A Bayesian network approach to support environmental flow restoration decisions in the Yarra River, Australia

Many rivers in Australia and elsewhere around the world are flow stressed. There is now considerable activity to restore crucial components of the natural flow regime of such rivers-known as environmental flows, but methods underpinning the decision-making often lack rigour and transparency. The restoration of environmental flow regimes is usually determined using historic flow data, available expertise, sometimes ad-hoc compromise between conflicting and spatio-temporally variable water needs, and then implemented through special environmental allocations or water sharing plans. However, ecological flow requirements are poorly integrated into the decision-making process. This paper reports an eco-hydrological Bayesian network model developed to assist in risk assessment and decision-making on improving the flow conditions for Australian Grayling (Prototroctes maraena) in the Yarra River in south-eastern Australia. The Yarra environmental flow model was adapted from previous work on Australian Grayling in the Latrobe River system, and contained the same ecological model as for the Latrobe River, but with a new hydrological sub-model developed specifically for the Yarra River. The model was used to assess the relative ecological benefits (to Australian Grayling) of restoring specific components of the natural flow regime, in particular autumn freshes and spring freshes.