Development of a decision support approach for sustainable urban water management

Abstract

The challenges of climate change, increasing population and drought have motivated water authorities around Australia to deliver water services more sustainably. Many of them have recently devised long-term water strategies using multi-criteria decision aid (MCDA) to assess their strategies. This thesis identified four shortcomings in the current practice of MCDA used in urban water management: (1) double counting, (2) judgement uncertainty, (3) interaction and (4) range sensitivity. The aim of this research was to account for these shortcomings, thereby aiding decision makers in their planning of more sustainable water strategies.

A decision support approach was developed to address the shortcomings, and utilised: (1) value focused thinking, to carefully structure the criteria value tree; (2) fuzzy sets, to account for uncertainty in decision makers’ judgements on criteria weights and criteria scores; (3) the Choquet integral, to model interaction between criteria in conjunction with a linguistic preference elicitation technique; and (4) a novel technique to elicit criteria weights and to normalise criteria scores based on the concept of mitigation.

The developed framework was applied to an empirical case study, the Gold Coast Waterfuture project to illustrate the approach. This involved interviewing three groups of decision makers (DMs)—water users, water experts and water managers—to elicit their preferences in relation to different water strategies. The results obtained using the framework were closer to the decision makers’ preferences, as compared to the conventional which neglects the identified shortcomings.

The main finding was that the selection of an appropriate preference model is critically related to the DMs’ level of understanding of the decision problem. An averaging function can serve as a good initial approximation when no uncertainty is considered. More computationally demanding preference models which use fuzzy set theory and Choquet integral should only be used if the DMs have to make imprecise judgement with insufficient information and/or the DMs have a sufficient understanding of the interaction between the selected criteria. The practical application of this research is in providing decision makers with a stronger set of tools that include a modified and more rigorous MCDA which overcomes the shortcomings in the current practice.