Multi-wave calibration of computer models with application to the NEMO ocean GCM Seminar

Calibrating an expensive computer model, such as a climate model, is an extremely challenging and involved problem, yet it is one that some would claim is, in principle, solved. Application of the Kennedy and O’Hagan framework for probabilistic Bayesian calibration, allowing simultaneous tuning of input parameters and of systematic model discrepancy given a one-off space filling design in the input parameters and a set of observations of the real world, is ubiquitous and is, perhaps, considered “off the shelf”. In this talk I will highlight some of the known dangers and drawbacks of taking this approach and argue for an alternative multi-wave approach to calibration known as history matching. The multi-wave approach raises a number of design questions and we will examine some of these in the context of calibrating the NEMO ocean model.