Current State of the Community Ice Sheet Model
Stephen Price1, Bill Lipscomb1, Jesse Johnson2
Tony Payne3, Ian Rutt4, Magnus Hagdorn5
1COSIM
Project, Fluid Dynamics and Solid Mechanics Group, Los Alamos National Laboratory
2Department
3Bristol
of Computer Sciences, University of Montana
Glaciology Centre, School of Geographical Sciences, University of Bristol
4School
of the Environment and Society, Swansea University
5School
of GeoSciences, University of Edinburgh
Climate, Ocean, and Sea Ice Modeling Project
Motivation for improved Ice Sheet Models
Current models do not capture observed ice sheet behaviour1, because:
(1) Fundamental physics are missing (e.g. simplified gov. equations, negating
realistic simulation of outlet glaciers and ice streams)
(2) Processes of fundamental importance are not accounted for (e.g. simplified,
static treatment of basal boundary conditions, ignoring atmos. and ocean
coupling, explicit hydrology, etc.)
Accurate predictions of future sea-level rise will require advanced models which can
demonstrate skill at reproducing and explaining recent dramatic ice sheet
behaviors.
1 IPCC
(2007)
Summary of Glimmer-CISM and development plans
Model validation (higher-order dynamics)
Tuning for initial conditions (Greenland)
Summary
Illulisat, Western Greenland
Summary of Glimmer-CISM and development plans
Model validation (higher-order dynamics)
Tuning for initial conditions (Greenland)
Summary
Illulisat, Western Greenland
Glimmer-CISM
•
The Glimmer ice sheet model (Rutt et al. 2009) was developed by Tony
Payne et al. in the U.K. with the goal of coupling dynamic ice sheets to global
climate models.
•
NSF has supported development of a Community Ice Sheet Model (CISM)
based on Glimmer.
– Jesse Johnson’s wiki: http://websrv.cs.umt.edu/isis
•
DOE has supported coupling of Glimmer to the Community Climate System
Model (CCSM).
– CCSM Land ice working group (meeting Feb. 2010 in Boulder):
http://www.ccsm.ucar.edu/working_groups/Land+Ice/
•
The U.S. and U.K. groups have recently combined efforts:
– BerliOS repository: http://developer.berlios.de/projects/glimmer-cism/
– 6-member steering committee (M. Hagdorn, J. Johnson, W. Lipscomb, T.
Payne, S. Price, I. Rutt)
Ice sheets in the Community Climate
System Model (CCSM)
•
Glimmer has been coupled to CCSM version 4 and will be used for IPCC
runs with a dynamic Greenland ice sheet.
•
The surface mass balance of ice sheets is computed by the land surface
model on a coarse grid (~100 km) in multiple elevation classes, passed to
Glimmer via the coupler, and downscaled to the ice sheet grid (~10 km).
Atmosphere
Land surface
(Ice sheet surface
mass balance)
Coupler
Ice sheet
(Dynamics)
Ocean
Sea Ice
DOE IMPACTS
IMPACTS - Investigation of the Magnitudes and Probabilities of Abrupt Climate
Transitions
•
•
5-year program on abrupt climate change
One part of the project (LANL and NYU) focuses on the potential instability of
the West Antarctic ice sheet due to ice/ocean interactions
– Couple Glimmer-CISM to the HYPOP ocean model on regional scales
– Model ocean circulation beneath dynamic ice shelves
DOE ISICLES
ISICLES - Ice Sheet Initiative for CLimate at Extreme Scales
•
•
•
New 3-year program to develop advanced ice sheet models using efficient,
scalable computational methods
Dynamical cores to be incorporated as options in Glimmer-CISM
Six projects with similar goals, different tools:
- Matrix solvers and preconditioners (e.g., PETSc, Trilinos)
- Adaptive mesh refinement (e.g., Chombo)
- Unstructured grids (e.g., Voronoi meshes)
Summary of Glimmer-CISM and development plans
Model validation (higher-order dynamics)
Tuning for initial conditions (Greenland)
Summary
Illulisat, Western Greenland
ice stream with plastic till
model vel (m/yr)
observed vel (m/yr)
Model
2
Vmax (m/a)
Bremerhavn1
3605
1379
Bremerhavn2
12518
1663
Chicago1
5114
1497
Chicago2
5125
1497
Grenoble
5237
1508
U. Mont.
4962
1495
LANL / UoB
5458
1543
Summary of Glimmer-CISM and development plans
Model validation (higher-order dynamics)
Tuning for initial conditions (Greenland)
Summary
field camp, central Greenland
Tuning Procedure
- Skipping the details, we have a simple procedure that allows us
to “tune” basal sliding parameters to match a target velocity field
- The resulting ice velocity, viscosity, and temperature fields are in
a steady-state with modern day geometry, temperature, and
geothermal flux fields (as well as those are known)
- The result should be useful as an initial condition for a range of
perturbation studies
How do model and target1 velocity fields compare?
1Here,
GIS balance velocities from Bamber et al. (J.Glac., 46, 2000)
Summary
• We have a 3D ice sheet model (Glimmer-CISM) that can simulate the
flow of inland ice, outlet glaciers and ice streams, and ice shelves
using either of two higher-order flow schemes.
• The model is nearly ready for coupled climate experiments using
CCSM. We plan to do experiments with dynamic, higher-order ice
sheets in time to contribute to IPCC AR5.
• The model is evolving rapidly. Within a few years, scalable solvers
and adaptive grids could make high-resolution, whole-ice-sheet
simulations routine.
• We welcome collaborators.
EXTRA
Model framework
Proposed CCSM4 experiments with GLIMMER
(0.9o x 1.25o atm, 1o ocn)
1. Control
• Pre-industrial control, 230+ yrs
• Pre-industrial control, 0.5o, ~100
yrs
• 20th century (1870-2005)
2. IPCC AR5 scenarios
• RCP4.5, 100-300 yrs
• RCP8.5, 100-300 yrs
3. Long-term (asynchronous)
• Continuation of RCP4.5, 200 yrs
(AOGCM), 2000 yrs (ice sheet)
• Branch runs of RCP4.5 and/or
RCP8.5 (study irreversibility)
• Eemian interglacial: 1000 yr
AOGCM w/ 10x accelerated
Milankovich; 10,000 yr ice sheet
Miren Vizcaino (UC Berkeley) et al. will analyze these runs.
Governing Equations
Conservation of Momentum
- 1st-order SIA (e.g. Blatter-Pattyn) using FDM (current)
- 1st-order SIA (Dukowicz+) using FEM (under development)
- full Stokes (next 2-4 yrs?)
Conservation of Energy
- standard heat equation solved for ice sheets
(e.g. horiz diff neglected)
- scalable improvement planned
Conservation of Mass
- standard, nonlinear diffusion equation in H (0-order SIA)
- incremental remapping (1st-order SIA)
Equations of Stress Equilibrium in
Cartesian Coordinates (Stokes Flow)
Assume static balance of forces by ignoring acceleration
x:
y:
z:
 xx
 xy
 xz
 yy
 xy
 yz
 zz
 zy
 xz
P



0
x
y
z x
P



0
y
x
z
y
P



 g
z
y
x z
… apply some scaling based on H/L …
1  u w 
1  u 
&xz      &xz   
2  z x 
2  z 
 xz
x

 yz
y
0 
 zz
 zz
P


 g
z
z z
First-Order SIA
x:
 xx
z:
 zz
 xy
 xz
P



0
x
y
z x
P
 g 
z
z
… integrate w.r.t. z to get
explicit expression for P…
First Order SIA
…integrate in vertical from upper sfc through depth …
 zz P
z:

 g
z z
P   g  s  z    zz ( z )
First Order SIA
…substitute vertical relation for P into horizontal balance …
 xx  xy  xz P
x:



x
y
z
x
 xx  xy  xz 
x:


   g  s  z    zz ( z ) 
x
y
z
x
 xx  zz  xy  xz
s
x:



 g
x
x
y
z
x
First Order SIA
…use definition of deviatoric stress to eliminate vertical-normal
stress deviator in horiz. equations …
 xx  zz  xy  xz
s
x:



 g
x
x
y
z
x
 zz   xx   yy
  xx  yy  xy  xz
s
x: 2



 g
x
x
y
z
x