Hydrologic Reference Stations

Analytical Methodology

The Bureau held a streamflow trend analysis workshop in December 2011 attended by hydrologists and water information experts from the Bureau of Meteorology, CSIRO, The University of Melbourne and Sinclair Knight Merz. The workshop:

  1. Explored the merits of the available fit-for-purpose scientific methods for trend analysis on long-term streamflow datasets.
  2. Developed the trend analysis methodology and identified the time frame for the delivery of products.

Outcomes of the trend analysis workshop identified the key streamflow variables and trend tests (as described in Chiew et al., 2005) that have been applied in the time-series analysis of streamflow data at each station. The key streamflow variables and statistical tests for the detection and identification of change are given in the table below.

Key streamflow variables and statistical tests
Analysis Component or Purpose Variable
Streamflow Variables
  • Annual total flow (Volume/year)
  • Daily maximum flow (Volume/day)
  • Q90 - 90th percentile daily flow per year (Volume/day)
  • Q50 - 50th percentile daily flow per year (Volume/day)
  • Q10 - 10th percentile daily flow per year (Volume/day)
  • Summer flow (Dec-Jan-Feb) (Volume/season)
  • Autumn flow (Mar-Apr-May) (Volume/season)
  • Winter flow (Jun-Jul-Aug) (Volume/season)
  • Spring flow (Sep-Oct-Nov) (Volume/season)
  • Percentage ceased to flow (percentage of the year during which there is no flow in the river)
  • Annual baseflow (Volume/year)
Test for independence (randomness)
  • Median Crossing (non-parametric)
  • Rank Difference (non-parametric)
Trend Test
  • Mann-Kendall test (non-parametric) including Sen's slope method
Test for step change in mean
  • The Distribution Free CUSUM (non-parametric) method
Test for difference in median in data periods
  • Rank-sum (median) (non-parametric)
Test for variability in daily flow duration between different decades
  • Linear and lognormal probability flow duration curves

Missing streamflow data was filled using the GR4J model (Perrin et al, 2003) which is a daily, lumped, four parameter rainfall-runoff model that accounts for soil moisture. Baseflow was separated from daily total streamflow using a digital filter based on theory developed by Lyne and Hollick (1979) and applied by Nathan and McMahon (1990).

It is important to note that the scope of the Hydrologic Reference Stations (HRS) project includes only the identification and attribution of decadal variability and change in long-term streamflow. Other variables such as rainfall, temperature and radiation are not included. Please see the Bureau of Meteorology web portal for high quality climate reference stations for further information on the long term trends in climate variables.

The Hydrological Reference Stations are now included in the Australian Hydrological Geospatial Fabric (Geofabric V2.1) within the Surface Network product. The catchment areas of Hydrologic Reference Stations were derived from the Geofabric, and were used for land use/forest cover change analysis and modelling. The station facts were also derived from Geofabric data.

Creative Commons By Attribution logo
Unless otherwise noted, all material on this page is licensed under the Creative Commons Attribution 4.0 International Licence