/[MITgcm]/MITgcm_contrib/articles/ceaice/ceaice.tex
ViewVC logotype

Annotation of /MITgcm_contrib/articles/ceaice/ceaice.tex

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


Revision 1.13 - (hide annotations) (download) (as text)
Mon Feb 25 23:45:46 2008 UTC (17 years, 4 months ago) by dimitri
Branch: MAIN
Changes since 1.12: +38 -31 lines
File MIME type: application/x-tex
updated ice dynamics eqs, fixed some typos, added reference to z* article

1 dimitri 1.13 % $Header: /u/gcmpack/MITgcm_contrib/articles/ceaice/ceaice.tex,v 1.12 2008/02/25 22:06:17 dimitri Exp $
2 dimitri 1.10 % $Name: $
3 dimitri 1.1 \documentclass[12pt]{article}
4 mlosch 1.4
5 mlosch 1.5 \usepackage[]{graphicx}
6     \usepackage{subfigure}
7 dimitri 1.1
8     \usepackage[round,comma]{natbib}
9 dimitri 1.2 \bibliographystyle{bib/agu04}
10 dimitri 1.1
11     \usepackage{amsmath,amssymb}
12     \newcommand\bmmax{10} \newcommand\hmmax{10}
13     \usepackage{bm}
14    
15     % math abbreviations
16     \newcommand{\vek}[1]{\ensuremath{\mathbf{#1}}}
17     \newcommand{\mat}[1]{\ensuremath{\mathbf{#1}}}
18     \newcommand{\vtau}{\bm{{\tau}}}
19    
20     \newcommand{\degree}{\ensuremath{^\circ}}
21     \newcommand{\degC}{\,\ensuremath{\degree}C}
22     \newcommand{\degE}{\ensuremath{\degree}\,E}
23     \newcommand{\degS}{\ensuremath{\degree}\,S}
24     \newcommand{\degN}{\ensuremath{\degree}\,N}
25     \newcommand{\degW}{\ensuremath{\degree}\,W}
26    
27     % cross reference scheme
28     \newcommand{\reffig}[1]{Figure~\ref{fig:#1}}
29     \newcommand{\reftab}[1]{Table~\ref{tab:#1}}
30     \newcommand{\refapp}[1]{Appendix~\ref{app:#1}}
31     \newcommand{\refsec}[1]{Section~\ref{sec:#1}}
32     \newcommand{\refeq}[1]{\,(\ref{eq:#1})}
33     \newcommand{\refeqs}[2]{\,(\ref{eq:#1})--(\ref{eq:#2})}
34    
35     \newlength{\stdfigwidth}\setlength{\stdfigwidth}{20pc}
36     %\newlength{\stdfigwidth}\setlength{\stdfigwidth}{\columnwidth}
37     \newlength{\mediumfigwidth}\setlength{\mediumfigwidth}{39pc}
38     %\newlength{\widefigwidth}\setlength{\widefigwidth}{39pc}
39     \newlength{\widefigwidth}\setlength{\widefigwidth}{\textwidth}
40 mlosch 1.4 \newcommand{\fpath}{figs}
41    
42     % commenting scheme
43     \newcommand{\ml}[1]{\textsf{\slshape #1}}
44 dimitri 1.1
45     \title{A Dynamic-Thermodynamic Sea ice Model for Ocean Climate
46     Estimation on an Arakawa C-Grid}
47    
48     \author{Martin Losch, Dimitris Menemenlis, Patrick Heimbach, \\
49     Jean-Michel Campin, and Chris Hill}
50     \begin{document}
51    
52     \maketitle
53    
54     \begin{abstract}
55 dimitri 1.10
56     As part of ongoing efforts to obtain a best possible synthesis of most
57 dimitri 1.12 available, global-scale, ocean and sea ice data, a dynamic and thermodynamic
58     sea-ice model has been coupled to the Massachusetts Institute of Technology
59     general circulation model (MITgcm). Ice mechanics follow a viscous plastic
60     rheology and the ice momentum equations are solved numerically using either
61     line successive relaxation (LSR) or elastic-viscous-plastic (EVP) dynamic
62     models. Ice thermodynamics are represented using either a zero-heat-capacity
63     formulation or a two-layer formulation that conserves enthalpy. The model
64     includes prognostic variables for snow and for sea-ice salinity. The above
65     sea ice model components were borrowed from current-generation climate models
66     but they were reformulated on an Arakawa C-grid in order to match the MITgcm
67     oceanic grid and they were modified in many ways to permit efficient and
68     accurate automatic differentiation. This paper describes the MITgcm sea ice
69     model; it presents example Arctic and Antarctic results from a realistic,
70     eddy-permitting, global ocean and sea-ice configuration; it compares B-grid
71     and C-grid dynamic solvers in a regional Arctic configuration; and it presents
72     example results from coupled ocean and sea-ice adjoint-model integrations.
73 dimitri 1.10
74 dimitri 1.1 \end{abstract}
75    
76     \section{Introduction}
77     \label{sec:intro}
78    
79     Traditionally, probably for historical reasons and the ease of
80     treating the Coriolis term, most standard sea-ice models are
81     discretized on Arakawa-B-grids \citep[e.g.,][]{hibler79, harder99,
82     kreyscher00, zhang98, hunke97}. From the perspective of coupling a
83     sea ice-model to a C-grid ocean model, the exchange of fluxes of heat
84     and fresh-water pose no difficulty for a B-grid sea-ice model
85     \citep[e.g.,][]{timmermann02a}. However, surface stress is defined at
86     velocities points and thus needs to be interpolated between a B-grid
87     sea-ice model and a C-grid ocean model. While the smoothing implicitly
88     associated with this interpolation may mask grid scale noise, it may
89     in two-way coupling lead to a computational mode as will be shown. By
90 dimitri 1.13 choosing a C-grid for the sea-ice model, we circumvent this difficulty
91 dimitri 1.1 altogether and render the stress coupling as consistent as the
92     buoyancy coupling.
93    
94     A further advantage of the C-grid formulation is apparent in narrow
95     straits. In the limit of only one grid cell between coasts there is no
96     flux allowed for a B-grid (with no-slip lateral boundary counditions),
97     whereas the C-grid formulation allows a flux of sea-ice through this
98     passage for all types of lateral boundary conditions. We (will)
99     demonstrate this effect in the Candian archipelago.
100    
101 dimitri 1.13 \section{Model}
102     \label{sec:model}
103    
104 dimitri 1.1 \subsection{Dynamics}
105     \label{sec:dynamics}
106    
107 dimitri 1.13 The momentum equation of the sea-ice model is
108 dimitri 1.1 \begin{equation}
109     \label{eq:momseaice}
110     m \frac{D\vek{u}}{Dt} = -mf\vek{k}\times\vek{u} + \vtau_{air} +
111 dimitri 1.13 \vtau_{ocean} - mg \nabla{\phi(0)} + \vek{F},
112 dimitri 1.1 \end{equation}
113 dimitri 1.13 where $m=m_{i}+m_{s}$ is the ice and snow mass per unit area;
114     $\vek{u}=u\vek{i}+v\vek{j}$ is the ice velocity vector;
115     $\vek{i}$, $\vek{j}$, and $\vek{k}$ are unit vectors in the $x$, $y$, and $z$
116     directions, respectively;
117     $f$ is the Coriolis parameter;
118     $\vtau_{air}$ and $\vtau_{ocean}$ are the wind-ice and ocean-ice stresses,
119     respectively;
120     $g$ is the gravity accelation;
121     $\nabla\phi(0)$ is the gradient (or tilt) of the sea surface height;
122     $\phi(0)$ is the sea surface height potential in response to ocean dynamics
123     and to atmospheric pressure loading;
124     and $\vek{F}=\nabla\cdot\sigma$ is the divergence of the internal ice stress
125     tensor $\sigma_{ij}$.
126     When using the rescaled vertical coordinate system, z$^\ast$, of
127     \citet{cam08}, $\phi(0)$ also includes a term due to snow and ice loading, $mg$.
128     Advection of sea-ice momentum is neglected. The wind and ice-ocean stress
129     terms are given by
130 dimitri 1.1 \begin{align*}
131 dimitri 1.13 \vtau_{air} = & \rho_{air} C_{air} |\vek{U}_{air} -\vek{u}|
132     R_{air} (\vek{U}_{air} -\vek{u}), \\
133     \vtau_{ocean} = & \rho_{ocean}C_{ocean} |\vek{U}_{ocean}-\vek{u}|
134 dimitri 1.1 R_{ocean}(\vek{U}_{ocean}-\vek{u}), \\
135     \end{align*}
136     where $\vek{U}_{air/ocean}$ are the surface winds of the atmosphere
137 dimitri 1.13 and surface currents of the ocean, respectively; $C_{air/ocean}$ are
138     air and ocean drag coefficients; $\rho_{air/ocean}$ are reference
139     densities; and $R_{air/ocean}$ are rotation matrices that act on the
140     wind/current vectors.
141 dimitri 1.1
142     For an isotropic system this stress tensor can be related to the ice
143     strain rate and strength by a nonlinear viscous-plastic (VP)
144     constitutive law \citep{hibler79, zhang98}:
145     \begin{equation}
146     \label{eq:vpequation}
147     \sigma_{ij}=2\eta(\dot{\epsilon}_{ij},P)\dot{\epsilon}_{ij}
148     + \left[\zeta(\dot{\epsilon}_{ij},P) -
149     \eta(\dot{\epsilon}_{ij},P)\right]\dot{\epsilon}_{kk}\delta_{ij}
150     - \frac{P}{2}\delta_{ij}.
151     \end{equation}
152     The ice strain rate is given by
153     \begin{equation*}
154     \dot{\epsilon}_{ij} = \frac{1}{2}\left(
155     \frac{\partial{u_{i}}}{\partial{x_{j}}} +
156     \frac{\partial{u_{j}}}{\partial{x_{i}}}\right).
157     \end{equation*}
158 mlosch 1.5 The maximum ice pressure $P_{\max}$, a measure of ice strength, depends on
159     both thickness $h$ and compactness (concentration) $c$:
160 mlosch 1.4 \begin{equation}
161 mlosch 1.5 P_{\max} = P^{*}c\,h\,e^{[C^{*}\cdot(1-c)]},
162 mlosch 1.9 \label{eq:icestrength}
163 mlosch 1.4 \end{equation}
164     with the constants $P^{*}$ and $C^{*}$. The nonlinear bulk and shear
165     viscosities $\eta$ and $\zeta$ are functions of ice strain rate
166     invariants and ice strength such that the principal components of the
167     stress lie on an elliptical yield curve with the ratio of major to
168     minor axis $e$ equal to $2$; they are given by:
169 dimitri 1.1 \begin{align*}
170 mlosch 1.5 \zeta =& \min\left(\frac{P_{\max}}{2\max(\Delta,\Delta_{\min})},
171     \zeta_{\max}\right) \\
172     \eta =& \frac{\zeta}{e^2} \\
173 dimitri 1.1 \intertext{with the abbreviation}
174     \Delta = & \left[
175     \left(\dot{\epsilon}_{11}^2+\dot{\epsilon}_{22}^2\right)
176     (1+e^{-2}) + 4e^{-2}\dot{\epsilon}_{12}^2 +
177     2\dot{\epsilon}_{11}\dot{\epsilon}_{22} (1-e^{-2})
178     \right]^{-\frac{1}{2}}
179     \end{align*}
180 mlosch 1.5 The bulk viscosities are bounded above by imposing both a minimum
181     $\Delta_{\min}=10^{-11}\text{\,s}^{-1}$ (for numerical reasons) and a
182     maximum $\zeta_{\max} = P_{\max}/\Delta^*$, where
183     $\Delta^*=(5\times10^{12}/2\times10^4)\text{\,s}^{-1}$. For stress
184 dimitri 1.13 tensor computation the replacement pressure $P = 2\,\Delta\zeta$
185 mlosch 1.5 \citep{hibler95} is used so that the stress state always lies on the
186     elliptic yield curve by definition.
187    
188 mlosch 1.6 In the so-called truncated ellipse method the shear viscosity $\eta$
189     is capped to suppress any tensile stress \citep{hibler97, geiger98}:
190     \begin{equation}
191     \label{eq:etatem}
192     \eta = \min(\frac{\zeta}{e^2}
193     \frac{\frac{P}{2}-\zeta(\dot{\epsilon}_{11}+\dot{\epsilon}_{22})}
194     {\sqrt{(\dot{\epsilon}_{11}+\dot{\epsilon}_{22})^2
195     +4\dot{\epsilon}_{12}^2}}
196     \end{equation}
197    
198 dimitri 1.1 In the current implementation, the VP-model is integrated with the
199     semi-implicit line successive over relaxation (LSOR)-solver of
200     \citet{zhang98}, which allows for long time steps that, in our case,
201     is limited by the explicit treatment of the Coriolis term. The
202     explicit treatment of the Coriolis term does not represent a severe
203     limitation because it restricts the time step to approximately the
204     same length as in the ocean model where the Coriolis term is also
205     treated explicitly.
206    
207     \citet{hunke97}'s introduced an elastic contribution to the strain
208 dimitri 1.13 rate elastic-viscous-plastic in order to regularize
209 dimitri 1.1 Eq.\refeq{vpequation} in such a way that the resulting
210 dimitri 1.13 elastic-viscous-plastic (EVP) and VP models are identical at steady
211 dimitri 1.1 state,
212     \begin{equation}
213     \label{eq:evpequation}
214     \frac{1}{E}\frac{\partial\sigma_{ij}}{\partial{t}} +
215     \frac{1}{2\eta}\sigma_{ij}
216     + \frac{\eta - \zeta}{4\zeta\eta}\sigma_{kk}\delta_{ij}
217     + \frac{P}{4\zeta}\delta_{ij}
218     = \dot{\epsilon}_{ij}.
219     \end{equation}
220     %In the EVP model, equations for the components of the stress tensor
221     %$\sigma_{ij}$ are solved explicitly. Both model formulations will be
222     %used and compared the present sea-ice model study.
223     The EVP-model uses an explicit time stepping scheme with a short
224     timestep. According to the recommendation of \citet{hunke97}, the
225     EVP-model is stepped forward in time 120 times within the physical
226     ocean model time step (although this parameter is under debate), to
227     allow for elastic waves to disappear. Because the scheme does not
228     require a matrix inversion it is fast in spite of the small timestep
229     \citep{hunke97}. For completeness, we repeat the equations for the
230     components of the stress tensor $\sigma_{1} =
231     \sigma_{11}+\sigma_{22}$, $\sigma_{2}= \sigma_{11}-\sigma_{22}$, and
232     $\sigma_{12}$. Introducing the divergence $D_D =
233     \dot{\epsilon}_{11}+\dot{\epsilon}_{22}$, and the horizontal tension
234     and shearing strain rates, $D_T =
235     \dot{\epsilon}_{11}-\dot{\epsilon}_{22}$ and $D_S =
236 dimitri 1.13 2\dot{\epsilon}_{12}$, respectively, and using the above abbreviations,
237 dimitri 1.1 the equations can be written as:
238     \begin{align}
239     \label{eq:evpstresstensor1}
240     \frac{\partial\sigma_{1}}{\partial{t}} + \frac{\sigma_{1}}{2T} +
241     \frac{P}{2T} &= \frac{P}{2T\Delta} D_D \\
242     \label{eq:evpstresstensor2}
243     \frac{\partial\sigma_{2}}{\partial{t}} + \frac{\sigma_{2} e^{2}}{2T}
244     &= \frac{P}{2T\Delta} D_T \\
245     \label{eq:evpstresstensor12}
246     \frac{\partial\sigma_{12}}{\partial{t}} + \frac{\sigma_{12} e^{2}}{2T}
247     &= \frac{P}{4T\Delta} D_S
248     \end{align}
249     Here, the elastic parameter $E$ is redefined in terms of a damping timescale
250     $T$ for elastic waves \[E=\frac{\zeta}{T}.\]
251     $T=E_{0}\Delta{t}$ with the tunable parameter $E_0<1$ and
252     the external (long) timestep $\Delta{t}$. \citet{hunke97} recommend
253     $E_{0} = \frac{1}{3}$.
254    
255     For details of the spatial discretization, the reader is referred to
256     \citet{zhang98, zhang03}. Our discretization differs only (but
257     importantly) in the underlying grid, namely the Arakawa C-grid, but is
258     otherwise straightforward. The EVP model in particular is discretized
259     naturally on the C-grid with $\sigma_{1}$ and $\sigma_{2}$ on the
260     center points and $\sigma_{12}$ on the corner (or vorticity) points of
261     the grid. With this choice all derivatives are discretized as central
262     differences and averaging is only involved in computing $\Delta$ and
263     $P$ at vorticity points.
264    
265     For a general curvilinear grid, one needs in principle to take metric
266 dimitri 1.13 terms into account that arise in the transformation of a curvilinear grid
267     on the sphere. For now, however, we can neglect these metric terms
268 dimitri 1.1 because they are very small on the cubed sphere grids used in this
269     paper; in particular, only near the edges of the cubed sphere grid, we
270     expect them to be non-zero, but these edges are at approximately
271     35\degS\ or 35\degN\ which are never covered by sea-ice in our
272     simulations. Everywhere else the coordinate system is locally nearly
273     cartesian. However, for last-glacial-maximum or snowball-earth-like
274     simulations the question of metric terms needs to be reconsidered.
275     Either, one includes these terms as in \citet{zhang03}, or one finds a
276 dimitri 1.13 vector-invariant formulation for the sea-ice internal stress term that
277 dimitri 1.1 does not require any metric terms, as it is done in the ocean dynamics
278     of the MITgcm \citep{adcroft04:_cubed_sphere}.
279    
280     Moving sea ice exerts a stress on the ocean which is the opposite of
281     the stress $\vtau_{ocean}$ in Eq.\refeq{momseaice}. This stess is
282     applied directly to the surface layer of the ocean model. An
283     alternative ocean stress formulation is given by \citet{hibler87}.
284     Rather than applying $\vtau_{ocean}$ directly, the stress is derived
285     from integrating over the ice thickness to the bottom of the oceanic
286     surface layer. In the resulting equation for the \emph{combined}
287     ocean-ice momentum, the interfacial stress cancels and the total
288     stress appears as the sum of windstress and divergence of internal ice
289     stresses: $\delta(z) (\vtau_{air} + \vek{F})/\rho_0$, \citep[see also
290     Eq.\,2 of][]{hibler87}. The disadvantage of this formulation is that
291     now the velocity in the surface layer of the ocean that is used to
292     advect tracers, is really an average over the ocean surface
293     velocity and the ice velocity leading to an inconsistency as the ice
294     temperature and salinity are different from the oceanic variables.
295    
296     Sea ice distributions are characterized by sharp gradients and edges.
297     For this reason, we employ a positive 3rd-order advection scheme
298     \citep{hundsdorfer94} for the thermodynamic variables discussed in the
299     next section.
300    
301     \subparagraph{boundary conditions: no-slip, free-slip, half-slip}
302    
303     \begin{itemize}
304     \item transition from existing B-Grid to C-Grid
305     \item boundary conditions: no-slip, free-slip, half-slip
306     \item fancy (multi dimensional) advection schemes
307     \item VP vs.\ EVP \citep{hunke97}
308     \item ocean stress formulation \citep{hibler87}
309     \end{itemize}
310    
311     \subsection{Thermodynamics}
312     \label{sec:thermodynamics}
313    
314     In the original formulation the sea ice model \citep{menemenlis05}
315     uses simple thermodynamics following the appendix of
316     \citet{semtner76}. This formulation does not allow storage of heat
317     (heat capacity of ice is zero, and this type of model is often refered
318     to as a ``zero-layer'' model). Upward heat flux is parameterized
319     assuming a linear temperature profile and together with a constant ice
320     conductivity. It is expressed as $(K/h)(T_{w}-T_{0})$, where $K$ is
321     the ice conductivity, $h$ the ice thickness, and $T_{w}-T_{0}$ the
322     difference between water and ice surface temperatures. The surface
323     heat budget is computed in a similar way to that of
324     \citet{parkinson79} and \citet{manabe79}.
325    
326     There is considerable doubt about the reliability of such a simple
327     thermodynamic model---\citet{semtner84} found significant errors in
328     phase (one month lead) and amplitude ($\approx$50\%\,overestimate) in
329     such models---, so that today many sea ice models employ more complex
330     thermodynamics. A popular thermodynamics model of \citet{winton00} is
331     based on the 3-layer model of \citet{semtner76} and treats brine
332     content by means of enthalphy conservation. This model requires in
333     addition to ice-thickness and compactness (fractional area) additional
334     state variables to be advected by ice velocities, namely enthalphy of
335     the two ice layers and the thickness of the overlying snow layer.
336    
337 mlosch 1.9
338 dimitri 1.1 \subsection{C-grid}
339     \begin{itemize}
340     \item no-slip vs. free-slip for both lsr and evp;
341     "diagnostics" to look at and use for comparison
342     \begin{itemize}
343     \item ice transport through Fram Strait/Denmark Strait/Davis
344     Strait/Bering strait (these are general)
345     \item ice transport through narrow passages, e.g.\ Nares-Strait
346     \end{itemize}
347     \item compare different advection schemes (if lsr turns out to be more
348     effective, then with lsr otherwise I prefer evp), eg.
349     \begin{itemize}
350     \item default 2nd-order with diff1=0.002
351     \item 1st-order upwind with diff1=0.
352     \item DST3FL (SEAICEadvScheme=33 with diff1=0., should work, works for me)
353     \item 2nd-order wit flux limiter (SEAICEadvScheme=77 with diff1=0.)
354     \end{itemize}
355     That should be enough. Here, total ice mass and location of ice edge
356     is interesting. However, this comparison can be done in an idealized
357     domain, may not require full Arctic Domain?
358     \item
359     Do a little study on the parameters of LSR and EVP
360     \begin{enumerate}
361     \item convergence of LSR, how many iterations do you need to get a
362     true elliptic yield curve
363     \item vary deltaTevp and the relaxation parameter for EVP and see when
364     the EVP solution breaks down (relative to the forcing time scale).
365     For this, it is essential that the evp solver gives use "stripeless"
366     solutions, that is your dtevp = 1sec solutions/or 10sec solutions
367     with SEAICE\_evpDampC = 615.
368     \end{enumerate}
369     \end{itemize}
370    
371     \section{Forward sensitivity experiments}
372     \label{sec:forward}
373    
374     A second series of forward sensitivity experiments have been carried out on an
375     Arctic Ocean domain with open boundaries. Once again the objective is to
376     compare the old B-grid LSR dynamic solver with the new C-grid LSR and EVP
377     solvers. One additional experiment is carried out to illustrate the
378     differences between the two main options for sea ice thermodynamics in the MITgcm.
379    
380     \subsection{Arctic Domain with Open Boundaries}
381     \label{sec:arctic}
382    
383 dimitri 1.12 The Arctic domain of integration is illustrated in Fig.~\ref{fig:arctic1}. It
384 mlosch 1.6 is carved out from, and obtains open boundary conditions from, the
385     global cubed-sphere configuration of the Estimating the Circulation
386     and Climate of the Ocean, Phase II (ECCO2) project
387     \citet{menemenlis05}. The domain size is 420 by 384 grid boxes
388     horizontally with mean horizontal grid spacing of 18 km.
389 dimitri 1.1
390 dimitri 1.12 \begin{figure}
391     %\centerline{{\includegraphics*[width=0.44\linewidth]{\fpath/arctic1.eps}}}
392     \caption{Bathymetry of Arctic Domain.\label{fig:arctic1}}
393     \end{figure}
394    
395 dimitri 1.1 There are 50 vertical levels ranging in thickness from 10 m near the surface
396     to approximately 450 m at a maximum model depth of 6150 m. Bathymetry is from
397     the National Geophysical Data Center (NGDC) 2-minute gridded global relief
398     data (ETOPO2) and the model employs the partial-cell formulation of
399 mlosch 1.6 \citet{adcroft97:_shaved_cells}, which permits accurate representation of the bathymetry. The
400 dimitri 1.1 model is integrated in a volume-conserving configuration using a finite volume
401     discretization with C-grid staggering of the prognostic variables. In the
402 mlosch 1.6 ocean, the non-linear equation of state of \citet{jackett95}. The ocean model is
403 dimitri 1.1 coupled to a sea-ice model described hereinabove.
404    
405 mlosch 1.6 This particular ECCO2 simulation is initialized from rest using the
406     January temperature and salinity distribution from the World Ocean
407     Atlas 2001 (WOA01) [Conkright et al., 2002] and it is integrated for
408     32 years prior to the 1996--2001 period discussed in the study. Surface
409     boundary conditions are from the National Centers for Environmental
410     Prediction and the National Center for Atmospheric Research
411     (NCEP/NCAR) atmospheric reanalysis [Kistler et al., 2001]. Six-hourly
412     surface winds, temperature, humidity, downward short- and long-wave
413     radiations, and precipitation are converted to heat, freshwater, and
414     wind stress fluxes using the \citet{large81, large82} bulk formulae.
415     Shortwave radiation decays exponentially as per Paulson and Simpson
416     [1977]. Additionally the time-mean river run-off from Large and Nurser
417     [2001] is applied and there is a relaxation to the monthly-mean
418     climatological sea surface salinity values from WOA01 with a
419     relaxation time scale of 3 months. Vertical mixing follows
420     \citet{large94} with background vertical diffusivity of
421     $1.5\times10^{-5}\text{\,m$^{2}$\,s$^{-1}$}$ and viscosity of
422     $10^{-3}\text{\,m$^{2}$\,s$^{-1}$}$. A third order, direct-space-time
423     advection scheme with flux limiter is employed \citep{hundsdorfer94}
424     and there is no explicit horizontal diffusivity. Horizontal viscosity
425     follows \citet{lei96} but
426     modified to sense the divergent flow as per Fox-Kemper and Menemenlis
427     [in press]. Shortwave radiation decays exponentially as per Paulson
428     and Simpson [1977]. Additionally, the time-mean runoff of Large and
429     Nurser [2001] is applied near the coastline and, where there is open
430     water, there is a relaxation to monthly-mean WOA01 sea surface
431     salinity with a time constant of 45 days.
432 dimitri 1.1
433     Open water, dry
434     ice, wet ice, dry snow, and wet snow albedo are, respectively, 0.15, 0.85,
435     0.76, 0.94, and 0.8.
436    
437     \begin{itemize}
438     \item Configuration
439     \item OBCS from cube
440     \item forcing
441     \item 1/2 and full resolution
442     \item with a few JFM figs from C-grid LSR no slip
443     ice transport through Canadian Archipelago
444     thickness distribution
445     ice velocity and transport
446     \end{itemize}
447    
448     \begin{itemize}
449     \item Arctic configuration
450     \item ice transport through straits and near boundaries
451     \item focus on narrow straits in the Canadian Archipelago
452     \end{itemize}
453    
454     \begin{itemize}
455     \item B-grid LSR no-slip
456     \item C-grid LSR no-slip
457     \item C-grid LSR slip
458     \item C-grid EVP no-slip
459     \item C-grid EVP slip
460 mlosch 1.6 \item C-grid LSR + TEM (truncated ellipse method, no tensile stress, new flag)
461 dimitri 1.1 \item C-grid LSR no-slip + Winton
462     \item speed-performance-accuracy (small)
463     ice transport through Canadian Archipelago differences
464     thickness distribution differences
465     ice velocity and transport differences
466     \end{itemize}
467    
468     We anticipate small differences between the different models due to:
469     \begin{itemize}
470     \item advection schemes: along the ice-edge and regions with large
471     gradients
472 mlosch 1.6 \item C-grid: less transport through narrow straits for no slip
473     conditons, more for free slip
474 dimitri 1.1 \item VP vs.\ EVP: speed performance, accuracy?
475     \item ocean stress: different water mass properties beneath the ice
476     \end{itemize}
477    
478     \section{Adjoint sensiivities of the MITsim}
479    
480     \subsection{The adjoint of MITsim}
481    
482     The ability to generate tangent linear and adjoint model components
483     of the MITsim has been a main design task.
484     For the ocean the adjoint capability has proven to be an
485     invaluable tool for sensitivity analysis as well as state estimation.
486     In short, the adjoint enables very efficient computation of the gradient
487     of scalar-valued model diagnostics (called cost function or objective function)
488     with respect to many model "variables".
489     These variables can be two- or three-dimensional fields of initial
490     conditions, model parameters such as mixing coefficients, or
491     time-varying surface or lateral (open) boundary conditions.
492     When combined, these variables span a potentially high-dimensional
493     (e.g. O(10$^8$)) so-called control space. Performing parameter perturbations
494     to assess model sensitivities quickly becomes prohibitive at these scales.
495     Alternatively, (time-varying) sensitivities of the objective function
496     to any element of the control space can be computed very efficiently in
497     one single adjoint
498     model integration, provided an efficient adjoint model is available.
499    
500     [REFERENCES]
501    
502    
503     The adjoint operator (ADM) is the transpose of the tangent linear operator (TLM)
504     of the full (in general nonlinear) forward model, i.e. the MITsim.
505     The TLM maps perturbations of elements of the control space
506     (e.g. initial ice thickness distribution)
507     via the model Jacobian
508     to a perturbation in the objective function
509     (e.g. sea-ice export at the end of the integration interval).
510     \textit{Tangent} linearity ensures that the derivatives are evaluated
511     with respect to the underlying model trajectory at each point in time.
512     This is crucial for nonlinear trajectories and the presence of different
513     regimes (e.g. effect of the seaice growth term at or away from the
514     freezing point of the ocean surface).
515     Ensuring tangent linearity can be easily achieved by integrating
516     the full model in sync with the TLM to provide the underlying model state.
517     Ensuring \textit{tangent} adjoints is equally crucial, but much more
518     difficult to achieve because of the reverse nature of the integration:
519     the adjoint accumulates sensitivities backward in time,
520     starting from a unit perturbation of the objective function.
521     The adjoint model requires the model state in reverse order.
522     This presents one of the major complications in deriving an
523     exact, i.e. \textit{tangent} adjoint model.
524    
525     Following closely the development and maintenance of TLM and ADM
526     components of the MITgcm we have relied heavily on the
527     autmomatic differentiation (AD) tool
528     "Transformation of Algorithms in Fortran" (TAF)
529     developed by Fastopt (Giering and Kaminski, 1998)
530     to derive TLM and ADM code of the MITsim.
531     Briefly, the nonlinear parent model is fed to the AD tool which produces
532     derivative code for the specified control space and objective function.
533     Following this approach has (apart from its evident success)
534     several advantages:
535     (1) the adjoint model is the exact adjoint operator of the parent model,
536     (2) the adjoint model can be kept up to date with respect to ongoing
537     development of the parent model, and adjustments to the parent model
538     to extend the automatically generated adjoint are incremental changes
539     only, rather than extensive re-developments,
540     (3) the parallel structure of the parent model is preserved
541     by the adjoint model, ensuring efficient use in high performance
542     computing environments.
543    
544     Some initial code adjustments are required to support dependency analysis
545     of the flow reversal and certain language limitations which may lead
546     to irreducible flow graphs (e.g. GOTO statements).
547     The problem of providing the required model state in reverse order
548     at the time of evaluating nonlinear or conditional
549     derivatives is solved via balancing
550     storing vs. recomputation of the model state in a multi-level
551     checkpointing loop.
552     Again, an initial code adjustment is required to support TAFs
553     checkpointing capability.
554 mlosch 1.6 The code adjustments are sufficiently simple so as not to cause
555 dimitri 1.1 major limitations to the full nonlinear parent model.
556     Once in place, an adjoint model of a new model configuration
557     may be derived in about 10 minutes.
558    
559     [HIGHLIGHT COUPLED NATURE OF THE ADJOINT!]
560    
561     \subsection{Special considerations}
562    
563     * growth term(?)
564    
565     * small active denominators
566    
567     * dynamic solver (implicit function theorem)
568    
569     * approximate adjoints
570    
571    
572     \subsection{An example: sensitivities of sea-ice export through Fram Strait}
573    
574     We demonstrate the power of the adjoint method
575     in the context of investigating sea-ice export sensitivities through Fram Strait
576     (for details of this study see Heimbach et al., 2007).
577 mlosch 1.6 %\citep[for details of this study see][]{heimbach07}. %Heimbach et al., 2007).
578 dimitri 1.1 The domain chosen is a coarsened version of the Arctic face of the
579     high-resolution cubed-sphere configuration of the ECCO2 project
580 mlosch 1.6 \citep[see][]{menemenlis05}. It covers the entire Arctic,
581 dimitri 1.1 extends into the North Pacific such as to cover the entire
582     ice-covered regions, and comprises parts of the North Atlantic
583     down to XXN to enable analysis of remote influences of the
584     North Atlantic current to sea-ice variability and export.
585     The horizontal resolution varies between XX and YY km
586     with 50 unevenly spaced vertical levels.
587     The adjoint models run efficiently on 80 processors
588     (benchmarks have been performed both on an SGI Altix as well as an
589     IBM SP5 at NASA/ARC).
590    
591 mlosch 1.6 Following a 1-year spinup, the model has been integrated for four
592     years between 1992 and 1995. It is forced using realistic 6-hourly
593     NCEP/NCAR atmospheric state variables. Over the open ocean these are
594     converted into air-sea fluxes via the bulk formulae of
595     \citet{large04}. Derivation of air-sea fluxes in the presence of
596     sea-ice is handled by the ice model as described in \refsec{model}.
597 dimitri 1.1 The objective function chosen is sea-ice export through Fram Strait
598 mlosch 1.6 computed for December 1995. The adjoint model computes sensitivities
599     to sea-ice export back in time from 1995 to 1992 along this
600     trajectory. In principle all adjoint model variable (i.e., Lagrange
601     multipliers) of the coupled ocean/sea-ice model are available to
602     analyze the transient sensitivity behaviour of the ocean and sea-ice
603     state. Over the open ocean, the adjoint of the bulk formula scheme
604     computes sensitivities to the time-varying atmospheric state. Over
605     ice-covered parts, the sea-ice adjoint converts surface ocean
606     sensitivities to atmospheric sensitivities.
607    
608     \reffig{4yradjheff}(a--d) depict sensitivities of sea-ice export
609     through Fram Strait in December 1995 to changes in sea-ice thickness
610     12, 24, 36, 48 months back in time. Corresponding sensitivities to
611     ocean surface temperature are depicted in
612     \reffig{4yradjthetalev1}(a--d). The main characteristics is
613     consistency with expected advection of sea-ice over the relevant time
614     scales considered. The general positive pattern means that an
615     increase in sea-ice thickness at location $(x,y)$ and time $t$ will
616     increase sea-ice export through Fram Strait at time $T_e$. Largest
617     distances from Fram Strait indicate fastest sea-ice advection over the
618     time span considered. The ice thickness sensitivities are in close
619     correspondence to ocean surface sentivitites, but of opposite sign.
620     An increase in temperature will incur ice melting, decrease in ice
621     thickness, and therefore decrease in sea-ice export at time $T_e$.
622 dimitri 1.1
623     The picture is fundamentally different and much more complex
624     for sensitivities to ocean temperatures away from the surface.
625 mlosch 1.6 \reffig{4yradjthetalev10??}(a--d) depicts ice export sensitivities to
626 dimitri 1.1 temperatures at roughly 400 m depth.
627     Primary features are the effect of the heat transport of the North
628     Atlantic current which feeds into the West Spitsbergen current,
629     the circulation around Svalbard, and ...
630    
631     \begin{figure}[t!]
632     \centerline{
633     \subfigure[{\footnotesize -12 months}]
634 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim072_cmax2.0E+02.eps}}
635 dimitri 1.1 %\includegraphics*[width=.3\textwidth]{H_c.bin_res_100_lev1.pdf}
636     %
637     \subfigure[{\footnotesize -24 months}]
638 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim145_cmax2.0E+02.eps}}
639 dimitri 1.1 }
640    
641     \centerline{
642     \subfigure[{\footnotesize
643     -36 months}]
644 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim218_cmax2.0E+02.eps}}
645 dimitri 1.1 %
646     \subfigure[{\footnotesize
647     -48 months}]
648 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim292_cmax2.0E+02.eps}}
649 dimitri 1.1 }
650     \caption{Sensitivity of sea-ice export through Fram Strait in December 2005 to
651     sea-ice thickness at various prior times.
652     \label{fig:4yradjheff}}
653     \end{figure}
654    
655    
656     \begin{figure}[t!]
657     \centerline{
658     \subfigure[{\footnotesize -12 months}]
659 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim072_cmax5.0E+01.eps}}
660 dimitri 1.1 %\includegraphics*[width=.3\textwidth]{H_c.bin_res_100_lev1.pdf}
661     %
662     \subfigure[{\footnotesize -24 months}]
663 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim145_cmax5.0E+01.eps}}
664 dimitri 1.1 }
665    
666     \centerline{
667     \subfigure[{\footnotesize
668     -36 months}]
669 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim218_cmax5.0E+01.eps}}
670 dimitri 1.1 %
671     \subfigure[{\footnotesize
672     -48 months}]
673 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim292_cmax5.0E+01.eps}}
674 dimitri 1.1 }
675 mlosch 1.6 \caption{Same as \reffig{4yradjheff} but for sea surface temperature
676 dimitri 1.1 \label{fig:4yradjthetalev1}}
677     \end{figure}
678    
679    
680    
681     \section{Discussion and conclusion}
682     \label{sec:concl}
683    
684     The story of the paper could be:
685     B-grid ice model + C-grid ocean model does not work properly for us,
686     therefore C-grid ice model with advantages:
687     \begin{enumerate}
688     \item stress coupling simpler (no interpolation required)
689     \item different boundary conditions
690     \item advection schemes carry over trivially from main code
691     \end{enumerate}
692     LSR/EVP solutions are similar with appropriate bcs, evp parameters as
693     a function of forcing time scale (when does VP solution break
694     down). Same for LSR solver, provided that it works (o:
695     Which scheme is more efficient for the resolution/time stepping
696     parameters that we use here. What about other resolutions?
697    
698     \paragraph{Acknowledgements}
699     We thank Jinlun Zhang for providing the original B-grid code and many
700 mlosch 1.6 helpful discussions. ML thanks Elizabeth Hunke for multiple explanations.
701 dimitri 1.1
702 dimitri 1.7 \bibliography{bib/journal_abrvs,bib/seaice,bib/genocean,bib/maths,bib/mitgcmuv,bib/fram}
703 dimitri 1.1
704     \end{document}
705    
706     %%% Local Variables:
707     %%% mode: latex
708     %%% TeX-master: t
709     %%% End:
710 mlosch 1.4
711    
712     A Dynamic-Thermodynamic Sea ice Model for Ocean Climate
713     Estimation on an Arakawa C-Grid
714    
715     Introduction
716    
717     Ice Model:
718     Dynamics formulation.
719     B-C, LSR, EVP, no-slip, slip
720     parallellization
721     Thermodynamics formulation.
722     0-layer Hibler salinity + snow
723     3-layer Winton
724    
725     Idealized tests
726     Funnel Experiments
727     Downstream Island tests
728     B-grid LSR no-slip
729     C-grid LSR no-slip
730     C-grid LSR slip
731     C-grid EVP no-slip
732     C-grid EVP slip
733    
734     Arctic Setup
735     Configuration
736     OBCS from cube
737     forcing
738     1/2 and full resolution
739     with a few JFM figs from C-grid LSR no slip
740     ice transport through Canadian Archipelago
741     thickness distribution
742     ice velocity and transport
743    
744     Arctic forward sensitivity experiments
745     B-grid LSR no-slip
746     C-grid LSR no-slip
747     C-grid LSR slip
748     C-grid EVP no-slip
749     C-grid EVP slip
750     C-grid LSR no-slip + Winton
751     speed-performance-accuracy (small)
752     ice transport through Canadian Archipelago differences
753     thickness distribution differences
754     ice velocity and transport differences
755    
756     Adjoint sensitivity experiment on 1/2-res setup
757     Sensitivity of sea ice volume flow through Fram Strait
758     *** Sensitivity of sea ice volume flow through Canadian Archipelago
759    
760     Summary and conluding remarks

  ViewVC Help
Powered by ViewVC 1.1.22