SCEC Community Modeling Environment (SCEC/CME):

SCEC TeraShake Platform: Dynamic Rupture and Wave Propagation Simulations

Seismological Society of America Meeting18 April 2006

SCEC/CME ProjectGoal: To develop a cyberinfrastructure that can support system-level earthquake science – the SCEC Community Modeling Environment (CME)

Support: 5-yr project funded by the NSF/ITR program under the CISE and Geosciences Directorates

Oct 1, 2001 – Sept 30, 2006

SCEC/ITRProject

NSFCISE GEO

SCECInstitutions

IRIS

USGSISI

SDSCInformation

ScienceEarth

Science

www.scec.org/cme

SCEC/CME – All Hands Meeting

SCEC/CME Focus On Seismic Hazard Analysis

SCEC/CME System aims to extend and enhance geosciences work already performed in the area of seismic hazard analysis.

Metadata for Map:IMT: Peak AccelerationPOE: 10%TimeSpan: 50 YearsIMR: 1) Boore, Joyner, and Fumal (BJF; 1993, 1994a) with later modifications to differentiate thrust and strike-slip faulting (Boore et al., 1994b), 2) Sadigh et al. (1993) and 3) Campbell and Bozorgnia (1994).

ERF: … (and more)

SCEC/CME Computational Pathways

IntensityMeasures

Earthquake Forecast Model

AttenuationRelationship

1

Pathway 1: Standard Seismic Hazard Analysis

AWMGroundMotionsSRM

Unified Structural RepresentationFaults Motions Stresses Anelastic model

2

AWP = Anelastic Wave PropagationSRM = Site Response Model

Pathway 2: Ground motion simulation

RDM

FSM

3

FSM = Fault System ModelRDM = Rupture Dynamics Model

Pathway 3: Physics-based earthquake forecasting

Invert

Other DataGeologyGeodesy

4

Pathway 4: Ground motion inverse problem

SCEC Computational Platform Concept

• Definition– A large-scale implementation of computational pathways

within a specific computational system (hardware + software + expertise) for producing specific knowledge

• Implied components– Validated simulation software and geophysical models– Broadly useful simulation capabilities– Imports from other systems. Exports to other Systems– IT/geoscience collaboration involved in operation– Access to High-performance hardware– May use Workflow management tools

Large Scale Simulation-based Seismic Hazard Computational Platform Development

• 2003– OpenSHA

• 2004– OpenSHA, TeraShake

• 2005– OpenSHA, TeraShake, CyberShake

• 2006– OpenSHA, TeraShake, CyberShake, Earthworks

TeraShake Platform Anelastic Wave Propagation Capabilities

• Investigate ground motion effects for several different large Southern San Andreas ruptures.

SCEC/CME

•TeraShake Simulation area

•600 km x 300 km x 80 km

•dx=200m

•Mesh of 1.8 Billion cubes

•0.011 sec time step, 20,000 time steps: 3 min

Kinematic source: Cajon Creek to Bombay Beach (or back-60 sec source duration-18,886 point sources, each 6,800 time steps in duration

Scenario Earthquake Simulations - TeraShake

TeraShake 1.3 Cumulative PGV and SA 3.0 Maps – Kim Olsen et al (AWM), Amit Chourasia et al (Viz)

Scenario Earthquake Simulations - TeraShake

TeraShake 1.2 and 1.3 Cumulative PGV – Kim Olsen et al (AWM), Amit Chourasia et al (Viz)

Peak Displacements

TeraShake N to S Rupture (left) TeraShake S to N Rupture (right)

Particle Velocities Along N50E Profile

Largest Peak Motions above ridge between SG and LA Basins

Causes of ‘Hotspot’ in Los Angeles– Buildup of forward directivity pulse

– Excitation of guided waves upon striking S.B. basin

– Amplification as sedimentary waveguide narrows between Puente Hills and San Gabriel Mtns.

– Lateral and vertical focusing effects from 2D horizontally incident plane waves produce amplification patterns similar to those from TS1.3 and TS1.4

– Additional effects possibly from Airy phase of Love wave

– Nonlinear effects likely to decrease peak motions

TeraShake Platform Dynamic Rupture Capabilities

• Incorporate more physics into the TeraShake simulations by introduction of a dynamic rupture-based source description.

TeraShake-2 Dynamic Simulations

• RDM run on an inner domain, containing the entire rupture as well as the relevant surrounding environment….So, in addition to rupture physics parameters (yield friction, d0, etc)

• Need to include inhomogeneous crust on either side of rupture– This will require validation of rupture simulations

• Need absorbing boundaries (PML)– To run the rupture to completion, without need for a large inner domain

• Need validation of code coupling scheme– Current plan is to save the rupture and treat it a kinematic source for AWP– This raises issues of rescaling, smoothing, registering– Need to choose a rupture representation format that AWP can ingest

Large Scale Dynamic Rupture Simulations – Terashake 2

TeraShake 2 Simulation Area

Spontaneous Rupture

Simulation Domain

Olsen Dynamic Rupture Simulations

Olsen Dynamic Rupture Simulation Movies

Slip Animation

TeraShake 1 vs TeraShake 2 PGV Map

Planned Application and Development for TeraShake Platform

TeraShake 3 Simulations for 2006

0-3 Hz Deterministic Earthquake Ground Motion Prediction

Grid points along east, north, and up

Grid points (billions)

dh (m)

fmax (Hz) @ Vsmin (m/s)

200 x 200 x 150 0.006 200 0.5 @ 500

400 x 400 x 300 0.048 100 1.0 @ 500, 0.5 @ 250

800 x 800 x 600 0.38 50 2.0 @ 500, 1.0 @ 250, 0.5 @ 125

1,333 x 1,333 x 1,000

1.78 30 3.33 @ 500, 1.67 @ 250, 0.83 @ 125

CMU Hercules Tool chain

Highly scalable AWM software tools that uses new technique called In-situ parallel mesh generation

David O’Hallaron et al (CMU) Etree Mesh RepresentationJacobo Bielak et al (CMU) AWM

Finite Element Dynamic Rupture CodesPeak Surface Velocity Ratio (Topography/No Topography)

Scenario Earthquake Simulations – Puente Hills

Puente Hills Simulation Scenario Earthquake (10 Hz)Robert Graves (AWM), Amit Chourasia et al (Viz)

Velocity Y Component Animation Peak SA 2.0 magnitude Map

Full 3-D Tomography Model (Preliminary)

Po Chen – Inversion-based updates to SCEC CVM3.0 Velocity Model.

SCEC Education and Outreach

Outreach To Geosciences, Computer Sciences, and Public• SC04

•Invited Presentations USC, SDSC Booths• AGU 2004

•Booth, Presentations, Posters• Globus World 2005

• Invited Presentation• SSA 2005

•Booth, Invited Presentations, Posters• GRIDS Workshop 2005

• Invited Presentation• Unavco/IRIS Annual Meeting 2005

• CIG Workshop Presentation, Web Services Workshop, Posters• SCEC Earthquake Spectrum Press Conference 2005• Earthscope Meeting 2005

• Poster, IT Workshop• GEON All Hands Meeting 2005

• Posters• National Forum for Geoscience Information Technology (FGIT) 2005

• Poster• GSA 2005

• Booth, Poster• SC05

•Invited Presentations USC, SDSC, TeraGrid Booths

Computational Platform Concept

Data Intensive Capacity Computing

Data Intensive CapabilityComputing

Capability Computing

CyberShake Platform

PetaShake Platform

Observation Platform

Seismic Hazard (Application)

Large Scale Scenarios(Full Inner/Outer Scales)

New Physics-based Simulations (Validation)

Community Access(Science Gateways)

Validates

Applies

SCEC ScienceSCEC Platforms

Computing Requirements

Cyberinfrastructure Layering

Intelligent services (smart assistants)Intelligent services (smart assistants)

Integrated system tools (workbench/dashboard)Integrated system tools (workbench/dashboard)

Workflow managementWorkflow management

Domain applications (webservices/applications)Domain applications (webservices/applications)

Resource sharing (grids)Resource sharing (grids)

Hardware (computing, networking, storage)Hardware (computing, networking, storage)1

2

3

4

5

6

Vertical

integration

SCEC Community Modeling EnvironmentA grid-enabled collaboratory for system-level earthquake science