Development of a national 7-day ensemble streamflow forecasting service for Australia

Abstract. Reliable streamflow forecasts with associated uncertainty estimates are essential to manage and make better use of Australia's scarce surface water resources. Here we present the development an operational 7 d ensemble forecasting service for Australia meet growing needs users, primarily river managers, probabilistic support their decision making. We test modelling methodology 100 catchments learn characteristics different rainfall from Numerical Weather Prediction (NWP) models, effect statistical processing on forecasts, optimal size, parameters a bootstrapping technique calculating forecast skill. A conceptual rainfall–runoff model, GR4H (hourly), lag route channel routing model that in-built in Short-term Water Information Forecasting Tools (SWIFT) hydrologic package used simulate input potential evaporation. The catchment pre-processor (CHyPP) is calibrating error reduction representation stages (ERRIS) reduce hydrological errors quantify uncertainty. Calibrating raw CHyPP efficient method significantly bias improve reliability up lead days. demonstrate ERRIS improves skill Forecast skills highest temperate perennially flowing rivers, while it lowest intermittently rivers. sensitivity analysis optimising number members shows more than 200 needed represent show block size sensitive calculation. 1 month recommended capture maximum possible benchmark criteria accepting locations public service. Based criteria, 209 out 283 selected hydro-climatic regions across service, which has been since 2019, provides daily updates graphical tabular products along performance information,