Bayesian calibration and emulation of geophysical computer models''
Monday, November 28, 3:30PM – 5PM
Serge Guillas, Department of Statistical Science, University College London
TBAIn this talk, we demonstrate a procedure for calibrating and emulating complex computer simulation models having uncertain inputs and internal parameters, with application to the NCAR Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM), and also illustrate further findings for Computational Fluid Dynamics (CFD) and landslide-generated tsunami wave modeling. In the case of TIE-GCM, we compare simulated magnetic perturbations with observations at two ground locations for various combinations of calibration parameters. These calibration parameters are: the amplitude of the semidiurnal tidal perturbation in the height of a constant-pressure surface at the TIE-GCM lower boundary, the local time at which this maximizes and the minimum night-time electron density. A fully Bayesian approach, that describes correlations in time and in the calibration input space is implemented. A Markov Chain Monte Carlo (MCMC) approach leads to potential optimal values for the amplitude and phase. Enhancements in terms of sequential design of experiments are shown for the calibration of key parameters, and emulation, of the k-epsilon CFD model for flows in street canyons. In terms of emulator speed, the so-called outer product emulator avoids the use of MCMC to enable fast tsunami wave modeling for real-time warnings according to uncertain speed, position and shape of the landslide.
Hosted by Omar Ghattas