Los Alamos National Laboratories, Advanced Computing Laboratory

Los Alamos National Laboratories, Institute for Geophysics & Planetary Physics

US Department of Energy, Office of Basic Energy Sciences, Division of Geosciences

The Southern California Earthquake Center (A Science & Technology Center funded by the National Science Foundation and the US Geological Survey)

The Colorado Center for Chaos & Complexity

John Rundle, University of Colorado

J. Bernard Minster, University of California San Diego

Andrew White, Los Alamos National Laboratory, ACL

Charles Keller, Los Alamos National Laboratory, IGPP

Introduction: A workshop on "General Earthquake Models" (GEM's) was held at Santa Fe, NM, during October 23-25, 1997. The primary objective was to explore the possibility of developing the computational capability to carry out large scale numerical simulations of the physics of earthquakes in southern California and elsewhere. These simulations would be capable of producing detailed temporal and spatial patterns of earthquakes, surface deformation and gravity change, seismicity, stress, as well as, in principle, other variables, including pore fluid and thermal changes, for comparison to field and laboratory data. To construct the simulations, a state-of-the-art problem solving environment must at some point be developed that will facilitate: 1) Construction of numerical and computational algorithms and specific environment(s) needed to carry out large scale simulations of these nonlinear physical processes over a geographically distributed, heterogeneous computing network; and 2) Development of a testbed for earthquake "forecasting" & "prediction" methodologies which uses modern Object Oriented techniques and scalable systems, software and algorithms which are efficient for both people and computational execution time.

Summary of Events: On friday, after discussion of international programs initiated by Australia (Peter Mora: ACES) and Japan (Mitsuhiro Matsu'ura: CAMP), a series of talks were given by Andy White (LANL) and Geoffrey Fox (Syracuse) describing the hardware and software challenges that need to be overcome in order to construct a viable GEM model. After poster presentations and displays, two more talks were given by Charles Sammis (USC) and Steve Day (SDSU) on the scientific rationale for GEM simulations. Saturday's morning session featured a presentation by Tom Henyey (SCEC and USC) about the organization, accomplishments, and future of SCEC, followed by a discussion period. Results of the discussion are summarized at the end of this report. In brief, it was decided that a major GEM effort should be incorporated into SCEC as one of a few central foci in the new State of California Earthquake Center proposal to the new NSF Science and Technology Center competition. The GEM efforts complements the anticipated focus of the new Center on the physics of earthquakes, by providing a "numerical laboratory" for developing, testing, and refining hypotheses about the earthquake source that are only now emerging from a variety of sophisticated new data types. These data include TRINET, SCIGN, and ISAR facilities in Southern California. Moreover, the temporal intersection of large quantities of high quality data, new high performance computational capability, and software make the GEM project extremely timely. These points are discussed in more detail in the following.

Scientific and Computational Needs for GEM: Discussions were focussed on plans to develop the capability to carry out large scale simulations of complex, multiple, interacting fault systems using a software environment optimized for rapid prototyping of new phenomenological models. The software environment would require: 1) Developing algorithms for solving computationally difficult nonlinear problems involving ("discontinuous") thresholds and nucleation events in a networked parallel (super) computing environment, adapting new "fast multipole" methods previously developed for general N-body problems; 2) Leveraging the Los Alamos ACL Infrastructure, including the POOMA object oriented framework, which is already being applied to computationally related adaptive particle mesh problems; and 3) Developing a modern problem solving environment to allow researchers to rapidly integrate simulation data with field and laboratory data (visually and quantitatively). This task is becoming increasingly important with the exponential increase in seismic, as well as space-based deformation data.

Scientific Significance of GEM: A GEM approach would allow the physics of large networks of earthquake faults to be analyzed within a general theoretical framework for the first time. We believe that it would stimulate changes in earthquake science in much the same way that General Circulation Models of the atmosphere have changed the way atmospheric science is carried out. The computational techniques developed by the project would, as a by-product, have important applications in many other large, computationally difficult problems, such as 1) Large particle-in-cell systems in ASCI; 2) Statistical physics approaches to random field spin systems; 3) Simulating large neural networks with learning and cognition ; as well as 4) Modeling the stability and safety of nuclear waste repositories in fractured rock.

Participants: Scientists included:

Jacobo Bialek, Carnegie Mellon University

Eric Blough, University of Colorado

William Bosl, Stanford University

David Bowman, University of Southern California

Steven Day, San Diego State University

Geoffrey Fox, NPAC, Syracuse University

Rajan Gupta, ACL, Los Alamos National Laboratory

Tom Henyey, SCEC and University of Southern California

Rebecca Johnson, Los Alamos National Laboratory

Charles Keller, IGPP, Los Alamos National Laboratory

William Klein, Boston University

Mitsuhiro Matsu'ura, University of Tokyo, Japan

Seth McGinnis, University of Colorado

John McRaney, SCEC and University of Southern California

J. Bernard Minster, SCEC and University of California San Diego

Peter Mora, University of Queensland, Australia

David O'Halloran, Carnegie Mellon University

David Oglesby, University of California Santa Barbara

Jon Pelletier, California Institute of Technology

John Rundle, University of Colorado

Charles Sammis, University of Southern California

Steven Shkoller, CNLS Los Alamos National Laboratory

Stewart Smith, University of Washington

Leon Teng, University of Southern California

Donald Turcotte, Cornell University

Andrew White, ACL. Los Alamos National Laboratory

Yehuda Ben-Zion, University of Southern California

Bryant York, Northeastern University

During the saturday morning breakout sessions, the following questions were considered by the group as a whole. Following this discussion, individual breakout groups were organized to address each question. Their conclusions are appended following each question.

Scientific goals

-physics of material failure, physics of earthquakes

-assimilation of laboratory data

-assimilation of field data

-assimilation of new theoretical frameworks

-realistic simulator of earth system with predictive capabilities

-ensemble behavior in long term

-specific behavior in short term

Other goals

-insertion of state of the art computational tools (HW and SW)

-workbench for testing and evaluating

-science components

-algorithms

-concepts

GEM/SCEC should seek to initiate research on short & long term problems:

-must be realizable in the finite future

-must illustrate the scale issues:

-Same problem for micro/macro ends of the spectrum?

-What does a proper validation consist of?

-When are we happy?

Answer:

The scientific objective is to develop a physically based model which simulates seismicity and the associated deformation field at the plate boundary scale over time intervals on the order of tens of years. We envision models which are consistent with recent historical seismicity and deformation and can be used to project these quantities tens of years into the future.

There is scientific justification to divide the computational problem into two distinct but interacting modules. The first is the physics of the nucleation, slip and arrest of a single earthquake on a single fault segment which incorporates current knowledge of the effects of friction, heterogeneity, fault geometry, and fluids. A network of such faults would then be introduced into a model which reproduces the observed regional geometry from the largest faults to as small a size as can be accurately resolved by geological data. The second scientific module of a regional model is then to develop and parameterize an effective medium representation of the deformation associated with smaller events on the myriad of smaller faults which control the behavior of the crust between those larger faults which are explicitly included in the model. Such an effective medium would necessarily be non-linear and probably time dependent.

The smallest scale at which individual faults are explicitly modeled may be determined by either computational limitations or by the geological resolution of the fault network. It is also important to point out that, in addition to realistic faults in an effective medium, the regional model should also include the known vertical rheology of a brittle surface layer over a ductile substrate.

2. Question: What should be the relationship of GEM to ongoing organized research efforts...specifically SCEC? (Subgroup Chair was T. Henyey, USC)

SCEC ongoing/renewed efforts in earthquake physics research

-support by contributing tools and research environment

-draw from SCEC multi-disc work?

-SCEC data base to be used for validation & model development

NPACI and ASCI

-formulate grand challenge

-provide scientific motivation

-draw from physical and intellectual resources

ACES and other international efforts

-scientific exchanges - software, people, meetings, ?

-coordination of problem permutation

-forum for broadening international efforts

-experiment with long distance cooperation/computation

Answer:

GEM Principal Investigators should request that the Southern California Earthquake Center (SCEC) invite the GEM Project (GEM) to be the major integrative, high-performance computational arm of SCEC, and perhaps the coordinating entity for computational earthquake physics within the Center. GEM would draw from, and participate in, SCEC's multi-disciplinary scientific and outreach activities. GEM scientists could be part of the SCEC Working Group on Earthquake Physics unless it was determined by the SCEC Board of Directors that a separate Working Group on high-performance computing be established. GEM academic scientists could receive funding from SCEC for science, but would be encouraged to seek outside funding for computational resources and other activities. It is further proposed that the GEM Steering Committee constitute a Standing Committee of SCEC with the chairman of that committee a member of SCEC's Steering Committee.

GEM/SCEC should be strongly encouraged to develop a coupled and cohesive international effort, with particular emphasis on the Pacific Rim. As such, GEM/SCEC should be designated as the official U.S. component of ACES (APEC Cooperation for Earthquake Simulation). Scientists from participating ACES countries would be invited to GEM workshops to be informed on the development of GEM's operating plan. Scientific exchanges would be encouraged vis-a-vis the ACES planning document as GEM matures into a fully operational entity. Interactions with other national and international groups in earthquake science, engineering, etc. should be developed through SCEC.

SCEC ( statisticainteractions with NPACI (through both NSF and the centers themselves) and ASCI to explore access to high-performance computational resources -- both hardware and software. In particular, GEM/SCEC should explore with ASCI the mutual benefits that might accrue through this interaction.

In order for the SCEC Board of Directors to entertain this proposal for incorporating GEM under SCEC's aegis, GEM Principal Investigators should prepare a one to two page project summary to include (but not limited to) the following items:

1) Lead PI (presumably also the chair of the GEM Steering Committee)

2) Members of the GEM Steering Committee

3) List of Co-PIs with their affiliations and brief statement of their expertise

4) Description of the project

5) Problem(s) in earthquake physics to be tackled (both short term and long term)

6) Proposed interfaces with SCEC's existing science and outreach programs

7) Anticipated scientific/computational products

8) Anticipated outreach products as they pertain to earthquake hazard reduction

9) Proposed timelines for (7) and (8)

10) Anticipated funding and computational resources external to SCEC

11) Anticipated international collaboration

12) Location and organization of the GEM "activity center"

Much of this information is already included in the GEM proposal "Development of an Object Oriented Simulation System with application to Earthquake Physics in Southern California", and simply needs to be pulled together and summarized.

3. Question: Should we organize an annual GEM workshop? (Answers in BOLD) (Subgroup Chairs were J. McRaney and C. Keller)

YES!

Duration 1 week?

( 2 weeks? 1 week + 1 week

3 months

Venue ( LANL? Initially

variable?

abroad?

summer 98 - late July

Activities ( lectures?

short course?

( hands on computing?

shoot-outs? Benchmarks

Size ( limited? agenda driven

no limits?

( students?

Intended ( U.S.?

Audience anyone else? as appropriate

observers

Agenda on a per workshop basis? YES

Support IGPP?

SCEC?

others? NSF, DOE

4. Question: What is the simplest, fastest, cheapest simulation that would lead to a significant scientific question and pose a significant computational challenge? (Subgroup Chairs were S. Day and G. Fox)

-must pose a significant computational challenge

-must address a substantial scientil/probabilities

-what relation to SCEC ( physical modeling effort )

-Cost of earthquake disasters (urban) will grow even higher

-Advanced computing/networking technology

accessible and more affordable

forward looking approach

-Advances in computational science

permit a new generation of simulations

-Interest and willingness in the scientific community

SCEC

DOE - ASCI

NPACI

APEC - ACES

Challenges

-money

-access to HPCC-ASCI capabilities

-customers for proposals

-leverage SCEC communities for adequate funding

-insert EQ science in DOE agenda

-network to Australia, Japan, elsewhere

-develop common language with computational scientists

-software architecture: test bench for components

-software validation

-averaging small scale (subgrid - subparticle) complexity

justify - validate

-testing of physical components

physical correctness - validity

Misc.:

Important Examples Affecting the Irreversible Evolution of Faults and Earthquakes

Missing ingredient in large scale modeling

-transition from distributed deformation to localized faulting

Missing ingredient in small scale physics

-Energy partition at rupture tip (mechanical damage, heat, seismic radiation,...)

Modeling should be physically correct, computationally feasible, and fundable

Answer:

Problem:

Quasi-static evolutions of stress field - California

Model:

viscoelastic

fixed faults

stress interaction via Green’s function

Inputs:

fault geometry

initial stress field (past events)

tectonic loading

Outputs:

events

stress fields

Relevance:

hazard models

earthquake scenarios

5. Question: How do we organize ourselves? (Subgroup Chairs were JB Minster and JB Rundle)

Task Forces

Computational Task Force:

identify computational technologies and challenges

EQ Physics Task Force:

formulate and prioritize ? and problems

Data and Parameters Task Force:

Validation Task Force:

design and run convincing tests

GEM Workgroup:

communications by web page, list servers, news groups, and bulletin boards; raise issues; conduct debate; find students to workshop; AGU sessions?; liaison with SCEC, ACES, NPACI, etc.

GEM Workshop Committee

-select dates, venues

-set agenda yearly

GEM Proposal Executive Committee

-identify proposal funding sources

-coordinate proposals

Timeliness of GEM:

-New technology

new data, Broadband seismology, GPS, SAR

-Multidisciplinary approach tested

Answer (Immediate Action Items):

1. J.B. Rundle elected chair of GEM Workgroup. Will consult with meeting participants and others to organize GEM effort in most efficient and effective way possible.

2. G.C. Fox to serve as chair of Computational Task Force.

3. Meeting at Stanford on November 20 to begin preparation of new STC Center proposal to NSF. Rundle will attend and discuss how to incoporate GEM effort into new State of California Earthquake Center activities.

4. J.B. Rundle will synthesize a short document prior to Stanford meeting describing how GEM might be incorporated into new Earthquake Center proposal.