2002 – Use of a Monte Carlo Framework to Characterise Hydrologic Risk

R.J. Nathan, P.E. Weinmann and P.I. Hill

Current practice for estimation of design floods is typically based on the “design event” approach, in which all parameters other than rainfall are input as fixed, single values. Considerable effort is made to ensure that the single values of the adopted parameters are “AEP-neutral”, that is, they are selected with the objective of ensuring that the resulting flood has the same annual exceedance probability as its causative rainfall. While this approach represents current best practice in Australia (and overseas), it does suffer from a number of limitations.

This paper describes the development and application of a Monte Carlo (or joint probability) framework which offer an alternative to the design event method. This technique recognises that any design flood characteristics (e.g. peakflow) could result from a variety of combinations of flood producing factors, rather than from a single combination. The approach mimics “mother nature” in that the influence of all probability distributed inputs are explicitly considered, thereby providing a more realistic representation of the flood generation processes.

The advantages of the technique are illustrated by application to a hypothetical dam located on a real catchment. The manner in which standard design inputs are incorporated are discussed, as is the relationship of the approach to current guidelines.


Want a discount? Become a member.