Our contribution to the KI:STE project combines the geohazards modeling expertise of the Computational Geoscience group at the Geoscience Centre at Georg-August University of Göttingen led by Prof. Julia Kowalski, and its associated Geofluiddynamics group at the AICES Graduate School at RWTH Aachen. More generally, our research is centered around developing physics-based and data-driven computational models to predictively model complex geosystems such as earth surface or cryospheric phase-change processes. We formulate process-based mathematical models, develop tailored numerical solution techniques, and investigate strategies for physics-informed meta modeling as well as its application to conduct compute-intensive tasks, suchas uncertainty quantification, Bayesian parameter estimation and model selection as well as optimal experimental design.
A special focus lies on approaching cryosphere physics research questions by means of computational models and on developing model-based decision support tools to assess and mitigate geohazards and climate change impact.
Our research project affiliated to KI:STE, is located in the area of the latter. This is particularly important, as geohazards such as landslides and avalanches pose risks to society and infrastructure worldwide. In many regions, these risks are likely to even increase due to climate change and continued urbanization. It is our vision to provide geohazard practitioners with physics-based and data-driventools that eventually will enable them to conduct a reliable, transparent, and adaptable risk assessment as well as to plan sustainably mitigation measures. In the past, we worked extensively on the development of computational methods to simulate the impact of rapid gravity-driven mass movements such as landslides and avalanches. For these processes, we investigate fundamental mathematical models and develop predictive computational models for simulation scenarios in complex terrain. We are furthermore investigating and developing computationally feasible approaches (based on combining forward-simulations with meta modeling techniques) to conduct uncertainty quantification, sensitivity analysis, parameters estimation, model selection and optimal designin the context of geohazards. Our work for KI:STE builds on and will continue this previous work, supplemented by scientific insights from machine learning methods. Specifically, it aims at the development of a generic workflow for adynamic hazard mapping process based on data science.
Author: Ann-Kathrin Edrich RWTH Aachen