Loss models for design flood estimation : toward applications within a Monte Carlo environment

Abstract

This thesis focusses on more holistic approaches to flood modelling. For instance, joint probability/Monte Carlo approaches have received a great deal of attention in recent years, particularly since their use was advocated in Australian Rainfall and Runoff (ARR) 2015. In Monte Carlo simulation, rainfall-runoff model inputs are described by probability distributions, rather than fixed inputs as previously carried out with the Design Event Approach (DEA). There is no compelling evidence of the most appropriate loss model for flood estimation within a Monte Carlo framework. Furthermore, the functional form of key stochastic variables is in dispute. Previous recommendations on loss models were made for traditional techniques; however, since the advent of Monte Carlo simulation the most suitable loss model has not been rigorously assessed. Additionally, while many have investigated the probability density function of key parameters, there is some dispute over the most suitable model. Therefore, due to the lack of guidance currently available, models of runoff generation processes and parameter variability require thorough investigation for more rigorous rainfall-runoff modelling within a Monte Carlo environment.

Date of Award	2016
Original language	English