General Earthquake Models:

Numerical Laboratories for Understanding

The Physics of Earthquakes

Southern California Earthquake Center (SCEC)

California Earthquake Research Center (CERC)

Colorado Center for Chaos & Complexity

University of Colorado, Boulder, CO 80309

IGPP, SIO

Unversity of California, San Diego 92093

A Consensus Proposal from the

Members:

Yehuda Ben-Zion, University of Southern California

Jacobo Bialek, Carnegie Mellon University

William Bosl, Stanford University

David Bowman, University of Southern California

Charles Carrigan, Lawrence Livermore National Laboratory, Livermore, CA

James Crutchfield, Santa Fe Institute, Santa Fe, NM

Julian Cummings, ACL, Los Alamos National Laboratory, Los Alamos, NM

Steven Day, San Diego State University

Geoffrey Fox, NPAC, Syracuse University, Syracuse, NY

William Foxall, Lawrence Livermore National Laboratory, Livermore, CA

Roscoe Giles, Boston University, Boston, MA

Rajan Gupta, ACL, Los Alamos National Laboratory

Tom Henyey, SCEC and University of Southern California

Thomas H. Jordan, Massachusetts Institute of Technology, Cambridge, MA

Hiroo Kanamori, California Institute of Technology, Pasadena, CA

Steven Karmesin, ACL, Los Alamos National Laboratory, Los Alamos, NM

Charles Keller, IGPP, Los Alamos National Laboratory

William Klein, Boston University, Boston, MA

Karen Carter Krogh, EES, Los Alamos National Laboratory, NM

Shawn Larsen, Lawrence Livermore National Laboratory, Livermore, CA

Christopher J. Marone, Massachusetts Institute of Technology, Cambridge, MA

John McRaney, SCEC and University of Southern California

Paul Messina, CACR, California Institute of Technology, Pasadena, CA

J.-Bernard Minster, University of California, San Diego, CA

David O'Halloran, Carnegie Mellon University

Lawrence Hutchings, Lawrence Livermore National Laboratory, Livermore, CA

Jon Pelletier, California Institute of Technology

John Reynders, ACL, Los Alamos National Laboratory, Los Alamos, NM

John B. Rundle, University of Colorado, Boulder, CO

John Salmon, CACR, California Institute of Technology, Pasadena, CA

Charles Sammis, University of Southern California

Steven Shkoller, CNLS Los Alamos National Laboratory

Stewart Smith, University of Washington

Ross Stein, United States Geological Survey, Menlo Park, CA

Leon Teng, University of Southern California

Donald Turcotte, Cornell University, Ithaca, NY

Michael Warren, ACL, Los Alamos National Laboratory, Los Alamos, NM

Andrew White, ACL. Los Alamos National Laboratory

Bryant York, Northeastern University, Boston, MA

1. What are they?

2. What is the purpose?

1. How is present m < 3 seismicity in the Mojave related to the possibility of m ~ 6 events in Palos Verdes?
2. Are there any detectable signatures of impending moderate-to-large earthquakes in crustal deformation or seismicity data? etc.

3. How would the process work?

1. advances in theory;
2. advances in hardware and software;
3. specific laboratory data to fix parameters in simulations;
4. specific field data to validate the simulations, to provide hypotheses to test, and to provide initialization data for possible "predictions" of various kinds (as opposed to "EARTHQUAKE PREDICTIONS")

1. Earthquakes occur on time and spatial scales that are prohibitive to many direct observations
2. Impossible to know state of stress, deformation, pore pressure, etc. on natural faults everywhere and for all times
3. Need for carrying out some experiments that may be too dangerous to ever test in the field, e.g., can water pumped down a fault release stress in a series of small events rather than one big one?
4. System is known to be fundamentally nonlinear -- thus has some kind of dynamical attractor -- question is, what is the geometry and nature of this attractor

1. Earthquake forecasting or prediction methodologies cannot be used without at first testing, evaluating, and "scoring" them
2. How to "initialize" an NEF model with prior data?
3. What is the role of unmodeled, sub-grid scale processes?

1. Provides one means of "attaching priorities" to the various observational, laboratory, numerical, and theoretical tasks

2. A focus and an organizing principle for earthquake science: 

• Unexpected new observations can be interrelated and explained using simulations
• Introducing new physics in simulations can be used to develop hypotheses to test with observations

3. GEM's are fundamentally an earthquake science activity, the results of which represent necessary inputs to engineering hazard calculations

4. GEM's are timely, in the sense that:

5. Interactions with parallel efforts in Japan (CAMP) and Australia (ACES) will be enabled by GEM-type approach

6. Advances in earthquake science will now be coupled to advances in computational science, enabling access to a broader array of scientific talent and resources

1. Simulations can only reproduce previously "known" results, i.e., "You can only get out what you put in..."

Counterexamples:

1. Lorenz equations and deterministic chaos, first discovered in the computer in 1963, later verified in atmospheric observations, wind shear experiments, electronic circuits, orbits of planets, etc.
2. Discovery of period doubling route to chaos, discovered in the computer, later observed in laboratory experiments
3. Boeing 777 designed entirely by simulations (no wind tunnel or field experiments in initial design)
4. See for example, recent issues of Physical Review, books (Casti)

2. "GCM's are fundamentally unlike the earthquake problem, since they have Navier-Stokes equations" 

Ah, yes,but they DON'T have, for example:

1. A viable model for clouds (warming or cooling?)
2. Adequate models for air-sea, or land-atmosphere interactions (how much vapor exchanged?)
3. Adequate models for ocean currents, surface winds, solar fluctuations, dust particles, etc.
4. Adequate constraints on the history of the atmosphere

While we DO have:

1. Interactions via elasticity & viscoelasticity
2. A (small) number of candidate friction laws
3. Substantial data on the configuration & history of faulting, etc.

SCEC and "CERC"

Two Basic Possibilities:

GEM to be a subgroup, or "Working Group" within Center, similar to role of SCIGN within SCEC

GEM to be integrated into the core of the Center, as one of the "alphabet soup" groups, similar to Seismology Group, Deformation Group, Master Model Group, etc.

1. How Center is administered: Managerial structure, use of matrix managment methods, etc.
2. Whether computational activities are seen to be "centered" in Southern California, Northern California, or both
3. Relationship of non-California institutions (core institutions or not?) to the California universities
   Note: This topic needs discussion, because computational facilities and personnel will in any case be distributed all over the country and the world

1: "Lead Investigator" or "Group Chair" or "Coordinator" or whatever...

2: "Working Group", or "Steering Committee"

3: "Activities", or "Task Forces"

Short Term Testbed Problem

Our workshop identified a focus problem: Quasi-static evolutions of stress field - California

1: Model:

2: Inputs:

3: Outputs:

4: Relevance:

Generality vs. Specificity

Statistical Physics vs. Deterministic

• Recognize that much "data" will always be unobservable, and we must evolve ways to deal with this problem
• Focus on space-time patterns & correlations...essentials of the nonlinear dynamics of complex high dimensional systems
• "Ensemble" predictions vs "Deterministic" predictions

Integrate Models & Simulations with Data Collection

• Two activities are not separate enterprises
• Data most useful with models, models must be validated by data

Statewide Director & PI, plan for rotating director

One Institution should be "Managing Institution" through which money flows...USC?

• Director to have a temporary faculty appointment at MI if not permanent faculty
• Define specific role and influence of MI
• "Center Manager" to be located at MI, reports to Director

Northern & Southern "Coordinators"

Build carefully on success of SCEC, don't expand too much, too fast!

New name to symbolically differentiate New Center from Old

Less frequent topical seminars:

• Every ~ 3 - 6 months, instead of monthly
• Pick nice locations for meetings

Projects should not be focussed on field sites, but on scientific problems...project teams should include personnel from both north & south