Advances in probabilistic hydrological modelling, including in ephemeral catchments

Hydrological models are widely used as engineering tools to help anticipate events such as floods, future water availability, and so forth. This talk will review advances in the use of probabilistic methods to characterise the uncertainty in streamflow predictions, with a particular focus on ephemeral catchments, where flows cease during certain times of the year. We consider the use of streamflow transformations in conjunction with censored Gaussian distributions for the model residual errors. Tradeoffs between reliability, precision and bias of the probabilistic streamflow predictions are reported. Results and case studies presented will include outcomes from the collaboration with the Australian Bureau of Meteorology on seasonal streamflow forecasting.