| 1 |
% $Header: /u/gcmpack/manual/part3/case_studies/held_suarez_cs/held_suarez_cs.tex,v 1.6 2006/06/27 19:08:22 molod Exp $ |
| 2 |
% $Name: $ |
| 3 |
|
| 4 |
\section[Held-Suarez Atmosphere MITgcm Example]{Held-Suarez atmospheric simulation |
| 5 |
on cube-sphere grid with 32 square cube faces.} |
| 6 |
\label{www:tutorials} |
| 7 |
\label{sect:eg-hs} |
| 8 |
\begin{rawhtml} |
| 9 |
<!-- CMIREDIR:eg-hs: --> |
| 10 |
\end{rawhtml} |
| 11 |
|
| 12 |
\bodytext{bgcolor="#FFFFFFFF"} |
| 13 |
|
| 14 |
%\begin{center} |
| 15 |
%{\Large \bf Using MITgcm to Simulate Global Climatological Ocean Circulation |
| 16 |
%At Four Degree Resolution with Asynchronous Time Stepping} |
| 17 |
% |
| 18 |
%\vspace*{4mm} |
| 19 |
% |
| 20 |
%\vspace*{3mm} |
| 21 |
%{\large May 2001} |
| 22 |
%\end{center} |
| 23 |
|
| 24 |
This example illustrates the use of the MITgcm as an Atmospheric GCM, |
| 25 |
using simple \cite{held-suar:94} forcing |
| 26 |
to simulate Atmospheric Dynamics on global scale. |
| 27 |
The set-up use the rescaled pressure coordinate ($p^*$)\cite[]{adcroft:04a} |
| 28 |
in the vertical direction, with 20 equaly-spaced levels, and |
| 29 |
the conformal cube-sphere grid (C32) \cite[]{adcroft:04b}. |
| 30 |
The files for this experiment can be found in the verification directory |
| 31 |
under tutorial\_held\_suarez\_cs. |
| 32 |
|
| 33 |
\subsection{Overview} |
| 34 |
\label{www:tutorials} |
| 35 |
|
| 36 |
This example demonstrates using the MITgcm to simulate |
| 37 |
the planetary atmospheric circulation, with flat orography |
| 38 |
and simplified forcing. |
| 39 |
In particular, only dry air processes are considered and |
| 40 |
radiation effects are represented by a simple newtownien cooling, |
| 41 |
Thus this example does not rely on any particular atmospheric |
| 42 |
physics package. |
| 43 |
This kind of simplified atmospheric simulation has been widely |
| 44 |
used in GFD-type experiments and in intercomparison projects of |
| 45 |
AGCM dynamical cores \cite[]{held-suar:94}. |
| 46 |
|
| 47 |
The horizontal grid is obtain from the projection of a uniform gridded cube |
| 48 |
to the sphere. Each of the 6 faces has the same resolution, with |
| 49 |
$32 \times 32$ grid points. The equator line coincide with a grid line |
| 50 |
and crosses, right in the midle, 4 of the 6 faces, leaving 2 faces |
| 51 |
for the Northern and Southern polar regions. |
| 52 |
This curvilinear grid requires the use of the 2nd generation exchange |
| 53 |
topology ({\it pkg/exch2}) to connect tile and face edges, |
| 54 |
but without any limitation on the number of processors. |
| 55 |
|
| 56 |
The use of the $p^*$ coordinate with 20 equally spaced levels |
| 57 |
($20 \times 50\,{\rm mb}$, from $p^*=1000,{\rm mb}$ to $0$ at the |
| 58 |
top of the atmosphere) follows the choice of \cite{held-suar:94}. |
| 59 |
Note that without topography, the $p^*$ coordinate and the normalized |
| 60 |
pressure coordinate ($\sigma_p$) coincide exactly. |
| 61 |
No viscosity and zero diffusion are used here, but |
| 62 |
a $8^th$ order \cite{Shapiro_70} filter is applied to both momentum and |
| 63 |
potential temperature, to remove selectively grid scale noise. |
| 64 |
Apart from the horizontal grid, this experiment is made very similar to |
| 65 |
the grid-point model case used in \cite{held-suar:94} study. |
| 66 |
|
| 67 |
At this resolution, the configuration can be integrated forward |
| 68 |
for many years on a single processor desktop computer. |
| 69 |
\\ |
| 70 |
|
| 71 |
\subsection{Forcing} |
| 72 |
\label{www:tutorials} |
| 73 |
|
| 74 |
The model is forced by relaxation to a radiative equilibrium temperature from |
| 75 |
\cite{held-suar:94}. |
| 76 |
A linear frictional drag (Rayleigh damping) is applied in the lower |
| 77 |
part of the atmosphere and account from surface friction and momentum |
| 78 |
dissipation in the boundary layer. |
| 79 |
Altogether, this yields the following forcing |
| 80 |
\cite[from][]{held-suar:94} that is applied to the fluid: |
| 81 |
|
| 82 |
\begin{eqnarray} |
| 83 |
\label{EQ:eg-hs-global_forcing} |
| 84 |
\label{EQ:eg-hs-global_forcing_fv} |
| 85 |
\vec{{\cal F}_\mathbf{v}} & = & -k_\mathbf{v}(p)\vec{\mathbf{v}}_h |
| 86 |
\\ |
| 87 |
\label{EQ:eg-hs-global_forcing_ft} |
| 88 |
{\cal F}_{\theta} & = & -k_{\theta}(\varphi,p)[\theta-\theta_{eq}(\varphi,p)] |
| 89 |
\end{eqnarray} |
| 90 |
|
| 91 |
\noindent where $\vec{\cal F}_\mathbf{v}$, ${\cal F}_{\theta}$, |
| 92 |
are the forcing terms in the zonal and meridional |
| 93 |
momentum and in the potential temperature equations respectively. |
| 94 |
The term $k_\mathbf{v}$ in equation (\ref{EQ:eg-hs-global_forcing_fv}) applies a |
| 95 |
Rayleigh damping that is active within the planetary boundary layer. |
| 96 |
It is defined so as to decay as pressure decreases according to |
| 97 |
\begin{eqnarray*} |
| 98 |
\label{EQ:eg-hs-define_kv} |
| 99 |
k_\mathbf{v} & = & k_{f}~\max[0,~(p^*/P^{0}_{s}-\sigma_{b})/(1-\sigma_{b})] |
| 100 |
\\ |
| 101 |
\sigma_{b} & = & 0.7 ~~{\rm and}~~ |
| 102 |
k_{f} = 1/86400 ~{\rm s}^{-1} |
| 103 |
\end{eqnarray*} |
| 104 |
|
| 105 |
where $p^*$ is the pressure level of the cell center |
| 106 |
and $P^{0}_{s}$ is the pressure at the base of the atmospheric column, |
| 107 |
which is constant and uniform here ($= 10^5 {\rm Pa}$), in the absence |
| 108 |
of topography. |
| 109 |
|
| 110 |
The Equilibrium temperature $\theta_{eq}$ and relaxation time scale $k_{\theta}$ |
| 111 |
are set to: |
| 112 |
\begin{eqnarray} |
| 113 |
\label{EQ:eg-hs-define_Teq} |
| 114 |
\theta_{eq}(\varphi,p^*) & = & \max \{ 200.K (P^{0}_{s}/p^*)^\kappa,\\ |
| 115 |
\nonumber |
| 116 |
& & \hspace{8mm} 315.K - \Delta T_y~\sin^2(\varphi) |
| 117 |
- \Delta \theta_z \cos^2(\varphi) \log(p^*/P^{0}_s) \} |
| 118 |
\\ |
| 119 |
\label{EQ:eg-hs-define_kT} |
| 120 |
k_{\theta}(\varphi,p^*) & = & |
| 121 |
k_a + (k_s -k_a)~\cos^4(\varphi)~\max[0,(p^*/P^{0}_{s}-\sigma_{b})/(1-\sigma_{b})] |
| 122 |
\end{eqnarray} |
| 123 |
with: |
| 124 |
\begin{eqnarray*} |
| 125 |
\Delta T_y = 60.K & k_a = 1/(40 \cdot 86400) ~{\rm s}^{-1}\\ |
| 126 |
\Delta \theta_z = 10.K & k_s = 1/(4 \cdot 86400) ~{\rm s}^{-1} |
| 127 |
\end{eqnarray*} |
| 128 |
|
| 129 |
Initial conditions correspond to a resting state with horizontally uniform |
| 130 |
stratified fluid. The initial temperature profile is simply the |
| 131 |
horizontally average of the radiative equilibrium temperature. |
| 132 |
|
| 133 |
\subsection{Set-up description} |
| 134 |
%\subsection{Discrete Numerical Configuration} |
| 135 |
\label{www:tutorials} |
| 136 |
|
| 137 |
The model is configured in hydrostatic form, using non-boussinesq |
| 138 |
$p^*$ coordinate. |
| 139 |
The vertical resolution is uniform, $\Delta p^* = 50.10^2 Pa$, |
| 140 |
with 20 levels, from $p^*=10^5 Pa$ to $0$ at the top. |
| 141 |
The domain is discretised using C32 cube-sphere grid \cite[]{adcroft:04b} |
| 142 |
that cover the whole sphere with a relatively uniform grid-spacing. |
| 143 |
The resolution at the equator or along the Greenwitch meridian |
| 144 |
is similar to the $128 \times 64$ equaly spaced longitude-latitude grid, |
| 145 |
but requires $25\%$ less grid points. |
| 146 |
Grid spacing and grid-point location are not computed by the model but |
| 147 |
read from files. |
| 148 |
|
| 149 |
The vector-invariant form of the momentum equation (see section |
| 150 |
\ref{sect:vect-inv_momentum_equations}) is used so that no explicit |
| 151 |
metrics are necessary. |
| 152 |
|
| 153 |
Applying the vector-invariant discretization to the |
| 154 |
atmospheric equations \ref{eq:atmos-prime}, and adding the |
| 155 |
forcing term |
| 156 |
(\ref{EQ:eg-hs-global_forcing_fv}, \ref{EQ:eg-hs-global_forcing_ft}) |
| 157 |
on the right-hand-side, |
| 158 |
leads to the set of equations that are solved in this configuration: |
| 159 |
|
| 160 |
%the The set of equations solved here is der |
| 161 |
%Wind-stress forcing is added to the momentum equations for both |
| 162 |
%the zonal flow, $u$ and the meridional flow $v$, according to equations |
| 163 |
%(\ref{EQ:eg-hs-global_forcing_fv}) and (\ref{EQ:eg-hs-global_forcing_fv}). |
| 164 |
%Thermodynamic forcing inputs are added to the equations for |
| 165 |
%potential temperature, $\theta$, and salinity, $S$, according to equations |
| 166 |
%(\ref{EQ:eg-hs-global_forcing_ft}) and (\ref{EQ:eg-hs-global_forcing_fs}). |
| 167 |
|
| 168 |
\begin{eqnarray} |
| 169 |
\label{EQ:eg-hs-model_equations} |
| 170 |
\frac{\partial \vec{\mathbf{v}}_h}{\partial t} |
| 171 |
+(f + \zeta)\hat{\mathbf{k}} \times \vec{\mathbf{v}}_h |
| 172 |
%+\mathbf{\nabla }_{p} ({\rm KE}) |
| 173 |
+\mathbf{\nabla }_{p} (\mbox{\sc ke}) |
| 174 |
+ \omega \frac{\partial \vec{\mathbf{v}}_h }{\partial p} |
| 175 |
+\mathbf{\nabla }_p \Phi ^{\prime } |
| 176 |
&=& |
| 177 |
% \vec{{\cal F}_v} = |
| 178 |
-k_\mathbf{v}\vec{\mathbf{v}}_h |
| 179 |
\\ |
| 180 |
\frac{\partial \Phi ^{\prime }}{\partial p} |
| 181 |
+\frac{\partial \Pi }{\partial p}\theta ^{\prime } &=&0 |
| 182 |
\\ |
| 183 |
\mathbf{\nabla }_{p}\cdot \vec{\mathbf{v}}_h+\frac{\partial \omega }{ |
| 184 |
\partial p} &=&0 |
| 185 |
\\ |
| 186 |
\frac{\partial \theta }{\partial t} |
| 187 |
+ \mathbf{\nabla }_{p}\cdot (\theta \vec{\mathbf{v}}_h) |
| 188 |
+ \frac{\partial (\theta \omega)}{\partial p} |
| 189 |
%= \frac{\mathcal{Q}}{\Pi } |
| 190 |
&=& -k_{\theta}[\theta-\theta_{eq}] |
| 191 |
\end{eqnarray} |
| 192 |
|
| 193 |
%\begin{equation} |
| 194 |
%\partial_t \vec{v} + ( 2\vec{\Omega} + \vec{\zeta}) \wedge \vec{v} |
| 195 |
%- b \hat{r} |
| 196 |
%+ \vec{\nabla} B = \vec{\nabla} \cdot \vec{\bf \tau} |
| 197 |
%\end{equation} |
| 198 |
%{\cal F}_{\theta} & = & -k_{\theta}(\varphi,p)[\theta-\theta_{eq}(\varphi,p)] |
| 199 |
|
| 200 |
\noindent where $\vec{\mathbf{v}}_h$ and $\omega = \frac{Dp}{Dt}$ |
| 201 |
are the horizontal velocity vector and the vertical velocity in pressure coordinate, |
| 202 |
$\zeta$ is the relative vorticity and $f$ the Coriolis parameter, |
| 203 |
$\hat{\mathbf{k}}$ is the vertical unity vector, |
| 204 |
{\sc ke} is the kinetic energy, $\Phi$ is the geopotential |
| 205 |
and $\Pi$ the Exner function |
| 206 |
($\Pi = C_p (p/p_c)^\kappa ~{\rm with}~ p_c = 10^5 Pa$). |
| 207 |
Variables marked with ' corresponds to anomaly from |
| 208 |
the resting, uniformly stratified state. |
| 209 |
|
| 210 |
As described in MITgcm Numerical Solution Procedure \ref{chap:discretization}, |
| 211 |
the continuity equation is integrated vertically, to give a prognostic |
| 212 |
equation for the surface pressure $p_s$: |
| 213 |
\begin{equation} |
| 214 |
\frac{\partial p_s}{\partial t} + \nabla_{h}\cdot \int_{0}^{p_s} \vec{\mathbf{v}}_h dp |
| 215 |
= 0 |
| 216 |
\end{equation} |
| 217 |
|
| 218 |
The implicit free surface form of the pressure equation described in |
| 219 |
\cite{marshall:97a} is employed to solve for $p_s$; |
| 220 |
Integrating vertically the hydrostatic balance |
| 221 |
gives the geopotential $\Phi'$ and allow to step forward the momentum equation |
| 222 |
\ref{EQ:eg-hs-model_equations}. |
| 223 |
The potential temperature, $\theta$, is stepped forward using the |
| 224 |
new velocity field ({\it staggered time-step}, section |
| 225 |
\ref{sect:adams-bashforth-staggered}). |
| 226 |
\\ |
| 227 |
|
| 228 |
\subsubsection{Numerical Stability Criteria} |
| 229 |
\label{www:tutorials} |
| 230 |
|
| 231 |
\noindent The numerical stability for inertial oscillations |
| 232 |
\cite{adcroft:95} |
| 233 |
|
| 234 |
\begin{eqnarray} |
| 235 |
\label{EQ:eg-hs-inertial_stability} |
| 236 |
S_{i} = f^{2} {\Delta t}^2 |
| 237 |
\end{eqnarray} |
| 238 |
|
| 239 |
\noindent evaluates to $4.\times10^{-3}$ at the poles, |
| 240 |
for $f=2\Omega\sin(\pi / 2) =1.45\times10^{-4}~{\rm s}^{-1}$, |
| 241 |
which is well below the $S_{i} < 1$ upper limit for stability. |
| 242 |
\\ |
| 243 |
|
| 244 |
\noindent The advective CFL \cite{adcroft:95} |
| 245 |
for a extreme maximum horizontal flow speed of $ | \vec{u} | = 90. {\rm m/s}$~ |
| 246 |
and the smallest horizontal grid spacing $ \Delta x = 1.1\times10^5 {\rm m}$~: |
| 247 |
|
| 248 |
\begin{eqnarray} |
| 249 |
\label{EQ:eg-hs-cfl_stability} |
| 250 |
S_{a} = \frac{| \vec{u} | \Delta t}{ \Delta x} |
| 251 |
\end{eqnarray} |
| 252 |
|
| 253 |
\noindent evaluates to $0.37$, which is close to the stability |
| 254 |
limit of 0.5. |
| 255 |
\\ |
| 256 |
|
| 257 |
\noindent The stability parameter for internal gravity waves propagating |
| 258 |
with a maximum speed of $c_{g}=100~{\rm m/s}$ |
| 259 |
\cite{adcroft:95} |
| 260 |
|
| 261 |
\begin{eqnarray} |
| 262 |
\label{EQ:eg-hs-gfl_stability} |
| 263 |
S_{c} = \frac{c_{g} \Delta t}{ \Delta x} |
| 264 |
\end{eqnarray} |
| 265 |
|
| 266 |
\noindent evaluates to $4 \times 10^{-1}$. This is close to the linear |
| 267 |
stability limit of 0.5. |
| 268 |
|
| 269 |
\subsection{Experiment Configuration} |
| 270 |
\label{www:tutorials} |
| 271 |
\label{SEC:eg-hs_examp_exp_config} |
| 272 |
|
| 273 |
The model configuration for this experiment resides under the |
| 274 |
directory {\it verification/tutorial\_held\_suarez\_cs}. The experiment files |
| 275 |
\begin{itemize} |
| 276 |
\item {\it input/data} |
| 277 |
\item {\it input/data.pkg} |
| 278 |
\item {\it input/eedata}, |
| 279 |
\item {\it input/data.shap}, |
| 280 |
\item {\it code/packages.conf}, |
| 281 |
\item {\it code/CPP\_OPTIONS.h}, |
| 282 |
\item {\it code/SIZE.h}, |
| 283 |
\item {\it code/DIAGNOSTICS\_SIZE.h}, |
| 284 |
\item {\it code/external\_forcing.F}, |
| 285 |
\end{itemize} |
| 286 |
contain the code customizations and parameter settings for these |
| 287 |
experiments. Below we describe the customizations |
| 288 |
to these files associated with this experiment. |
| 289 |
|
| 290 |
\subsubsection{File {\it input/data}} |
| 291 |
\label{www:tutorials} |
| 292 |
|
| 293 |
\input{part3/case_studies/held_suarez_cs/inp_data} |
| 294 |
|
| 295 |
\subsubsection{File {\it input/data.pkg}} |
| 296 |
\label{www:tutorials} |
| 297 |
|
| 298 |
\input{part3/case_studies/held_suarez_cs/inp_data.pkg} |
| 299 |
|
| 300 |
\subsubsection{File {\it input/eedata}} |
| 301 |
\label{www:tutorials} |
| 302 |
|
| 303 |
This file uses standard default values except line 6: |
| 304 |
\begin{verbatim} |
| 305 |
useCubedSphereExchange=.TRUE., |
| 306 |
\end{verbatim} |
| 307 |
This line selects the cubed-sphere specific exchanges to |
| 308 |
to connect tiles and faces edges. |
| 309 |
|
| 310 |
\subsubsection{File {\it input/data.shap}} |
| 311 |
\label{www:tutorials} |
| 312 |
|
| 313 |
\input{part3/case_studies/held_suarez_cs/inp_data.shap} |
| 314 |
|
| 315 |
\subsubsection{File {\it code/SIZE.h}} |
| 316 |
\label{www:tutorials} |
| 317 |
|
| 318 |
Four lines are customized in this file for the current experiment |
| 319 |
|
| 320 |
\begin{itemize} |
| 321 |
|
| 322 |
\item Line 39, |
| 323 |
\begin{verbatim} sNx=32, \end{verbatim} |
| 324 |
sets the lateral domain extent in grid points along the x-direction, |
| 325 |
for 1 face. |
| 326 |
|
| 327 |
\item Line 40, |
| 328 |
\begin{verbatim} sNy=32, \end{verbatim} |
| 329 |
sets the lateral domain extent in grid points along the y-direction, |
| 330 |
for 1 face. |
| 331 |
|
| 332 |
\item Line 43, |
| 333 |
\begin{verbatim} nSx=6, \end{verbatim} |
| 334 |
sets the number of tiles in the x-direction, for the model domain |
| 335 |
decomposition. In this simple case (one processor and 1 tile per |
| 336 |
face), this number corresponds to the total number of faces. |
| 337 |
|
| 338 |
\item Line 49, |
| 339 |
\begin{verbatim} Nr=20, \end{verbatim} |
| 340 |
sets the vertical domain extent in grid points. |
| 341 |
|
| 342 |
\end{itemize} |
| 343 |
|
| 344 |
%\begin{small} |
| 345 |
%\input{part3/case_studies/held_suarez_cs/code/SIZE.h} |
| 346 |
%\end{small} |
| 347 |
|
| 348 |
\subsubsection{File {\it code/packages.conf}} |
| 349 |
\label{www:tutorials} |
| 350 |
|
| 351 |
\input{part3/case_studies/held_suarez_cs/cod_packages.conf} |
| 352 |
|
| 353 |
\subsubsection{File {\it code/CPP\_OPTIONS.h}} |
| 354 |
\label{www:tutorials} |
| 355 |
|
| 356 |
This file uses standard default except for Line 40\\ |
| 357 |
({\it diff CPP\_OPTIONS.h ../../../model/inc}): |
| 358 |
\begin{verbatim} |
| 359 |
#define NONLIN_FRSURF |
| 360 |
\end{verbatim} |
| 361 |
This line allow to use the non-linear free-surface part of the code, |
| 362 |
which is required for the $p^*$ coordinate formulation. |
| 363 |
|
| 364 |
\subsubsection{Other Files } |
| 365 |
\label{www:tutorials} |
| 366 |
|
| 367 |
Other files relevant to this experiment are |
| 368 |
\begin{itemize} |
| 369 |
\item {\it code/external\_forcing.F} |
| 370 |
\item {\it input/grid\_cs32.face00[n].bin}, with $n=1,2,3,4,5,6$ |
| 371 |
\end{itemize} |
| 372 |
contain the code customisations and binary input files for this |
| 373 |
experiments. Below we describe the customisations |
| 374 |
to these files associated with this experiment.\\ |
| 375 |
|
| 376 |
The file {\it code/external\_forcing.F} contains 4 subroutines |
| 377 |
that calculate the forcing terms (Right-Hand side term) in the |
| 378 |
momentum equation (\ref{EQ:eg-hs-global_forcing_fv}, |
| 379 |
{\it S/R EXTERNAL\_FORCING\_U} and {\it EXTERNAL\_FORCING\_V}) |
| 380 |
and in the potential temperature equation |
| 381 |
(\ref{EQ:eg-hs-global_forcing_ft}, {\it S/R EXTERNAL\_FORCING\_T}). |
| 382 |
The water-vapour forcing subroutine ({\it S/R EXTERNAL\_FORCING\_S}) |
| 383 |
is left empty for this experiment.\\ |
| 384 |
|
| 385 |
The grid-files {\it input/grid\_cs32.face00[n].bin}, with $n=1,2,3,4,5,6$, |
| 386 |
are binary files (direct-access, big-endian 64.bits real) that |
| 387 |
contains all the cubed-sphere grid lengths, areas and grid-point |
| 388 |
positions, with one file per face. |
| 389 |
Each file contains 18 2-D arrays (dimension $33 \times 33$) that correspond |
| 390 |
to the model variables: |
| 391 |
{\it |
| 392 |
XC YC DXF DYF RA XG YG DXV DYU RAZ DXC DYC RAW RAS DXG DYG AngleCS AngleSN |
| 393 |
} |
| 394 |
(see {\it GRID.h} file) |
| 395 |
|
Parent Directory
| 
Revision Log
| 
Revision Graph