1. To develop realistic numerical simulation models

2. To foster collaboration

3. To foster development of infrastructure & programs

Develop a unified simulation model for earthquake generation and earthquake

cycles

The ACES international cooperation:advances and challenges

The ACES international cooperation:advances and challenges

By Peter Mora

AcknowledgementsAcknowledgements

ACES was established under the Asia PacificEconomic Cooperation (APEC) sponsored by

Australia, China, Japan and USA

• Australia: ISR and ARC• China: MOST, NSFC and CSB• Japan: RIST and JSPS• USA: NASA &NSF (3-rd workshop sponsors)

Thanks to US LOC for organisation of 3-rd WS

The scientific challenge

• Science involves the feedback between a predictive theory and observations

• Such a theory does not exist for earthquakesSuch a theory does not exist for earthquakes

• The earth is a complex systemThe earth is a complex system

WHY?

nonlinear elements + interactions = cooperativebehaviour qualitatively different from elements

Observns Model

Analysis

Observations Theory

SimulationA powerful tool to fuel breakthroughs

SimulationA powerful tool to fuel breakthroughs

Observn’s Model

Analysis

ComputationalVirtual earth laboratory

History & program of activitiesHistory & program of activities

1994 Discussions

1995 AESF Proposal

1996 NSPE Proposal

1997 ACES Proposal endorsed

1998 Planning meeting

1999 Workshop (AUS)

2000 WG Mtg & Workshop (JPN) +

Visitors

2001 WG Mtg (USA)

2002 Workshop (USA)

2004 WG Mtg (AUS)

2004 Workshop (CHINA)

Selected collaborative projects 2000Selected collaborative projects 2000

Australia/Japan

• Fault zone evolution• LSMearth/GeoFEM interface

Australia/USA

• Fault constitutive laws/rate-state friction• Strong motion/teleseismic obs

Australia/China

• Mesoscopic mechanism for catastrophic rupture• Physical mechanism for LURR• Probe relation CP/LURR and critical scaling

Infrastructure developmentsInfrastructure developments

Japan• Earth Simulator – 40 Tflops (1997-2002)• GeoFEM macro-scale software system

Australia• Solid Earth Simulator – 200 Gflops (2000-2002)• LSMearth micro-scale software system• ACcESS MNRF – 2 Tflops (2002-2007)• ESS multi-scale software system

A few selected advances and

highlights from ACES and

its participants

A few selected advances and

highlights from ACES and

its participants

Japan: Earth Simulator ProjectJapan: Earth Simulator Project

MEXT (STA), 1997-2001 F/Y

Objectives• Forecast earth dynamics by "Virtual Earth"• Enhance information science & technology

Development of• Parallel computer “Earth Simulator”• Advanced software Fluid earth : ocean-atmosphere dynamics Solid earth : crust-mantle-core dynamics

Inside of the Earth Simulator Buildinghttp://www.es.jamstec.go.jp

- 640 nodes ( 8 vector processors / node ) - single-stage crossbar network - max 16 Gbytes/sec x 2

SOLID EARTH DYNAMICSSIMULATION SYSTEM

Core-Mantle Dynamics Modelling

Solid Earth Simulation Platform

Earth Simulator: High performance massively parallel processingcomputer system (10TB Memory, 40TFlops Peak Performance)

Crustal Activity Modelling

Earthquake Rupture &

Seismic Wave PropagationSimulation Software

Strong Motion Modelling

Solid Earth ResponseSimulation Software

Earthquake Cycle

Plate Motion

Crustal Activity DataAssimilation Software

Core-Mantle ConvectionSimulation Model

Earthquake RuptueSimulation Model

Crustal ActivitySimulation Model

GeoFEM: Multi-PurposeParallel FEM Code

PIM PIMPIM PIMPIM PIM

Courtesy of Mitsuhiro Matsu’ura

“Solid Earth” simulator“Solid Earth” simulator

Global scale （ 104 km, 106 ～ 108 y ）Mantle/Core Dynamics and InteractionInterior Earth Structure

Regional scale （ 103 km, 103 ～ 104

y ）Quasi-Static Earthquake Generation CycleDynamic RuptureGPS Tectonic Data Assimilation

Local scale （ 10 km, 102 s ）Earthquake GenerationSeismic Wave Propagation

Visualization dataGPPView

One-domain mesh

PEs

Partitioner

構造計算（ Static linear ）構造計算（ Dynamic linear ）構造計算

（ Contact ）

Partitioned mesh

SolverI/F

Comm.I/F

Vis.I/F

UtilitiesStructureFluid

Wave

Pluggable Analysis Modules

Equationsolvers

VisualizerParallelI/O

Platform

System Configuration of GeoFEMSystem Configuration of GeoFEM

Courtesy of Hiroshi Okuda

GeodynamoGeodynamo

Entire mesh Earth’s interior

64 domains P isosurface Magnetic force lines

Electrically conductive fluid in Earth's outer core

Enlarged view

Courtesy of Matsui & Okuda

Modeling of South West JapanModeling of South West Japan

xy

z

FE model of Philippine Sea Plate and Pacific Ocean Plate

ShikokuIzu

Philippine Sea Plate Pacific Ocean Plate

Mantle

Mantle

Courtesy of Iizuka and Hirahara

Long-term Crustal Deformation Caused by Steady Plate Subduction

(a) 3-D geometry of plate interfaces (b) Computed crustal uplift rates

Courtesy of Hashimoto and Matsu’ura

(a) Quasi-static stress accumulation (b) Dynamic rupture propagation

0 Shear Stress (MPa) 3

0 Slip Deficits (m) 2

Shear Stress Slip Deficits Shear Stress Fault Slip

Shear Stress (MPa)

Fault Slip (m)

3-D Simulation of Earthquake Generation Cycles at a Plate Boundary

Courtesy of Fukuyama and Matsu’ura

3-D Simulation of Strong Ground Motion in the 1995 Kobe Earthquake

Seismic wave radiationSeismic wave radiation

Strong ground motion beltsStrong ground motion belts

Rupture initiationRupture initiation

Seismic wave propagationSeismic wave propagation

Courtesy of Furumura and Koketsu

Australia

QUAKES to ACcESSLSMearth micro-model

Australia

QUAKES to ACcESSLSMearth micro-model

• Object oriented “plug-and-play” system

• New virtual environment for micro-model

Courtesy of David Place and Peter Mora

Physics of fault zonesLocalisation phenomena

Physics of fault zonesLocalisation phenomena

From Place & Mora, 2000PAGEOPH, 157, 1821-1845.

Fault constitutive relation

Australia/USAQUAKES/USGS

Fault constitutive relation

Australia/USAQUAKES/USGS

From Abe, Dieterich, Mora and Place, 2002, PAGEOPH, 159, No. 10

Multi-scale simulator development

LSMearth to GeoFEM & EFG

(Australia-Japan: QUAKES/RIST/Yokohama/Tokyo U)

Multi-scale simulator development

LSMearth to GeoFEM & EFG

(Australia-Japan: QUAKES/RIST/Yokohama/Tokyo U)

Courtesy of David Place

The Australian Computational Earth Systems Simulator (ACcESS)

Major National Research Facility

The Australian Computational Earth Systems Simulator (ACcESS)

Major National Research Facility

A multi-scale multi-physics ESS• Achieve a holistic virtual earth simulation capability

• Provide a computational virtual earth serving Australia’s national needs

• Empower Australian geosciences with a never before seen computational capacity

• Act as a focal point for Earth Systems Simulation

Centre for ComputationalEarth Systems Simulation

http://www.access.edu.au

Multi-institutional, multi-disciplinaryMulti-institutional, multi-disciplinary

Queensland (MNRF HQ)

Micro-models, LSMearth software

Comp. ES, earthquakes

QUAKES

Western Australia

Nonlinear rheologies, geodynamics

Comp. mech, mining,

Solid Mech, CSIRO; UWA

VictoriaACRC/Mon, VPAC, Melb, RMIT

Geology, tectonics reconstruction

Min. exploration, SE/Vis/IT

ParticleModels

…

Communication Substrate

ContinuumModels

DataAssimilation

PostProcessing

Visualisation

Observ’s Theory

ComputationalVirtual earth laboratory

Facility developmentFacility development

ParticleModels

…

Communication Substrate

ContinuumModels

DataAssimilation

PostProcessing

Visualisation

Usable software platformPowerful extendable models

Development of software & models and establishment of thematic supercomputer needed for research outcomes.

ObjectivesObjectives

Novel multi-scale simulation methodology:

• Macro-models: Mountains, folding, faulting

• Chemical processes: Mineralisation

• Micro-models: Localisation, fracture, friction

• Virtual earth reconstructions for exploration

• Stress field reconstruction

Broad range of applicability:

• Global scale minerals exploration

• Geohazards mitigation and forecasting

• Mining excavation stability and safety

• Virtual prototyping in mining

Courtesy of/by (top to bottom & L to R): Moresi and Muhlhaus, Mora & Place, Moresi, Sakaguchi, Coutel & Mora

USAUSA

Courtesy of Kim Olsen

Strong ground motion in LA basinStrong ground motion in LA basin

Courtesy of Kim Olsen

Pattern Recognition Techniques Show Promise for Earthquake

Forecasting

Pattern Recognition Techniques Show Promise for Earthquake

Forecasting

Courtesy of Kristy Tiampo and John Rundle

• Red regions indicate anomalies detected through Principal Component Analysis.

• Blue triangles and circles are earthquakes

• Recent earthquakes have occurred in the anomalous regions.

Analysis of observations indicates strong correlations

Analysis of observations indicates strong correlations

Courtesy of Kristy Tiampo & John Rundle

AustraliaCorrelation function evolution in simulations

AustraliaCorrelation function evolution in simulations

TimeFrom Mora and Place, 2002, PAGEOPH, 159, No. 10

p/pc

.001 .01 .1 1

S

0

5

10

15

20

25

30

35

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.60

5

10

15

20

25

30

35

40 = 0.1 sensitivity energy rate(*10) damage rate(*50)

time(t/0)

0 100 200 300 400 500 600 700

-50

0

50

100

150

200

250

failure point

time (s)

P sensitivity

ChinaCritical sensitivity

ChinaCritical sensitivity

Courtesy of Mengfen Xia

China/Australia: Load-unload response ratio c.f. AMR

China/Australia: Load-unload response ratio c.f. AMR

1960 1970 1980 1990

AMR observations

Predicted CP

Newcastle earthquake1.2

0.8

0.4

0.0

Ben

ioff

str

ain

(x

107 )

NewcastleEarthquake

time

LURR value5

4

3

2

1

1980 1985 1990

X

XLURR

From Yin and Mora et al, 2002, PAGEOPH, 159, No. 10

LURR vs AMR critical region size(CH/AU … USA)

LURR vs AMR critical region size(CH/AU … USA)

5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5

1.8

2.0

2.2

2.4

2.6

2.8

3.0

log(r)=0.08724+0.3375m

log(

radi

us/k

m)

magnitude

Courtesy of Can Yin (left) and Xiang-chu Yin (right)

Simulation of intact material (AU/CH)Simulation of intact material (AU/CH)

From Mora & Wang et al, 2002, PAGEOPH, 159, No. 10

Exploring system complexity with CA(AU/US/CH)

Exploring system complexity with CA(AU/US/CH)

Courtesy of Dion Weatherley

USA

Simulation of the CA interacting fault system

USA

Simulation of the CA interacting fault system

Southern California Seismicity Space-time Stress Diagram

Courtesy of P.B. Rundle and J.B. Rundle

A Comparison at C-Band:InSAR Difference Fringes Tend to Define the

Rupture Extent

A Comparison at C-Band:InSAR Difference Fringes Tend to Define the

Rupture Extent

Courtesy of John Rundle and Louise Kellogg

The difference fringes are small (red = positive and blue = negative regions), and are concentrated along the portions of the San Andreas that are about to initiate sliding, either in the main shock or the pre-and post-shocks.

ChallengesChallenges

• Understanding the complex system

• Computer speed limited to explore complex system dynamics

• Computational models need to be enhanced• Software effort needs to be increased• Data assimilation effort is large• Cooperation needs to be enhanced

ACES: Towards predictive modelling of earthquake phenomena - a new era of

earthquake science

ACES: Towards predictive modelling of earthquake phenomena - a new era of

earthquake science

A vision future earth systems scienceA vision future earth systems science

Advances in understanding earth physics, numericalsimulation methodology & supercomputer technology

are bringing the vision within reach

A predictive capability for earth system dynamics

Observns Model

Analysis

ComputationalVirtual earth laboratory

c.f. GCM’s

The endAcknowledgements

Contributors (incomplete list)

Rundle, Donnellan, Tiampo, Olsen, GEM & SCEC Place, Mora, Weatherley, Abe, Wang, Yin, Muhlhaus,

MoresiMatsu’ura, Okuda, Nakajima, Matsui, RIST group

Yin, Peng, Xia, CSB/LNM Group

The endAcknowledgements

Contributors (incomplete list)

Rundle, Donnellan, Tiampo, Olsen, GEM & SCEC Place, Mora, Weatherley, Abe, Wang, Yin, Muhlhaus,

MoresiMatsu’ura, Okuda, Nakajima, Matsui, RIST group

Yin, Peng, Xia, CSB/LNM Group