Regulations often require structures such as sea defences, reservoir walls, oil rigs and tall buildings to be strong enough to withstand the worst conditions likely to be encountered during a specified period, usually longer than that over which observations have been made. Statistical methods give a rational approach to such extrapolation problems, not claiming clairvoyance but indicating the best use of past data and attempting to estimate the uncertainties in predictions. The talk will describe some environmental and engineering applications of extremal statistical theory.

Meeting contact: