/[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 1.16 - (hide annotations) (download) (as text)
Tue Feb 26 19:14:36 2008 UTC (17 years, 4 months ago) by mlosch
Branch: MAIN
Changes since 1.15: +34 -31 lines
File MIME type: application/x-tex
rewrite first and third paragraph of introduction before Dimitris
tears nice tex file appart

1 mlosch 1.16 % $Header: /u/gcmpack/MITgcm_contrib/articles/ceaice/ceaice.tex,v 1.15 2008/02/26 17:21:48 mlosch 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 As part of ongoing efforts to obtain a best possible synthesis of most
56 dimitri 1.12 available, global-scale, ocean and sea ice data, a dynamic and thermodynamic
57     sea-ice model has been coupled to the Massachusetts Institute of Technology
58     general circulation model (MITgcm). Ice mechanics follow a viscous plastic
59     rheology and the ice momentum equations are solved numerically using either
60     line successive relaxation (LSR) or elastic-viscous-plastic (EVP) dynamic
61     models. Ice thermodynamics are represented using either a zero-heat-capacity
62     formulation or a two-layer formulation that conserves enthalpy. The model
63     includes prognostic variables for snow and for sea-ice salinity. The above
64     sea ice model components were borrowed from current-generation climate models
65     but they were reformulated on an Arakawa C-grid in order to match the MITgcm
66     oceanic grid and they were modified in many ways to permit efficient and
67     accurate automatic differentiation. This paper describes the MITgcm sea ice
68     model; it presents example Arctic and Antarctic results from a realistic,
69     eddy-permitting, global ocean and sea-ice configuration; it compares B-grid
70     and C-grid dynamic solvers in a regional Arctic configuration; and it presents
71     example results from coupled ocean and sea-ice adjoint-model integrations.
72 dimitri 1.10
73 dimitri 1.1 \end{abstract}
74    
75     \section{Introduction}
76     \label{sec:intro}
77    
78 mlosch 1.16 The availability of an adjoint model as a powerful research tool
79     complementary to an ocean model was a major design requirement early
80     on in the development of the MIT general circulation model (MITgcm)
81     [Marshall et al. 1997a, Marotzke et al. 1999, Adcroft et al. 2002]. It
82     was recognized that the adjoint model permitted computing the
83     gradients of various scalar-valued model diagnostics, norms or,
84     generally, objective functions with respect to external or independent
85     parameters very efficiently. The information associtated with these
86     gradients is useful in at least two major contexts. First, for state
87     estimation problems, the objective function is the sum of squared
88     differences between observations and model results weighted by the
89     inverse error covariances. The gradient of such an objective function
90     can be used to reduce this measure of model-data misfit to find the
91     optimal model solution in a least-squares sense. Second, the
92     objective function can be a key oceanographic quantity such as
93     meridional heat or volume transport, ocean heat content or mean
94     surface temperature index. In this case the gradient provides a
95     complete set of sensitivities of this quantity to all independent
96     variables simultaneously. These sensitivities can be used to address
97     the cause of, say, changing net transports accurately.
98 dimitri 1.14
99     References to existing sea-ice adjoint models, explaining that they are either
100     for simplified configurations, for ice-only studies, or for short-duration
101     studies to motivate the present work.
102    
103 dimitri 1.1 Traditionally, probably for historical reasons and the ease of
104     treating the Coriolis term, most standard sea-ice models are
105     discretized on Arakawa-B-grids \citep[e.g.,][]{hibler79, harder99,
106 mlosch 1.16 kreyscher00, zhang98, hunke97}. From the perspective of coupling a
107 dimitri 1.1 sea ice-model to a C-grid ocean model, the exchange of fluxes of heat
108     and fresh-water pose no difficulty for a B-grid sea-ice model
109     \citep[e.g.,][]{timmermann02a}. However, surface stress is defined at
110     velocities points and thus needs to be interpolated between a B-grid
111 mlosch 1.16 sea-ice model and a C-grid ocean model. Smoothing implicitly
112     associated with this interpolation may mask grid scale noise and may
113     contribute to stabilizing the solution. On the other hand, by
114     smoothing the stress signals are damped which could lead to reduced
115     variability of the system. By choosing a C-grid for the sea-ice model,
116     we circumvent this difficulty altogether and render the stress
117     coupling as consistent as the buoyancy coupling.
118 dimitri 1.1
119     A further advantage of the C-grid formulation is apparent in narrow
120     straits. In the limit of only one grid cell between coasts there is no
121     flux allowed for a B-grid (with no-slip lateral boundary counditions),
122 mlosch 1.16 and models have used topographies artificially widened straits to
123     avoid this problem \citep{holloway07}. The C-grid formulation on the
124     other hand allows a flux of sea-ice through narrow passages if
125     free-slip along the boundaries is allowed. We demonstrate this effect
126     in the Candian archipelago.
127 dimitri 1.1
128 dimitri 1.14 Talk about problems that make the sea-ice-ocean code very sensitive and
129     changes in the code that reduce these sensitivities.
130    
131     This paper describes the MITgcm sea ice
132     model; it presents example Arctic and Antarctic results from a realistic,
133     eddy-permitting, global ocean and sea-ice configuration; it compares B-grid
134     and C-grid dynamic solvers in a regional Arctic configuration; and it presents
135     example results from coupled ocean and sea-ice adjoint-model integrations.
136    
137 dimitri 1.13 \section{Model}
138     \label{sec:model}
139    
140 dimitri 1.1 \subsection{Dynamics}
141     \label{sec:dynamics}
142    
143 dimitri 1.13 The momentum equation of the sea-ice model is
144 dimitri 1.1 \begin{equation}
145     \label{eq:momseaice}
146     m \frac{D\vek{u}}{Dt} = -mf\vek{k}\times\vek{u} + \vtau_{air} +
147 mlosch 1.15 \vtau_{ocean} - m \nabla{\phi(0)} + \vek{F},
148 dimitri 1.1 \end{equation}
149 dimitri 1.13 where $m=m_{i}+m_{s}$ is the ice and snow mass per unit area;
150     $\vek{u}=u\vek{i}+v\vek{j}$ is the ice velocity vector;
151     $\vek{i}$, $\vek{j}$, and $\vek{k}$ are unit vectors in the $x$, $y$, and $z$
152     directions, respectively;
153     $f$ is the Coriolis parameter;
154     $\vtau_{air}$ and $\vtau_{ocean}$ are the wind-ice and ocean-ice stresses,
155     respectively;
156     $g$ is the gravity accelation;
157     $\nabla\phi(0)$ is the gradient (or tilt) of the sea surface height;
158 mlosch 1.15 $\phi(0) = g\eta + p_{a}/\rho_{0}$ is the sea surface height potential
159     in response to ocean dynamics ($g\eta$) and to atmospheric pressure
160     loading ($p_{a}/\rho_{0}$, where $\rho_{0}$ is a reference density);
161 dimitri 1.13 and $\vek{F}=\nabla\cdot\sigma$ is the divergence of the internal ice stress
162     tensor $\sigma_{ij}$.
163     When using the rescaled vertical coordinate system, z$^\ast$, of
164 mlosch 1.15 \citet{cam08}, $\phi(0)$ also includes a term due to snow and ice
165     loading, $mg/\rho_{0}$.
166 dimitri 1.13 Advection of sea-ice momentum is neglected. The wind and ice-ocean stress
167     terms are given by
168 dimitri 1.1 \begin{align*}
169 dimitri 1.13 \vtau_{air} = & \rho_{air} C_{air} |\vek{U}_{air} -\vek{u}|
170     R_{air} (\vek{U}_{air} -\vek{u}), \\
171     \vtau_{ocean} = & \rho_{ocean}C_{ocean} |\vek{U}_{ocean}-\vek{u}|
172 dimitri 1.1 R_{ocean}(\vek{U}_{ocean}-\vek{u}), \\
173     \end{align*}
174     where $\vek{U}_{air/ocean}$ are the surface winds of the atmosphere
175 dimitri 1.13 and surface currents of the ocean, respectively; $C_{air/ocean}$ are
176     air and ocean drag coefficients; $\rho_{air/ocean}$ are reference
177     densities; and $R_{air/ocean}$ are rotation matrices that act on the
178     wind/current vectors.
179 dimitri 1.1
180 mlosch 1.15 For an isotropic system the stress tensor $\sigma_{ij}$ ($i,j=1,2$) can
181     be related to the ice strain rate and strength by a nonlinear
182     viscous-plastic (VP) constitutive law \citep{hibler79, zhang98}:
183 dimitri 1.1 \begin{equation}
184     \label{eq:vpequation}
185     \sigma_{ij}=2\eta(\dot{\epsilon}_{ij},P)\dot{\epsilon}_{ij}
186     + \left[\zeta(\dot{\epsilon}_{ij},P) -
187     \eta(\dot{\epsilon}_{ij},P)\right]\dot{\epsilon}_{kk}\delta_{ij}
188     - \frac{P}{2}\delta_{ij}.
189     \end{equation}
190     The ice strain rate is given by
191     \begin{equation*}
192     \dot{\epsilon}_{ij} = \frac{1}{2}\left(
193     \frac{\partial{u_{i}}}{\partial{x_{j}}} +
194     \frac{\partial{u_{j}}}{\partial{x_{i}}}\right).
195     \end{equation*}
196 mlosch 1.5 The maximum ice pressure $P_{\max}$, a measure of ice strength, depends on
197     both thickness $h$ and compactness (concentration) $c$:
198 mlosch 1.4 \begin{equation}
199 mlosch 1.5 P_{\max} = P^{*}c\,h\,e^{[C^{*}\cdot(1-c)]},
200 mlosch 1.9 \label{eq:icestrength}
201 mlosch 1.4 \end{equation}
202     with the constants $P^{*}$ and $C^{*}$. The nonlinear bulk and shear
203     viscosities $\eta$ and $\zeta$ are functions of ice strain rate
204     invariants and ice strength such that the principal components of the
205     stress lie on an elliptical yield curve with the ratio of major to
206     minor axis $e$ equal to $2$; they are given by:
207 dimitri 1.1 \begin{align*}
208 mlosch 1.5 \zeta =& \min\left(\frac{P_{\max}}{2\max(\Delta,\Delta_{\min})},
209     \zeta_{\max}\right) \\
210     \eta =& \frac{\zeta}{e^2} \\
211 dimitri 1.1 \intertext{with the abbreviation}
212     \Delta = & \left[
213     \left(\dot{\epsilon}_{11}^2+\dot{\epsilon}_{22}^2\right)
214     (1+e^{-2}) + 4e^{-2}\dot{\epsilon}_{12}^2 +
215     2\dot{\epsilon}_{11}\dot{\epsilon}_{22} (1-e^{-2})
216     \right]^{-\frac{1}{2}}
217     \end{align*}
218 mlosch 1.5 The bulk viscosities are bounded above by imposing both a minimum
219     $\Delta_{\min}=10^{-11}\text{\,s}^{-1}$ (for numerical reasons) and a
220     maximum $\zeta_{\max} = P_{\max}/\Delta^*$, where
221     $\Delta^*=(5\times10^{12}/2\times10^4)\text{\,s}^{-1}$. For stress
222 dimitri 1.13 tensor computation the replacement pressure $P = 2\,\Delta\zeta$
223 mlosch 1.5 \citep{hibler95} is used so that the stress state always lies on the
224     elliptic yield curve by definition.
225    
226 mlosch 1.6 In the so-called truncated ellipse method the shear viscosity $\eta$
227     is capped to suppress any tensile stress \citep{hibler97, geiger98}:
228     \begin{equation}
229     \label{eq:etatem}
230 mlosch 1.15 \eta = \min\left(\frac{\zeta}{e^2},
231 mlosch 1.6 \frac{\frac{P}{2}-\zeta(\dot{\epsilon}_{11}+\dot{\epsilon}_{22})}
232     {\sqrt{(\dot{\epsilon}_{11}+\dot{\epsilon}_{22})^2
233 mlosch 1.15 +4\dot{\epsilon}_{12}^2}}\right).
234 mlosch 1.6 \end{equation}
235    
236 dimitri 1.1 In the current implementation, the VP-model is integrated with the
237     semi-implicit line successive over relaxation (LSOR)-solver of
238     \citet{zhang98}, which allows for long time steps that, in our case,
239 mlosch 1.15 are limited by the explicit treatment of the Coriolis term. The
240 dimitri 1.1 explicit treatment of the Coriolis term does not represent a severe
241     limitation because it restricts the time step to approximately the
242     same length as in the ocean model where the Coriolis term is also
243     treated explicitly.
244    
245     \citet{hunke97}'s introduced an elastic contribution to the strain
246 mlosch 1.15 rate in order to regularize Eq.\refeq{vpequation} in such a way that
247     the resulting elastic-viscous-plastic (EVP) and VP models are
248     identical at steady state,
249 dimitri 1.1 \begin{equation}
250     \label{eq:evpequation}
251     \frac{1}{E}\frac{\partial\sigma_{ij}}{\partial{t}} +
252     \frac{1}{2\eta}\sigma_{ij}
253     + \frac{\eta - \zeta}{4\zeta\eta}\sigma_{kk}\delta_{ij}
254     + \frac{P}{4\zeta}\delta_{ij}
255     = \dot{\epsilon}_{ij}.
256     \end{equation}
257     %In the EVP model, equations for the components of the stress tensor
258     %$\sigma_{ij}$ are solved explicitly. Both model formulations will be
259     %used and compared the present sea-ice model study.
260     The EVP-model uses an explicit time stepping scheme with a short
261     timestep. According to the recommendation of \citet{hunke97}, the
262     EVP-model is stepped forward in time 120 times within the physical
263     ocean model time step (although this parameter is under debate), to
264     allow for elastic waves to disappear. Because the scheme does not
265     require a matrix inversion it is fast in spite of the small timestep
266     \citep{hunke97}. For completeness, we repeat the equations for the
267     components of the stress tensor $\sigma_{1} =
268     \sigma_{11}+\sigma_{22}$, $\sigma_{2}= \sigma_{11}-\sigma_{22}$, and
269     $\sigma_{12}$. Introducing the divergence $D_D =
270     \dot{\epsilon}_{11}+\dot{\epsilon}_{22}$, and the horizontal tension
271     and shearing strain rates, $D_T =
272     \dot{\epsilon}_{11}-\dot{\epsilon}_{22}$ and $D_S =
273 dimitri 1.13 2\dot{\epsilon}_{12}$, respectively, and using the above abbreviations,
274 dimitri 1.1 the equations can be written as:
275     \begin{align}
276     \label{eq:evpstresstensor1}
277     \frac{\partial\sigma_{1}}{\partial{t}} + \frac{\sigma_{1}}{2T} +
278     \frac{P}{2T} &= \frac{P}{2T\Delta} D_D \\
279     \label{eq:evpstresstensor2}
280     \frac{\partial\sigma_{2}}{\partial{t}} + \frac{\sigma_{2} e^{2}}{2T}
281     &= \frac{P}{2T\Delta} D_T \\
282     \label{eq:evpstresstensor12}
283     \frac{\partial\sigma_{12}}{\partial{t}} + \frac{\sigma_{12} e^{2}}{2T}
284     &= \frac{P}{4T\Delta} D_S
285     \end{align}
286     Here, the elastic parameter $E$ is redefined in terms of a damping timescale
287     $T$ for elastic waves \[E=\frac{\zeta}{T}.\]
288     $T=E_{0}\Delta{t}$ with the tunable parameter $E_0<1$ and
289     the external (long) timestep $\Delta{t}$. \citet{hunke97} recommend
290     $E_{0} = \frac{1}{3}$.
291    
292     For details of the spatial discretization, the reader is referred to
293     \citet{zhang98, zhang03}. Our discretization differs only (but
294     importantly) in the underlying grid, namely the Arakawa C-grid, but is
295 mlosch 1.15 otherwise straightforward. The EVP model, in particular, is discretized
296 dimitri 1.1 naturally on the C-grid with $\sigma_{1}$ and $\sigma_{2}$ on the
297     center points and $\sigma_{12}$ on the corner (or vorticity) points of
298     the grid. With this choice all derivatives are discretized as central
299     differences and averaging is only involved in computing $\Delta$ and
300     $P$ at vorticity points.
301    
302     For a general curvilinear grid, one needs in principle to take metric
303 mlosch 1.15 terms into account that arise in the transformation of a curvilinear
304     grid on the sphere. For now, however, we can neglect these metric
305     terms because they are very small on the \ml{[modify following
306     section3:] cubed sphere grids used in this paper; in particular,
307     only near the edges of the cubed sphere grid, we expect them to be
308     non-zero, but these edges are at approximately 35\degS\ or 35\degN\
309     which are never covered by sea-ice in our simulations. Everywhere
310     else the coordinate system is locally nearly cartesian.} However, for
311     last-glacial-maximum or snowball-earth-like simulations the question
312     of metric terms needs to be reconsidered. Either, one includes these
313     terms as in \citet{zhang03}, or one finds a vector-invariant
314     formulation for the sea-ice internal stress term that does not require
315     any metric terms, as it is done in the ocean dynamics of the MITgcm
316     \citep{adcroft04:_cubed_sphere}.
317    
318     Lateral boundary conditions are naturally ``no-slip'' for B-grids, as
319     the tangential velocities points lie on the boundary. For C-grids, the
320     lateral boundary condition for tangential velocities is realized via
321     ``ghost points'', allowing alternatively no-slip or free-slip
322     conditions. In ocean models free-slip boundary conditions in
323     conjunction with piecewise-constant (``castellated'') coastlines have
324     been shown to reduce in effect to no-slip boundary conditions
325     \citep{adcroft98:_slippery_coast}; for sea-ice models the effects of
326     lateral boundary conditions have not yet been studied.
327 dimitri 1.1
328     Moving sea ice exerts a stress on the ocean which is the opposite of
329     the stress $\vtau_{ocean}$ in Eq.\refeq{momseaice}. This stess is
330     applied directly to the surface layer of the ocean model. An
331     alternative ocean stress formulation is given by \citet{hibler87}.
332     Rather than applying $\vtau_{ocean}$ directly, the stress is derived
333     from integrating over the ice thickness to the bottom of the oceanic
334     surface layer. In the resulting equation for the \emph{combined}
335     ocean-ice momentum, the interfacial stress cancels and the total
336     stress appears as the sum of windstress and divergence of internal ice
337     stresses: $\delta(z) (\vtau_{air} + \vek{F})/\rho_0$, \citep[see also
338     Eq.\,2 of][]{hibler87}. The disadvantage of this formulation is that
339     now the velocity in the surface layer of the ocean that is used to
340     advect tracers, is really an average over the ocean surface
341     velocity and the ice velocity leading to an inconsistency as the ice
342     temperature and salinity are different from the oceanic variables.
343    
344     Sea ice distributions are characterized by sharp gradients and edges.
345 mlosch 1.15 For this reason, we employ positive, multidimensional 2nd and 3rd-order
346     advection scheme with flux limiter \citep{roe85, hundsdorfer94} for the
347     thermodynamic variables discussed in the next section.
348 dimitri 1.1
349     \subparagraph{boundary conditions: no-slip, free-slip, half-slip}
350    
351     \begin{itemize}
352     \item transition from existing B-Grid to C-Grid
353     \item boundary conditions: no-slip, free-slip, half-slip
354     \item fancy (multi dimensional) advection schemes
355     \item VP vs.\ EVP \citep{hunke97}
356     \item ocean stress formulation \citep{hibler87}
357     \end{itemize}
358    
359     \subsection{Thermodynamics}
360     \label{sec:thermodynamics}
361    
362     In the original formulation the sea ice model \citep{menemenlis05}
363     uses simple thermodynamics following the appendix of
364     \citet{semtner76}. This formulation does not allow storage of heat
365     (heat capacity of ice is zero, and this type of model is often refered
366     to as a ``zero-layer'' model). Upward heat flux is parameterized
367     assuming a linear temperature profile and together with a constant ice
368     conductivity. It is expressed as $(K/h)(T_{w}-T_{0})$, where $K$ is
369     the ice conductivity, $h$ the ice thickness, and $T_{w}-T_{0}$ the
370     difference between water and ice surface temperatures. The surface
371     heat budget is computed in a similar way to that of
372     \citet{parkinson79} and \citet{manabe79}.
373    
374     There is considerable doubt about the reliability of such a simple
375     thermodynamic model---\citet{semtner84} found significant errors in
376     phase (one month lead) and amplitude ($\approx$50\%\,overestimate) in
377     such models---, so that today many sea ice models employ more complex
378     thermodynamics. A popular thermodynamics model of \citet{winton00} is
379     based on the 3-layer model of \citet{semtner76} and treats brine
380     content by means of enthalphy conservation. This model requires in
381     addition to ice-thickness and compactness (fractional area) additional
382     state variables to be advected by ice velocities, namely enthalphy of
383     the two ice layers and the thickness of the overlying snow layer.
384 mlosch 1.15 \ml{[Jean-Michel, your turf: ]Care must be taken in advecting these
385     quantities in order to ensure conservation of enthalphy. Currently
386     this can only be accomplished with a 2nd-order advection scheme with
387     flux limiter \citep{roe85}.}
388 dimitri 1.1
389 mlosch 1.9
390 dimitri 1.1 \subsection{C-grid}
391     \begin{itemize}
392     \item no-slip vs. free-slip for both lsr and evp;
393     "diagnostics" to look at and use for comparison
394     \begin{itemize}
395     \item ice transport through Fram Strait/Denmark Strait/Davis
396     Strait/Bering strait (these are general)
397     \item ice transport through narrow passages, e.g.\ Nares-Strait
398     \end{itemize}
399     \item compare different advection schemes (if lsr turns out to be more
400     effective, then with lsr otherwise I prefer evp), eg.
401     \begin{itemize}
402     \item default 2nd-order with diff1=0.002
403     \item 1st-order upwind with diff1=0.
404     \item DST3FL (SEAICEadvScheme=33 with diff1=0., should work, works for me)
405     \item 2nd-order wit flux limiter (SEAICEadvScheme=77 with diff1=0.)
406     \end{itemize}
407     That should be enough. Here, total ice mass and location of ice edge
408     is interesting. However, this comparison can be done in an idealized
409     domain, may not require full Arctic Domain?
410     \item
411     Do a little study on the parameters of LSR and EVP
412     \begin{enumerate}
413     \item convergence of LSR, how many iterations do you need to get a
414     true elliptic yield curve
415     \item vary deltaTevp and the relaxation parameter for EVP and see when
416     the EVP solution breaks down (relative to the forcing time scale).
417     For this, it is essential that the evp solver gives use "stripeless"
418     solutions, that is your dtevp = 1sec solutions/or 10sec solutions
419     with SEAICE\_evpDampC = 615.
420     \end{enumerate}
421     \end{itemize}
422    
423     \section{Forward sensitivity experiments}
424     \label{sec:forward}
425    
426     A second series of forward sensitivity experiments have been carried out on an
427     Arctic Ocean domain with open boundaries. Once again the objective is to
428     compare the old B-grid LSR dynamic solver with the new C-grid LSR and EVP
429     solvers. One additional experiment is carried out to illustrate the
430     differences between the two main options for sea ice thermodynamics in the MITgcm.
431    
432     \subsection{Arctic Domain with Open Boundaries}
433     \label{sec:arctic}
434    
435 dimitri 1.12 The Arctic domain of integration is illustrated in Fig.~\ref{fig:arctic1}. It
436 mlosch 1.6 is carved out from, and obtains open boundary conditions from, the
437     global cubed-sphere configuration of the Estimating the Circulation
438     and Climate of the Ocean, Phase II (ECCO2) project
439     \citet{menemenlis05}. The domain size is 420 by 384 grid boxes
440     horizontally with mean horizontal grid spacing of 18 km.
441 dimitri 1.1
442 dimitri 1.12 \begin{figure}
443     %\centerline{{\includegraphics*[width=0.44\linewidth]{\fpath/arctic1.eps}}}
444     \caption{Bathymetry of Arctic Domain.\label{fig:arctic1}}
445     \end{figure}
446    
447 dimitri 1.1 There are 50 vertical levels ranging in thickness from 10 m near the surface
448     to approximately 450 m at a maximum model depth of 6150 m. Bathymetry is from
449     the National Geophysical Data Center (NGDC) 2-minute gridded global relief
450     data (ETOPO2) and the model employs the partial-cell formulation of
451 mlosch 1.6 \citet{adcroft97:_shaved_cells}, which permits accurate representation of the bathymetry. The
452 dimitri 1.1 model is integrated in a volume-conserving configuration using a finite volume
453     discretization with C-grid staggering of the prognostic variables. In the
454 mlosch 1.6 ocean, the non-linear equation of state of \citet{jackett95}. The ocean model is
455 dimitri 1.1 coupled to a sea-ice model described hereinabove.
456    
457 mlosch 1.6 This particular ECCO2 simulation is initialized from rest using the
458     January temperature and salinity distribution from the World Ocean
459     Atlas 2001 (WOA01) [Conkright et al., 2002] and it is integrated for
460     32 years prior to the 1996--2001 period discussed in the study. Surface
461     boundary conditions are from the National Centers for Environmental
462     Prediction and the National Center for Atmospheric Research
463     (NCEP/NCAR) atmospheric reanalysis [Kistler et al., 2001]. Six-hourly
464     surface winds, temperature, humidity, downward short- and long-wave
465     radiations, and precipitation are converted to heat, freshwater, and
466     wind stress fluxes using the \citet{large81, large82} bulk formulae.
467     Shortwave radiation decays exponentially as per Paulson and Simpson
468     [1977]. Additionally the time-mean river run-off from Large and Nurser
469     [2001] is applied and there is a relaxation to the monthly-mean
470     climatological sea surface salinity values from WOA01 with a
471     relaxation time scale of 3 months. Vertical mixing follows
472     \citet{large94} with background vertical diffusivity of
473     $1.5\times10^{-5}\text{\,m$^{2}$\,s$^{-1}$}$ and viscosity of
474     $10^{-3}\text{\,m$^{2}$\,s$^{-1}$}$. A third order, direct-space-time
475     advection scheme with flux limiter is employed \citep{hundsdorfer94}
476     and there is no explicit horizontal diffusivity. Horizontal viscosity
477     follows \citet{lei96} but
478     modified to sense the divergent flow as per Fox-Kemper and Menemenlis
479     [in press]. Shortwave radiation decays exponentially as per Paulson
480     and Simpson [1977]. Additionally, the time-mean runoff of Large and
481     Nurser [2001] is applied near the coastline and, where there is open
482     water, there is a relaxation to monthly-mean WOA01 sea surface
483     salinity with a time constant of 45 days.
484 dimitri 1.1
485     Open water, dry
486     ice, wet ice, dry snow, and wet snow albedo are, respectively, 0.15, 0.85,
487     0.76, 0.94, and 0.8.
488    
489     \begin{itemize}
490     \item Configuration
491     \item OBCS from cube
492     \item forcing
493     \item 1/2 and full resolution
494     \item with a few JFM figs from C-grid LSR no slip
495     ice transport through Canadian Archipelago
496     thickness distribution
497     ice velocity and transport
498     \end{itemize}
499    
500     \begin{itemize}
501     \item Arctic configuration
502     \item ice transport through straits and near boundaries
503     \item focus on narrow straits in the Canadian Archipelago
504     \end{itemize}
505    
506     \begin{itemize}
507     \item B-grid LSR no-slip
508     \item C-grid LSR no-slip
509     \item C-grid LSR slip
510     \item C-grid EVP no-slip
511     \item C-grid EVP slip
512 mlosch 1.6 \item C-grid LSR + TEM (truncated ellipse method, no tensile stress, new flag)
513 dimitri 1.1 \item C-grid LSR no-slip + Winton
514     \item speed-performance-accuracy (small)
515     ice transport through Canadian Archipelago differences
516     thickness distribution differences
517     ice velocity and transport differences
518     \end{itemize}
519    
520     We anticipate small differences between the different models due to:
521     \begin{itemize}
522     \item advection schemes: along the ice-edge and regions with large
523     gradients
524 mlosch 1.6 \item C-grid: less transport through narrow straits for no slip
525     conditons, more for free slip
526 dimitri 1.1 \item VP vs.\ EVP: speed performance, accuracy?
527     \item ocean stress: different water mass properties beneath the ice
528     \end{itemize}
529    
530     \section{Adjoint sensiivities of the MITsim}
531    
532     \subsection{The adjoint of MITsim}
533    
534     The ability to generate tangent linear and adjoint model components
535     of the MITsim has been a main design task.
536     For the ocean the adjoint capability has proven to be an
537     invaluable tool for sensitivity analysis as well as state estimation.
538     In short, the adjoint enables very efficient computation of the gradient
539     of scalar-valued model diagnostics (called cost function or objective function)
540     with respect to many model "variables".
541     These variables can be two- or three-dimensional fields of initial
542     conditions, model parameters such as mixing coefficients, or
543     time-varying surface or lateral (open) boundary conditions.
544     When combined, these variables span a potentially high-dimensional
545     (e.g. O(10$^8$)) so-called control space. Performing parameter perturbations
546     to assess model sensitivities quickly becomes prohibitive at these scales.
547     Alternatively, (time-varying) sensitivities of the objective function
548     to any element of the control space can be computed very efficiently in
549     one single adjoint
550     model integration, provided an efficient adjoint model is available.
551    
552     [REFERENCES]
553    
554    
555     The adjoint operator (ADM) is the transpose of the tangent linear operator (TLM)
556     of the full (in general nonlinear) forward model, i.e. the MITsim.
557     The TLM maps perturbations of elements of the control space
558     (e.g. initial ice thickness distribution)
559     via the model Jacobian
560     to a perturbation in the objective function
561     (e.g. sea-ice export at the end of the integration interval).
562     \textit{Tangent} linearity ensures that the derivatives are evaluated
563     with respect to the underlying model trajectory at each point in time.
564     This is crucial for nonlinear trajectories and the presence of different
565     regimes (e.g. effect of the seaice growth term at or away from the
566     freezing point of the ocean surface).
567     Ensuring tangent linearity can be easily achieved by integrating
568     the full model in sync with the TLM to provide the underlying model state.
569     Ensuring \textit{tangent} adjoints is equally crucial, but much more
570     difficult to achieve because of the reverse nature of the integration:
571     the adjoint accumulates sensitivities backward in time,
572     starting from a unit perturbation of the objective function.
573     The adjoint model requires the model state in reverse order.
574     This presents one of the major complications in deriving an
575     exact, i.e. \textit{tangent} adjoint model.
576    
577     Following closely the development and maintenance of TLM and ADM
578     components of the MITgcm we have relied heavily on the
579     autmomatic differentiation (AD) tool
580     "Transformation of Algorithms in Fortran" (TAF)
581     developed by Fastopt (Giering and Kaminski, 1998)
582     to derive TLM and ADM code of the MITsim.
583     Briefly, the nonlinear parent model is fed to the AD tool which produces
584     derivative code for the specified control space and objective function.
585     Following this approach has (apart from its evident success)
586     several advantages:
587     (1) the adjoint model is the exact adjoint operator of the parent model,
588     (2) the adjoint model can be kept up to date with respect to ongoing
589     development of the parent model, and adjustments to the parent model
590     to extend the automatically generated adjoint are incremental changes
591     only, rather than extensive re-developments,
592     (3) the parallel structure of the parent model is preserved
593     by the adjoint model, ensuring efficient use in high performance
594     computing environments.
595    
596     Some initial code adjustments are required to support dependency analysis
597     of the flow reversal and certain language limitations which may lead
598     to irreducible flow graphs (e.g. GOTO statements).
599     The problem of providing the required model state in reverse order
600     at the time of evaluating nonlinear or conditional
601     derivatives is solved via balancing
602     storing vs. recomputation of the model state in a multi-level
603     checkpointing loop.
604     Again, an initial code adjustment is required to support TAFs
605     checkpointing capability.
606 mlosch 1.6 The code adjustments are sufficiently simple so as not to cause
607 dimitri 1.1 major limitations to the full nonlinear parent model.
608     Once in place, an adjoint model of a new model configuration
609     may be derived in about 10 minutes.
610    
611     [HIGHLIGHT COUPLED NATURE OF THE ADJOINT!]
612    
613     \subsection{Special considerations}
614    
615     * growth term(?)
616    
617     * small active denominators
618    
619     * dynamic solver (implicit function theorem)
620    
621     * approximate adjoints
622    
623    
624     \subsection{An example: sensitivities of sea-ice export through Fram Strait}
625    
626     We demonstrate the power of the adjoint method
627     in the context of investigating sea-ice export sensitivities through Fram Strait
628     (for details of this study see Heimbach et al., 2007).
629 mlosch 1.6 %\citep[for details of this study see][]{heimbach07}. %Heimbach et al., 2007).
630 dimitri 1.1 The domain chosen is a coarsened version of the Arctic face of the
631     high-resolution cubed-sphere configuration of the ECCO2 project
632 mlosch 1.6 \citep[see][]{menemenlis05}. It covers the entire Arctic,
633 dimitri 1.1 extends into the North Pacific such as to cover the entire
634     ice-covered regions, and comprises parts of the North Atlantic
635     down to XXN to enable analysis of remote influences of the
636     North Atlantic current to sea-ice variability and export.
637     The horizontal resolution varies between XX and YY km
638     with 50 unevenly spaced vertical levels.
639     The adjoint models run efficiently on 80 processors
640     (benchmarks have been performed both on an SGI Altix as well as an
641     IBM SP5 at NASA/ARC).
642    
643 mlosch 1.6 Following a 1-year spinup, the model has been integrated for four
644     years between 1992 and 1995. It is forced using realistic 6-hourly
645     NCEP/NCAR atmospheric state variables. Over the open ocean these are
646     converted into air-sea fluxes via the bulk formulae of
647     \citet{large04}. Derivation of air-sea fluxes in the presence of
648     sea-ice is handled by the ice model as described in \refsec{model}.
649 dimitri 1.1 The objective function chosen is sea-ice export through Fram Strait
650 mlosch 1.6 computed for December 1995. The adjoint model computes sensitivities
651     to sea-ice export back in time from 1995 to 1992 along this
652     trajectory. In principle all adjoint model variable (i.e., Lagrange
653     multipliers) of the coupled ocean/sea-ice model are available to
654     analyze the transient sensitivity behaviour of the ocean and sea-ice
655     state. Over the open ocean, the adjoint of the bulk formula scheme
656     computes sensitivities to the time-varying atmospheric state. Over
657     ice-covered parts, the sea-ice adjoint converts surface ocean
658     sensitivities to atmospheric sensitivities.
659    
660     \reffig{4yradjheff}(a--d) depict sensitivities of sea-ice export
661     through Fram Strait in December 1995 to changes in sea-ice thickness
662     12, 24, 36, 48 months back in time. Corresponding sensitivities to
663     ocean surface temperature are depicted in
664     \reffig{4yradjthetalev1}(a--d). The main characteristics is
665     consistency with expected advection of sea-ice over the relevant time
666     scales considered. The general positive pattern means that an
667     increase in sea-ice thickness at location $(x,y)$ and time $t$ will
668     increase sea-ice export through Fram Strait at time $T_e$. Largest
669     distances from Fram Strait indicate fastest sea-ice advection over the
670     time span considered. The ice thickness sensitivities are in close
671     correspondence to ocean surface sentivitites, but of opposite sign.
672     An increase in temperature will incur ice melting, decrease in ice
673     thickness, and therefore decrease in sea-ice export at time $T_e$.
674 dimitri 1.1
675     The picture is fundamentally different and much more complex
676     for sensitivities to ocean temperatures away from the surface.
677 mlosch 1.6 \reffig{4yradjthetalev10??}(a--d) depicts ice export sensitivities to
678 dimitri 1.1 temperatures at roughly 400 m depth.
679     Primary features are the effect of the heat transport of the North
680     Atlantic current which feeds into the West Spitsbergen current,
681     the circulation around Svalbard, and ...
682    
683     \begin{figure}[t!]
684     \centerline{
685     \subfigure[{\footnotesize -12 months}]
686 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim072_cmax2.0E+02.eps}}
687 dimitri 1.1 %\includegraphics*[width=.3\textwidth]{H_c.bin_res_100_lev1.pdf}
688     %
689     \subfigure[{\footnotesize -24 months}]
690 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim145_cmax2.0E+02.eps}}
691 dimitri 1.1 }
692    
693     \centerline{
694     \subfigure[{\footnotesize
695     -36 months}]
696 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim218_cmax2.0E+02.eps}}
697 dimitri 1.1 %
698     \subfigure[{\footnotesize
699     -48 months}]
700 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJheff_arc_lev1_tim292_cmax2.0E+02.eps}}
701 dimitri 1.1 }
702     \caption{Sensitivity of sea-ice export through Fram Strait in December 2005 to
703     sea-ice thickness at various prior times.
704     \label{fig:4yradjheff}}
705     \end{figure}
706    
707    
708     \begin{figure}[t!]
709     \centerline{
710     \subfigure[{\footnotesize -12 months}]
711 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim072_cmax5.0E+01.eps}}
712 dimitri 1.1 %\includegraphics*[width=.3\textwidth]{H_c.bin_res_100_lev1.pdf}
713     %
714     \subfigure[{\footnotesize -24 months}]
715 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim145_cmax5.0E+01.eps}}
716 dimitri 1.1 }
717    
718     \centerline{
719     \subfigure[{\footnotesize
720     -36 months}]
721 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim218_cmax5.0E+01.eps}}
722 dimitri 1.1 %
723     \subfigure[{\footnotesize
724     -48 months}]
725 mlosch 1.4 {\includegraphics*[width=0.44\linewidth]{\fpath/run_4yr_ADJtheta_arc_lev1_tim292_cmax5.0E+01.eps}}
726 dimitri 1.1 }
727 mlosch 1.6 \caption{Same as \reffig{4yradjheff} but for sea surface temperature
728 dimitri 1.1 \label{fig:4yradjthetalev1}}
729     \end{figure}
730    
731    
732    
733     \section{Discussion and conclusion}
734     \label{sec:concl}
735    
736     The story of the paper could be:
737     B-grid ice model + C-grid ocean model does not work properly for us,
738     therefore C-grid ice model with advantages:
739     \begin{enumerate}
740     \item stress coupling simpler (no interpolation required)
741     \item different boundary conditions
742     \item advection schemes carry over trivially from main code
743     \end{enumerate}
744     LSR/EVP solutions are similar with appropriate bcs, evp parameters as
745     a function of forcing time scale (when does VP solution break
746     down). Same for LSR solver, provided that it works (o:
747     Which scheme is more efficient for the resolution/time stepping
748     parameters that we use here. What about other resolutions?
749    
750     \paragraph{Acknowledgements}
751     We thank Jinlun Zhang for providing the original B-grid code and many
752 mlosch 1.6 helpful discussions. ML thanks Elizabeth Hunke for multiple explanations.
753 dimitri 1.1
754 dimitri 1.7 \bibliography{bib/journal_abrvs,bib/seaice,bib/genocean,bib/maths,bib/mitgcmuv,bib/fram}
755 dimitri 1.1
756     \end{document}
757    
758     %%% Local Variables:
759     %%% mode: latex
760     %%% TeX-master: t
761     %%% End:
762 mlosch 1.4
763    
764     A Dynamic-Thermodynamic Sea ice Model for Ocean Climate
765     Estimation on an Arakawa C-Grid
766    
767     Introduction
768    
769     Ice Model:
770     Dynamics formulation.
771     B-C, LSR, EVP, no-slip, slip
772     parallellization
773     Thermodynamics formulation.
774     0-layer Hibler salinity + snow
775     3-layer Winton
776    
777     Idealized tests
778     Funnel Experiments
779     Downstream Island tests
780     B-grid LSR no-slip
781     C-grid LSR no-slip
782     C-grid LSR slip
783     C-grid EVP no-slip
784     C-grid EVP slip
785    
786     Arctic Setup
787     Configuration
788     OBCS from cube
789     forcing
790     1/2 and full resolution
791     with a few JFM figs from C-grid LSR no slip
792     ice transport through Canadian Archipelago
793     thickness distribution
794     ice velocity and transport
795    
796     Arctic forward sensitivity experiments
797     B-grid LSR no-slip
798     C-grid LSR no-slip
799     C-grid LSR slip
800     C-grid EVP no-slip
801     C-grid EVP slip
802     C-grid LSR no-slip + Winton
803     speed-performance-accuracy (small)
804     ice transport through Canadian Archipelago differences
805     thickness distribution differences
806     ice velocity and transport differences
807    
808     Adjoint sensitivity experiment on 1/2-res setup
809     Sensitivity of sea ice volume flow through Fram Strait
810     *** Sensitivity of sea ice volume flow through Canadian Archipelago
811    
812     Summary and conluding remarks

  ViewVC Help
Powered by ViewVC 1.1.22