Seasonal Streamflow Forecasts
Historical and probability distribution for Swan Creek at Swanfels
Return to catchment list
Product list for Swan Creek at Swanfels
Download forecast data
|Average (1950+) (GL)||Last year (2012) (GL)||Observed (2013) (GL)||Minimum (2007) (GL)||10 yr average (2003+) (GL)||Maximum (1971) (GL)|
- The minimum and maximum are determined from the total for the 3 month period.
(3 month total flow in GL)
|Histogram Bin of Historical Reference (GL)||Frequency of Historical Reference (%)|
About the historical and probability distribution product
- The forecast is a simulation from the Bayesian Joint Probability (BJP) Model. The simulation comprises 5000 members.
- The forecast skill of the BJP model is different for different forecast periods of the year. Please look at the Model Validation results to assess model skill for this forecast period.
- The historical reference is a probabilistic representation of the historical data.
- The forecast location name is displayed in the graph title. Site forecast locations are followed by the Australian Water Resources Council (AWRC) river station number in brackets e.g. (405219).
- The streamflow data used for new forecasts and for verification is from realtime data sources which are not rigorously quality controlled.
- The historical reference plot and the historical data plots are derived from the available historical record. The legend shows the year associated with the start month of the forecast period.
- The RMSEP skill scores have been defined as less than 10 very low skill, 10-20 low skill, 20-40 moderate skill, greater than 40 high skill.
- The 25% quartile, also defined as the first quartile cuts off the lowest 25% of data. The 75% quartile, also defined as the third quartile cuts off the lowest 75%. The interquartile range, also called the middle fifty, is a measure of statistical dispersion, being equal to the difference between the third and first quartiles.
- The statistics are not generated from the complete ensemble of historical reference distribution. The top 1% of ensemble members are removed for plotting purposes and some difference may be observed between the means from the observed streamflow record and the historical reference.
- Further explanation of some technical terms is provided under the FAQ.
Unless otherwise noted, all material on this page is licensed under the Creative Commons Attribution Australia Licence