Summary

GEM Workshop

4/22/99

Jet Propulsion Laboratory

Pasadena, California

 

I. The meeting began with brief introductions and discussion of the need for a science focus within GEM. We also discussed the NASA ESE RFI due on May 18th and plan to respond to it.

II. The indroductory remarks were followed by tutorials on various topics which filled most of the remainder of the meeting.

  1. John Rundle - Quasi-static simulations of interactions of vertical faults in an elastic layer over a viscoelastic half-space. The simulations allow exploration over long timescales whereas data cover short timescales. Long term rates of slip are specified as constraints and a simple sliding friction law results in slip occurring when a stress criteria is exceeded.
  2. Seth McGinnis - Three-dimensional fault interactions and Green's functions. An extension of Rundle's method and generalization to dipping faults.
  3. Terry Tullis - Single fault instability models using rate and state friction, the effects of approximations in calculations, and fifth order Runga Kutta integration of partial differential equations with adaptive step-size.
  4. John Salmon - Fast multipole methods. Unneeded short wavelength parts of interactions are removed. For example, points far apart only see long wavelength interactions of the stress-field, therefore only a sparse matrix is required for distant points. Stress is smoothly varying and is thus well-suited to this type of approximation.
  5. Jay Parker - Three-dimensional viscoelastic finite element code and its relevance to GEM. Viscoelastic code with unstructured mesh generator, stress or slip criteria on faults, 3D structure and rheology. Useful for model validation and comparison to real data. Can be used for establishing boundary conditions, rheological parameters, and slip and stress rates for GEM simulations.
  6. Jin-Fa Lee - Three-dimensional finite element mesh generator and adaptive meshing technologies.
  7. Geoffrey Fox - GEM computational environment. Distributed object technology. Three-tiered compuing model: user -> web server -> program. Take advantage of existing technologies (e.g. XML, web technologies) to develop object based codes for running simulations. Need standardization of data, and codes to allow sharing of information and easy retrieval of data.

III. Actions: several committees were set up to begin work on GEM.

  1. Code assessment - Geoffrey Fox (chair), Ken Hurst, Andrea Donnellan Inventory existing codes, catalogue what they do in terms of inputs, output, and assumptions. Assess how well they will mesh with other codes. Develop standards for codes (e.g. should have a test case, documentation, proto-objects). Finish by July 1 deadline and submit article for earthquake physics AGU book.
  2. Simulation initiative - Terry Tullis and John Rundle (chairs), John Salmon, Jin-Fa Lee, Jay Parker, Yehuda Ben-Zion, Steve Ward, Jim Dieterich Begin developing general simulations using object oriented approach.
  3. Integration of data with simulations - Andrea Donnellan and Mike Heflin (chairs), Jay Parker, Jim Dieterich, Gilles Peltzer Develop data standards, explore exporting data to XML format for easy retrieval by codes.
  4. Computational environments and visualization - Geoffrey Fox (chair), Jay Parker, Susanna Gross Develop computational and visualization environment standards
  5. GEM program committee - John Rundle and Andrea Donnellan (chairs), Geoffrey Fox, Jay Parker, Bernard Minster, Tom Jordan Develop GEM program, respond to NASA RFI, pursue JPL DDF money, develop AGU special session on GEM.

To date, financial support in the amount of $100K has been provided this year by the NSF through the Southern California Earthquake Center. Conversations with NSF and SCEC scientists indicate that more support is likely to be forthcoming through this avenue, and that there is considerable interest in these approaches among participants in the planned new Earthquake Center. NASA is likely to be another major sponsor for the GEM effort. It became clear during the course of the workshop that surface geodetic data provide the constraints for GEM simulations and that GEM simulations are likely to drive future data collection efforts, making this fit well into the NASA solid earth and geodynamics program.

There are two major lines within the GEM effort.

  1. Development of algorithms and simulations including data assimilation
  2. Interface development including integration of data with the models.