Detailed studies of natural hazards such as floods, droughts, tropical cyclones and earthquakes often require estimates of the frequencies of occurrence of particular extreme events However, the techniques for making these estimates are sometimes misunderstood and misapplied, particularly with regard to the forms of frequency distribution that are assumed or implied The most appropriate distribution and estimation technique depend on several factors, including the physical nature of the event, the available data and the intended application of the estimate. The present paper attempts to review these matters, giving attention to a range of techniques from simple graphical extrapolation to reasonably rigorous analytical procedures. It is pointed out that, no matter how sophisticated the technique, there is always some inaccuracy and uncertainty in such estimates. Attention is therefore also given to relatively simple methods of assessing the degree of uncertainty.
展开▼