Planning the SCEC Pathways

1
Planning the SCEC Pathways

• Ewa Deelman, Yolanda Gil, Sridhar Gullapalli,
  Carl Kesselman, Jihie Kim, John McGee, Gaurang
  Mehta, Gurmeet Singh, Marc Spraragen, Mei-Hui Su,
  Karan Vahi
• USC Information Sciences Institute
• Vipin Gupta, Phil Maechling
• Southern California Earthquake Center
• Maureen Dougherty, Brian Mendenhall, Garrick
  Staples
• USC

2
Applying Information Technology (IT) to Seismic
Hazard Analysis
KNOWLEDGE REPRESENTATION REASONING Knowledge
Server Knowledge base access, Inference Translatio
n Services Syntactic semantic translation
Knowledge Base
Ontologies Curated taxonomies, Relations
constraints
Pathway Models Pathway templates, Models of
simulation codes
DIGITAL LIBRARIES Navigation Queries Versioning
, Topic maps Mediated Collections Federated acces
s
KNOWLEDGE ACQUISITION Acquisition
Interfaces Dialog planning, Pathway
construction strategies Pathway Assembly Template
instantiation, Resource selection, Constraint
checking
Code Repositories
FSM
RDM
AWM
SRM
Users
Data Simulation Products
Data Collections
GRID Pathway Execution Policy, Data ingest,
Repository access Grid Services Compute storage
management, Security
Pathway Instantiations
Storage
Computing
3
Seismic Hazard Analysis

• Definition Specification of the maximum
  intensity of shaking expected at a site during a
  fixed time interval
• Example National seismic hazard maps

• Intensity measure peak ground acceleration (PGA)
• Interval 50 years
• Probability of exceedance 2

4
SHA Computational Pathways
Standardized Seismic Hazard Analysis Ground
motion simulation Physics-based earthquake
forecasting Ground-motion inverse problem
1
2
3
Other Data Geology Geodesy
4
Unified Structural Representation
Invert
4
Faults Motions Stresses
Anelastic model
Ground Motions
AWM
SRM
RDM
FSM
3
2
Intensity Measures
Earthquake Forecast Model
Attenuation Relationship
1
AWP Anelastic Wave Propagation SRM Site
Response Model
FSM Fault System Model RDM Rupture Dynamics
Model
5
Pathway 1 OpenSHA
Time Span
OpenSHA A Community Modeling Environment
for Seismic Hazard Analysis
Earthquake- Rupture Forecast
IM
Rupn,i
Site
Type, Level
Sourcei
Intensity-Measure Relationship
6
Example Application of Pathway 1Scenarios for M
7.4 Southern San Andreas Rupture
Without soil basin effects
With soil basin effects
Courtesy of Ned Field, USGS, Pasadena
7
Pathway 2 Ground Motion Simulation
K. Olsen (2002)
Numerical Simulation of 1994 Northridge Earthquake
P and S waves superimposed on the mountain
topography for the San Fernando basin and the
vertical ground motion on the topography for 25
sec of rupture and wave propagation
8
Pathway 4 Data Inversion Assimilation
Pathway 4 inversions techniques are used to
update the geological models needed for
simulations in the other pathways
X-phase Fréchet Kernels 09/04/02 Yorba Linda
Earthquake
9
SCEC Analysis (Pathways)

• Workflow Creation and Execution
• produce a workflow template, populate with data
• produce an abstract workflow
• hand the abstract workflow to the GriPhyN VDS
• map the workflow (with MPI-based and single
  processor jobs) using Pegasus onto the SCEC and
  USC grid.
• use DAGMan and Condor-G to execute the workflow
  components.

SCE develops the ability for scientists to
improve computer models of how the Earth is
structured and how the ground moves during
earthquakes.
Images courtesy of SCEC
10
SCEC Analysis

• Used the GriPhyN and iVDGL-developed software
  suite--Virtual Data Toolkit (http//www.cs.wisc.ed
  u/vdt/)
• Consisted of three different Seismic
  Hazard-related Calculations (Pathway 1, Pathway
  2, and Pathway 4)
• Used Pegasus to uniformly handle different
  seismic hazard programs
• Previously, each SCEC pathway ran in an
  independent and dissimilar manner.
• Software used in the demonstrations NMI, VDT
  (including Pegasus and DAGMan), and MCS (the
  Metadata Catalog Service.)

11
SCEC Computations and Grid Testbed

• Pathway 2 example
• Preparation prepares input for Pathway 2
  simulation.
• Pathway 2 is Fortran based, wave propagation MPI
  code.
• Pathway2PGV reads in binary output file
  generated by Pathway2 and converts it into hazard
  map that can be visualized.

SDSC
USC
SCEC
12 CPUs
1,700 CPUs
1,200 CPUs
ISI
PSC
4 CPUs
1 CPU
12
SCECs portal job view

• Uses Pegasus portal solution for displaying job
  status
• Displays performance and status information about
  individual workflow tasks and entire workflows

13
Results

• SCEC workflows consisted of a mixture of single
  processor jobs as well as MPI-based computations
• 3 different types of SCEC workflows were
  demonstrated, the execution time ranged from 10
  to 45 mins, some generated ½ GB of data to be
  visualized
• The SCEC portal was used to set up and launch
  each type of analysis
• Input data for the analysis was selected based on
  metadata attributes (via the portal)
• Future directions
• Expand the analysis to use Teragrid resources
• Expand the CAT knowledge representations to
  support a variety of SCEC pathways

14
Information about the relevant projects and
software components

• SCEC www.scec.org/cme
• GriPhyN www.griphyn.org
• VDT www.ivdgl.org
• CAT www.isi.edu/ikcap/cat/
• Chimera www.griphyn.org/chimera
• Pegasus pegasus.isi.edu