The performance of a new ensemble flood and short-term streamflow forecasting system for Australia
David Robertson, JC Bennett, Durga Lal Shrestha, Quan J. Wang, DJ Enever, Hap Hapuarachchi, NK Tuteja
Abstract
David Robertson, JC Bennett, Durga Lal Shrestha, Quan J. Wang, DJ Enever, Hap Hapuarachchi, NK Tuteja
Abstract
Forecasts of streamflows with lead times up to 10 days can support optimal management of water resources for consumptive and environmental purposes and inform emergency management agencies of impending floods. At present, most flood forecasting services in Australia are based on event models which are limited to providing forecast information as events are in progress. CSIRO and the Bureau of Meteorology have collaboratively developed a new ensemble flood and short-term streamflow forecasting system for Australia that integrates real-time observations of rainfall and streamflows with post-processed NWP model rainfall forecasts and continuous hydrological modelling. The system produces streamflow forecasts for the entire hydrograph with lead times of up to 10 days. Uncertainty in streamflow forecasts is represented using ensembles. The forecasts produced by the new ensemble short-term streamflow forecasting system are verified for 10 catchments covering a range of Australian climatic and hydrological conditions. We evaluate forecasts produced by running the forecasting system at several different time steps and find that a one hour time step produced forecasts with the lowest errors, particularly at short lead times. The streamflow forecasts are most skilful for the perennial streams of southern and eastern Australia and display skill for lead times of many days. The reliability of the forecast ensemble can be improved, particularly at short lead times. A pilot deterministic short-term forecasting service for the 10 catchments described in this paper is being established by the Bureau of Meteorology and will soon be accessible to interested users.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Forecasts of streamflows with lead times up to 10 days can support optimal management of water resources for consumptive and environmental purposes and inform emergency management agencies of impending floods. At present, most flood forecasting services in Australia are based on event models which are limited to providing forecast information as events are in progress. CSIRO and the Bureau of Meteorology have collaboratively developed a new ensemble flood and short-term streamflow forecasting system for Australia that integrates real-time observations of rainfall and streamflows with post-processed NWP model rainfall forecasts and continuous hydrological modelling. The system produces streamflow forecasts for the entire hydrograph with lead times of up to 10 days. Uncertainty in streamflow forecasts is represented using ensembles. The forecasts produced by the new ensemble short-term streamflow forecasting system are verified for 10 catchments covering a range of Australian climatic and hydrological conditions. We evaluate forecasts produced by running the forecasting system at several different time steps and find that a one hour time step produced forecasts with the lowest errors, particularly at short lead times. The streamflow forecasts are most skilful for the perennial streams of southern and eastern Australia and display skill for lead times of many days. The reliability of the forecast ensemble can be improved, particularly at short lead times. A pilot deterministic short-term forecasting service for the 10 catchments described in this paper is being established by the Bureau of Meteorology and will soon be accessible to interested users.
Key concepts: Streamflow, Flood forecasting, Lead time, Meteorology, Environmental science, Consensus forecast, Flood myth, Forecast period