Geological and geophysical data sets convey observations of physical processes governing the Earth’s evolution. Such data sets are widely varied and range from the internal structure of the Earth, plate kinematics, composition of geomaterials, estimation of physical conditions, dating of key geological events, thermal state of the Earth to more shallow processes such as natural and "engineered" reservoir dynamics in the subsurface.
The complexity in the physics of geological processes arises from their multi-physics nature, as they combine hydrological, thermal, chemical and mechanical processes. Multi-physics couplings are prone to nonlinear interactions ultimately leading to spontaneous localisation of flow and deformation. Understanding the couplings among those processes therefore requires the development of appropriate numerical tools.
Integrating high-quality data into physics-based predictive numerical simulations may lead to further constraining unknown key parameters within the models. Innovative inversion strategies, linking forward dynamic models with observables, and combining PDE solvers with machine-learning via differentiable programming is therefore an important research topic that will improve our knowledge of the governing physical parameters.
We invite contributions from the following two complementary themes:
#1 Computational advances associated with
- Alternative spatial and/or temporal discretisation for existing forward/inverse models
- Scalable HPC implementations of new and existing methodologies (GPUs / multi-core)
- Solver and preconditioner developments
- Combining PDEs with AI / Machine learning-based approaches (physics-informed ML)
- Automatic differentiation (AD) and differentiable programming
- Code and methodology comparisons (benchmarks)
#2 Physics advances associated with
- Development of partial differential equations to describe geological processes
- Inversion strategies and adjoint-based modelling
- Numerical model validation through comparison with observables (data)
- Scientific discovery enabled by 2D and 3D modelling
- Utilisation of coupled models to explore nonlinear interactions
The research output presented in this session can be submitted to the ongoing Special Issue (SI) in the EGU journal of Geoscientific Model Development (GMD): https://www.geoscientific-model-development.net/articles_and_preprints/scheduled_sis.html
The complexity in the physics of geological processes arises from their multi-physics nature, as they combine hydrological, thermal, chemical and mechanical processes. Multi-physics couplings are prone to nonlinear interactions ultimately leading to spontaneous localisation of flow and deformation. Understanding the couplings among those processes therefore requires the development of appropriate numerical tools.
Integrating high-quality data into physics-based predictive numerical simulations may lead to further constraining unknown key parameters within the models. Innovative inversion strategies, linking forward dynamic models with observables, and combining PDE solvers with machine-learning via differentiable programming is therefore an important research topic that will improve our knowledge of the governing physical parameters.
We invite contributions from the following two complementary themes:
#1 Computational advances associated with
- Alternative spatial and/or temporal discretisation for existing forward/inverse models
- Scalable HPC implementations of new and existing methodologies (GPUs / multi-core)
- Solver and preconditioner developments
- Combining PDEs with AI / Machine learning-based approaches (physics-informed ML)
- Automatic differentiation (AD) and differentiable programming
- Code and methodology comparisons (benchmarks)
#2 Physics advances associated with
- Development of partial differential equations to describe geological processes
- Inversion strategies and adjoint-based modelling
- Numerical model validation through comparison with observables (data)
- Scientific discovery enabled by 2D and 3D modelling
- Utilisation of coupled models to explore nonlinear interactions
The research output presented in this session can be submitted to the ongoing Special Issue (SI) in the EGU journal of Geoscientific Model Development (GMD): https://www.geoscientific-model-development.net/articles_and_preprints/scheduled_sis.html