Streamflow variability, seasonal forecasting and water resources systems

The interannual variability of streamflow in Australia is greater than elsewhere in the world. Reliable forecasts of streamflow would go a long way towards improving the management of water resources systems by enabling them to cope better with the high inter-annual variability. In the first part of this paper, a clear link between runoff in Australian catchments and ENSO is established using lag-correlation analysis. This suggests that runoff in many areas in Australia can be forecast over certain seasons with some success using indicators of ENSO and the serial correlation in runoff. The second part of the paper presents statistical methods for forecasting streamflow. Probabilistic methods for forecasting categorical values of streamflow are well developed, although their use is limited by the short historical data and the varying strength in the streamflow-ENSO relationship over the last century. There have been far fewer studies on forecasting probabilities of exceedance of streamflow amounts, although these forecasts are required to evaluate the benefits and risks of using ENSO-based forecasts for different levels of conservatism in water resources management. The third part of the paper discusses the present and potential uses of seasonal streamflow forecasts for managing urban water and rural irrigation systems.