the stress to zero on boundaries.  The
- no-slip condition is implemented as an additional term acting on top
- of the interior and free-slip stresses. Bottom drag represents
- additional friction, in addition to that imposed by the no-slip
- condition at the bottom. The drag is cast as a stress expressed as a
- linear or quadratic function of the mean flow in the layer above the
- topography:
- \begin{eqnarray}
- \tau_{13}^{bottom-drag} & = &
- \left(
-A_v \frac{1}{\Delta r_c}
- + r_b
- + C_d \sqrt{ \overline{2 KE}^i }
- \right) u \\
- \tau_{23}^{bottom-drag} & = &
- \left(
-A_v \frac{1}{\Delta r_c}
- + r_b
- + C_d \sqrt{ \overline{2 KE}^j }
- \right) v
- \end{eqnarray}
- where these terms are only evaluated immediately above topography.
- $r_b$ ({\bf bottomDragLinear}) has units of $m s^{-1}$ and a typical value
- of the order 0.0002 $m s^{-1}$. $C_d$ ({\bf bottomDragQuadratic}) is
- dimensionless with typical values in the range 0.001--0.003.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_BOTTOMDRAG} ({\em mom\_u\_bottomdrag.F})
- {\em S/R MOM\_V\_BOTTOMDRAG} ({\em mom\_v\_bottomdrag.F})
- $\tau_{13}^{bottom-drag}$, $\tau_{23}^{bottom-drag}$: {\bf vf} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsection{Tracer equations}
- The tracer equations are discretized consistantly with the continuity
- equation to facilitate conservation properties analogous to the
- continuum:
- \begin{equation}
- {\cal A}_c \Delta r_f h_c \partial_\theta
- + \delta_i U \overline{ \theta }^i
- + \delta_j V \overline{ \theta }^j
- + \delta_k W \overline{ \theta }^k
- = {\cal A}_c \Delta r_f h_c {\cal S}_\theta + \theta {\cal A}_c \delta_k (P-E)_{r=0}
- \end{equation}
- The quantities $U$, $V$ and $W$ are volume fluxes defined:
- \marginpar{$U$: {\bf uTrans} }
- \marginpar{$V$: {\bf vTrans} }
- \marginpar{$W$: {\bf rTrans} }
- \begin{eqnarray}
- U & = & \Delta y_g \Delta r_f h_w u \\
- V & = & \Delta x_g \Delta r_f h_s v \\
- W & = & {\cal A}_c w
- \end{eqnarray}
- ${\cal S}$ represents the ``parameterized'' SGS processes and
- physics associated with the tracer. For instance, potential
- temperature equation in the ocean has is forced by surface and
- partially penetrating heat fluxes:
- \begin{equation}
- {\cal A}_c \Delta r_f h_c {\cal S}_\theta = \frac{1}{c_p \rho_o} \delta_k {\cal A}_c {\cal Q}
- \end{equation}
- while the salt equation has no real sources, ${\cal S}=0$, which
- leaves just the $P-E$ term.
- The continuity equation can be recovered by setting ${\cal Q}=0$ and
- $\theta=1$. The term $\theta (P-E)_{r=0}$ is required to retain local
- conservation of $\theta$. Global conservation is not possible using
- the flux-form (as here) and a linearized free-surface
- (\cite{Griffies00,Campin02}).
- \subsection{Derivation of discrete energy conservation}
- These discrete equations conserve kinetic plus potential energy using the
- following definitions:
- \begin{equation}
- KE = \frac{1}{2} \left( \overline{ u^2 }^i + \overline{ v^2 }^j +
- \epsilon_{nh} \overline{ w^2 }^k \right)
- \end{equation}
- \subsection{Vector invariant momentum equations}
- The finite volume method lends itself to describing the continuity and
- tracer equations in curvilinear coordinate systems but the appearance
- of new metric terms in the flux-form momentum equations makes
- generalizing them far from elegant. The vector invariant form of the
- momentum equations are exactly that; invariant under coordinate
- transformations.
- The non-hydrostatic vector invariant equations read:
- \begin{equation}
- \partial_t \vec{v} + ( 2\vec{\Omega} + \vec{\zeta}) \wedge \vec{v}
- - b \hat{r}
- + \vec{\nabla} B = \vec{\nabla} \cdot \vec{\bf \tau}
- \end{equation}
- which describe motions in any orthogonal curvilinear coordinate
- system. Here, $B$ is the Bernoulli function and $\vec{\zeta}=\nabla
- \wedge \vec{v}$ is the vorticity vector. We can take advantage of the
- elegance of these equations when discretizing them and use the
- discrete definitions of the grad, curl and divergence operators to
- satisfy constraints. We can also consider the analogy to forming
- derived equations, such as the vorticity equation, and examine how the
- discretization can be adjusted to give suitable vorticity advection
- among other things.
- The underlying algorithm is the same as for the flux form
- equations. All that has changed is the contents of the ``G's''. For
- the time-being, only the hydrostatic terms have been coded but we will
- indicate the points where non-hydrostatic contributions will enter:
- \begin{eqnarray}
- G_u & = & G_u^{fv} + G_u^{\zeta_3 v} + G_u^{\zeta_2 w} + G_u^{\partial_x B}
- + G_u^{\partial_z \tau^x} + G_u^{h-dissip} + G_u^{v-dissip} \\
- G_v & = & G_v^{fu} + G_v^{\zeta_3 u} + G_v^{\zeta_1 w} + G_v^{\partial_y B}
- + G_v^{\partial_z \tau^y} + G_v^{h-dissip} + G_v^{v-dissip} \\
- G_w & = & G_w^{fu} + G_w^{\zeta_1 v} + G_w^{\zeta_2 u} + G_w^{\partial_z B}
- + G_w^{h-dissip} + G_w^{v-dissip}
- \end{eqnarray}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R CALC\_MOM\_RHS} ({\em pkg/mom\_vecinv/calc\_mom\_rhs.F})
- $G_u$: {\bf Gu} ({\em DYNVARS.h})
- $G_v$: {\bf Gv} ({\em DYNVARS.h})
- $G_w$: {\bf Gw} ({\em DYNVARS.h})
- \end{minipage} }
- \subsubsection{Relative vorticity}
- The vertical component of relative vorticity is explicitly calculated
- and use in the discretization. The particular form is crucial for
- numerical stablility; alternative definitions break the conservation
- properties of the discrete equations.
- Relative vorticity is defined:
- \begin{equation}
- \zeta_3 = \frac{\Gamma}{A_\zeta}
- = \frac{1}{{\cal A}_\zeta} ( \delta_i \Delta y_c v - \delta_j \Delta x_c u )
- \end{equation}
- where ${\cal A}_\zeta$ is the area of the vorticity cell presented in
- the vertical and $\Gamma$ is the circulation about that cell.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_CALC\_RELVORT3} ({\em mom\_vi\_calc\_relvort3.F})
- $\zeta_3$: {\bf vort3} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Kinetic energy}
- The kinetic energy, denoted $KE$, is defined:
- \begin{equation}
- KE = \frac{1}{2} ( \overline{ u^2 }^i + \overline{ v^2 }^j
- + \epsilon_{nh} \overline{ w^2 }^k )
- \end{equation}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_CALC\_KE} ({\em mom\_vi\_calc\_ke.F})
- $KE$: {\bf KE} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Coriolis terms}
- The potential enstrophy conserving form of the linear Coriolis terms
- are written:
- \begin{eqnarray}
- G_u^{fv} & = &
- \frac{1}{\Delta x_c}
- \overline{ \frac{f}{h_\zeta} }^j \overline{ \overline{ \Delta x_g h_s v }^j }^i \\
- G_v^{fu} & = & -
- \frac{1}{\Delta y_c}
- \overline{ \frac{f}{h_\zeta} }^i \overline{ \overline{ \Delta y_g h_w u }^i }^j
- \end{eqnarray}
- Here, the Coriolis parameter $f$ is defined at vorticity (corner)
- points.
- \marginpar{$f$: {\bf fCoriG}}
- \marginpar{$h_\zeta$: {\bf hFacZ}}
- The potential enstrophy conserving form of the non-linear Coriolis
- terms are written:
- \begin{eqnarray}
- G_u^{\zeta_3 v} & = &
- \frac{1}{\Delta x_c}
- \overline{ \frac{\zeta_3}{h_\zeta} }^j \overline{ \overline{ \Delta x_g h_s v }^j }^i \\
- G_v^{\zeta_3 u} & = & -
- \frac{1}{\Delta y_c}
- \overline{ \frac{\zeta_3}{h_\zeta} }^i \overline{ \overline{ \Delta y_g h_w u }^i }^j
- \end{eqnarray}
- \marginpar{$\zeta_3$: {\bf vort3}}
- The Coriolis terms can also be evaluated together and expressed in
- terms of absolute vorticity $f+\zeta_3$. The potential enstrophy
- conserving form using the absolute vorticity is written:
- \begin{eqnarray}
- G_u^{fv} + G_u^{\zeta_3 v} & = &
- \frac{1}{\Delta x_c}
- \overline{ \frac{f + \zeta_3}{h_\zeta} }^j \overline{ \overline{ \Delta x_g h_s v }^j }^i \\
- G_v^{fu} + G_v^{\zeta_3 u} & = & -
- \frac{1}{\Delta y_c}
- \overline{ \frac{f + \zeta_3}{h_\zeta} }^i \overline{ \overline{ \Delta y_g h_w u }^i }^j
- \end{eqnarray}
- \marginpar{Run-time control needs to be added for these options} The
- disctinction between using absolute vorticity or relative vorticity is
- useful when constructing higher order advection schemes; monotone
- advection of relative vorticity behaves differently to monotone
- advection of absolute vorticity. Currently the choice of
- relative/absolute vorticity, centered/upwind/high order advection is
- available only through commented subroutine calls.
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_CORIOLIS} ({\em mom\_vi\_coriolis.F})
- {\em S/R MOM\_VI\_U\_CORIOLIS} ({\em mom\_vi\_u\_coriolis.F})
- {\em S/R MOM\_VI\_V\_CORIOLIS} ({\em mom\_vi\_v\_coriolis.F})
- $G_u^{fv}$, $G_u^{\zeta_3 v}$: {\bf uCf} (local to {\em calc\_mom\_rhs.F})
- $G_v^{fu}$, $G_v^{\zeta_3 u}$: {\bf vCf} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Shear terms}
- The shear terms ($\zeta_2w$ and $\zeta_1w$) are are discretized to
- guarantee that no spurious generation of kinetic energy is possible;
- the horizontal gradient of Bernoulli function has to be consistent
- with the vertical advection of shear:
- \marginpar{N-H terms have not been tried!}
- \begin{eqnarray}
- G_u^{\zeta_2 w} & = &
- \frac{1}{ {\cal A}_w \Delta r_f h_w } \overline{
- \overline{ {\cal A}_c w }^i ( \delta_k u - \epsilon_{nh} \delta_j w )
- }^k \\
- G_v^{\zeta_1 w} & = &
- \frac{1}{ {\cal A}_s \Delta r_f h_s } \overline{
- \overline{ {\cal A}_c w }^i ( \delta_k u - \epsilon_{nh} \delta_j w )
- }^k
- \end{eqnarray}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_U\_VERTSHEAR} ({\em mom\_vi\_u\_vertshear.F})
- {\em S/R MOM\_VI\_V\_VERTSHEAR} ({\em mom\_vi\_v\_vertshear.F})
- $G_u^{\zeta_2 w}$: {\bf uCf} (local to {\em calc\_mom\_rhs.F})
- $G_v^{\zeta_1 w}$: {\bf vCf} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Gradient of Bernoulli function}
- \begin{eqnarray}
- G_u^{\partial_x B} & = &
- \frac{1}{\Delta x_c} \delta_i ( \phi' + KE ) \\
- G_v^{\partial_y B} & = &
- \frac{1}{\Delta x_y} \delta_j ( \phi' + KE )
- %G_w^{\partial_z B} & = &
- %\frac{1}{\Delta r_c} h_c \delta_k ( \phi' + KE )
- \end{eqnarray}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_U\_GRAD\_KE} ({\em mom\_vi\_u\_grad\_ke.F})
- {\em S/R MOM\_VI\_V\_GRAD\_KE} ({\em mom\_vi\_v\_grad\_ke.F})
- $G_u^{\partial_x KE}$: {\bf uCf} (local to {\em calc\_mom\_rhs.F})
- $G_v^{\partial_y KE}$: {\bf vCf} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Horizontal dissipation}
- The horizontal divergence, a complimentary quantity to relative
- vorticity, is used in parameterizing the Reynolds stresses and is
- discretized:
- \begin{equation}
- D = \frac{1}{{\cal A}_c h_c} (
-   \delta_i \Delta y_g h_w u
- + \delta_j \Delta x_g h_s v )
- \end{equation}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_CALC\_HDIV} ({\em mom\_vi\_calc\_hdiv.F})
- $D$: {\bf hDiv} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Horizontal dissipation}
- The following discretization of horizontal dissipation conserves
- potential vorticity (thickness weighted relative vorticity) and
- divergence and dissipates energy, enstrophy and divergence squared:
- \begin{eqnarray}
- G_u^{h-dissip} & = &
-   \frac{1}{\Delta x_c} \delta_i ( A_D D - A_{D4} D^*)
- - \frac{1}{\Delta y_u h_w} \delta_j h_\zeta ( A_\zeta \zeta - A_{\zeta4} \zeta^* )
- \\
- G_v^{h-dissip} & = &
-   \frac{1}{\Delta x_v h_s} \delta_i h_\zeta ( A_\zeta \zeta - A_\zeta \zeta^* )
- + \frac{1}{\Delta y_c} \delta_j ( A_D D - A_{D4} D^* )
- \end{eqnarray}
- where
- \begin{eqnarray}
- D^* & = & \frac{1}{{\cal A}_c h_c} (
-   \delta_i \Delta y_g h_w \nabla^2 u
- + \delta_j \Delta x_g h_s \nabla^2 v ) \\
- \zeta^* & = & \frac{1}{{\cal A}_\zeta} (
-   \delta_i \Delta y_c \nabla^2 v
- - \delta_j \Delta x_c \nabla^2 u )
- \end{eqnarray}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_VI\_HDISSIP} ({\em mom\_vi\_hdissip.F})
- $G_u^{h-dissip}$: {\bf uDiss} (local to {\em calc\_mom\_rhs.F})
- $G_v^{h-dissip}$: {\bf vDiss} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }
- \subsubsection{Vertical dissipation}
- Currently, this is exactly the same code as the flux form equations.
- \begin{eqnarray}
- G_u^{v-diss} & = &
- \frac{1}{\Delta r_f h_w} \delta_k \tau_{13} \\
- G_v^{v-diss} & = &
- \frac{1}{\Delta r_f h_s} \delta_k \tau_{23}
- \end{eqnarray}
- represents the general discrete form of the vertical dissipation terms.
- In the interior the vertical stresses are discretized:
- \begin{eqnarray}
- \tau_{13} & = & A_v \frac{1}{\Delta r_c} \delta_k u \\
- \tau_{23} & = & A_v \frac{1}{\Delta r_c} \delta_k v
- \end{eqnarray}
- \fbox{ \begin{minipage}{4.25in}
- {\em S/R MOM\_U\_RVISCLFUX} ({\em mom\_u\_rviscflux.F})
- {\em S/R MOM\_V\_RVISCLFUX} ({\em mom\_v\_rviscflux.F})
- $\tau_{13}$: {\bf urf} (local to {\em calc\_mom\_rhs.F})
- $\tau_{23}$: {\bf vrf} (local to {\em calc\_mom\_rhs.F})
- \end{minipage} }

 Legend:



Removed from v.1.1
 


changed lines


 
Added in v.1.6
 Legend:



Removed from v.1.1
 


changed lines


 
Added in v.1.6
-Removed from v.1.1
+Added in v.1.6

	ViewVC Help
Powered by ViewVC 1.1.22