Reliable estimates of expected extreme flood events are required for design and operation of vital infrastructure such as flood defenses, bridges and culverts, and also for more general flood risk management and planning, e.g. emergency planning, flood risk mapping, and for defining flood insurance premiums. In practice, this information is obtained through the use of flood frequency estimation techniques based on the principle of analyzing series of observed events to infer a probabilistic behavior, which is then extrapolated to provide estimates of the likely magnitude of future extreme events (e.g. the magnitude of the flood expected to be exceeded on average once every 100-year).
By nature, extreme flood events are seldom observed locally and hydrologists have little or no chance of gathering an adequate sample of catastrophes. This raises the question of how best to extrapolate to extreme events when no or only short se-ries of recent events are available. Where methods do exists they are often simple and their ability to accurately predict the effect of environmental change (e.g. urbanization, land-use change, river training and climate change) is unknown. Also, the problem of consistent estimates of extreme floods for trans-boundary rivers is rarely considered.