Newcastle University

Uncertainty is an inescapable aspect of ESE decision-making because of the complexity of the systems involved, the time-scales of appraisal and the fundamental role of human agency in system behaviour. Uncertainty analysis therefore forms a large component of the decision-support research theme, with a focus upon treatment of severe uncertainty in complex decision problems. We focus upon methods for robust decision making under conditions of severe uncertainty and in the presence of imprecise information. Sensitivity analysis permits us to identify the most critical influences on the management outcome of a system, while scenario analysis allows a direct comparison between potential management options.

Of fundamental importance is the problem of assimilating observations into simulation models, through a process of calibration. The Bayesian version of this calibration problem is therefore another focal point of our research on uncertainty analysis. This has been achieved for calibration of steady-state models with spatial data; work is continuing on calibration of time-varying models.

The use of uncertainty analysis has been demonstrated in a number of applications; for example in studies on the flood risk management of the Thames estuary under conditions of uncertain long term sea level rise and economic development, and in studies incorporating the uncertainty inherent in climate predictions into assessment of water resources for London. Demonstration of the utilisation of the UKCP09 climate predictions is continuing.