CIG Business Meeting 2009 Presen by liwenting


									CIG Business Meeting
         December 15, 2009

               - Welcome; GWP (Olson)
               - CIG Proposal to NSF (Buffett)
               - Our incoming director (Kellogg)
Agenda         - Our outgoing director (Gurnis)
               - Working group chairs (5 minutes ea):
Items            - Long term tectonics
                 - Short term tectonics
                 - Mantle convection
                 - Magma migration
                 - Seismology
                 - Dynamo
               - Discussion/Q&A
        Executive Committee
•   Peter Olson, Chair, (Sept. 2010), Johns Hopkins
•   Bill Appelbe, (Sept. 2011), Victorian Partnership for
    Advanced Computing (VPAC), Australia
•   Alan Levander, (Sept. 2010), Rice University
•   Carolina Lithgow-Bertelloni, (Sept. 2010), University
    College, London
•   Marc Parmentier, (Sept. 2012), Brown University
•   Michael Aivazis, Ex officio
•   Michael Gurnis, Ex officio
•   Louise Kellogg, Ex officio

    Science Steering Committee
•    Wolfgang Bangerth, (Sept. 2011), Texas A&M
•    Magali Billen, (Sept. 2012), University of California,
•    Bruce Buffett, (Sept. 2010), UC Berkeley
•    Andy Freed, (Sept. 2011), Purdue
•    Omar Ghattas, (Sept. 2010), University of Texas at
•    Garrett Ito, (Sept. 2012), University of Hawaii
•    Luc Lavier, (Sept. 2011), University of Texas at Austin
•    Jeroen Tromp, (Sept. 2010), Princeton
•    Peter Olson, (Sept. 2010), Johns Hopkins, Ex officio

\                                                          3
                          Geodynamics White Paper
              10 Grand Challenge Problems in Geodynamics



   David         Lucy      Bruce      Rick      Diana              Mousumi   Peter
                 Flesch                                   Phipps
  Bercovici                Buffett   Carlson   Valencia              Roy     Olson

 • For NRC/NSF “New Research Opportunities in the Earth Sciences”
 • Comments, additions, suggestions, all welcome
 • CIG INPUT NEEDED to identify key instrumentation and facilities !!!

         Send to:
                   Chronology of Events

12/15/08   SSC committee posts “Vision for CIG-II” for comment

12/16/08   Proposal writing committee announced at Business Meeting
02/15/09 Solicit one-page abstracts on work enabled by CIG

03/30/09 Two-day Workshop on “Opportunities & Challenges in
         Computational Geophysics”, Pasadena, CA

04/01/09 PWC meets to review input and begin writing
05/05/09 Draft proposal distributed for community input
07/08/09 Proposal submitted to NSF
Administrative Issues
       i) Maintain centralized team of software developers
       ii) Enable specialized development through subcontracts
       iii) Shift location of CIG to UC Davis
Science Issues
      i) Explore problems at interface of disciplines
      ii) Emphasis on interactions of the Earth’s subsystems
      iii) Identify and exploit overlapping interest

Figure 1 from Proposal
                  Science Questions

Science Questions Organized by Scale

         Global Scale                  Plate-Boundary Scale
           i) Mantle dynamics             i) Lithospheric tectonics
          ii) Core dynamics              ii) Fault system dynamics

                    Integration with Observations
                       i) seismic imaging
                      ii) inverse problems
          Emerging Methodologies

                Adaptive Mesh Refinement

                Adjoints & Inversion

                Multiphysics Coupling

                Scientific Visualization
Promote reliance on existing packages using proven technology
               Proposed Development

Ongoing Development
1. Maintain CitcomS
  - more equations of state, improve solver, add adjoint capability

2. Maintain Gale
      - increase flexibility of meshing and boundary conditions
      - interface to surface process codes
      - linking of deformation to regional-scale seismic imaging
 3. Continue development of PyLith
      - achieve goal of resolving entire seismic cycle
                Proposed Development

New Development
 1. Plate Tectonic/Mantle Dynamics
       - new code development based on deal.II
       - adds AMR, adjoints, melting model, flexible geometry
 2. Lithospheric Tectonics & Surface Processes
        - new code built from same components as mantle dynamics
        - improved flexibility of model components
        - new strategies for tracking damage/plastic failure
 3. Magma Dynamics
       - integrated into mantle dynamics/lithospheric tectonics code
               Proposed Development
New Development

 4. Fault System Dynamics
      - identified challenges in achieving resolution and scalability

 5. Seismic Imaging
       - emphasis on regional forward and inverse problems
       - integration with regional geodynamic models
       - future needs can be met with development of SpecFem
Proposed Development
       Leadership Transition
•   Michael Gurnis, outgoing Director
•   Louise Kellogg, incoming Director

Computational Infrastructure in Geodynamics:
   A virtual organization for geodynamics
               Louise Kellogg, Director-designate
                 University of California, Davis

•     Established 5 years ago as a member-governed
•     Funded by an NSF cooperative agreement to Caltech
•     55 participating institutions:
    •    45 US member institutions and 10 foreign affiliates.
    •    3 new US members and 1 foreign affiliate joined
         this fall.
    •    Governed by an Executive Committee, with input
         from a Science Steering Committee, and 6 working
    •    47 people from 36 institutions current serve on CIG
         committees and working groups.

                Transition to UC Davis
•     Proposal submitted to NSF Geoinformatics - July 2009
•     Funding anticipated to start in early 2010
•     Preparation for transition is underway led by Director,
      Director-Designate, EC, and SSC.
•     CIG’s new home at UC Davis:
    •    Located in proximity to Computer Science, Geology,
         and Applied Math, with meeting rooms and
         administrative staff.
    •    IT support includes computing and networking
         resources and staff who are expert in parallel
         computing on clusters.
    •    Geophysical/computational visualization through the
    •    UC Davis’s strong affiliation with LLNL and LBL
         provides access to additional computational experts.

CIG Long-Term Tectonics Working Group:
       „State of the Group‟ Report
                        Fall AGU, 2009

                      Group Members:

        Todd Ehlers, co-lead (U. Tübingen, Germany)
          Mousumi Roy, co-lead (U. New Mexico)
            Thorsten Becker (U. South California)
                    Noah Fay (U. Arizona)
       Dennis Harry (Colorado State U., former co-lead)
           Ritske Huismans (Bergen U., Norway)
           Carolina Lithgow-Bertelloni (UCL, UK)
            Dietmar Muller (U. Sidney, Australia)
              Patrice Rey (U. Sidney, Australia)
               Jolante van Wijke (U. Houston)

                With Valuable Input From:
             Louis Moresi (Monash U, Australia)
             Steve Quenette (VPAC, Australia)
    LTT Workgroup Activities This Past

•   March 2009: CIG-II meeting identified problems and achievements within
    the work group (well attended)

•   August 2009: „Developer‟ meeting: provided more in-depth training on
    GALE (well attended, including VPAC and L. Moresi).

•   Fall 2009 -> 0 Ma: Walter and VPAC initiated a re-integration of GALE into
    underworld (will provide increased compatibility with underworld

•   November 2009: Submission of GALE short course proposal for the GSA
    2010 meeting. Will cover basics of GALE use, as well as (hopefully) user
    modification of GALE.

•   September-December 2009: Increased activity on LTT email list serve -
    appears to be increased use/interest in GALE.
        Activities Proposed in CIG-II

•   Outstanding Scientific Problems Addressed:
    •   Fluids: What role do magmatic and hydrous fluids
        play in controlling lithospheric deformation?
    •   Rheology: What is the rheologic structure of the
    •   Memory: How does structural „memory‟ associated
        with previous deformation influence modern
    •   Interconnections: To what degree are climate,
        tectonics, and surface processes coupled?
             Activities Proposed in CIG-II
•   Future Directions of the LTT Workgroup in CIG-II:
    •   Increase functionality and portability, for example:
        •   addition of magma migration.
        •   implement more general boundary conditions
        •   interface for coupling with surface process models
    •   Coupling with other CIG codes, for example:
        •   coupling with seismic imaging (SPECFEM).
    •   Improve the numerics: Add AMR to improve fault
        localization, allow for fluid transport.
    •   Education/Outreach: Continue offering short courses,
        provide „advanced user‟ training for code modification.
LTT Example I Magma Injection & Faulting

              G. Ito & M. Behn
              See also Behn and Ito G3 and Ito and Behn G3, 2008.
LTT Example II Shear Band Formation & Intrusions

                           After J. Van Wijk et al. , In Press EPSL
Example III: Surface Processes & Faulting

                             Cruz et al. , In Prep or Review
Example IV: Surface Processes & GALE

                         courtesy Steve Quenette & VPAC
Example IV: Surface Processes - Transient Erosion

                 Ehlers unpublished:
                 modified CASCADE model (Braun & Sambridge 1997)
                 Work in progress with L. Moresi and S. Quenette
                 to couple with GALE/Underworld.
Example V: GALE in the classroom (UNM, Univ. Houston)


   Strike Slip Faulting          Normal Faulting, weak lower crust

                                       courtesy Mousumi Roy
Short-Term Crustal Dynamics
       Working Group
        Brad Aagaard
                               2009 Activities
•       Numerical Modeling of Crustal Deformation workshop, June 22–26, 2009,
        Colorado School of Mines
    •      8th annual workshop, sponsored by CIG, SCEC, NSF, NASA
    •      60 participants
    •      Mixture of science talks and tutorials

•       PyLith (Aagaard, Williams, Knepley)
    •      Parallel, 2-D and 3-D finite-element code for modeling crustal deformation
           associated with earthquake faulting and elastic and/or viscoelastic
    •      1 feature release + 2 bugfix releases in last 12 months
    •      Version 1.4.0 released Jul 1, 2009
           •   Power-law viscoelastic bulk rheology
           •   Automatic nondimensionalization of parameters
           •   Time-dependent boundary conditions
           •   Under-the-hood improvements (Pyrex → SWIG)
    •      ∼200 downloads per quarter and growing
                      PyLith Development
             Aagaard, Williams, Knepley, and Somala

•     Version 1.5 (Spring 2010)
    •    Fault friction (50% complete)
    •    Large deformations (100% complete)
    •    Drucker-Prager elastoplastic bulk rheology (5% complete)
    •    Optimized solver for explicit time-stepping (80% complete)
    •    Laplacian preconditioner (10% complete)
    •    Improvements to manual (5% complete)
         •   Expand discussion of governing equations and implementation
         •   Reorganize tutorials

•     Long-term Development
    •    Multi-scale: Coupling between interseismic and seismic scales
    •    Multi-physics: Integration with heat float and fluid flow
                             2010 Activities
•     Numerical Modeling of Crustal Deformation workshop, June 14–18, 2009,
      Colorado School of Mines
    •    Seeking funding from same sponsors
    •    Number of attendees (∼60) limited by meeting facility
    •    Tutorials: Day 1 (beginners), Day 2 (intermediate)
    •    Science talks and discussion: Days 3–5
    •    Strengthen ties to longer-term deformation (SCEC CDM & LAD)

•     Collect semi-analytic codes for simple problems
    •     Centralize dissemination of useful codes for simple problems
    •     Efficient tools for problems that don‟t need PyLith
    •     Useful for benchmarking codes for complex problems

•     Benchmark crustal deformation codes
    •    Continue to define benchmark suite as capabilities expand
    •    Using CIGMA to quantify errors and differences

CIG Business Meeting
    December 15, 2009

   CIG Activities in Mantle Convection
    Community-Driven activities:
        Working groups
•   On compressible mantle convection (King,
    van Keken, Tan, Zhong, …).
•   On development of 1-D analytical codes
    (Becker, O’Neil, Steinberger and others).
•   Since the objectives for Mantle Convection
    within CIG have largely been meet, we’ll
    probably need a new Mantle Convection
    working (structure) group.
      Mantle Convection Codes:
          Available at CIG
•   CitcomS (3D spherical convection:
    compressibility, thermochemical, and non-
    Newtonian rheology, …)
    Latest release version 3.0.3 in September, 2008
•   CitcomCU (3D Cartesian and regional spherical)
    Latest release version 1.0.2 in May, 2007.
•   Ellipsis3d (3D Cartesian)
    Latest release version 1.0.2 in April, 2007.
•   ConMan (2D Cartesian)
    Latest release in September, 2008.
• Newest release: v3.1.1 (Jul 2009)
• Allowing tracers in global model
  with nprocx,nprocy > 2
  •   better load balancing
  •   allowing higher resolution
• CBF topography
• Self gravitation geoid
• ~10% speed up
           CitcomS +
• Linkage through CIG Seismology
  Web Portal
  •   CitcomS outputs Vp, Vs, ρ
  •   upload Vp, Vs, ρ to the portal
  •   choose mesh/resolution/length of
      seismogram, submit job
  •   download seismograms
•   Newest release: v1.0.3 (Nov 2009)
•   Allowing models using more than 100
•   Scripts to convert model result to VTK
•   Example input files for benchmarks
    (Busse et al. 1993)
•   Several bug fixes
     Geodynamics of the Lithosphere and
            Deep Earth, or GLADE
Venue: Scripps Institution of Oceanography, in La Jolla,
    California, from July 26-29, 2010.
SSC Approved Co-Sponsoring of Workshop
The organizing committee :John Hernlund & Dave
    Stegman (co-chairs); Steve Cande, Alessandro Forte,
    Allen McNamara, Leigh Royden & Shijie Zhong

CIG Business Meeting
December 15, 2009                                          39
•          Being developed as a collaboration between U. Texas, Caltech, &
           Colorado and Funded by the NSF PetaApps Program
•          Rhea is a parallel, adaptive-mesh finite-element-based mantle
           convection solver
•          Builds on the ALPS framework for the adaptive solution of partial
           differential equations.
•          The management and distribution of the elements is handled by the
           forest-of-octree library p4est.
•          The solution of the pre-conditioned Stokes equation is handled with
           Algebraic Multi-Grid (AMG)
•          Strong scaling of the energy equation and the AMR is almost perfect
           out to 62K processors
•          Weak Scaling of the Stokes equation is reasonably good (for
           example efficiency on 16,384 cores is about 25% compared to 1
•          A open source version of p4est and Rhea will be available 2010

CIG Business Meeting
December 15, 2009                                                            40

CIG Business Meeting
December 15, 2009             41
               Future Plans for Mantle
•            Maintain CitcomS. Add multiple realistic equations of
             state to existing compressible-thermo chemical code.
             Add new Dave May pre-conditioners (CitcomSX uses
•            New code for very high resolution locally, such as
             plate boundaries (AMR), full inverse capabilities, very
             robust solvers, and linkage to multi-physics
         •        Rhea would be an interim solution
         •        Develop new code based on deal.II which will be enhanced with
                  Rhea technology (such as p4est)
         •        Long-term tectonic and Magma will based on the same
                  approach. All of the physics and capabilities of long-term and
                  magma will be embedded, self-consistently with this approach.
         •        Cartesian and spherical versions

CIG Business Meeting
December 15, 2009                                                              42
Magma Dynamics
 Marc Spiegelman
                 Magma Dynamics
• The Magma Dynamics Demonstration Suite: (MADDs):
• Components
    High resolution Stokes problem for Ridge geometries with resolved
    Constant porosity melt flow benchmark
    2-D/3-D Solitary wave benchmark code
    Isoviscous ridge model with forced adiabatic melting
• Two existing implementations at CIG
    stgMADDs: CIG/VPAC collaboration, same code base as Gale
    MADDs-FP: FEniCS/PETSc hybrid codes for “Flexible
     Programming” in magma dynamics
• Other partial implementations in COMSOL, deal.ii
 MADDs 1&2       MADDs 4

MADDs 5a         Features:
                 •FEniCS form compiler for PDE based
                 code generation (easily include new
                 •PETSc access to full range of advanced
                 solvers/preconditioners e.g (AMG
                 •Semi-Lagrangian advection on fully
                 unstructured meshes
                  Future directions
• Full integration with the next
  generation AMR Solid Flow codes
  (MC,LTT) likely deal.ii
• MADDs-FP will continue as part of a
  MARGINS sponsored project
  (seeking post-doc) and be released
  through CIG
    More general thermodynamics for
                                                      3-D AMR Ridge Stokes Calculation
     reactive fluid flow                              Deal.ii (Bangerth, 2009, pers comm)

    Geochemical transport
• Regional modeling toolkit
• Requires detailed technical
  discussion across CIG
• Expand MD-working group
                   Fluid migration
                   Nicaragua subduction
                   Zone (Spiegelman, van Keken, Hacker)
Seismology Working Group
  Compiled by Alan Levander
    CIG Seismology Working Group

•   New Additions & Updates
    •   Flexwin: for windowing seismograms
    •   Seismic_CPML: 2D and 3D FD codes
    •   SPECFEM2D with adjoint capabilities
    •   New SPECFEM3D in 2010
•   Integration of CitcomS and SPECFEM to
    produce synthetic seismograms from
    CitcomS outputs
  CIG Seismology Working Group
• Downloads of FLEXWIN, MINEOS,
  SPECFEM_GLOBE accounted for ~40% of CIG
  downloads in 4th quarter 2009.

• We are discussing having large US participation
  at a QUEST workshop in Sardinia in September
  2010. [QUEST is a recently funded European
  seismic imaging initiative that is a follow on to SPICE.
  CIG/IRIS/SPICE held a joint workshop in Jackson,
  New Hampshire in 2007].
              Bruce Buffett
•   Current status of community

To top