Statistics Seminars: Dynamic emulation for structured chaotic time series, with application to large climate models
29 October 2012 14:00 in CM221
In this talk I will develop Bayesian dynamic linear model Gaussian processes for emulation of time series output for computer models that may exhibit chaotic behaviour. The statistical technology is particularly suited to emulating time series output of large climate models that exhibit this feature and where we want samples from the posterior of the emulator to evolve in the same way as dynamic processes in the computer model do. I'll apply this methodology to emulating the Atlantic Meridional Overturning Circulation (AMOC) as a time series output of the fully coupled non-flux-adjusted atmosphere-ocean general circulation model, HadCM3, and illustrate some methods of obtaining prior judgements required to build the emulator when working with a large ensemble of runs of a climate model. I will present a block metropolis-within-gibbs MCMC algorithm to obtain posterior samples for the parameters and discuss the value of such an analysis when an emulator is eventually adopted as part of wider analyses.
Contact email@example.com for more information