Operational modelling system development for short-term and seasonal streamflow forecasting services

The Bureau of Meteorology has been issuing seasonal probabilistic streamflow forecasts since 2010. Currently forecasts are provided for more than 100 sites every month, with plans to expand to approximately 200 sites by December 2015. In parallel, the Bureau has developed a new short-term forecasting service out to seven days, updated daily, at 100 sites across Australia. Three operational systems have been developed to support these services and are described in this paper. WAFARi (Water Availability Forecasts of Australian Rivers) is an extendable modelling system providing both an interactive modelling environment and a programming interface for seasonal streamflow modelling. Integrating kernels for the Bayesian Joint Probability and Bayesian Total Error Analysis, the system supports both, a statistical and dynamic modelling approach as well as data ingestion, rigorous cross-validation and web product generation. MSDM (Modified Statistical Downscaling Method) provides catchment-scale rainfall forecasts by downscaling seasonal rainfall forecasts from the POAMA (Predictive Ocean Atmosphere Model for Australia) global climate model. STFS (Short-Term Forecasting System) drives the short-term streamflow modelling with functionality to calibrate, verify and operate semi-distributed catchment models. The system has SWIFT (Shortterm Water Information Forecasting Tools) as the kernel for rainfall-runoff modelling and is integrated with the Bureau's next-generation flood forecasting platform: HyFS (Hydrological Forecasting System). In the near future, a new WAFARi component will merge the statistical and dynamic probabilistic forecasts to a single forecast. Upgraded MSDM will make use of high resolution climate forecasts available in 2017. STFS will be enhanced to produce ensemble streamflow forecasts using improved rainfall forecasts. The Bureau continues to invest in system development by adopting more advanced modelling methods, which will lead to increased forecast accuracy and more efficient service operations.