1 | % ================================================================ |
---|
2 | % Chapter Ñ Appendix C : Discrete Invariants of the Equations |
---|
3 | % ================================================================ |
---|
4 | \chapter{Discrete Invariants of the Equations} |
---|
5 | \label{Apdx_C} |
---|
6 | \minitoc |
---|
7 | |
---|
8 | %%% Appendix put in gmcomment as it has not been updated for z* and s coordinate |
---|
9 | I'm writting this appendix. It will be available in a forthcoming release of the documentation |
---|
10 | |
---|
11 | %\gmcomment{ |
---|
12 | |
---|
13 | % ================================================================ |
---|
14 | % Conservation Properties on Ocean Dynamics |
---|
15 | % ================================================================ |
---|
16 | \section{Conservation Properties on Ocean Dynamics} |
---|
17 | \label{Apdx_C.1} |
---|
18 | |
---|
19 | |
---|
20 | First, the boundary condition on the vertical velocity (no flux through the surface and the bottom) is established for the discrete set of momentum equations. Then, it is shown that the non linear terms of the momentum equation are written such that the potential enstrophy of a horizontally non divergent flow is preserved while all the other non-diffusive terms preserve the kinetic energy: the energy is also preserved in practice. In addition, an option is also offer for the vorticity term discretization which provides |
---|
21 | a total kinetic energy conserving discretization for that term. |
---|
22 | |
---|
23 | Nota Bene: these properties are established here in the rigid-lid case and for the 2nd order centered scheme. A forthcoming update will be their generalisation to the free surface case |
---|
24 | and higher order scheme. |
---|
25 | |
---|
26 | % ------------------------------------------------------------------------------------------------------------- |
---|
27 | % Bottom Boundary Condition on Vertical Velocity Field |
---|
28 | % ------------------------------------------------------------------------------------------------------------- |
---|
29 | \subsection{Bottom Boundary Condition on Vertical Velocity Field} |
---|
30 | \label{Apdx_C.1.1} |
---|
31 | |
---|
32 | |
---|
33 | The discrete set of momentum equations used in rigid lid approximation |
---|
34 | automatically satisfies the surface and bottom boundary conditions |
---|
35 | (no flux through the surface and the bottom: $w_{surface} =w_{bottom} =~0$). |
---|
36 | Indeed, taking the discrete horizontal divergence of the vertical sum of the |
---|
37 | horizontal momentum equations (Eqs. (II.2.1) and (II.2.2)~) wheighted by the |
---|
38 | vertical scale factors, it becomes: |
---|
39 | \begin{flalign*} |
---|
40 | \frac{\partial } {\partial t} \left( \sum\limits_k \chi \right) |
---|
41 | \equiv |
---|
42 | \frac{\partial } {\partial t} \left( w_{surface} -w_{bottom} \right)&&&\\ |
---|
43 | \end{flalign*} |
---|
44 | \begin{flalign*} |
---|
45 | \equiv \frac{1} {e_{1T}\,e_{2T}\,e_{3T}} |
---|
46 | \biggl\{ \quad |
---|
47 | \delta_i |
---|
48 | &\left[ |
---|
49 | e_{2u}\,H_u |
---|
50 | \left( |
---|
51 | M_u - M_u - \frac{1} {H_u\,e_{2u}} \delta_j |
---|
52 | \left[ \partial_t\, \psi \right] |
---|
53 | \right) |
---|
54 | \right] && |
---|
55 | \biggr. \\ |
---|
56 | \biggl. |
---|
57 | + \delta_j |
---|
58 | &\left[ |
---|
59 | e_{1v}\,H_v |
---|
60 | \left( M_v - M_v - \frac{1} {H_v\,e_{1v}} \delta_i |
---|
61 | \left[ \partial_i\, \psi \right] |
---|
62 | \right) |
---|
63 | \right] |
---|
64 | \biggr\}&& \\ |
---|
65 | \end{flalign*} |
---|
66 | \begin{flalign*} |
---|
67 | \equiv \frac{1} {e_{1T} \,e_{2T} \,e_{3T}} \; |
---|
68 | \biggl\{ |
---|
69 | - \delta_i |
---|
70 | \Bigl[ |
---|
71 | \delta_j |
---|
72 | \left[ \partial_t \psi \right] |
---|
73 | \Bigr] |
---|
74 | + \delta_j |
---|
75 | \Bigl[ |
---|
76 | \delta_i |
---|
77 | \left[ \partial_t \psi \right] |
---|
78 | \Bigr]\; |
---|
79 | \biggr\}\; |
---|
80 | \equiv 0 |
---|
81 | &&&\\ |
---|
82 | \end{flalign*} |
---|
83 | |
---|
84 | |
---|
85 | The surface boundary condition associated with the rigid lid approximation ($w_{surface} =0)$ is imposed in the computation of the vertical velocity (II.2.5). Therefore, it turns out to be: |
---|
86 | \begin{equation*} |
---|
87 | \frac{\partial } {\partial t}w_{bottom} \equiv 0 |
---|
88 | \end{equation*} |
---|
89 | As the bottom velocity is initially set to zero, it remains zero all the time. Symmetrically, if $w_{bottom} =0$ is used in the computation of the vertical velocity (upward integral of the horizontal divergence), the same computation leads to $w_{surface} =0$ as soon as the surface vertical velocity is initially set to zero. |
---|
90 | |
---|
91 | % ------------------------------------------------------------------------------------------------------------- |
---|
92 | % Coriolis and advection terms: vector invariant form |
---|
93 | % ------------------------------------------------------------------------------------------------------------- |
---|
94 | \subsection{Coriolis and advection terms: vector invariant form} |
---|
95 | \label{Apdx_C_vor_zad} |
---|
96 | |
---|
97 | % ------------------------------------------------------------------------------------------------------------- |
---|
98 | % Vorticity Term |
---|
99 | % ------------------------------------------------------------------------------------------------------------- |
---|
100 | \subsubsection{Vorticity Term} |
---|
101 | \label{Apdx_C_vor} |
---|
102 | |
---|
103 | Potential vorticity is located at $f$-points and defined as: $\zeta / e_{3f}$. The standard discrete formulation of the relative vorticity term obviously conserves potential vorticity (ENS scheme). It also conserves the potential enstrophy for a horizontally non-divergent flow (i.e. $\chi $=0) but not the total kinetic energy. Indeed, using the symmetry or skew symmetry properties of the operators (Eqs \eqref{DOM_mi_adj} and \eqref{DOM_di_adj}), it can be shown that: |
---|
104 | \begin{equation} \label{Apdx_C_1.1} |
---|
105 | \int_D {\zeta / e_3\,\;{\textbf{k}}\cdot \frac{1} {e_3} \nabla \times \left( {\zeta \;{\textbf{k}}\times {\textbf{U}}_h } \right)\;dv} \equiv 0 |
---|
106 | \end{equation} |
---|
107 | where $dv=e_1\,e_2\,e_3 \; di\,dj\,dk$ is the volume element. Indeed, using |
---|
108 | \eqref{Eq_dynvor_ens}, the discrete form of the right hand side of \eqref{Apdx_C_1.1} |
---|
109 | can be transformed as follow: |
---|
110 | \begin{flalign*} |
---|
111 | &\int_D \zeta / e_3\,\; \textbf{k} \cdot \frac{1} {e_3 } \nabla \times |
---|
112 | \left( |
---|
113 | \zeta \; \textbf{k} \times \textbf{U}_h |
---|
114 | \right)\; |
---|
115 | dv |
---|
116 | &&& \displaybreak[0] \\ |
---|
117 | % |
---|
118 | \equiv& \sum\limits_{i,j,k} |
---|
119 | \frac{\zeta / e_{3f}} {e_{1f}\,e_{2f}\,e_{3f}} |
---|
120 | \biggl\{ \quad |
---|
121 | \delta_{i+1/2} |
---|
122 | \left[ |
---|
123 | - \overline {\left( {\zeta / e_{3f}} \right)}^{\,i}\; |
---|
124 | \overline{\overline {\left( e_{2u}\,e_{3u}\,u \right)}}^{\,i,j+1/ 2} |
---|
125 | \right] |
---|
126 | && \\ & \qquad \qquad \qquad \;\; |
---|
127 | - \delta_{j+1/2} |
---|
128 | \left[ \;\;\; |
---|
129 | \overline {\left( \zeta / e_{3f} \right)}^{\,j}\; |
---|
130 | \overline{\overline {\left( e_{1v}\,e_{3v}\,v \right)}}^{\,i+1/2,j} |
---|
131 | \right] |
---|
132 | \;\;\biggr\} \; e_{1f}\,e_{2f}\,e_{3f} && \displaybreak[0] \\ |
---|
133 | % |
---|
134 | \equiv& \sum\limits_{i,j,k} |
---|
135 | \biggl\{ \delta_i \left[ \frac{\zeta} {e_{3f}} \right] \; |
---|
136 | \overline{ \left( \frac{\zeta} {e_{3f}} \right) }^{\,i}\; |
---|
137 | \overline{ \overline{ \left( e_{1u}\,e_{3u}\,u \right) } }^{\,i,j+1/2} |
---|
138 | + \delta_j \left[ \frac{\zeta} {e_{3f}} \right] \; |
---|
139 | \overline{ \left( \frac{\zeta} {e_{3f}} \right) }^{\,j} \; |
---|
140 | \overline{\overline {\left( e_{2v}\,e_{3v}\,v \right)}}^{\,i+1/2,j} \biggr\} |
---|
141 | &&&& \displaybreak[0] \\ |
---|
142 | % |
---|
143 | \equiv& \frac{1} {2} \sum\limits_{i,j,k} |
---|
144 | \biggl\{ \delta_i \Bigl[ \left( \frac{\zeta} {e_{3f}} \right)^2 \Bigr]\; |
---|
145 | \overline{\overline {\left( e_{2u}\,e_{3u}\,u \right)}}^{\,i,j+1/2} |
---|
146 | + \delta_j \Bigl[ \left( \zeta / e_{3f} \right)^2 \Bigr]\; |
---|
147 | \overline{\overline {\left( e_{1v}\,e_{3v}\,v \right)}}^{\,i+1/2,j} |
---|
148 | \biggr\} |
---|
149 | && \displaybreak[0] \\ |
---|
150 | % |
---|
151 | \equiv& - \frac{1} {2} \sum\limits_{i,j,k} \left( \frac{\zeta} {e_{3f}} \right)^2\; |
---|
152 | \biggl\{ \delta_{i+1/2} |
---|
153 | \left[ \overline{\overline {\left( e_{2u}\,e_{3u}\,u \right)}}^{\,i,j+1/2} \right] |
---|
154 | + \delta_{j+1/2} |
---|
155 | \left[ \overline{\overline {\left( e_{1v}\,e_{3v}\,v \right)}}^{\,i+1/2,j} \right] |
---|
156 | \biggr\} && \\ |
---|
157 | \end{flalign*} |
---|
158 | Since $\overline {\;\cdot \;} $ and $\delta $ operators commute: $\delta_{i+1/2} |
---|
159 | \left[ {\overline a^{\,i}} \right] = \overline {\delta_i \left[ a \right]}^{\,i+1/2}$, |
---|
160 | and introducing the horizontal divergence $\chi $, it becomes: |
---|
161 | \begin{align*} |
---|
162 | \equiv& \sum\limits_{i,j,k} - \frac{1} {2} \left( \frac{\zeta} {e_{3f}} \right)^2 \; \overline{\overline{ e_{1T}\,e_{2T}\,e_{3T}\, \chi}}^{\,i+1/2,j+1/2} \;\;\equiv 0 |
---|
163 | \qquad \qquad \qquad \qquad \qquad \qquad \qquad &&&&\\ |
---|
164 | \end{align*} |
---|
165 | |
---|
166 | Note that the demonstration is done here for the relative potential |
---|
167 | vorticity but it still hold for the planetary ($f/e_3$) and the total |
---|
168 | potential vorticity $((\zeta +f) /e_3 )$. Another formulation of |
---|
169 | the two components of the vorticity term is optionally offered (ENE scheme) : |
---|
170 | \begin{equation*} |
---|
171 | - \zeta \;{\textbf{k}}\times {\textbf {U}}_h |
---|
172 | \equiv |
---|
173 | \left( {{ |
---|
174 | \begin{array} {*{20}c} |
---|
175 | + \frac{1} {e_{1u}} \; |
---|
176 | \overline {\left( \zeta / e_{3f} \right) |
---|
177 | \overline {\left( e_{1v}\,e_{3v}\,v \right)}^{\,i+1/2}} ^{\,j} |
---|
178 | \hfill \\ |
---|
179 | - \frac{1} {e_{2v}} \; |
---|
180 | \overline {\left( \zeta / e_{3f} \right) |
---|
181 | \overline {\left( e_{2u}\,e_{3u}\,u \right)}^{\,j+1/2}} ^{\,i} |
---|
182 | \hfill \\ |
---|
183 | \end{array}} } |
---|
184 | \right) |
---|
185 | \end{equation*} |
---|
186 | |
---|
187 | This formulation does not conserve the enstrophy but the total kinetic |
---|
188 | energy. It is also possible to mix the two formulations in order to conserve |
---|
189 | enstrophy on the relative vorticity term and energy on the Coriolis term. |
---|
190 | \begin{flalign*} |
---|
191 | &\int\limits_D - \textbf{U}_h \cdot \left( \zeta \;\textbf{k} \times \textbf{U}_h \right) \; dv && \\ |
---|
192 | \equiv& \sum\limits_{i,j,k} \biggl\{ |
---|
193 | \overline {\left( \frac{\zeta} {e_{3f}} \right) |
---|
194 | \overline {\left( e_{1v}e_{3v}v \right)}^{\,i+1/2}} ^{\,j} \, e_{2u}e_{3u}u |
---|
195 | - \overline {\left( \frac{\zeta} {e_{3f}} \right) |
---|
196 | \overline {\left( e_{2u}e_{3u}u \right)}^{\,j+1/2}} ^{\,i} \, e_{1v}e_{3v}v \; |
---|
197 | \biggr\} |
---|
198 | \\ |
---|
199 | \equiv& \sum\limits_{i,j,k} \frac{\zeta} {e_{3f}} |
---|
200 | \biggl\{ \overline {\left( e_{1v}e_{3v} v \right)}^{\,i+1/2}\; |
---|
201 | \overline {\left( e_{2u}e_{3u} u \right)}^{\,j+1/2} |
---|
202 | - \overline {\left( e_{2u}e_{3u} u \right)}^{\,j+1/2}\; |
---|
203 | \overline {\left( e_{1v}e_{3v} v \right)}^{\,i+1/2} |
---|
204 | \biggr\} |
---|
205 | \equiv 0 |
---|
206 | \end{flalign*} |
---|
207 | |
---|
208 | |
---|
209 | % ------------------------------------------------------------------------------------------------------------- |
---|
210 | % Gradient of Kinetic Energy / Vertical Advection |
---|
211 | % ------------------------------------------------------------------------------------------------------------- |
---|
212 | \subsubsection{Gradient of Kinetic Energy / Vertical Advection} |
---|
213 | \label{Apdx_C_zad} |
---|
214 | |
---|
215 | The change of Kinetic Energy (KE) due to the vertical advection is exactly |
---|
216 | balanced by the change of KE due to the horizontal gradient of KE~: |
---|
217 | \begin{equation*} |
---|
218 | \int_D \textbf{U}_h \cdot \nabla_h \left( \frac{1}{2}\;{\textbf{U}_h}^2 \right)\;dv |
---|
219 | = - \int_D \textbf{U}_h \cdot w \frac{\partial \textbf{U}_h} {\partial k}\;dv |
---|
220 | \end{equation*} |
---|
221 | Indeed, using successively \eqref{DOM_di_adj} ($i.e.$ the skew symmetry property of the $\delta$ operator) and the incompressibility, then again \eqref{DOM_di_adj}, then |
---|
222 | the commutativity of operators $\overline {\,\cdot \,}$ and $\delta$, and finally \eqref{DOM_mi_adj} ($i.e.$ the symmetry property of the $\overline {\,\cdot \,}$ operator) applied in the horizontal and vertical direction, it becomes: |
---|
223 | \begin{flalign*} |
---|
224 | &\int_D \textbf{U}_h \cdot \nabla_h \left( \frac{1}{2}\;{\textbf{U}_h}^2 \right)\;dv &&&\\ |
---|
225 | \equiv& \frac{1}{2} \sum\limits_{i,j,k} |
---|
226 | \biggl\{ |
---|
227 | \frac{1} {e_{1u}} \delta_{i+1/2} |
---|
228 | \left[ \overline {u^2}^{\,i} + \overline {v^2}^{\,j} \right] u\,e_{1u}e_{2u}e_{3u} |
---|
229 | + \frac{1} {e_{2v}} \delta_{j+1/2} |
---|
230 | \left[ \overline {u^2}^{\,i} + \overline {v^2}^{\,j} \right] v\,e_{1v}e_{2v}e_{3v} |
---|
231 | \biggr\} |
---|
232 | &&& \displaybreak[0] \\ |
---|
233 | % |
---|
234 | \equiv& \frac{1}{2} \sum\limits_{i,j,k} |
---|
235 | \left( \overline {u^2}^{\,i} + \overline {v^2}^{\,j} \right)\; |
---|
236 | \delta_k \left[ e_{1T}\,e_{2T} \,w \right] |
---|
237 | % |
---|
238 | \;\; \equiv -\frac{1}{2} \sum\limits_{i,j,k} \delta_{k+1/2} |
---|
239 | \left[ |
---|
240 | \overline{ u^2}^{\,i} |
---|
241 | + \overline{ v^2}^{\,j} |
---|
242 | \right] \; |
---|
243 | e_{1v}\,e_{2v}\,w |
---|
244 | &&& \displaybreak[0]\\ |
---|
245 | % |
---|
246 | \equiv &\frac{1} {2} \sum\limits_{i,j,k} |
---|
247 | \left( \overline {\delta_{k+1/2} \left[ u^2 \right]}^{\,i} |
---|
248 | + \overline {\delta_{k+1/2} \left[ v^2 \right]}^{\,j} \right) \; e_{1T}\,e_{2T} \,w |
---|
249 | && \displaybreak[0] \\ |
---|
250 | % |
---|
251 | \equiv &\frac{1} {2} \sum\limits_{i,j,k} |
---|
252 | \biggl\{ \overline {e_{1T}\,e_{2T} \,w}^{\,i+1/2}\;2 |
---|
253 | \overline {u}^{\,k+1/2}\; \delta_{k+1/2} \left[ u \right] %&&& \\ |
---|
254 | + \overline {e_{1T}\,e_{2T} \,w}^{\,j+1/2}\;2 \overline {v}^{\,k+1/2}\; \delta_{k+1/2} \left[ v \right] \; |
---|
255 | \biggr\} |
---|
256 | &&\displaybreak[0] \\ |
---|
257 | % |
---|
258 | \equiv& -\sum\limits_{i,j,k} |
---|
259 | \biggl\{ |
---|
260 | \quad \frac{1} {b_u } \; |
---|
261 | \overline {\Bigl\{ \overline {e_{1T}\,e_{2T} \,w}^{\,i+1/2}\,\delta_{k+1/2} |
---|
262 | \left[ u \right] |
---|
263 | \Bigr\} }^{\,k} \;u\;e_{1u}\,e_{2u}\,e_{3u} |
---|
264 | && \\ |
---|
265 | &\qquad \quad\; + \frac{1} {b_v } \; |
---|
266 | \overline {\Bigl\{ \overline {e_{1T}\,e_{2T} \,w}^{\,j+1/2} \delta_{k+1/2} |
---|
267 | \left[ v \right] |
---|
268 | \Bigr\} }^{\,k} \;v\;e_{1v}\,e_{2v}\,e_{3v} \; |
---|
269 | \biggr\} |
---|
270 | && \\ |
---|
271 | \equiv& -\int\limits_D \textbf{U}_h \cdot w \frac{\partial \textbf{U}_h} {\partial k}\;dv &&&\\ |
---|
272 | \end{flalign*} |
---|
273 | |
---|
274 | The main point here is that the satisfaction of this property links the choice of the discrete formulation of vertical advection and of horizontal gradient of KE. Choosing one imposes the other. For example KE can also be discretized as $1/2\,({\overline u^{\,i}}^2 + {\overline v^{\,j}}^2)$. This leads to the following expression for the vertical advection: |
---|
275 | \begin{equation*} |
---|
276 | \frac{1} {e_3 }\; w\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
277 | \equiv \left( {{\begin{array} {*{20}c} |
---|
278 | \frac{1} {e_{1u}\,e_{2u}\,e_{3u}} \; |
---|
279 | \overline{\overline {e_{1T}\,e_{2T} \,w\;\delta_{k+1/2} |
---|
280 | \left[ \overline u^{\,i+1/2} \right]}}^{\,i+1/2,k} \hfill \\ |
---|
281 | \frac{1} {e_{1v}\,e_{2v}\,e_{3v}} \; |
---|
282 | \overline{\overline {e_{1T}\,e_{2T} \,w\;\delta_{k+1/2} |
---|
283 | \left[ \overline v^{\,j+1/2} \right]}}^{\,j+1/2,k} \hfill \\ |
---|
284 | \end{array}} } \right) |
---|
285 | \end{equation*} |
---|
286 | a formulation that requires a additional horizontal mean compare to the one used in NEMO. Nine velocity points have to be used instead of 3. This is the reason why it has not been choosen. |
---|
287 | |
---|
288 | % ------------------------------------------------------------------------------------------------------------- |
---|
289 | % Coriolis and advection terms: flux form |
---|
290 | % ------------------------------------------------------------------------------------------------------------- |
---|
291 | \subsection{Coriolis and advection terms: flux form} |
---|
292 | \label{Apdx_C.1.3} |
---|
293 | |
---|
294 | % ------------------------------------------------------------------------------------------------------------- |
---|
295 | % Coriolis plus ``metric'' Term |
---|
296 | % ------------------------------------------------------------------------------------------------------------- |
---|
297 | \subsubsection{Coriolis plus ``metric'' Term} |
---|
298 | \label{Apdx_C.1.3.1} |
---|
299 | |
---|
300 | In flux from the vorticity term reduces to a Coriolis term in which the Coriolis parameter has been modified to account for the ``metric'' term. This altered Coriolis parameter is discretised at F-point. It is given by: |
---|
301 | \begin{equation*} |
---|
302 | f+\frac{1} {e_1 e_2 } |
---|
303 | \left( v \frac{\partial e_2 } {\partial i} - u \frac{\partial e_1 } {\partial j}\right)\; |
---|
304 | \equiv \; |
---|
305 | f+\frac{1} {e_{1f}\,e_{2f}} |
---|
306 | \left( \overline v^{\,i+1/2} \delta_{i+1/2} \left[ e_{2u} \right] |
---|
307 | -\overline u^{\,j+1/2} \delta_{j+1/2} \left[ e_{1u} \right] \right) |
---|
308 | \end{equation*} |
---|
309 | |
---|
310 | The ENE scheme is then applied to obtain the vorticity term in flux form. It therefore conserves the total KE. The demonstration is same as for the vorticity term in vector invariant form (\S\ref{Apdx_C_vor}). |
---|
311 | |
---|
312 | % ------------------------------------------------------------------------------------------------------------- |
---|
313 | % Flux form advection |
---|
314 | % ------------------------------------------------------------------------------------------------------------- |
---|
315 | \subsubsection{Flux form advection} |
---|
316 | \label{Apdx_C.1.3.2} |
---|
317 | |
---|
318 | The flux form operator of the momentum advection is evaluated using a centered second order finite difference scheme. Because of the flux form, the discrete operator does not contribute to the global budget of linear momentum. Because of the centered second order scheme, it conserves the horizontal kinetic energy, that is : |
---|
319 | |
---|
320 | \begin{equation} \label{Apdx_C_I.3.10} |
---|
321 | \int_D \textbf{U}_h \cdot |
---|
322 | \left( {{\begin{array} {*{20}c} |
---|
323 | \nabla \cdot \left( \textbf{U}\,u \right) \hfill \\ |
---|
324 | \nabla \cdot \left( \textbf{U}\,v \right) \hfill \\ |
---|
325 | \end{array}} } \right)\;dv =\;0 |
---|
326 | \end{equation} |
---|
327 | |
---|
328 | Let us demonstrate this property for the first term of the scalar product (i.e. considering just the the terms associated with the i-component of the advection): |
---|
329 | \begin{flalign*} |
---|
330 | &\int_D u \cdot \nabla \cdot \left( \textbf{U}\,u \right) \; dv &&&\\ |
---|
331 | % |
---|
332 | \equiv& \sum\limits_{i,j,k} |
---|
333 | \biggl\{ \frac{1} {e_{1u}\, e_{2u}\,e_{3u}} \biggl( |
---|
334 | \delta_{i+1/2} \left[ \overline {e_{2u}\,e_{3u}\,u}^{\,i} \;\overline u^{\,i} \right] |
---|
335 | + \delta_j \left[ \overline {e_{1u}\,e_{3u}\,v}^{\,i+1/2}\;\overline u^{\,j+1/2} \right] |
---|
336 | &&& \\ & \qquad \qquad \qquad \qquad \qquad \qquad |
---|
337 | + \delta_k \left[ \overline {e_{1w}\,e_{2w}\,w}^{\,i+1/2}\;\overline u^{\,k+1/2} \right] |
---|
338 | \biggr) \; \biggr\} \, e_{1u}\,e_{2u}\,e_{3u} \;u |
---|
339 | &&& \displaybreak[0] \\ |
---|
340 | % |
---|
341 | \equiv& \sum\limits_{i,j,k} |
---|
342 | \biggl\{ |
---|
343 | \delta_{i+1/2} \left[ \overline {e_{2u}\,e_{3u}\,u}^{\,i}\; \overline u^{\,i} \right] |
---|
344 | + \delta_j \left[ \overline {e_{1u}\,e_{3u}\,v}^{\,i+1/2}\;\overline u^{\,j+1/2} \right] |
---|
345 | &&& \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad |
---|
346 | + \delta_k \left[ \overline {e_{1w}\,e_{2w}\,w}^{\,i+12}\;\overline u^{\,k+1/2} \right] |
---|
347 | \; \biggr\} \; u &&& \displaybreak[0] \\ |
---|
348 | % |
---|
349 | \equiv& - \sum\limits_{i,j,k} |
---|
350 | \biggl\{ |
---|
351 | \overline {e_{2u}\,e_{3u}\,u}^{\,i}\; \overline u^{\,i} \delta_i |
---|
352 | \left[ u \right] |
---|
353 | + \overline {e_{1u}\,e_{3u}\,v}^{\,i+1/2}\; \overline u^{\,j+1/2} \delta_{j+1/2} |
---|
354 | \left[ u \right] |
---|
355 | &&& \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad |
---|
356 | + \overline {e_{1w}\,e_{2w}\,w}^{\,i+1/2}\; \overline u^{\,k+1/2} \delta_{k+1/2} \left[ u \right] \biggr\} && \displaybreak[0] \\ |
---|
357 | % |
---|
358 | \equiv& - \sum\limits_{i,j,k} |
---|
359 | \biggl\{ |
---|
360 | \overline {e_{2u}\,e_{3u}\,u}^{\,i} \delta_i \left[ u^2 \right] |
---|
361 | + \overline {e_{1u}\,e_{3u}\,v}^{\,i+1/2} \delta_{j+/2} \left[ u^2 \right] |
---|
362 | + \overline {e_{1w}\,e_{2w}\,w}^{\,i+1/2} \delta_{k+1/2} \left[ u^2 \right] |
---|
363 | \biggr\} |
---|
364 | && \displaybreak[0] \\ |
---|
365 | % |
---|
366 | \equiv& \sum\limits_{i,j,k} |
---|
367 | \bigg\{ |
---|
368 | e_{2u}\,e_{3u}\,u\; \delta_{i+1/2} \left[ \overline {u^2}^{\,i} \right] |
---|
369 | + e_{1u}\,e_{3u}\,v\; \delta_{j+1/2} \; \left[ \overline {u^2}^{\,i} \right] |
---|
370 | + e_{1w}\,e_{2w}\,w\; \delta_{k+1/2} \left[ \overline {u^2}^{\,i} \right] |
---|
371 | \biggr\} |
---|
372 | && \displaybreak[0] \\ |
---|
373 | % |
---|
374 | \equiv& \sum\limits_{i,j,k} |
---|
375 | \overline {u^2}^{\,i} |
---|
376 | \biggl\{ |
---|
377 | \delta_{i+1/2} \left[ e_{2u}\,e_{3u}\,u \right] |
---|
378 | + \delta_{j+1/2} \left[ e_{1u}\,e_{3u}\,v \right] |
---|
379 | + \delta_{k+1/2} \left[ e_{1w}\,e_{2w}\,w \right] |
---|
380 | \biggr\} \;\; \equiv 0 |
---|
381 | &&& \\ |
---|
382 | \end{flalign*} |
---|
383 | |
---|
384 | When the UBS scheme is used to evaluate the flux form momentum advection, the discrete operator does not contribute to the global budget of linear momentum (flux form). The horizontal kinetic energy is not conserved, but forced to decrease (the scheme is diffusive). |
---|
385 | |
---|
386 | % ------------------------------------------------------------------------------------------------------------- |
---|
387 | % Hydrostatic Pressure Gradient Term |
---|
388 | % ------------------------------------------------------------------------------------------------------------- |
---|
389 | \subsection{Hydrostatic Pressure Gradient Term} |
---|
390 | \label{Apdx_C.1.4} |
---|
391 | |
---|
392 | |
---|
393 | A pressure gradient has no contribution to the evolution of the vorticity as the curl of a gradient is zero. In $z$-coordinate, this properties is satisfied locally on a C-grid with 2nd order finite differences (property \eqref{Eq_DOM_curl_grad}). When the equation of state is linear ($i.e.$ when an advective-diffusive equation for density can be derived from those of temperature and salinity) the change of KE due to the work of pressure forces is balanced by the change of potential energy due to buoyancy forces: |
---|
394 | \begin{equation*} |
---|
395 | \int_D - \frac{1} {\rho_o} \left. \nabla p^h \right|_z \cdot \textbf{U}_h \;dv |
---|
396 | = \int_D \nabla \cdot \left( \rho \,\textbf {U} \right)\;g\;z\;\;dv |
---|
397 | \end{equation*} |
---|
398 | |
---|
399 | This property can be satisfied in discrete sense for both $z$- and $s$-coordinates. Indeed, defining the depth of a $T$-point, $z_T$ defined as the sum of the vertical scale factors at $w$-points starting from the surface, the workof pressure forces can be written as: |
---|
400 | \begin{flalign*} |
---|
401 | &\int_D - \frac{1} {\rho_o} \left. \nabla p^h \right|_z \cdot \textbf{U}_h \;dv &&& \\ |
---|
402 | \equiv& \sum\limits_{i,j,k} \biggl\{ \; - \frac{1} {\rho_o e_{1u}} \Bigl( |
---|
403 | \delta_{i+1/2} \left[ p^h \right] - g\;\overline \rho^{\,i+1/2}\;\delta_{i+1/2} \left[ z_T \right] |
---|
404 | \Bigr) \; u\;e_{1u}\,e_{2u}\,e_{3u} |
---|
405 | && \\ & \qquad \qquad |
---|
406 | - \frac{1} {\rho_o e_{2v}} \Bigl( |
---|
407 | \delta_{j+1/2} \left[ p^h \right] - g\;\overline \rho^{\,j+1/2}\delta_{j+1/2} \left[ z_T \right] |
---|
408 | \Bigr) \; v\;e_{1v}\,e_{2v}\,e_{3v} \; |
---|
409 | \biggr\} && \\ |
---|
410 | \end{flalign*} |
---|
411 | |
---|
412 | Using \eqref{DOM_di_adj}, $i.e.$ the skew symmetry property of the $\delta$ operator, \eqref{Eq_wzv}, the continuity equation), and \eqref{Eq_dynhpg_sco}, the hydrostatic |
---|
413 | equation in $s$-coordinate, it turns out to be: |
---|
414 | \begin{flalign*} |
---|
415 | \equiv& \frac{1} {\rho_o} \sum\limits_{i,j,k} \biggl\{ |
---|
416 | e_{2u}\,e_{3u} \;u\;g\; \overline \rho^{\,i+1/2}\;\delta_{i+1/2}[ z_T] |
---|
417 | + e_{1v}\,e_{3v} \;v\;g\; \overline \rho^{\,j+1/2}\;\delta_{j+1/2}[ z_T] |
---|
418 | && \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \;\;\, |
---|
419 | +\Bigl( \delta_i[ e_{2u}\,e_{3u}\,u] + \delta_j [ e_{1v}\,e_{3v}\,v] \Bigr)\;p^h \biggr\} &&\\ |
---|
420 | % |
---|
421 | \equiv& \frac{1} {\rho_o } \sum\limits_{i,j,k} |
---|
422 | \biggl\{ |
---|
423 | e_{2u}\,e_{3u} \;u\;g\; \overline \rho^{\,i+1/2} \delta_{i+1/2} \left[ z_T \right] |
---|
424 | + e_{1v}\,e_{3v} \;v\;g\; \overline \rho^{\,j+1/2} \delta_{j+1/2} \left[ z_T \right] |
---|
425 | &&&\\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \;\;\, |
---|
426 | - \delta_k \left[ e_{1w} e_{2w}\,w \right]\;p^h \biggr\} &&&\\ |
---|
427 | % |
---|
428 | \equiv& \frac{1} {\rho_o } \sum\limits_{i,j,k} |
---|
429 | \biggl\{ |
---|
430 | e_{2u}\,e_{3u} \;u\;g\; \overline \rho^{\,i+1/2}\;\delta_{i+1/2} \left[ z_T \right] |
---|
431 | + e_{1v}\,e_{3v} \;v\;g\; \overline \rho^{\,j+1/2} \;\delta_{j+1/2} \left[ z_T \right] |
---|
432 | &&& \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \;\;\, |
---|
433 | + e_{1w} e_{2w} \;w\;\delta_{k+1/2} \left[ p_h \right] |
---|
434 | \biggr\} &&&\\ |
---|
435 | % |
---|
436 | \equiv& \frac{g} {\rho_o} \sum\limits_{i,j,k} |
---|
437 | \biggl\{ |
---|
438 | e_{2u}\,e_{3u} \;u\; \overline \rho^{\,i+1/2}\;\delta_{i+1/2} \left[ z_T \right] |
---|
439 | + e_{1v}\,e_{3v} \;v\; \overline \rho^{\,j+1/2}\;\delta_{j+1/2} \left[ z_T \right] |
---|
440 | &&& \\ & \qquad \qquad \qquad \qquad \qquad \qquad \qquad \;\;\, |
---|
441 | - e_{1w} e_{2w} \;w\;e_{3w} \overline \rho^{\,k+1/2} |
---|
442 | \biggr\} &&&\\ |
---|
443 | \end{flalign*} |
---|
444 | noting that by definition of $z_T$, $\delta_{k+1/2} \left[ z_T \right] \equiv - e_{3w} $, thus: |
---|
445 | \begin{multline*} |
---|
446 | \equiv \frac{g} {\rho_o} \sum\limits_{i,j,k} |
---|
447 | \biggl\{ |
---|
448 | e_{2u}\,e_{3u} \;u\; \overline \rho^{\,i+1/2}\;\delta_{i+1/2} \left[ z_T \right] |
---|
449 | + e_{1v}\,e_{3v} \;v\; \overline \rho^{\,j+1/2} \delta_{j+1/2} \left[ z_T \right] |
---|
450 | \biggr. \\ |
---|
451 | \shoveright{ |
---|
452 | \biggl. |
---|
453 | + e_{1w} e_{2w} \;w\; \overline \rho^{\,k+1/2}\;\delta_{k+1/2} \left[ z_T \right] |
---|
454 | \biggr\} } \\ |
---|
455 | \end{multline*} |
---|
456 | Using \eqref{DOM_di_adj}, it becomes: |
---|
457 | \begin{flalign*} |
---|
458 | \equiv& - \frac{g} {\rho_o} \sum\limits_{i,j,k} z_T |
---|
459 | \biggl\{ |
---|
460 | \delta_i \left[ e_{2u}\,e_{3u}\,u\; \overline \rho^{\,i+1/2} \right] |
---|
461 | + \delta_j \left[ e_{1v}\,e_{3v}\,v\; \overline \rho^{\,j+1/2} \right] |
---|
462 | + \delta_k \left[ e_{1w} e_{2w}\,w\; \overline \rho^{\,k+1/2} \right] |
---|
463 | \biggr\} |
---|
464 | &&& \\ |
---|
465 | % |
---|
466 | \equiv& -\int_D \nabla \cdot \left( \rho \, \textbf{U} \right)\;g\;z\;\;dv &&& \\ |
---|
467 | \end{flalign*} |
---|
468 | |
---|
469 | Note that this property strongly constraints the discrete expression of both |
---|
470 | the depth of $T-$points and of the term added to the pressure gradient in |
---|
471 | $s$-coordinate. Nevertheless, it is almost never satisfied as a linear equation of state |
---|
472 | is rarely used. |
---|
473 | |
---|
474 | % ------------------------------------------------------------------------------------------------------------- |
---|
475 | % Surface Pressure Gradient Term |
---|
476 | % ------------------------------------------------------------------------------------------------------------- |
---|
477 | \subsection{Surface Pressure Gradient Term} |
---|
478 | \label{Apdx_C.1.5} |
---|
479 | |
---|
480 | |
---|
481 | The surface pressure gradient has no contribution to the evolution of the vorticity. This property is trivially satisfied locally as the equation verified by $\psi $ has been derived from the discrete formulation of the momentum equation and of the curl. But it has to be noticed that since the elliptic equation verified by $\psi $ is solved numerically by an iterative solver (preconditioned conjugate gradient or successive over relaxation), the |
---|
482 | property is only satisfied at the precision required on the solver used. |
---|
483 | |
---|
484 | With the rigid-lid approximation, the change of KE due to the work of surface pressure forces is exactly zero. This is satisfied in discrete form, at the precision required on the elliptic solver used to solve this equation. This can be demonstrated as follows: |
---|
485 | \begin{flalign*} |
---|
486 | \int\limits_D - \frac{1} {\rho_o} \nabla_h \left( p_s \right) \cdot \textbf{U}_h \;dv% &&& \\ |
---|
487 | % |
---|
488 | &\equiv \sum\limits_{i,j,k} \biggl\{ \; |
---|
489 | \left( - M_u - \frac{1} {H_u \,e_{2u}} \delta_j \left[ \partial_t \psi \right] \right)\; |
---|
490 | u\;e_{1u}\,e_{2u}\,e_{3u} |
---|
491 | &&&\\& \qquad \;\;\, |
---|
492 | + \left( - M_v + \frac{1} {H_v \,e_{1v}} \delta_i \left[ \partial_t \psi \right] \right)\; |
---|
493 | v\;e_{1v}\,e_{2v}\,e_{3v} \; \biggr\} |
---|
494 | &&&\\ |
---|
495 | \\ |
---|
496 | % |
---|
497 | &\equiv \sum\limits_{i,j} \Biggl\{ \; |
---|
498 | \biggl( - M_u - \frac{1} {H_u \,e_{2u}} \delta_j \left[ \partial_t \psi \right] \biggr) |
---|
499 | \biggl( \sum\limits_k u\;e_{3u} \biggr)\; e_{1u}\,e_{2u} |
---|
500 | &&&\\& \qquad \;\;\, |
---|
501 | + \biggl( - M_v + \frac{1} {H_v \,e_{1v}} \delta_i \left[ \partial_t \psi \right] \biggr) |
---|
502 | \biggl( \sum\limits_k v\;e_{3v} \biggr)\; e_{1v}\,e_{2v} \; \Biggr\} |
---|
503 | && \\ |
---|
504 | % |
---|
505 | \intertext{using the relation between \textit{$\psi $} and the vertically sum of the velocity, it becomes:} |
---|
506 | % |
---|
507 | &\equiv \sum\limits_{i,j} |
---|
508 | \biggl\{ \; |
---|
509 | \left( \;\;\, |
---|
510 | M_u + \frac{1} {H_u \,e_{2u}} \delta_j \left[ \partial_t \psi \right] |
---|
511 | \right)\; |
---|
512 | e_{1u} \,\delta_j |
---|
513 | \left[ \partial_t \psi \right] |
---|
514 | && \\ & \qquad \;\;\, |
---|
515 | + \left( |
---|
516 | - M_v + \frac{1} {H_v \,e_{1v}} \delta_i \left[ \partial_t \psi \right] |
---|
517 | \right)\; |
---|
518 | e_{2v} \,\delta_i \left[ \partial_t \psi \right] \; |
---|
519 | \biggr\} |
---|
520 | && \\ |
---|
521 | % |
---|
522 | \intertext{applying the adjoint of the $\delta$ operator, it is now:} |
---|
523 | % |
---|
524 | &\equiv \sum\limits_{i,j} - \partial_t \psi \; |
---|
525 | \biggl\{ \; |
---|
526 | \delta_{j+1/2} \left[ e_{1u} M_u \right] |
---|
527 | - \delta_{i+1/2} \left[ e_{1v} M_v \right] |
---|
528 | && \\ & \qquad \;\;\, |
---|
529 | + \delta_{i+1/2} |
---|
530 | \left[ \frac{e_{2v}} {H_v \,e_{2v}} \delta_i \left[ \partial_t \psi \right] |
---|
531 | \right] |
---|
532 | + \delta_{j+1/2} |
---|
533 | \left[ \frac{e_{1u}} {H_u \,e_{2u}} \delta_j \left[ \partial_t \psi \right] |
---|
534 | \right] |
---|
535 | \biggr\} &&&\\ |
---|
536 | &\equiv 0 && \\ |
---|
537 | \end{flalign*} |
---|
538 | |
---|
539 | The last equality is obtained using \eqref{Eq_dynspg_rl}, the discrete barotropic streamfunction time evolution equation. By the way, this shows that \eqref{Eq_dynspg_rl} is the only way do compute the streamfunction, otherwise the surface pressure forces will work. Nevertheless, since the elliptic equation verified by $\psi $ is solved numerically by an iterative solver, the property is only satisfied at the precision required on the solver. |
---|
540 | |
---|
541 | % ================================================================ |
---|
542 | % Conservation Properties on Tracers |
---|
543 | % ================================================================ |
---|
544 | \section{Conservation Properties on Tracers} |
---|
545 | \label{Apdx_C.2} |
---|
546 | |
---|
547 | |
---|
548 | All the numerical schemes used in NEMO are written such that the tracer content is conserved by the internal dynamics and physics (equations in flux form). For advection, only the CEN2 scheme ($i.e.$ 2nd order finite different scheme) conserves the global variance of tracer. Nevertheless the other schemes ensure that the global variance decreases ($i.e.$ they are at least slightly diffusive). For diffusion, all the schemes ensure the decrease of the total tracer variance, but the iso-neutral operator. There is generally no strict conservation of mass, as the equation of state is non linear with respect to $T$ and $S$. In practice, the mass is conserved with a very good accuracy. |
---|
549 | % ------------------------------------------------------------------------------------------------------------- |
---|
550 | % Advection Term |
---|
551 | % ------------------------------------------------------------------------------------------------------------- |
---|
552 | \subsection{Advection Term} |
---|
553 | \label{Apdx_C.2.1} |
---|
554 | |
---|
555 | Whatever the advection scheme considered it conserves of the tracer content as all the scheme are written in flux form. Let $\tau$ be the tracer interpolated at velocity point (whatever the interpolation is). The conservation of the tracer content is obtained as follows: |
---|
556 | \begin{flalign*} |
---|
557 | &\int_D \nabla \cdot \left( T \textbf{U} \right)\;dv &&&\\ |
---|
558 | &\equiv \sum\limits_{i,j,k} \biggl\{ |
---|
559 | \frac{1} {e_{1T}\,e_{2T}\,e_{3T}} |
---|
560 | \left( \delta_i \left[ e_{2u}\,e_{3u}\; u \;\tau_u \right] |
---|
561 | + \delta_j \left[ e_{1v}\,e_{3v}\; v \;\tau_v \right] \right) |
---|
562 | &&&\\& \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad |
---|
563 | + \frac{1} {e_{3T}} \delta_k \left[ w\;\tau \right] \biggl\} e_{1T}\,e_{2T}\,e_{3T} &&&\\ |
---|
564 | % |
---|
565 | &\equiv \sum\limits_{i,j,k} \left\{ |
---|
566 | \delta_i \left[ e_{2u}\,e_{3u} \,\overline T^{\,i+1/2}\,u \right] |
---|
567 | + \delta_j \left[ e_{1v}\,e_{3v} \,\overline T^{\,j+1/2}\,v \right] |
---|
568 | + \delta_k \left[ e_{1T}\,e_{2T} \,\overline T^{\,k+1/2}\,w \right] \right\} |
---|
569 | && \\ |
---|
570 | &\equiv 0 &&& |
---|
571 | \end{flalign*} |
---|
572 | |
---|
573 | The conservation of the variance of tracer can be achieved only with the CEN2 scheme. It can be demonstarted as follows: |
---|
574 | \begin{flalign*} |
---|
575 | &\int\limits_D T\;\nabla \cdot \left( T\; \textbf{U} \right)\;dv &&&\\ |
---|
576 | \equiv& \sum\limits_{i,j,k} T\; |
---|
577 | \left\{ |
---|
578 | \delta_i \left[ e_{2u}\,e_{3u} \overline T^{\,i+1/2}\,u \right] |
---|
579 | + \delta_j \left[ e_{1v}\,e_{3v} \overline T^{\,j+1/2}\,v \right] |
---|
580 | + \delta_k \left[ e_{1T}\,e_{2T} \overline T^{\,k+1/2}w \right] |
---|
581 | \right\} |
---|
582 | && \\ |
---|
583 | \equiv& \sum\limits_{i,j,k} |
---|
584 | \left\{ |
---|
585 | - e_{2u}\,e_{3u} \overline T^{\,i+1/2}\,u\,\delta_{i+1/2} \left[ T \right] \right. |
---|
586 | - e_{1v}\,e_{3v} \overline T^{\,j+1/2}\,v\;\delta_{j+1/2} \left[ T \right] |
---|
587 | &&&\\& \qquad \qquad \qquad \qquad \qquad \qquad \quad \; |
---|
588 | - \left. e_{1T}\,e_{2T} \overline T^{\,k+1/2}w\;\delta_{k+1/2} \left[ T \right] |
---|
589 | \right\} |
---|
590 | &&\\ |
---|
591 | \equiv& -\frac{1} {2} \sum\limits_{i,j,k} |
---|
592 | \Bigl\{ |
---|
593 | e_{2u}\,e_{3u} \; u\;\delta_{i+1/2} \left[ T^2 \right] |
---|
594 | + e_{1v}\,e_{3v} \; v\;\delta_{j+1/2} \left[ T^2 \right] |
---|
595 | + e_{1T}\,e_{2T} \;w\;\delta_{k+1/2} \left[ T^2 \right] |
---|
596 | \Bigr\} |
---|
597 | && \\ |
---|
598 | \equiv& \frac{1} {2} \sum\limits_{i,j,k} T^2 |
---|
599 | \Bigl\{ |
---|
600 | \delta_i \left[ e_{2u}\,e_{3u}\,u \right] |
---|
601 | + \delta_j \left[ e_{1v}\,e_{3v}\,v \right] |
---|
602 | + \delta_k \left[ e_{1T}\,e_{2T}\,w \right] |
---|
603 | \Bigr\} |
---|
604 | \quad \equiv 0 &&& |
---|
605 | \end{flalign*} |
---|
606 | |
---|
607 | |
---|
608 | % ================================================================ |
---|
609 | % Conservation Properties on Lateral Momentum Physics |
---|
610 | % ================================================================ |
---|
611 | \section{Conservation Properties on Lateral Momentum Physics} |
---|
612 | \label{Apdx_C.3} |
---|
613 | |
---|
614 | |
---|
615 | The discrete formulation of the horizontal diffusion of momentum ensures the |
---|
616 | conservation of potential vorticity and horizontal divergence and the |
---|
617 | dissipation of the square of these quantities (i.e. enstrophy and the |
---|
618 | variance of the horizontal divergence) as well as the dissipation of the |
---|
619 | horizontal kinetic energy. In particular, when the eddy coefficients are |
---|
620 | horizontally uniform, it ensures a complete separation of vorticity and |
---|
621 | horizontal divergence fields, so that diffusion (dissipation) of vorticity |
---|
622 | (enstrophy) does not generate horizontal divergence (variance of the |
---|
623 | horizontal divergence) and \textit{vice versa}. |
---|
624 | |
---|
625 | These properties of the horizontal diffusive operator are a direct |
---|
626 | consequence of properties \eqref{Eq_DOM_curl_grad} and \eqref{Eq_DOM_div_curl}. When the vertical curl of the horizontal diffusion of momentum (discrete sense) is taken, the term associated to the horizontal gradient of the divergence is zero locally. |
---|
627 | |
---|
628 | % ------------------------------------------------------------------------------------------------------------- |
---|
629 | % Conservation of Potential Vorticity |
---|
630 | % ------------------------------------------------------------------------------------------------------------- |
---|
631 | \subsection{Conservation of Potential Vorticity} |
---|
632 | \label{Apdx_C.3.1} |
---|
633 | |
---|
634 | The lateral momentum diffusion term conserves the potential vorticity : |
---|
635 | \begin{flalign*} |
---|
636 | &\int \limits_D \frac{1} {e_3 } \textbf{k} \cdot \nabla \times |
---|
637 | \Bigl[ \nabla_h |
---|
638 | \left( A^{\,lm}\;\chi \right) |
---|
639 | - \nabla_h \times |
---|
640 | \left( A^{\,lm}\;\zeta \; \textbf{k} \right) |
---|
641 | \Bigr]\;dv = 0 |
---|
642 | \end{flalign*} |
---|
643 | %%%%%%%%%% recheck here.... (gm) |
---|
644 | \begin{flalign*} |
---|
645 | = \int \limits_D -\frac{1} {e_3 } \textbf{k} \cdot \nabla \times |
---|
646 | \Bigl[ \nabla_h \times |
---|
647 | \left( A^{\,lm}\;\zeta \; \textbf{k} \right) |
---|
648 | \Bigr]\;dv &&& \\ |
---|
649 | \end{flalign*} |
---|
650 | \begin{flalign*} |
---|
651 | \equiv& \sum\limits_{i,j} |
---|
652 | \left\{ |
---|
653 | \delta_{i+1/2} |
---|
654 | \left[ |
---|
655 | \frac {e_{2v}} {e_{1v}\,e_{3v}} \delta_i |
---|
656 | \left[ A_f^{\,lm} e_{3f} \zeta \right] |
---|
657 | \right] |
---|
658 | + \delta_{j+1/2} |
---|
659 | \left[ |
---|
660 | \frac {e_{1u}} {e_{2u}\,e_{3u}} \delta_j |
---|
661 | \left[ A_f^{\,lm} e_{3f} \zeta \right] |
---|
662 | \right] |
---|
663 | \right\} |
---|
664 | && \\ |
---|
665 | % |
---|
666 | \intertext{Using \eqref{DOM_di_adj}, it follows:} |
---|
667 | % |
---|
668 | \equiv& \sum\limits_{i,j,k} |
---|
669 | -\,\left\{ |
---|
670 | \frac{e_{2v}} {e_{1v}\,e_{3v}} \delta_i |
---|
671 | \left[ A_f^{\,lm} e_{3f} \zeta \right]\;\delta_i \left[ 1\right] |
---|
672 | + \frac{e_{1u}} {e_{2u}\,e_{3u}} \delta_j |
---|
673 | \left[ A_f^{\,lm} e_{3f} \zeta \right]\;\delta_j \left[ 1\right] |
---|
674 | \right\} \quad \equiv 0 |
---|
675 | && \\ |
---|
676 | \end{flalign*} |
---|
677 | |
---|
678 | % ------------------------------------------------------------------------------------------------------------- |
---|
679 | % Dissipation of Horizontal Kinetic Energy |
---|
680 | % ------------------------------------------------------------------------------------------------------------- |
---|
681 | \subsection{Dissipation of Horizontal Kinetic Energy} |
---|
682 | \label{Apdx_C.3.2} |
---|
683 | |
---|
684 | |
---|
685 | The lateral momentum diffusion term dissipates the horizontal kinetic energy: |
---|
686 | %\begin{flalign*} |
---|
687 | \begin{equation*} |
---|
688 | \begin{split} |
---|
689 | \int_D \textbf{U}_h \cdot |
---|
690 | \left[ \nabla_h \right. & \left. \left( A^{\,lm}\;\chi \right) |
---|
691 | - \nabla_h \times \left( A^{\,lm}\;\zeta \;\textbf{k} \right) \right] \; dv \\ |
---|
692 | \\ %%% |
---|
693 | \equiv& \sum\limits_{i,j,k} |
---|
694 | \left\{ |
---|
695 | \frac{1} {e_{1u}} \delta_{i+1/2} \left[ A_T^{\,lm} \chi \right] |
---|
696 | - \frac{1} {e_{2u}\,e_{3u}} \delta_j \left[ A_f^{\,lm} e_{3f} \zeta \right] |
---|
697 | \right\} \; e_{1u}\,e_{2u}\,e_{3u} \;u \\ |
---|
698 | &\;\; + \left\{ |
---|
699 | \frac{1} {e_{2u}} \delta_{j+1/2} \left[ A_T^{\,lm} \chi \right] |
---|
700 | + \frac{1} {e_{1v}\,e_{3v}} \delta_i \left[ A_f^{\,lm} e_{3f} \zeta \right] |
---|
701 | \right\} \; e_{1v}\,e_{2u}\,e_{3v} \;v \qquad \\ |
---|
702 | \\ %%% |
---|
703 | \equiv& \sum\limits_{i,j,k} |
---|
704 | \Bigl\{ |
---|
705 | e_{2u}\,e_{3u} \;u\; \delta_{i+1/2} \left[ A_T^{\,lm} \chi \right] |
---|
706 | - e_{1u} \;u\; \delta_j \left[ A_f^{\,lm} e_{3f} \zeta \right] |
---|
707 | \Bigl\} |
---|
708 | \\ |
---|
709 | &\;\; + \Bigl\{ |
---|
710 | e_{1v}\,e_{3v} \;v\; \delta_{j+1/2} \left[ A_T^{\,lm} \chi \right] |
---|
711 | + e_{2v} \;v\; \delta_i \left[ A_f^{\,lm} e_{3f} \zeta \right] |
---|
712 | \Bigl\} \\ |
---|
713 | \\ %%% |
---|
714 | \equiv& \sum\limits_{i,j,k} |
---|
715 | - \Bigl( |
---|
716 | \delta_i \left[ e_{2u}\,e_{3u} \;u \right] |
---|
717 | + \delta_j \left[ e_{1v}\,e_{3v} \;v \right] |
---|
718 | \Bigr) \; A_T^{\,lm} \chi \\ |
---|
719 | &\;\; - \Bigl( |
---|
720 | \delta_{i+1/2} \left[ e_{2v} \;v \right] |
---|
721 | - \delta_{j+1/2} \left[ e_{1u} \;u \right] |
---|
722 | \Bigr)\; A_f^{\,lm} e_{3f} \zeta \\ |
---|
723 | \\ %%% |
---|
724 | \equiv& \sum\limits_{i,j,k} |
---|
725 | - A_T^{\,lm} \,\chi^2 \;e_{1T}\,e_{2T}\,e_{3T} |
---|
726 | - A_f ^{\,lm} \,\zeta^2 \;e_{1f }\,e_{2f }\,e_{3f} |
---|
727 | \quad \leq 0 \\ |
---|
728 | \end{split} |
---|
729 | \end{equation*} |
---|
730 | |
---|
731 | % ------------------------------------------------------------------------------------------------------------- |
---|
732 | % Dissipation of Enstrophy |
---|
733 | % ------------------------------------------------------------------------------------------------------------- |
---|
734 | \subsection{Dissipation of Enstrophy} |
---|
735 | \label{Apdx_C.3.3} |
---|
736 | |
---|
737 | |
---|
738 | The lateral momentum diffusion term dissipates the enstrophy when the eddy |
---|
739 | coefficients are horizontally uniform: |
---|
740 | \begin{flalign*} |
---|
741 | &\int\limits_D \zeta \; \textbf{k} \cdot \nabla \times |
---|
742 | \left[ |
---|
743 | \nabla_h |
---|
744 | \left( A^{\,lm}\;\chi \right) |
---|
745 | -\nabla_h \times |
---|
746 | \left( A^{\,lm}\;\zeta \; \textbf{k} \right) |
---|
747 | \right]\;dv &&&\\ |
---|
748 | &= A^{\,lm} \int \limits_D \zeta \textbf{k} \cdot \nabla \times |
---|
749 | \left[ |
---|
750 | \nabla_h \times |
---|
751 | \left( \zeta \; \textbf{k} \right) |
---|
752 | \right]\;dv &&&\displaybreak[0]\\ |
---|
753 | &\equiv A^{\,lm} \sum\limits_{i,j,k} \zeta \;e_{3f} |
---|
754 | \left\{ |
---|
755 | \delta_{i+1/2} |
---|
756 | \left[ |
---|
757 | \frac{e_{2v}} {e_{1v}\,e_{3v}} \delta_i |
---|
758 | \left[ e_{3f} \zeta \right] |
---|
759 | \right] |
---|
760 | + \delta_{j+1/2} |
---|
761 | \left[ |
---|
762 | \frac{e_{1u}} {e_{2u}\,e_{3u}} \delta_j |
---|
763 | \left[ e_{3f} \zeta \right] |
---|
764 | \right] |
---|
765 | \right\} |
---|
766 | &&&\\ |
---|
767 | % |
---|
768 | \intertext{Using \eqref{DOM_di_adj}, it follows:} |
---|
769 | % |
---|
770 | &\equiv - A^{\,lm} \sum\limits_{i,j,k} |
---|
771 | \left\{ |
---|
772 | \left( |
---|
773 | \frac{1} {e_{1v}\,e_{3v}} \delta_i |
---|
774 | \left[ e_{3f} \zeta \right] |
---|
775 | \right)^2 e_{1v}\,e_{2v}\,e_{3v} |
---|
776 | + \left( |
---|
777 | \frac{1} {e_{2u}\,e_{3u}} \delta_j |
---|
778 | \left[ e_{3f} \zeta \right] |
---|
779 | \right)^2 e_{1u}\,e_{2u}\,e_{3u} |
---|
780 | \right\} &&&\\ |
---|
781 | & \leq \;0 &&&\\ |
---|
782 | \end{flalign*} |
---|
783 | |
---|
784 | % ------------------------------------------------------------------------------------------------------------- |
---|
785 | % Conservation of Horizontal Divergence |
---|
786 | % ------------------------------------------------------------------------------------------------------------- |
---|
787 | \subsection{Conservation of Horizontal Divergence} |
---|
788 | \label{Apdx_C.3.4} |
---|
789 | |
---|
790 | When the horizontal divergence of the horizontal diffusion of momentum |
---|
791 | (discrete sense) is taken, the term associated to the vertical curl of the |
---|
792 | vorticity is zero locally, due to (II.1.8). The resulting term conserves the |
---|
793 | $\chi$ and dissipates $\chi^2$ when the eddy coefficients are |
---|
794 | horizontally uniform. |
---|
795 | \begin{flalign*} |
---|
796 | & \int\limits_D \nabla_h \cdot |
---|
797 | \Bigl[ |
---|
798 | \nabla_h |
---|
799 | \left( A^{\,lm}\;\chi \right) |
---|
800 | - \nabla_h \times |
---|
801 | \left( A^{\,lm}\;\zeta \;\textbf{k} \right) |
---|
802 | \Bigr] |
---|
803 | dv |
---|
804 | = \int\limits_D \nabla_h \cdot \nabla_h |
---|
805 | \left( A^{\,lm}\;\chi \right) |
---|
806 | dv |
---|
807 | &&&\\ |
---|
808 | &\equiv \sum\limits_{i,j,k} |
---|
809 | \left\{ |
---|
810 | \delta_i |
---|
811 | \left[ |
---|
812 | A_u^{\,lm} \frac{e_{2u}\,e_{3u}} {e_{1u}} \delta_{i+1/2} |
---|
813 | \left[ \chi \right] |
---|
814 | \right] |
---|
815 | + \delta_j |
---|
816 | \left[ |
---|
817 | A_v^{\,lm} \frac{e_{1v}\,e_{3v}} {e_{2v}} \delta_{j+1/2} |
---|
818 | \left[ \chi \right] |
---|
819 | \right] |
---|
820 | \right\} |
---|
821 | &&&\\ |
---|
822 | % |
---|
823 | \intertext{Using \eqref{DOM_di_adj}, it follows:} |
---|
824 | % |
---|
825 | &\equiv \sum\limits_{i,j,k} |
---|
826 | - \left\{ |
---|
827 | \frac{e_{2u}\,e_{3u}} {e_{1u}} A_u^{\,lm} \delta_{i+1/2} |
---|
828 | \left[ \chi \right] |
---|
829 | \delta_{i+1/2} |
---|
830 | \left[ 1 \right] |
---|
831 | + \frac{e_{1v}\,e_{3v}} {e_{2v}} A_v^{\,lm} \delta_{j+1/2} |
---|
832 | \left[ \chi \right] |
---|
833 | \delta_{j+1/2} |
---|
834 | \left[ 1 \right] |
---|
835 | \right\} \; |
---|
836 | \equiv 0 |
---|
837 | &&& \\ |
---|
838 | \end{flalign*} |
---|
839 | |
---|
840 | % ------------------------------------------------------------------------------------------------------------- |
---|
841 | % Dissipation of Horizontal Divergence Variance |
---|
842 | % ------------------------------------------------------------------------------------------------------------- |
---|
843 | \subsection{Dissipation of Horizontal Divergence Variance} |
---|
844 | \label{Apdx_C.3.5} |
---|
845 | |
---|
846 | \begin{flalign*} |
---|
847 | &\int\limits_D \chi \;\nabla_h \cdot |
---|
848 | \left[ \nabla_h \left( A^{\,lm}\;\chi \right) |
---|
849 | - \nabla_h \times \left( A^{\,lm}\;\zeta \;\textbf{k} \right) \right]\; dv |
---|
850 | = A^{\,lm} \int\limits_D \chi \;\nabla_h \cdot \nabla_h \left( \chi \right)\; dv &&&\\ |
---|
851 | % |
---|
852 | &\equiv A^{\,lm} \sum\limits_{i,j,k} \frac{1} {e_{1T}\,e_{2T}\,e_{3T}} \chi |
---|
853 | \left\{ |
---|
854 | \delta_i \left[ \frac{e_{2u}\,e_{3u}} {e_{1u}} \delta_{i+1/2} \left[ \chi \right] \right] |
---|
855 | + \delta_j \left[ \frac{e_{1v}\,e_{3v}} {e_{2v}} \delta_{j+1/2} \left[ \chi \right] \right] |
---|
856 | \right\} \; e_{1T}\,e_{2T}\,e_{3T} &&&\\ |
---|
857 | % |
---|
858 | \intertext{Using \eqref{DOM_di_adj}, it turns out to be:} |
---|
859 | % |
---|
860 | &\equiv - A^{\,lm} \sum\limits_{i,j,k} |
---|
861 | \left\{ |
---|
862 | \left( \frac{1} {e_{1u}} \delta_{i+1/2} \left[ \chi \right] \right)^2 e_{1u}\,e_{2u}\,e_{3u} |
---|
863 | + \left( \frac{1} {e_{2v}} \delta_{j+1/2} \left[ \chi \right] \right)^2 e_{1v}\,e_{2v}\,e_{3v} |
---|
864 | \right\} \; &&&\\ |
---|
865 | &\leq 0 &&&\\ |
---|
866 | \end{flalign*} |
---|
867 | |
---|
868 | % ================================================================ |
---|
869 | % Conservation Properties on Vertical Momentum Physics |
---|
870 | % ================================================================ |
---|
871 | \section{Conservation Properties on Vertical Momentum Physics} |
---|
872 | \label{Apdx_C_4} |
---|
873 | |
---|
874 | |
---|
875 | As for the lateral momentum physics, the continuous form of the vertical diffusion of momentum satisfies the several integral constraints. The first two are associated to the conservation of momentum and the dissipation of horizontal kinetic energy: |
---|
876 | \begin{align*} |
---|
877 | \int\limits_D |
---|
878 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
879 | \left( |
---|
880 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
881 | \right)\; |
---|
882 | dv \qquad \quad &= \vec{\textbf{0}} |
---|
883 | \\ |
---|
884 | % |
---|
885 | \intertext{and} |
---|
886 | % |
---|
887 | \int\limits_D |
---|
888 | \textbf{U}_h \cdot |
---|
889 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
890 | \left( |
---|
891 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
892 | \right)\; |
---|
893 | dv \quad &\leq 0 |
---|
894 | \\ |
---|
895 | \end{align*} |
---|
896 | The first property is obvious. The second results from: |
---|
897 | |
---|
898 | \begin{flalign*} |
---|
899 | \int\limits_D |
---|
900 | \textbf{U}_h \cdot |
---|
901 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
902 | \left( |
---|
903 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
904 | \right)\;dv |
---|
905 | &&&\\ |
---|
906 | \end{flalign*} |
---|
907 | \begin{flalign*} |
---|
908 | &\equiv \sum\limits_{i,j,k} |
---|
909 | \left( |
---|
910 | u\; \delta_k |
---|
911 | \left[ |
---|
912 | \frac{A_u^{\,vm}} {e_{3uw}} \delta_{k+1/2} |
---|
913 | \left[ u \right] |
---|
914 | \right]\; |
---|
915 | e_{1u}\,e_{2u} |
---|
916 | + v\;\delta_k |
---|
917 | \left[ |
---|
918 | \frac{A_v^{\,vm}} {e_{3vw}} \delta_{k+1/2} |
---|
919 | \left[ v \right] |
---|
920 | \right]\; |
---|
921 | e_{1v}\,e_{2v} |
---|
922 | \right) |
---|
923 | &&&\\ |
---|
924 | % |
---|
925 | \intertext{as the horizontal scale factor do not depend on $k$, it follows:} |
---|
926 | % |
---|
927 | &\equiv - \sum\limits_{i,j,k} |
---|
928 | \left( |
---|
929 | \frac{A_u^{\,vm}} {e_{3uw}} |
---|
930 | \left( |
---|
931 | \delta_{k+1/2} |
---|
932 | \left[ u \right] |
---|
933 | \right)^2\; |
---|
934 | e_{1u}\,e_{2u} |
---|
935 | + \frac{A_v^{\,vm}} {e_{3vw}} |
---|
936 | \left( \delta_{k+1/2} |
---|
937 | \left[ v \right] |
---|
938 | \right)^2\; |
---|
939 | e_{1v}\,e_{2v} |
---|
940 | \right) |
---|
941 | \quad \leq 0 |
---|
942 | &&&\\ |
---|
943 | \end{flalign*} |
---|
944 | |
---|
945 | The vorticity is also conserved. Indeed: |
---|
946 | \begin{flalign*} |
---|
947 | \int \limits_D |
---|
948 | \frac{1} {e_3 } \textbf{k} \cdot \nabla \times |
---|
949 | \left( |
---|
950 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
951 | \left( |
---|
952 | \frac{A^{\,vm}} {e_3}\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
953 | \right) |
---|
954 | \right)\; |
---|
955 | dv |
---|
956 | &&&\\ |
---|
957 | \end{flalign*} |
---|
958 | \begin{flalign*} |
---|
959 | \equiv \sum\limits_{i,j,k} \frac{1} {e_{3f}}\; \frac{1} {e_{1f}\,e_{2f}} |
---|
960 | \bigg\{ \biggr. \quad |
---|
961 | \delta_{i+1/2} |
---|
962 | &\left( |
---|
963 | \frac{e_{2v}} {e_{3v}} \delta_k |
---|
964 | \left[ |
---|
965 | \frac{1} {e_{3vw}} \delta_{k+1/2} |
---|
966 | \left[ v \right] |
---|
967 | \right] |
---|
968 | \right) |
---|
969 | &&\\ |
---|
970 | \biggl. |
---|
971 | - \delta_{j+1/2} |
---|
972 | &\left( |
---|
973 | \frac{e_{1u}} {e_{3u}} \delta_k |
---|
974 | \left[ |
---|
975 | \frac{1} {e_{3uw}}\delta_{k+1/2} |
---|
976 | \left[ u \right] |
---|
977 | \right] |
---|
978 | \right) |
---|
979 | \biggr\} \; |
---|
980 | e_{1f}\,e_{2f}\,e_{3f} \; \equiv 0 |
---|
981 | && \\ |
---|
982 | \end{flalign*} |
---|
983 | If the vertical diffusion coefficient is uniform over the whole domain, the |
---|
984 | enstrophy is dissipated, i.e. |
---|
985 | \begin{flalign*} |
---|
986 | \int\limits_D \zeta \, \textbf{k} \cdot \nabla \times |
---|
987 | \left( |
---|
988 | \frac{1} {e_3}\; \frac{\partial } {\partial k} |
---|
989 | \left( |
---|
990 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
991 | \right) |
---|
992 | \right)\; |
---|
993 | dv = 0 |
---|
994 | &&&\\ |
---|
995 | \end{flalign*} |
---|
996 | This property is only satisfied in $z$-coordinates: |
---|
997 | |
---|
998 | \begin{flalign*} |
---|
999 | \int\limits_D \zeta \, \textbf{k} \cdot \nabla \times |
---|
1000 | \left( |
---|
1001 | \frac{1} {e_3}\; \frac{\partial } {\partial k} |
---|
1002 | \left( |
---|
1003 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
1004 | \right) |
---|
1005 | \right)\; |
---|
1006 | dv |
---|
1007 | &&& \\ |
---|
1008 | \end{flalign*} |
---|
1009 | \begin{flalign*} |
---|
1010 | \equiv \sum\limits_{i,j,k} \zeta \;e_{3f} \; |
---|
1011 | \biggl\{ \biggr. \quad |
---|
1012 | \delta_{i+1/2} |
---|
1013 | &\left( |
---|
1014 | \frac{e_{2v}} {e_{3v}} \delta_k |
---|
1015 | \left[ |
---|
1016 | \frac{A_v^{\,vm}} {e_{3vw}} \delta_{k+1/2} |
---|
1017 | \left[ v \right] |
---|
1018 | \right] |
---|
1019 | \right) |
---|
1020 | &&\\ |
---|
1021 | - \delta_{j+1/2} |
---|
1022 | &\biggl. |
---|
1023 | \left( |
---|
1024 | \frac{e_{1u}} {e_{3u}} \delta_k |
---|
1025 | \left[ |
---|
1026 | \frac{A_u^{\,vm}} {e_{3uw}} \delta_{k+1/2} |
---|
1027 | \left[ u \right] |
---|
1028 | \right] |
---|
1029 | \right) |
---|
1030 | \biggr\} |
---|
1031 | &&\\ |
---|
1032 | \end{flalign*} |
---|
1033 | \begin{flalign*} |
---|
1034 | \equiv \sum\limits_{i,j,k} \zeta \;e_{3f} |
---|
1035 | \biggl\{ \biggr. \quad |
---|
1036 | \frac{1} {e_{3v}} \delta_k |
---|
1037 | &\left[ |
---|
1038 | \frac{A_v^{\,vm}} {e_{3vw}} \delta_{k+1/2} |
---|
1039 | \left[ \delta_{i+1/2} |
---|
1040 | \left[ e_{2v}\,v \right] |
---|
1041 | \right] |
---|
1042 | \right] |
---|
1043 | &&\\ |
---|
1044 | \biggl. |
---|
1045 | - \frac{1} {e_{3u}} \delta_k |
---|
1046 | &\left[ |
---|
1047 | \frac{A_u^{\,vm}} {e_{3uw}} \delta_{k+1/2} |
---|
1048 | \left[ \delta_{j+1/2} |
---|
1049 | \left[ e_{1u}\,u \right] |
---|
1050 | \right] |
---|
1051 | \right] |
---|
1052 | \biggr\} |
---|
1053 | &&\\ |
---|
1054 | \end{flalign*} |
---|
1055 | Using the fact that the vertical diffusive coefficients are uniform and that in $z$-coordinates, the vertical scale factors do not depends on $i$ and $j$ so that: $e_{3f} =e_{3u} =e_{3v} =e_{3T} $ and $e_{3w} =e_{3uw} =e_{3vw} $, it follows: |
---|
1056 | \begin{flalign*} |
---|
1057 | \equiv A^{\,vm} \sum\limits_{i,j,k} \zeta \;\delta_k |
---|
1058 | \left[ |
---|
1059 | \frac{1} {e_{3w}} \delta_{k+1/2} |
---|
1060 | \Bigl[ |
---|
1061 | \delta_{i+1/2} |
---|
1062 | \left[ e_{2v}\,v \right] |
---|
1063 | - \delta_{j+1/ 2} |
---|
1064 | \left[ e_{1u}\,u \right] |
---|
1065 | \Bigr] |
---|
1066 | \right] |
---|
1067 | &&&\\ |
---|
1068 | \end{flalign*} |
---|
1069 | \begin{flalign*} |
---|
1070 | \equiv - A^{\,vm} \sum\limits_{i,j,k} \frac{1} {e_{3w}} |
---|
1071 | \left( |
---|
1072 | \delta_{k+1/2} |
---|
1073 | \left[ \zeta \right] |
---|
1074 | \right)^2 \; |
---|
1075 | e_{1f}\,e_{2f} |
---|
1076 | \; \leq 0 |
---|
1077 | &&&\\ |
---|
1078 | \end{flalign*} |
---|
1079 | Similarly, the horizontal divergence is obviously conserved: |
---|
1080 | |
---|
1081 | \begin{flalign*} |
---|
1082 | \int\limits_D \nabla \cdot |
---|
1083 | \left( |
---|
1084 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
1085 | \left( |
---|
1086 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
1087 | \right) |
---|
1088 | \right)\; |
---|
1089 | dv = 0 |
---|
1090 | &&&\\ |
---|
1091 | \end{flalign*} |
---|
1092 | and the square of the horizontal divergence decreases (i.e. the horizontal divergence is dissipated) if vertical diffusion coefficient is uniform over the whole domain: |
---|
1093 | |
---|
1094 | \begin{flalign*} |
---|
1095 | \int\limits_D \chi \;\nabla \cdot |
---|
1096 | \left( |
---|
1097 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
1098 | \left( |
---|
1099 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
1100 | \right) |
---|
1101 | \right)\; |
---|
1102 | dv = 0 |
---|
1103 | &&&\\ |
---|
1104 | \end{flalign*} |
---|
1105 | This property is only satisfied in $z$-coordinates: |
---|
1106 | |
---|
1107 | \begin{flalign*} |
---|
1108 | \int\limits_D \chi \;\nabla \cdot |
---|
1109 | \left( |
---|
1110 | \frac{1} {e_3 }\; \frac{\partial } {\partial k} |
---|
1111 | \left( |
---|
1112 | \frac{A^{\,vm}} {e_3 }\; \frac{\partial \textbf{U}_h } {\partial k} |
---|
1113 | \right) |
---|
1114 | \right)\; |
---|
1115 | dv |
---|
1116 | &&&\\ |
---|
1117 | \end{flalign*} |
---|
1118 | \begin{flalign*} |
---|
1119 | \equiv \sum\limits_{i,j,k} \frac{\chi } {e_{1T}\,e_{2T}} |
---|
1120 | \biggl\{ \Biggr. \quad |
---|
1121 | \delta_{i+1/2} |
---|
1122 | &\left( |
---|
1123 | \frac{e_{2u}} {e_{3u}} \delta_k |
---|
1124 | \left[ |
---|
1125 | \frac{A_u^{\,vm}} {e_{3uw}} \delta_{k+1/2} |
---|
1126 | \left[ u \right] |
---|
1127 | \right] |
---|
1128 | \right) |
---|
1129 | &&\\ |
---|
1130 | \Biggl. |
---|
1131 | + \delta_{j+1/2} |
---|
1132 | &\left( |
---|
1133 | \frac{e_{1v}} {e_{3v}} \delta_k |
---|
1134 | \left[ |
---|
1135 | \frac{A_v^{\,vm}} {e_{3vw}} \delta_{k+1/2} |
---|
1136 | \left[ v \right] |
---|
1137 | \right] |
---|
1138 | \right) |
---|
1139 | \Biggr\} \; |
---|
1140 | e_{1T}\,e_{2T}\,e_{3T} |
---|
1141 | &&\\ |
---|
1142 | \end{flalign*} |
---|
1143 | |
---|
1144 | \begin{flalign*} |
---|
1145 | \equiv A^{\,vm} \sum\limits_{i,j,k} \chi \, |
---|
1146 | \biggl\{ \biggr. \quad |
---|
1147 | \delta_{i+1/2} |
---|
1148 | &\left( |
---|
1149 | \delta_k |
---|
1150 | \left[ |
---|
1151 | \frac{1} {e_{3uw}} \delta_{k+1/2} |
---|
1152 | \left[ e_{2u}\,u \right] |
---|
1153 | \right] |
---|
1154 | \right) |
---|
1155 | && \\ |
---|
1156 | \biggl. |
---|
1157 | + \delta_{j+1/2} |
---|
1158 | &\left( |
---|
1159 | \delta_k |
---|
1160 | \left[ |
---|
1161 | \frac{1} {e_{3vw}} \delta_{k+1/2} |
---|
1162 | \left[ e_{1v}\,v \right] |
---|
1163 | \right] |
---|
1164 | \right) |
---|
1165 | \biggr\} |
---|
1166 | && \\ |
---|
1167 | \end{flalign*} |
---|
1168 | |
---|
1169 | \begin{flalign*} |
---|
1170 | \equiv -A^{\,vm} \sum\limits_{i,j,k} |
---|
1171 | \frac{\delta_{k+1/2} \left[ \chi \right]} {e_{3w}}\; |
---|
1172 | \biggl\{ |
---|
1173 | \delta_{k+1/2} |
---|
1174 | \Bigl[ |
---|
1175 | \delta_{i+1/2} |
---|
1176 | \left[ e_{2u}\,u \right] |
---|
1177 | + \delta_{j+1/2} |
---|
1178 | \left[ e_{1v}\,v \right] |
---|
1179 | \Bigr] |
---|
1180 | \biggr\} |
---|
1181 | &&&\\ |
---|
1182 | \end{flalign*} |
---|
1183 | |
---|
1184 | \begin{flalign*} |
---|
1185 | \equiv -A^{\,vm} \sum\limits_{i,j,k} |
---|
1186 | \frac{1} {e_{3w}} |
---|
1187 | \delta_{k+1/2} |
---|
1188 | \left[ \chi \right]\; |
---|
1189 | \delta_{k+1/2} |
---|
1190 | \left[ e_{1T}\,e_{2T} \;\chi \right] |
---|
1191 | &&&\\ |
---|
1192 | \end{flalign*} |
---|
1193 | |
---|
1194 | \begin{flalign*} |
---|
1195 | \equiv -A^{\,vm} \sum\limits_{i,j,k} |
---|
1196 | \frac{e_{1T}\,e_{2T}} {e_{3w}}\; |
---|
1197 | \left( |
---|
1198 | \delta_{k+1/2} |
---|
1199 | \left[ \chi \right] |
---|
1200 | \right)^2 |
---|
1201 | \quad \equiv 0 |
---|
1202 | &&&\\ |
---|
1203 | \end{flalign*} |
---|
1204 | |
---|
1205 | % ================================================================ |
---|
1206 | % Conservation Properties on Tracer Physics |
---|
1207 | % ================================================================ |
---|
1208 | \section{Conservation Properties on Tracer Physics} |
---|
1209 | \label{Apdx_C.5} |
---|
1210 | |
---|
1211 | |
---|
1212 | |
---|
1213 | The numerical schemes used for tracer subgridscale physics are written such that the heat and salt contents are conserved (equations in flux form, second order centered finite differences). As a form flux is used to compute the temperature and salinity, the quadratic form of these quantities (i.e. their variance) globally tends to diminish. As for the advection term, there is generally no strict conservation of mass even if, in practice, the mass is conserved with a very good accuracy. |
---|
1214 | |
---|
1215 | % ------------------------------------------------------------------------------------------------------------- |
---|
1216 | % Conservation of Tracers |
---|
1217 | % ------------------------------------------------------------------------------------------------------------- |
---|
1218 | \subsection{Conservation of Tracers} |
---|
1219 | \label{Apdx_C.5.1} |
---|
1220 | |
---|
1221 | constraint of conservation of tracers: |
---|
1222 | \begin{flalign*} |
---|
1223 | &\int\limits_D \nabla \cdot \left( A\;\nabla T \right)\;dv &&&\\ |
---|
1224 | \\ |
---|
1225 | &\equiv \sum\limits_{i,j,k} |
---|
1226 | \biggl\{ \biggr. |
---|
1227 | \delta_i |
---|
1228 | \left[ |
---|
1229 | A_u^{\,lT} \frac{e_{2u}\,e_{3u}} {e_{1u}} \delta_{i+1/2} |
---|
1230 | \left[ T \right] |
---|
1231 | \right] |
---|
1232 | + \delta_j |
---|
1233 | \left[ |
---|
1234 | A_v^{\,lT} \frac{e_{1v}\,e_{3v}} {e_{2v}} \delta_{j+1/2} |
---|
1235 | \left[ T \right] |
---|
1236 | \right] |
---|
1237 | &&\\ & \qquad \qquad \qquad \qquad \qquad \qquad \quad \;\;\; |
---|
1238 | + \delta_k |
---|
1239 | \left[ |
---|
1240 | A_w^{\,vT} \frac{e_{1T}\,e_{2T}} {e_{3T}} \delta_{k+1/2} |
---|
1241 | \left[ T \right] |
---|
1242 | \right] |
---|
1243 | \biggr\} \quad \equiv 0 |
---|
1244 | &&\\ |
---|
1245 | \end{flalign*} |
---|
1246 | |
---|
1247 | In fact, this property is simply resulting from the flux form of the operator. |
---|
1248 | |
---|
1249 | % ------------------------------------------------------------------------------------------------------------- |
---|
1250 | % Dissipation of Tracer Variance |
---|
1251 | % ------------------------------------------------------------------------------------------------------------- |
---|
1252 | \subsection{Dissipation of Tracer Variance} |
---|
1253 | \label{Apdx_C.5.2} |
---|
1254 | |
---|
1255 | constraint of dissipation of tracer variance: |
---|
1256 | \begin{flalign*} |
---|
1257 | \int\limits_D T\;\nabla & \cdot \left( A\;\nabla T \right)\;dv &&&\\ |
---|
1258 | &\equiv \sum\limits_{i,j,k} \; T |
---|
1259 | \biggl\{ \biggr. |
---|
1260 | \delta_i \left[ A_u^{\,lT} \frac{e_{2u}\,e_{3u}} {e_{1u}} \delta_{i+1/2} \left[T\right] \right] |
---|
1261 | & + \delta_j \left[ A_v^{\,lT} \frac{e_{1v} \,e_{3v}} {e_{2v}} \delta_{j+1/2} \left[T\right] \right] |
---|
1262 | \quad&& \\ |
---|
1263 | \biggl. |
---|
1264 | &&+ \delta_k \left[A_w^{\,vT}\frac{e_{1T}\,e_{2T}} {e_{3T}}\delta_{k+1/2}\left[T\right]\right] |
---|
1265 | \biggr\} && |
---|
1266 | \end{flalign*} |
---|
1267 | \begin{flalign*} |
---|
1268 | \equiv - \sum\limits_{i,j,k} |
---|
1269 | \biggl\{ \biggr. \quad |
---|
1270 | & A_u^{\,lT} |
---|
1271 | \left( |
---|
1272 | \frac{1} {e_{1u}} \delta_{i+1/2} |
---|
1273 | \left[ T \right] |
---|
1274 | \right)^2 |
---|
1275 | e_{1u}\,e_{2u}\,e_{3u} |
---|
1276 | && \\ |
---|
1277 | & + A_v^{\,lT} |
---|
1278 | \left( |
---|
1279 | \frac{1} {e_{2v}} \delta_{j+1/2} |
---|
1280 | \left[ T \right] |
---|
1281 | \right)^2 |
---|
1282 | e_{1v}\,e_{2v}\,e_{3v} |
---|
1283 | && \\ |
---|
1284 | \biggl. |
---|
1285 | & + A_w^{\,vT} |
---|
1286 | \left( |
---|
1287 | \frac{1} {e_{3w}} \delta_{k+1/2} |
---|
1288 | \left[ T \right] |
---|
1289 | \right)^2 |
---|
1290 | e_{1w}\,e_{2w}\,e_{3w} |
---|
1291 | \biggr\} |
---|
1292 | \quad \leq 0 |
---|
1293 | && \\ |
---|
1294 | \end{flalign*} |
---|
1295 | |
---|
1296 | |
---|
1297 | %%%% end of appendix in gm comment |
---|
1298 | %} |
---|