Work Plan
Immediate, Urgent Goals
SOFTWARE DEVELOPMENT
- Continue PyLith development (For more details see the complete, current PyLith Development Plan).
- Establish interaction with the Seismology Working Group on meshing issues (e.g., keep up-to-date on the development of Geo-CUBIT).
COMMUNITY ACTIVITIES
- Provide training via virtual workshops
- We have another in-person workshop scheduled for June 2014. It will coincide with the release of PyLith 2.2 and will focus on intermediate and advanced topics. We will solicit ideas from users on additional online training.
- Continue development of a PyLith wiki to complement the cig-short email list and PyLith manual.
- Continue series of in-person workshops on biannual basis (even years).
Intermediate Goals
SOFTWARE DEVELOPMENT
- Inversion framework for geodetic, seismic, and combined inversions.
- We envision plug-ins for various inversion methods, including Bayesian approaches.
- Need a scientist to lead this effort. Responsibilities would involve getting funding, assembling a development team (might include a CIG developer), and overseeing software development.
SCIENTIFIC QUESTIONS
- Observationally constrained and internally consistent physics for the entire seismic cycle
- Resolve the entire seismic cycle in simulations that capture interseismic deformation, rupture nucleation and propagation, and postseismic deformation with realistic Earth models (geometrical complexity, material heterogeneity, and inelastic rheologies). Constraints on fault and bulk rheologies that are consistent with extensive geodetic, seismic, and geologic observations are critical to understanding the behavior of fault systems and improving the accuracy and precision of earthquake hazard assessments.
- Observationally constrained and internally consistent physics for tectonics of magmatic systems, geothermal systems, the cryosphere, and induced seismicity from fluid injection
- Integrate modeling tectonic processes with heat and fluid flow, thereby enabling complex rheologies with temperature dependent parameters. Incorporating heat and fluid flow into tectonic modeling significantly expands the range of problems that can be addressed (such as seismic tremor in geothermal areas) and permits direct application of additional geophysical constraints. Viscoelastic, elastoplastic, and viscoplastic rheologies are important for bridging between seismic and tectonic time scales.
- Observationally constrained modeling of crustal deformation associated with surface loads
- Constrain the bulk rheologies of the crust using geodetic and geologic observations of deformation arising from glacial rebound, reservoir impounding, and other surface loads.
POTENTIALLY RELEVANT COMPUTATIONAL TECHNIQUES
Need to assess the applicability and implications of using currently available and emerging computational techniques for earthquake modeling. Techniques may impose undesirable limitations on the geometry of the domain (e.g., topography) and faults or may introduce severe ill-conditioning of the system.
- Adaptive mesh refinement (e.g., deal.ii and p4est)
- Efficiently resolve evolving small length scale features through local refinement and coarsening of the mesh.
- Data assimilation and inversions
- Data assimilation aids in quantifying the uncertainty in parameters based on observations.
- Multi-scale techniques
- Introduction of multiple spatial and temporal scales through homogenization. Resolution of multiple time scales through slow/fast timescale coupling.
OTHER TOOLS
- Workflow management
- Streamline problem workflow using tools to manage inputs and outputs of the various stages of modeling (creating the geologic model, meshing the domain, simulating the physics, and post-processing the results).
- Establish community benchmarks for problems that cannot be solved by current software.
COMMUNITY BUILDING
- Software development
- Must find the proper balance between providing software that is accessible to new users but also provides the flexibility and extensibility required by expert users.
- Subcontracts for scientific driven cutting-edge development of community codes
- Provide funding for expert users to work with code developers to add new features to community codes in order to solve specific research problems.
- Training
- Regular, multi-day workshops are essential for training the community in the use of state-of-the-art modeling codes and tools.
- Additional complementary training is needed to provide different levels of training.
- Short workshops at larger scientific meetings to expand the user base
- Extended visits by computational scientists and software engineers to expert users and earth science developers and vice versa. This could be implemented via some form of travel grants for in-depth training.
- Focused online training for new releases, common problems, and introduction to advanced features
2014
CRUSTAL DEFORMATION MODELING WORKSHOP, JUNE 23-27, STANFORD UNIVERSITY
More information coming soon.
PYLITH DEVELOPMENT
For more detailed information, see the complete, current PyLith Development Plan.
- Version 2.0 (early March 2014)
- Replace C++ Sieve implementation of finite-element data structures with C DMPlex implementation.
- Switch from using Subversion to Git for version control.
- Add ability to recursively refine a mesh
- Version 2.1 (June 2014)
- . Improve fault formulation for spontaneous rupture
- Version 2.2 (Summer/Fall 2014)
Past Activity
2013
CRUSTAL DEFORMATION MODELING TUTORIAL, JUNE 24-28 This online tutorial, offered via Adobe Connect, provided training in the use of PyLith, CUBIT, and ParaView over six two-hour tutorial sessions. The sessions were offered at two different times to accommodate a wide variety of time zones. They focused on intermediate and advanced topics with beginner material provided by the June 2011 online tutorial. The sessions were recorded and archived for on-demand playback. See the PyLith Wiki to access the archived sessions.
PYLITH DEVELOPMENT Version 1.9 was released June 20. This release added a Newton-Raphson algorithm for spontaneous rupture simulations with explicit time-stepping to improve the convergence of the fault constitutive model, and fixed several bugs. Most development focused on version 2.0.
2012
CRUSTAL DEFORMATION MODELING WORKSHOP, JUNE 18-22, COLORADO SCHOOL OF MINES The 2012 Crustal Deformation Modeling workshop was held June 18-22 on the campus of Colorado School of Mines in Golden, CO. This meeting continued an ongoing series of workshops held over the past 10 years. The focus of these workshops is physically based models of the distribution of lithospheric stress in space and time via simulation of the strain accumulation, dynamic rupture propagation, and postseismic relaxation over multiple earthquake cycles. The workshop included participation from 62 scientists. As in previous workshops in this series, nearly two-thirds of the participants were graduate students and postdocs. In this workshop 44% of the participants were graduate students, 21% were postdocs, 18% were faculty, and 18% were researchers. About 80% of the attendees participated in both the tutorials on the first two days of the workshop (Monday and Tuesday) and the science talks and discussions over the following two and one-half days (Wednesday, Thursday, and Friday). For more details, see the workshop web page.
PYLITH DEVELOPMENT PyLith development continues to be the main focus of the working group activities. The PyLith development team completed two releases with new features as well as a bugfix release since June 1, 2012. These releases provided (1) a user-friendly interface for computing static Green’s functions for use in geodetic inversions with lateral variations in Earth structure, (2) the ability to output the displacement or velocity interpolated to arbitrary locations within the domain, (3) support for PETSc GPU solvers, (4) spatial and temporal variations in tractions for spontaneous earthquake rupture simulations, (5) a post-processing utility to compute fault slip information, such as seismic potency, seismic moment, and moment magnitude directly from PyLith fault output, (6) computation of the stable time step for explicit-time stepping based on the CFL condition and distortion of the finite-element cells from their ideal shapes, and (7) new CUBIT meshing example illustrating how to use a field over the domain to specify the discretization size. The releases fixed over ten minor bugs identified through benchmarking, full scale tests, and user feedback.