Recent Advances in Computational Geodynamics: from Large-scale Models to Magmatic Systems (Boris Kaus)

Most geological processes take place over long time and length-scales. Numerical models are helpful to simulate such (nonlinear) processes and compare models with (often sparse) data. Over the past few decades, large, parallel, community codes have driven significant advancements in computational geosciences (including software developed in Mainz). However, these monolithic codes, often written in low-level languages, optimized for CPU architectures, and comprising hundreds of thousands of lines of code, are difficult to modify and maintain. This hinders innovation, as much ongoing and future research requires incorporating additional physics or linking simulations with observations (for example, through adjoint-based inversion approaches). Here, I will give an overview of our recent work in developing modular software packages as a replacement for monolithic community codes. We apply these tools primarily to magmatic and volcanic systems, which involves wide ranges of scales and phase transitions and evolving chemistry. To connect simulations with observations and quantify model uncertainties, we use adjoint-based sensitivity kernels derived through automatic differentiation. Beyond volcanology, the same tools can be applied to industrial problems such as the long term safety underground repositories.