P9: Model adequacy under uncertainty

Supervisors

Prof. Barbara Wohlmuth, Dr. Laura Melas

Advisors (int.)

Dr. Bernhard Schuberth (LMU)

Advisors (ext.)

Prof. Georg Stadler (New York University)

Description

Several model uncertainties will affect the ability to retrodict mantle flow induced vertical motions of the lithosphere and will be considered in the context of the proposed RTG. They include the linkage of seismic tomography models (P5) and geodynamic Earth models (P6) which involves uncertain present-day state estimates of mantle heterogeneity due to tomographic filtering effects and conversions of seismic velocity anomalies into mantle density variations. They also include uncertainties in geodetically and geophysically derived estimates of the density structure and corresponding isostatic response of the lithosphere (P8). We will seek a systematic understanding of the various sources of uncertainties, for example, through sensitivity analyses for key modelling parameters, twin experiments and the introduction of probabilistic considerations. A systematic uncertainty quantification, from the seismic imaging process over geodynamic and lithospheric responses to geologically assessable predictions of vertical lithosphere motion, will provide important understanding on the propagation and evolution of errors in geodynamic retrodictions.

Main objectives

• Formulation of a forward uncertainty quantification model: from geodetic observations to a stochastic geodynamic process model
• Development of a probabilistic model of coupled physical processes on different scales
• Model validation and evaluation of model adequacy
• Investigating uncertainties and providing a sensitivity analysis, e.g., for tomography, plate reconstruction, mineralogical mapping
• Provision of multiple physical states for the geodynamic Earth model and quantification of uncertainties for the mantle flow evolution
• Formulation of a backward uncertainty quantification problem: application of active subspace strategies, and Bayesian techniques  

RTG coupling

Input: 

• Uncertainty quantification of geodetic observations (P1, P2)
• Uncertainty estimation from geologic data (P3, P4)
• Geodynamic mantle heterogeneity structure from P5
• P6 model trajectory.

Output:

• Forward uncertainty estimates to P5
• Backward uncertainty estimates to P6
• Uncertainty quantification of integrated process model (P6, P7, P8)