1 |
dimitri |
1.1 |
\documentclass[12pt]{article} |
2 |
|
|
\usepackage{epsfig} |
3 |
|
|
\usepackage{graphics} |
4 |
|
|
\usepackage{subfigure} |
5 |
|
|
|
6 |
|
|
\usepackage[round,comma]{natbib} |
7 |
dimitri |
1.2 |
\bibliographystyle{bib/agu04} |
8 |
dimitri |
1.1 |
|
9 |
|
|
\usepackage{amsmath,amssymb} |
10 |
|
|
\newcommand\bmmax{10} \newcommand\hmmax{10} |
11 |
|
|
\usepackage{bm} |
12 |
|
|
|
13 |
|
|
% math abbreviations |
14 |
|
|
\newcommand{\vek}[1]{\ensuremath{\mathbf{#1}}} |
15 |
|
|
\newcommand{\mat}[1]{\ensuremath{\mathbf{#1}}} |
16 |
|
|
\newcommand{\vtau}{\bm{{\tau}}} |
17 |
|
|
|
18 |
|
|
\newcommand{\degree}{\ensuremath{^\circ}} |
19 |
|
|
\newcommand{\degC}{\,\ensuremath{\degree}C} |
20 |
|
|
\newcommand{\degE}{\ensuremath{\degree}\,E} |
21 |
|
|
\newcommand{\degS}{\ensuremath{\degree}\,S} |
22 |
|
|
\newcommand{\degN}{\ensuremath{\degree}\,N} |
23 |
|
|
\newcommand{\degW}{\ensuremath{\degree}\,W} |
24 |
|
|
|
25 |
|
|
% cross reference scheme |
26 |
|
|
\newcommand{\reffig}[1]{Figure~\ref{fig:#1}} |
27 |
|
|
\newcommand{\reftab}[1]{Table~\ref{tab:#1}} |
28 |
|
|
\newcommand{\refapp}[1]{Appendix~\ref{app:#1}} |
29 |
|
|
\newcommand{\refsec}[1]{Section~\ref{sec:#1}} |
30 |
|
|
\newcommand{\refeq}[1]{\,(\ref{eq:#1})} |
31 |
|
|
\newcommand{\refeqs}[2]{\,(\ref{eq:#1})--(\ref{eq:#2})} |
32 |
|
|
|
33 |
|
|
\newlength{\stdfigwidth}\setlength{\stdfigwidth}{20pc} |
34 |
|
|
%\newlength{\stdfigwidth}\setlength{\stdfigwidth}{\columnwidth} |
35 |
|
|
\newlength{\mediumfigwidth}\setlength{\mediumfigwidth}{39pc} |
36 |
|
|
%\newlength{\widefigwidth}\setlength{\widefigwidth}{39pc} |
37 |
|
|
\newlength{\widefigwidth}\setlength{\widefigwidth}{\textwidth} |
38 |
|
|
\newcommand{\fpath}{.} |
39 |
|
|
|
40 |
|
|
\title{A Dynamic-Thermodynamic Sea ice Model for Ocean Climate |
41 |
|
|
Estimation on an Arakawa C-Grid} |
42 |
|
|
|
43 |
|
|
\author{Martin Losch, Dimitris Menemenlis, Patrick Heimbach, \\ |
44 |
|
|
Jean-Michel Campin, and Chris Hill} |
45 |
|
|
\begin{document} |
46 |
|
|
|
47 |
|
|
\maketitle |
48 |
|
|
|
49 |
|
|
\begin{abstract} |
50 |
|
|
Some blabla |
51 |
|
|
\end{abstract} |
52 |
|
|
|
53 |
|
|
\section{Introduction} |
54 |
|
|
\label{sec:intro} |
55 |
|
|
|
56 |
|
|
more blabla |
57 |
|
|
|
58 |
|
|
\section{Model} |
59 |
|
|
\label{sec:model} |
60 |
|
|
|
61 |
|
|
Traditionally, probably for historical reasons and the ease of |
62 |
|
|
treating the Coriolis term, most standard sea-ice models are |
63 |
|
|
discretized on Arakawa-B-grids \citep[e.g.,][]{hibler79, harder99, |
64 |
|
|
kreyscher00, zhang98, hunke97}. From the perspective of coupling a |
65 |
|
|
sea ice-model to a C-grid ocean model, the exchange of fluxes of heat |
66 |
|
|
and fresh-water pose no difficulty for a B-grid sea-ice model |
67 |
|
|
\citep[e.g.,][]{timmermann02a}. However, surface stress is defined at |
68 |
|
|
velocities points and thus needs to be interpolated between a B-grid |
69 |
|
|
sea-ice model and a C-grid ocean model. While the smoothing implicitly |
70 |
|
|
associated with this interpolation may mask grid scale noise, it may |
71 |
|
|
in two-way coupling lead to a computational mode as will be shown. By |
72 |
|
|
choosing a C-grid for the sea-ice model, we circumvene this difficulty |
73 |
|
|
altogether and render the stress coupling as consistent as the |
74 |
|
|
buoyancy coupling. |
75 |
|
|
|
76 |
|
|
A further advantage of the C-grid formulation is apparent in narrow |
77 |
|
|
straits. In the limit of only one grid cell between coasts there is no |
78 |
|
|
flux allowed for a B-grid (with no-slip lateral boundary counditions), |
79 |
|
|
whereas the C-grid formulation allows a flux of sea-ice through this |
80 |
|
|
passage for all types of lateral boundary conditions. We (will) |
81 |
|
|
demonstrate this effect in the Candian archipelago. |
82 |
|
|
|
83 |
|
|
\subsection{Dynamics} |
84 |
|
|
\label{sec:dynamics} |
85 |
|
|
|
86 |
|
|
The momentum equations of the sea-ice model are standard with |
87 |
|
|
\begin{equation} |
88 |
|
|
\label{eq:momseaice} |
89 |
|
|
m \frac{D\vek{u}}{Dt} = -mf\vek{k}\times\vek{u} + \vtau_{air} + |
90 |
|
|
\vtau_{ocean} - m \nabla{\phi(0)} + \vek{F}, |
91 |
|
|
\end{equation} |
92 |
|
|
where $\vek{u} = u\vek{i}+v\vek{j}$ is the ice velocity vectory, $m$ |
93 |
|
|
the ice mass per unit area, $f$ the Coriolis parameter, $g$ is the |
94 |
|
|
gravity accelation, $\nabla\phi$ is the gradient (tilt) of the sea |
95 |
|
|
surface height potential beneath the ice. $\phi$ is the sum of |
96 |
|
|
atmpheric pressure $p_{a}$ and loading due to ice and snow |
97 |
|
|
$(m_{i}+m_{s})g$. $\vtau_{air}$ and $\vtau_{ocean}$ are the wind and |
98 |
|
|
ice-ocean stresses, respectively. $\vek{F}$ is the interaction force |
99 |
|
|
and $\vek{i}$, $\vek{j}$, and $\vek{k}$ are the unit vectors in the |
100 |
|
|
$x$, $y$, and $z$ directions. Advection of sea-ice momentum is |
101 |
|
|
neglected. The wind and ice-ocean stress terms are given by |
102 |
|
|
\begin{align*} |
103 |
|
|
\vtau_{air} =& \rho_{air} |\vek{U}_{air}|R_{air}(\vek{U}_{air}) \\ |
104 |
|
|
\vtau_{ocean} =& \rho_{ocean} |\vek{U}_{ocean}-\vek{u}| |
105 |
|
|
R_{ocean}(\vek{U}_{ocean}-\vek{u}), \\ |
106 |
|
|
\end{align*} |
107 |
|
|
where $\vek{U}_{air/ocean}$ are the surface winds of the atmosphere |
108 |
|
|
and surface currents of the ocean, respectively. $C_{air/ocean}$ are |
109 |
|
|
air and ocean drag coefficients, $\rho_{air/ocean}$ reference |
110 |
|
|
densities, and $R_{air/ocean}$ rotation matrices that act on the |
111 |
|
|
wind/current vectors. $\vek{F} = \nabla\cdot\sigma$ is the divergence |
112 |
|
|
of the interal stress tensor $\sigma_{ij}$. |
113 |
|
|
|
114 |
|
|
For an isotropic system this stress tensor can be related to the ice |
115 |
|
|
strain rate and strength by a nonlinear viscous-plastic (VP) |
116 |
|
|
constitutive law \citep{hibler79, zhang98}: |
117 |
|
|
\begin{equation} |
118 |
|
|
\label{eq:vpequation} |
119 |
|
|
\sigma_{ij}=2\eta(\dot{\epsilon}_{ij},P)\dot{\epsilon}_{ij} |
120 |
|
|
+ \left[\zeta(\dot{\epsilon}_{ij},P) - |
121 |
|
|
\eta(\dot{\epsilon}_{ij},P)\right]\dot{\epsilon}_{kk}\delta_{ij} |
122 |
|
|
- \frac{P}{2}\delta_{ij}. |
123 |
|
|
\end{equation} |
124 |
|
|
The ice strain rate is given by |
125 |
|
|
\begin{equation*} |
126 |
|
|
\dot{\epsilon}_{ij} = \frac{1}{2}\left( |
127 |
|
|
\frac{\partial{u_{i}}}{\partial{x_{j}}} + |
128 |
|
|
\frac{\partial{u_{j}}}{\partial{x_{i}}}\right). |
129 |
|
|
\end{equation*} |
130 |
|
|
The pressure $P$, a measure of ice strength, depends on both thickness |
131 |
|
|
$h$ and compactness (concentration) $c$: \[P = |
132 |
|
|
P^{*}c\,h\,e^{[C^{*}\cdot(1-c)]},\] with the constants $P^{*}$ and |
133 |
|
|
$C^{*}$. The nonlinear bulk and shear viscosities $\eta$ and $\zeta$ |
134 |
|
|
are functions of ice strain rate invariants and ice strength such that |
135 |
|
|
the principal components of the stress lie on an elliptical yield |
136 |
|
|
curve with the ratio of major to minor axis $e$ equal to $2$; they are |
137 |
|
|
given by: |
138 |
|
|
\begin{align*} |
139 |
|
|
\zeta =& \frac{P}{2\Delta} \\ |
140 |
|
|
\eta =& \frac{P}{2\Delta{e}^2} \\ |
141 |
|
|
\intertext{with the abbreviation} |
142 |
|
|
\Delta = & \left[ |
143 |
|
|
\left(\dot{\epsilon}_{11}^2+\dot{\epsilon}_{22}^2\right) |
144 |
|
|
(1+e^{-2}) + 4e^{-2}\dot{\epsilon}_{12}^2 + |
145 |
|
|
2\dot{\epsilon}_{11}\dot{\epsilon}_{22} (1-e^{-2}) |
146 |
|
|
\right]^{-\frac{1}{2}} |
147 |
|
|
\end{align*} |
148 |
|
|
In the current implementation, the VP-model is integrated with the |
149 |
|
|
semi-implicit line successive over relaxation (LSOR)-solver of |
150 |
|
|
\citet{zhang98}, which allows for long time steps that, in our case, |
151 |
|
|
is limited by the explicit treatment of the Coriolis term. The |
152 |
|
|
explicit treatment of the Coriolis term does not represent a severe |
153 |
|
|
limitation because it restricts the time step to approximately the |
154 |
|
|
same length as in the ocean model where the Coriolis term is also |
155 |
|
|
treated explicitly. |
156 |
|
|
|
157 |
|
|
\citet{hunke97}'s introduced an elastic contribution to the strain |
158 |
|
|
rate elatic-viscous-plastic in order to regularize |
159 |
|
|
Eq.\refeq{vpequation} in such a way that the resulting |
160 |
|
|
elatic-viscous-plastic (EVP) and VP models are identical at steady |
161 |
|
|
state, |
162 |
|
|
\begin{equation} |
163 |
|
|
\label{eq:evpequation} |
164 |
|
|
\frac{1}{E}\frac{\partial\sigma_{ij}}{\partial{t}} + |
165 |
|
|
\frac{1}{2\eta}\sigma_{ij} |
166 |
|
|
+ \frac{\eta - \zeta}{4\zeta\eta}\sigma_{kk}\delta_{ij} |
167 |
|
|
+ \frac{P}{4\zeta}\delta_{ij} |
168 |
|
|
= \dot{\epsilon}_{ij}. |
169 |
|
|
\end{equation} |
170 |
|
|
%In the EVP model, equations for the components of the stress tensor |
171 |
|
|
%$\sigma_{ij}$ are solved explicitly. Both model formulations will be |
172 |
|
|
%used and compared the present sea-ice model study. |
173 |
|
|
The EVP-model uses an explicit time stepping scheme with a short |
174 |
|
|
timestep. According to the recommendation of \citet{hunke97}, the |
175 |
|
|
EVP-model is stepped forward in time 120 times within the physical |
176 |
|
|
ocean model time step (although this parameter is under debate), to |
177 |
|
|
allow for elastic waves to disappear. Because the scheme does not |
178 |
|
|
require a matrix inversion it is fast in spite of the small timestep |
179 |
|
|
\citep{hunke97}. For completeness, we repeat the equations for the |
180 |
|
|
components of the stress tensor $\sigma_{1} = |
181 |
|
|
\sigma_{11}+\sigma_{22}$, $\sigma_{2}= \sigma_{11}-\sigma_{22}$, and |
182 |
|
|
$\sigma_{12}$. Introducing the divergence $D_D = |
183 |
|
|
\dot{\epsilon}_{11}+\dot{\epsilon}_{22}$, and the horizontal tension |
184 |
|
|
and shearing strain rates, $D_T = |
185 |
|
|
\dot{\epsilon}_{11}-\dot{\epsilon}_{22}$ and $D_S = |
186 |
|
|
2\dot{\epsilon}_{12}$, respectively and using the above abbreviations, |
187 |
|
|
the equations can be written as: |
188 |
|
|
\begin{align} |
189 |
|
|
\label{eq:evpstresstensor1} |
190 |
|
|
\frac{\partial\sigma_{1}}{\partial{t}} + \frac{\sigma_{1}}{2T} + |
191 |
|
|
\frac{P}{2T} &= \frac{P}{2T\Delta} D_D \\ |
192 |
|
|
\label{eq:evpstresstensor2} |
193 |
|
|
\frac{\partial\sigma_{2}}{\partial{t}} + \frac{\sigma_{2} e^{2}}{2T} |
194 |
|
|
&= \frac{P}{2T\Delta} D_T \\ |
195 |
|
|
\label{eq:evpstresstensor12} |
196 |
|
|
\frac{\partial\sigma_{12}}{\partial{t}} + \frac{\sigma_{12} e^{2}}{2T} |
197 |
|
|
&= \frac{P}{4T\Delta} D_S |
198 |
|
|
\end{align} |
199 |
|
|
Here, the elastic parameter $E$ is redefined in terms of a damping timescale |
200 |
|
|
$T$ for elastic waves \[E=\frac{\zeta}{T}.\] |
201 |
|
|
$T=E_{0}\Delta{t}$ with the tunable parameter $E_0<1$ and |
202 |
|
|
the external (long) timestep $\Delta{t}$. \citet{hunke97} recommend |
203 |
|
|
$E_{0} = \frac{1}{3}$. |
204 |
|
|
|
205 |
|
|
For details of the spatial discretization, the reader is referred to |
206 |
|
|
\citet{zhang98, zhang03}. Our discretization differs only (but |
207 |
|
|
importantly) in the underlying grid, namely the Arakawa C-grid, but is |
208 |
|
|
otherwise straightforward. The EVP model in particular is discretized |
209 |
|
|
naturally on the C-grid with $\sigma_{1}$ and $\sigma_{2}$ on the |
210 |
|
|
center points and $\sigma_{12}$ on the corner (or vorticity) points of |
211 |
|
|
the grid. With this choice all derivatives are discretized as central |
212 |
|
|
differences and averaging is only involved in computing $\Delta$ and |
213 |
|
|
$P$ at vorticity points. |
214 |
|
|
|
215 |
|
|
For a general curvilinear grid, one needs in principle to take metric |
216 |
|
|
terms into account that arise in the transformation a curvilinear grid |
217 |
|
|
on the sphere. However, for now we can neglect these metric terms |
218 |
|
|
because they are very small on the cubed sphere grids used in this |
219 |
|
|
paper; in particular, only near the edges of the cubed sphere grid, we |
220 |
|
|
expect them to be non-zero, but these edges are at approximately |
221 |
|
|
35\degS\ or 35\degN\ which are never covered by sea-ice in our |
222 |
|
|
simulations. Everywhere else the coordinate system is locally nearly |
223 |
|
|
cartesian. However, for last-glacial-maximum or snowball-earth-like |
224 |
|
|
simulations the question of metric terms needs to be reconsidered. |
225 |
|
|
Either, one includes these terms as in \citet{zhang03}, or one finds a |
226 |
|
|
vector-invariant formulation fo the sea-ice internal stress term that |
227 |
|
|
does not require any metric terms, as it is done in the ocean dynamics |
228 |
|
|
of the MITgcm \citep{adcroft04:_cubed_sphere}. |
229 |
|
|
|
230 |
|
|
Moving sea ice exerts a stress on the ocean which is the opposite of |
231 |
|
|
the stress $\vtau_{ocean}$ in Eq.\refeq{momseaice}. This stess is |
232 |
|
|
applied directly to the surface layer of the ocean model. An |
233 |
|
|
alternative ocean stress formulation is given by \citet{hibler87}. |
234 |
|
|
Rather than applying $\vtau_{ocean}$ directly, the stress is derived |
235 |
|
|
from integrating over the ice thickness to the bottom of the oceanic |
236 |
|
|
surface layer. In the resulting equation for the \emph{combined} |
237 |
|
|
ocean-ice momentum, the interfacial stress cancels and the total |
238 |
|
|
stress appears as the sum of windstress and divergence of internal ice |
239 |
|
|
stresses: $\delta(z) (\vtau_{air} + \vek{F})/\rho_0$, \citep[see also |
240 |
|
|
Eq.\,2 of][]{hibler87}. The disadvantage of this formulation is that |
241 |
|
|
now the velocity in the surface layer of the ocean that is used to |
242 |
|
|
advect tracers, is really an average over the ocean surface |
243 |
|
|
velocity and the ice velocity leading to an inconsistency as the ice |
244 |
|
|
temperature and salinity are different from the oceanic variables. |
245 |
|
|
|
246 |
|
|
Sea ice distributions are characterized by sharp gradients and edges. |
247 |
|
|
For this reason, we employ a positive 3rd-order advection scheme |
248 |
|
|
\citep{hundsdorfer94} for the thermodynamic variables discussed in the |
249 |
|
|
next section. |
250 |
|
|
|
251 |
|
|
\subparagraph{boundary conditions: no-slip, free-slip, half-slip} |
252 |
|
|
|
253 |
|
|
\begin{itemize} |
254 |
|
|
\item transition from existing B-Grid to C-Grid |
255 |
|
|
\item boundary conditions: no-slip, free-slip, half-slip |
256 |
|
|
\item fancy (multi dimensional) advection schemes |
257 |
|
|
\item VP vs.\ EVP \citep{hunke97} |
258 |
|
|
\item ocean stress formulation \citep{hibler87} |
259 |
|
|
\end{itemize} |
260 |
|
|
|
261 |
|
|
\subsection{Thermodynamics} |
262 |
|
|
\label{sec:thermodynamics} |
263 |
|
|
|
264 |
|
|
In the original formulation the sea ice model \citep{menemenlis05} |
265 |
|
|
uses simple thermodynamics following the appendix of |
266 |
|
|
\citet{semtner76}. This formulation does not allow storage of heat |
267 |
|
|
(heat capacity of ice is zero, and this type of model is often refered |
268 |
|
|
to as a ``zero-layer'' model). Upward heat flux is parameterized |
269 |
|
|
assuming a linear temperature profile and together with a constant ice |
270 |
|
|
conductivity. It is expressed as $(K/h)(T_{w}-T_{0})$, where $K$ is |
271 |
|
|
the ice conductivity, $h$ the ice thickness, and $T_{w}-T_{0}$ the |
272 |
|
|
difference between water and ice surface temperatures. The surface |
273 |
|
|
heat budget is computed in a similar way to that of |
274 |
|
|
\citet{parkinson79} and \citet{manabe79}. |
275 |
|
|
|
276 |
|
|
There is considerable doubt about the reliability of such a simple |
277 |
|
|
thermodynamic model---\citet{semtner84} found significant errors in |
278 |
|
|
phase (one month lead) and amplitude ($\approx$50\%\,overestimate) in |
279 |
|
|
such models---, so that today many sea ice models employ more complex |
280 |
|
|
thermodynamics. A popular thermodynamics model of \citet{winton00} is |
281 |
|
|
based on the 3-layer model of \citet{semtner76} and treats brine |
282 |
|
|
content by means of enthalphy conservation. This model requires in |
283 |
|
|
addition to ice-thickness and compactness (fractional area) additional |
284 |
|
|
state variables to be advected by ice velocities, namely enthalphy of |
285 |
|
|
the two ice layers and the thickness of the overlying snow layer. |
286 |
|
|
|
287 |
|
|
\section{Funnel Experiments} |
288 |
|
|
\label{sec:funnel} |
289 |
|
|
|
290 |
|
|
\begin{itemize} |
291 |
|
|
\item B-grid LSR no-slip |
292 |
|
|
\item C-grid LSR no-slip |
293 |
|
|
\item C-grid LSR slip |
294 |
|
|
\item C-grid EVP no-slip |
295 |
|
|
\item C-grid EVP slip |
296 |
|
|
\end{itemize} |
297 |
|
|
|
298 |
|
|
\subsection{B-grid vs.\ C-grid} |
299 |
|
|
Comparison between: |
300 |
|
|
\begin{itemize} |
301 |
|
|
\item B-grid, lsr, no-slip |
302 |
|
|
\item C-grid, lsr, no-slip |
303 |
|
|
\item C-grid, evp, no-slip |
304 |
|
|
\end{itemize} |
305 |
|
|
all without ice-ocean stress, because ice-ocean stress does not work |
306 |
|
|
for B-grid. |
307 |
|
|
|
308 |
|
|
\subsection{C-grid} |
309 |
|
|
\begin{itemize} |
310 |
|
|
\item no-slip vs. free-slip for both lsr and evp; |
311 |
|
|
"diagnostics" to look at and use for comparison |
312 |
|
|
\begin{itemize} |
313 |
|
|
\item ice transport through Fram Strait/Denmark Strait/Davis |
314 |
|
|
Strait/Bering strait (these are general) |
315 |
|
|
\item ice transport through narrow passages, e.g.\ Nares-Strait |
316 |
|
|
\end{itemize} |
317 |
|
|
\item compare different advection schemes (if lsr turns out to be more |
318 |
|
|
effective, then with lsr otherwise I prefer evp), eg. |
319 |
|
|
\begin{itemize} |
320 |
|
|
\item default 2nd-order with diff1=0.002 |
321 |
|
|
\item 1st-order upwind with diff1=0. |
322 |
|
|
\item DST3FL (SEAICEadvScheme=33 with diff1=0., should work, works for me) |
323 |
|
|
\item 2nd-order wit flux limiter (SEAICEadvScheme=77 with diff1=0.) |
324 |
|
|
\end{itemize} |
325 |
|
|
That should be enough. Here, total ice mass and location of ice edge |
326 |
|
|
is interesting. However, this comparison can be done in an idealized |
327 |
|
|
domain, may not require full Arctic Domain? |
328 |
|
|
\item |
329 |
|
|
Do a little study on the parameters of LSR and EVP |
330 |
|
|
\begin{enumerate} |
331 |
|
|
\item convergence of LSR, how many iterations do you need to get a |
332 |
|
|
true elliptic yield curve |
333 |
|
|
\item vary deltaTevp and the relaxation parameter for EVP and see when |
334 |
|
|
the EVP solution breaks down (relative to the forcing time scale). |
335 |
|
|
For this, it is essential that the evp solver gives use "stripeless" |
336 |
|
|
solutions, that is your dtevp = 1sec solutions/or 10sec solutions |
337 |
|
|
with SEAICE\_evpDampC = 615. |
338 |
|
|
\end{enumerate} |
339 |
|
|
\end{itemize} |
340 |
|
|
|
341 |
|
|
\section{Forward sensitivity experiments} |
342 |
|
|
\label{sec:forward} |
343 |
|
|
|
344 |
|
|
A second series of forward sensitivity experiments have been carried out on an |
345 |
|
|
Arctic Ocean domain with open boundaries. Once again the objective is to |
346 |
|
|
compare the old B-grid LSR dynamic solver with the new C-grid LSR and EVP |
347 |
|
|
solvers. One additional experiment is carried out to illustrate the |
348 |
|
|
differences between the two main options for sea ice thermodynamics in the MITgcm. |
349 |
|
|
|
350 |
|
|
\subsection{Arctic Domain with Open Boundaries} |
351 |
|
|
\label{sec:arctic} |
352 |
|
|
|
353 |
|
|
The Arctic domain of integration is illustrated in Fig.~\ref{???}. It is |
354 |
|
|
carved out from, and obtains open boundary conditions from, the global |
355 |
|
|
cubed-sphere configuration of the Estimating the Circulation and Climate of |
356 |
|
|
the Ocean, Phase II (ECCO2) project \cite{men05a}. The domain size is 420 by |
357 |
|
|
384 grid boxes horizontally with mean horizontal grid spacing of 18 km. |
358 |
|
|
|
359 |
|
|
There are 50 vertical levels ranging in thickness from 10 m near the surface |
360 |
|
|
to approximately 450 m at a maximum model depth of 6150 m. Bathymetry is from |
361 |
|
|
the National Geophysical Data Center (NGDC) 2-minute gridded global relief |
362 |
|
|
data (ETOPO2) and the model employs the partial-cell formulation of |
363 |
|
|
\cite{adc97}, which permits accurate representation of the bathymetry. The |
364 |
|
|
model is integrated in a volume-conserving configuration using a finite volume |
365 |
|
|
discretization with C-grid staggering of the prognostic variables. In the |
366 |
|
|
ocean, the non-linear equation of state of \cite{jac95}. The ocean model is |
367 |
|
|
coupled to a sea-ice model described hereinabove. |
368 |
|
|
|
369 |
|
|
This particular ECCO2 simulation is initialized from rest using the January |
370 |
|
|
temperature and salinity distribution from the World Ocean Atlas 2001 (WOA01) |
371 |
|
|
[Conkright et al., 2002] and it is integrated for 32 years prior to the |
372 |
|
|
1996-2001 period discussed in the study. Surface boundary conditions are from |
373 |
|
|
the National Centers for Environmental Prediction and the National Center for |
374 |
|
|
Atmospheric Research (NCEP/NCAR) atmospheric reanalysis [Kistler et al., |
375 |
|
|
2001]. Six-hourly surface winds, temperature, humidity, downward short- and |
376 |
|
|
long-wave radiations, and precipitation are converted to heat, freshwater, and |
377 |
|
|
wind stress fluxes using the Large and Pond [1981, 1982] bulk |
378 |
|
|
formulae. Shortwave radiation decays exponentially as per Paulson and Simpson |
379 |
|
|
[1977]. Additionally the time-mean river run-off from Large and Nurser [2001] |
380 |
|
|
is applied and there is a relaxation to the monthly-mean climatological sea |
381 |
|
|
surface salinity values from WOA01 with a relaxation time scale of 3 |
382 |
|
|
months. Vertical mixing follows Large et al. [1994] with background vertical |
383 |
|
|
diffusivity of 1.5 × 10-5 m2 s-1 and viscosity of 10-3 m2 s-1. A third order, |
384 |
|
|
direct-space-time advection scheme with flux limiter is employed and there is |
385 |
|
|
no explicit horizontal diffusivity. Horizontal viscosity follows Leith [1996] |
386 |
|
|
but modified to sense the divergent flow as per Fox-Kemper and Menemenlis [in |
387 |
|
|
press]. Shortwave radiation decays exponentially as per Paulson and Simpson |
388 |
|
|
[1977]. Additionally, the time-mean runoff of Large and Nurser [2001] is |
389 |
|
|
applied near the coastline and, where there is open water, there is a |
390 |
|
|
relaxation to monthly-mean WOA01 sea surface salinity with a time constant of |
391 |
|
|
45 days. |
392 |
|
|
|
393 |
|
|
Open water, dry |
394 |
|
|
ice, wet ice, dry snow, and wet snow albedo are, respectively, 0.15, 0.85, |
395 |
|
|
0.76, 0.94, and 0.8. |
396 |
|
|
|
397 |
|
|
\begin{itemize} |
398 |
|
|
\item Configuration |
399 |
|
|
\item OBCS from cube |
400 |
|
|
\item forcing |
401 |
|
|
\item 1/2 and full resolution |
402 |
|
|
\item with a few JFM figs from C-grid LSR no slip |
403 |
|
|
ice transport through Canadian Archipelago |
404 |
|
|
thickness distribution |
405 |
|
|
ice velocity and transport |
406 |
|
|
\end{itemize} |
407 |
|
|
|
408 |
|
|
\begin{itemize} |
409 |
|
|
\item Arctic configuration |
410 |
|
|
\item ice transport through straits and near boundaries |
411 |
|
|
\item focus on narrow straits in the Canadian Archipelago |
412 |
|
|
\end{itemize} |
413 |
|
|
|
414 |
|
|
\begin{itemize} |
415 |
|
|
\item B-grid LSR no-slip |
416 |
|
|
\item C-grid LSR no-slip |
417 |
|
|
\item C-grid LSR slip |
418 |
|
|
\item C-grid EVP no-slip |
419 |
|
|
\item C-grid EVP slip |
420 |
|
|
\item C-grid LSR no-slip + Winton |
421 |
|
|
\item speed-performance-accuracy (small) |
422 |
|
|
ice transport through Canadian Archipelago differences |
423 |
|
|
thickness distribution differences |
424 |
|
|
ice velocity and transport differences |
425 |
|
|
\end{itemize} |
426 |
|
|
|
427 |
|
|
We anticipate small differences between the different models due to: |
428 |
|
|
\begin{itemize} |
429 |
|
|
\item advection schemes: along the ice-edge and regions with large |
430 |
|
|
gradients |
431 |
|
|
\item C-grid: more transport through narrow straits for no slip |
432 |
|
|
conditons, less for free slip |
433 |
|
|
\item VP vs.\ EVP: speed performance, accuracy? |
434 |
|
|
\item ocean stress: different water mass properties beneath the ice |
435 |
|
|
\end{itemize} |
436 |
|
|
|
437 |
|
|
\section{Adjoint sensiivities of the MITsim} |
438 |
|
|
|
439 |
|
|
\subsection{The adjoint of MITsim} |
440 |
|
|
|
441 |
|
|
The ability to generate tangent linear and adjoint model components |
442 |
|
|
of the MITsim has been a main design task. |
443 |
|
|
For the ocean the adjoint capability has proven to be an |
444 |
|
|
invaluable tool for sensitivity analysis as well as state estimation. |
445 |
|
|
In short, the adjoint enables very efficient computation of the gradient |
446 |
|
|
of scalar-valued model diagnostics (called cost function or objective function) |
447 |
|
|
with respect to many model "variables". |
448 |
|
|
These variables can be two- or three-dimensional fields of initial |
449 |
|
|
conditions, model parameters such as mixing coefficients, or |
450 |
|
|
time-varying surface or lateral (open) boundary conditions. |
451 |
|
|
When combined, these variables span a potentially high-dimensional |
452 |
|
|
(e.g. O(10$^8$)) so-called control space. Performing parameter perturbations |
453 |
|
|
to assess model sensitivities quickly becomes prohibitive at these scales. |
454 |
|
|
Alternatively, (time-varying) sensitivities of the objective function |
455 |
|
|
to any element of the control space can be computed very efficiently in |
456 |
|
|
one single adjoint |
457 |
|
|
model integration, provided an efficient adjoint model is available. |
458 |
|
|
|
459 |
|
|
[REFERENCES] |
460 |
|
|
|
461 |
|
|
|
462 |
|
|
The adjoint operator (ADM) is the transpose of the tangent linear operator (TLM) |
463 |
|
|
of the full (in general nonlinear) forward model, i.e. the MITsim. |
464 |
|
|
The TLM maps perturbations of elements of the control space |
465 |
|
|
(e.g. initial ice thickness distribution) |
466 |
|
|
via the model Jacobian |
467 |
|
|
to a perturbation in the objective function |
468 |
|
|
(e.g. sea-ice export at the end of the integration interval). |
469 |
|
|
\textit{Tangent} linearity ensures that the derivatives are evaluated |
470 |
|
|
with respect to the underlying model trajectory at each point in time. |
471 |
|
|
This is crucial for nonlinear trajectories and the presence of different |
472 |
|
|
regimes (e.g. effect of the seaice growth term at or away from the |
473 |
|
|
freezing point of the ocean surface). |
474 |
|
|
Ensuring tangent linearity can be easily achieved by integrating |
475 |
|
|
the full model in sync with the TLM to provide the underlying model state. |
476 |
|
|
Ensuring \textit{tangent} adjoints is equally crucial, but much more |
477 |
|
|
difficult to achieve because of the reverse nature of the integration: |
478 |
|
|
the adjoint accumulates sensitivities backward in time, |
479 |
|
|
starting from a unit perturbation of the objective function. |
480 |
|
|
The adjoint model requires the model state in reverse order. |
481 |
|
|
This presents one of the major complications in deriving an |
482 |
|
|
exact, i.e. \textit{tangent} adjoint model. |
483 |
|
|
|
484 |
|
|
Following closely the development and maintenance of TLM and ADM |
485 |
|
|
components of the MITgcm we have relied heavily on the |
486 |
|
|
autmomatic differentiation (AD) tool |
487 |
|
|
"Transformation of Algorithms in Fortran" (TAF) |
488 |
|
|
developed by Fastopt (Giering and Kaminski, 1998) |
489 |
|
|
to derive TLM and ADM code of the MITsim. |
490 |
|
|
Briefly, the nonlinear parent model is fed to the AD tool which produces |
491 |
|
|
derivative code for the specified control space and objective function. |
492 |
|
|
Following this approach has (apart from its evident success) |
493 |
|
|
several advantages: |
494 |
|
|
(1) the adjoint model is the exact adjoint operator of the parent model, |
495 |
|
|
(2) the adjoint model can be kept up to date with respect to ongoing |
496 |
|
|
development of the parent model, and adjustments to the parent model |
497 |
|
|
to extend the automatically generated adjoint are incremental changes |
498 |
|
|
only, rather than extensive re-developments, |
499 |
|
|
(3) the parallel structure of the parent model is preserved |
500 |
|
|
by the adjoint model, ensuring efficient use in high performance |
501 |
|
|
computing environments. |
502 |
|
|
|
503 |
|
|
Some initial code adjustments are required to support dependency analysis |
504 |
|
|
of the flow reversal and certain language limitations which may lead |
505 |
|
|
to irreducible flow graphs (e.g. GOTO statements). |
506 |
|
|
The problem of providing the required model state in reverse order |
507 |
|
|
at the time of evaluating nonlinear or conditional |
508 |
|
|
derivatives is solved via balancing |
509 |
|
|
storing vs. recomputation of the model state in a multi-level |
510 |
|
|
checkpointing loop. |
511 |
|
|
Again, an initial code adjustment is required to support TAFs |
512 |
|
|
checkpointing capability. |
513 |
|
|
The code adjustments are sufficiently simply so as not to cause |
514 |
|
|
major limitations to the full nonlinear parent model. |
515 |
|
|
Once in place, an adjoint model of a new model configuration |
516 |
|
|
may be derived in about 10 minutes. |
517 |
|
|
|
518 |
|
|
[HIGHLIGHT COUPLED NATURE OF THE ADJOINT!] |
519 |
|
|
|
520 |
|
|
\subsection{Special considerations} |
521 |
|
|
|
522 |
|
|
* growth term(?) |
523 |
|
|
|
524 |
|
|
* small active denominators |
525 |
|
|
|
526 |
|
|
* dynamic solver (implicit function theorem) |
527 |
|
|
|
528 |
|
|
* approximate adjoints |
529 |
|
|
|
530 |
|
|
|
531 |
|
|
\subsection{An example: sensitivities of sea-ice export through Fram Strait} |
532 |
|
|
|
533 |
|
|
We demonstrate the power of the adjoint method |
534 |
|
|
in the context of investigating sea-ice export sensitivities through Fram Strait |
535 |
|
|
(for details of this study see Heimbach et al., 2007). |
536 |
|
|
The domain chosen is a coarsened version of the Arctic face of the |
537 |
|
|
high-resolution cubed-sphere configuration of the ECCO2 project |
538 |
|
|
(see Menemenlis et al. 2005). It covers the entire Arctic, |
539 |
|
|
extends into the North Pacific such as to cover the entire |
540 |
|
|
ice-covered regions, and comprises parts of the North Atlantic |
541 |
|
|
down to XXN to enable analysis of remote influences of the |
542 |
|
|
North Atlantic current to sea-ice variability and export. |
543 |
|
|
The horizontal resolution varies between XX and YY km |
544 |
|
|
with 50 unevenly spaced vertical levels. |
545 |
|
|
The adjoint models run efficiently on 80 processors |
546 |
|
|
(benchmarks have been performed both on an SGI Altix as well as an |
547 |
|
|
IBM SP5 at NASA/ARC). |
548 |
|
|
|
549 |
|
|
Following a 1-year spinup, the model has been integrated for four years |
550 |
|
|
between 1992 and 1995. |
551 |
|
|
It is forced using realistic 6-hourly NCEP/NCAR atmospheric state variables. |
552 |
|
|
Over the open ocean these are converted into |
553 |
|
|
air-sea fluxes via the bulk formulae of Large and Yeager (2004). |
554 |
|
|
Derivation of air-sea fluxes in the presence of sea-ice is handled |
555 |
|
|
by the ice model as described in Section XXX. |
556 |
|
|
The objective function chosen is sea-ice export through Fram Strait |
557 |
|
|
computed for December 1995 |
558 |
|
|
The adjoint model computes sensitivities to sea-ice export back in time |
559 |
|
|
from 1995 to 1992 along this trajectory. |
560 |
|
|
In principle all adjoint model variable (i.e. Lagrange multipliers) |
561 |
|
|
of the coupled ocean/sea-ice model |
562 |
|
|
are available to analyze the transient sensitivity behaviour |
563 |
|
|
of the ocean and sea-ice state. |
564 |
|
|
Over the open ocean, the adjoint of the bulk formula scheme |
565 |
|
|
computes sensitivities to the time-varying atmospheric state. |
566 |
|
|
Over ice-covered parts, the sea-ice adjoint converts |
567 |
|
|
surface ocean sensitivities to atmospheric sensitivities. |
568 |
|
|
|
569 |
|
|
Fig. XXX(a--d) depict sensitivities of sea-ice export through Fram Strait |
570 |
|
|
in December 1995 to changes in sea-ice thickness |
571 |
|
|
12, 24, 36, 48 months back in time. |
572 |
|
|
Corresponding sensitivities to ocean surface temperature are |
573 |
|
|
depicted in Fig. XXX(a--d). |
574 |
|
|
The main characteristics is consistency with expected advection |
575 |
|
|
of sea-ice over the relevant time scales considered. |
576 |
|
|
The general positive pattern means that an increase in |
577 |
|
|
sea-ice thickness at location $(x,y)$ and time $t$ will increase |
578 |
|
|
sea-ice export through Fram Strait at time $T_e$. |
579 |
|
|
Largest distances from Fram Strait indicate fastest sea-ice advection |
580 |
|
|
over the time span considered. |
581 |
|
|
The ice thickness sensitivities are in close correspondence to |
582 |
|
|
ocean surface sentivitites, but of opposite sign. |
583 |
|
|
An increase in temperature will incur ice melting, decrease in ice thickness, |
584 |
|
|
and therefore decrease in sea-ice export at time $T_e$. |
585 |
|
|
|
586 |
|
|
The picture is fundamentally different and much more complex |
587 |
|
|
for sensitivities to ocean temperatures away from the surface. |
588 |
|
|
Fig. XXX (a--d) depicts ice export sensitivities to |
589 |
|
|
temperatures at roughly 400 m depth. |
590 |
|
|
Primary features are the effect of the heat transport of the North |
591 |
|
|
Atlantic current which feeds into the West Spitsbergen current, |
592 |
|
|
the circulation around Svalbard, and ... |
593 |
|
|
|
594 |
|
|
\begin{figure}[t!] |
595 |
|
|
\centerline{ |
596 |
|
|
\subfigure[{\footnotesize -12 months}] |
597 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJheff_arc_lev1_tim072_cmax2.0E+02.eps}} |
598 |
|
|
%\includegraphics*[width=.3\textwidth]{H_c.bin_res_100_lev1.pdf} |
599 |
|
|
% |
600 |
|
|
\subfigure[{\footnotesize -24 months}] |
601 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJheff_arc_lev1_tim145_cmax2.0E+02.eps}} |
602 |
|
|
} |
603 |
|
|
|
604 |
|
|
\centerline{ |
605 |
|
|
\subfigure[{\footnotesize |
606 |
|
|
-36 months}] |
607 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJheff_arc_lev1_tim218_cmax2.0E+02.eps}} |
608 |
|
|
% |
609 |
|
|
\subfigure[{\footnotesize |
610 |
|
|
-48 months}] |
611 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJheff_arc_lev1_tim292_cmax2.0E+02.eps}} |
612 |
|
|
} |
613 |
|
|
\caption{Sensitivity of sea-ice export through Fram Strait in December 2005 to |
614 |
|
|
sea-ice thickness at various prior times. |
615 |
|
|
\label{fig:4yradjheff}} |
616 |
|
|
\end{figure} |
617 |
|
|
|
618 |
|
|
|
619 |
|
|
\begin{figure}[t!] |
620 |
|
|
\centerline{ |
621 |
|
|
\subfigure[{\footnotesize -12 months}] |
622 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJtheta_arc_lev1_tim072_cmax5.0E+01.eps}} |
623 |
|
|
%\includegraphics*[width=.3\textwidth]{H_c.bin_res_100_lev1.pdf} |
624 |
|
|
% |
625 |
|
|
\subfigure[{\footnotesize -24 months}] |
626 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJtheta_arc_lev1_tim145_cmax5.0E+01.eps}} |
627 |
|
|
} |
628 |
|
|
|
629 |
|
|
\centerline{ |
630 |
|
|
\subfigure[{\footnotesize |
631 |
|
|
-36 months}] |
632 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJtheta_arc_lev1_tim218_cmax5.0E+01.eps}} |
633 |
|
|
% |
634 |
|
|
\subfigure[{\footnotesize |
635 |
|
|
-48 months}] |
636 |
|
|
{\includegraphics*[width=0.44\linewidth]{figs/run_4yr_ADJtheta_arc_lev1_tim292_cmax5.0E+01.eps}} |
637 |
|
|
} |
638 |
|
|
\caption{Same as Fig. XXX but for sea surface temperature |
639 |
|
|
\label{fig:4yradjthetalev1}} |
640 |
|
|
\end{figure} |
641 |
|
|
|
642 |
|
|
|
643 |
|
|
|
644 |
|
|
\section{Discussion and conclusion} |
645 |
|
|
\label{sec:concl} |
646 |
|
|
|
647 |
|
|
The story of the paper could be: |
648 |
|
|
B-grid ice model + C-grid ocean model does not work properly for us, |
649 |
|
|
therefore C-grid ice model with advantages: |
650 |
|
|
\begin{enumerate} |
651 |
|
|
\item stress coupling simpler (no interpolation required) |
652 |
|
|
\item different boundary conditions |
653 |
|
|
\item advection schemes carry over trivially from main code |
654 |
|
|
\end{enumerate} |
655 |
|
|
LSR/EVP solutions are similar with appropriate bcs, evp parameters as |
656 |
|
|
a function of forcing time scale (when does VP solution break |
657 |
|
|
down). Same for LSR solver, provided that it works (o: |
658 |
|
|
Which scheme is more efficient for the resolution/time stepping |
659 |
|
|
parameters that we use here. What about other resolutions? |
660 |
|
|
|
661 |
|
|
\paragraph{Acknowledgements} |
662 |
|
|
We thank Jinlun Zhang for providing the original B-grid code and many |
663 |
|
|
helpful discussions. |
664 |
|
|
|
665 |
|
|
\bibliography{bib/journal_abrvs,bib/seaice,bib/genocean,bib/maths,bib/mitgcmuv,bib/fram} |
666 |
|
|
|
667 |
|
|
\end{document} |
668 |
|
|
|
669 |
|
|
%%% Local Variables: |
670 |
|
|
%%% mode: latex |
671 |
|
|
%%% TeX-master: t |
672 |
|
|
%%% End: |