Management of water resources under uncertainty: what does the future hold?

Current predictions indicate that the climate in Australia will become more variable and extreme. This will increase pressure on fragile natural resources and sound water resource management will become increasingly important. Forecasting is therefore essential in both agriculture and natural resource management. Is the current hydrological science up to this task and what tools are available for Australia? Much of the current hydrological knowledge is grounded in the work on perennial and humid systems with virtually no real ‘Australian hydrology’, despite evidence of the unique nature of our semi-arid river systems. While simulation models have become ubiquitous in hydrological science, most are used for so-called backcasting, rather than forecasting. This means simulations are run based on existing daily climate data as future daily predictions are often unavailable and uncertain. Additionally, there is still a significant gap with respect to the data needs of distributed models and the actually available natural resource data. Models with fewer parameters have been advocated, taking full advantage of the information in the available streamflow data. However, the associated simplifications and assumptions also introduce model uncertainties, which should be based on sound systems knowledge. It is unclear how such uncertainties would propagate if the models are used for forecasting rather than backcasting and perturbed under different management scenarios. Research into the temporal variability of model parameters, combined with probabilistic approaches