IRTG Modern Inverse Problems (MIP)

Bayesian Model Selection for Complex Shallow Flow

13. Mai 2022 | von

Prof. Dr. Julia Kowalski


Shallow flow models are applied to a wide range of science and engineering fields including open channel hydraulics, weather forecasting, landslides in geo-hazard engineering, granular transport in chemical engineering, or coating processes in production engineering. Shallow flows have a much smaller height than length, which justifies depth-averaging yet comes at the price of losing vertical information. Many shallow flow flavors have been proposed in the past. However, it often is not obvious, which model candidate would be the best for a certain situation. In this project, we develop a Bayesian model selection approach to infer on the most plausible candidate process model given a collection of observational data. We will furthermore integrate goal-oriented Gaussian process emulation to increase computational feasibility.

The research goals of this project are:

  • Investigate Bayesian model selection approaches for complex shallow flow
  • Integrate goal-oriented Gaussian process emulation to increase computational feasibility
  • Investigate the impact of different observation data portfolios on selection result

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht.