Bayesian modelling for risk-based environmental water allocation
About the project
This sub-project, funded under the $641,000 Water Dependent Ecosystems Discussion Starters Part 1 project, reviewed the application of Bayesian models for use in the determination and management of environmental flow allocations.
A Bayesian belief network, in short, is simply a way of showing how things interact and cause specific outcomes.
Bayesian belief models are decision-aiding tools, based on conditional probabilities. In the context of aquatic ecosystems, Bayesian belief models aggregate the risks posed to critical aspects of the life cycle (e.g. recruitment) to assess the overall risk to the long-term viability of the asset. Once the risk is known for a particular flow scenario, it is then possible to specify ecological outcomes that provide for/protect the water requirements of identified ecological assets at an agreed level of risk.
The advantages of Bayesian belief networks is that they can explicitly lay out parameters in a decision process, they can explicitly depict uncertain possibilities, and can be used to identify key factors that most influence some outcome of interest, to help prioritise monitoring or research.
The project was undertaken by Water Science Pty Ltd and the Australian National University.
Commenced: 1 March 2008
Completion: 1 February 2009
Project benefits
The proponents of this project conducted a workshop on Bayesian belief modelling and risk based water allocation for aquatic ecosystems. They identified key issues related to Bayesian belief modelling in the ecosystem context, and developed an appropriate outline for future work on this topic. The key output from this project was a Waterlines report that serves as a discussion starter on this topic. This report is part of a series of papers commissioned on issues relating to Australian aquatic ecosystems.