Application of the open-source mantle convection code ASPECT to long-term tectonic simulations
Abstract
The open-source mantle convection code ASPECT (Computational Infrastructure for Geodynamics) provides a robust foundation for numerically examining a wide range of geodynamic processes. ASPECT's strength arises from its massive scalability, use of AMR (adaptive mesh refinement), advanced solvers, community support and modular design, the latter of which permits straightforward modification of discrete components of the code, such as constitutive relationships. Here, we present preliminary work that uses such adaptions to model long-term tectonic problems (lithospheric and upper mantle dynamics) with ASPECT. Crustal- and lithospheric-scale deformation is modeled using a recently implemented visco-plastic constitutive relationship, with viscous flow characterized by diffusion and/or dislocation creep and plastic failure following a Drucker-Prager yield criterion model. Distinct compositional layers and their associated material properties are tracked through field- or tracer-based methods, which also allow tracking of time-dependent properties such as accumulated strain. Preliminary 2-D models of long-term (> 10 Myr) continental extension successfully reproduce results of prior studies with rift basin structure largely dependent on initial lithospheric structure, extension velocity and strain-softening parameterization. Building on these preliminary results, our presentation will focus on ASPECT's viability and performance for modeling long-term tectonics within the context of 2-D vs. 3-D simulations, CPU scaling, distinct solver methods, adaptive mesh refinement and field- vs. tracer-based methods. Our primary goal in addressing these topics is to highlight ASPECT's current functionality and address key areas of future development associated with modeling long-term tectonics. Additionally, we hope to spur discussion regarding a long-term tectonics benchmark that addresses the strong resolution-dependence of simulations using plasticity formulations based on the Drucker-Prager yield criterion.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2016
- Bibcode:
- 2016AGUFMDI23A2586N
- Keywords:
-
- 1932 High-performance computing;
- INFORMATICSDE: 1990 Uncertainty;
- INFORMATICSDE: 3225 Numerical approximations and analysis;
- MATHEMATICAL GEOPHYSICSDE: 3260 Inverse theory;
- MATHEMATICAL GEOPHYSICS