Murray-Darling Basin flows: #1 The data

Start with the data, always.

There’s been more rubbish written and spoken about the what and why of our greatest river system than most topics get. As usual, I’m not very interested in what people have to say, even in the competing expert¹ reports². I want to see what the data says.



Before we get to the river flows, I want to share just one figure from the first of those reports¹. Folks are inclined to say it’s all due to drought, or variations of that. Obviously drought affects flow, but drought is not the only effect, nor is it as pervasive as many assume. This is weighted average rainfall across the Darling River system (the ‘northern basin’), with important droughts marked:

The caption there is confusing. The dotted black line is the 5-year moving average, plotted trailing. The red line is horizontal. It shows the 5-year average rainfall to the end of 2017, for comparison with the whole record (it turns out to be just a little below the whole-of-record average).

Recent years have not been exceptionally dry in the Darling catchment, except for 2017, which had the fourth-lowest rainfall in the 128-year record. Last year (2018) was also very dry, but is not included in that dataset. Here is the Australian Bureau of Meteorology’s similar plot for the whole Murray-Darling Basin, including for 2018 (the 5-year moving average here is plotted centred, and the chart is not zero-scaled):


Flow measurement

Streamflow is extensively measured across the Murray-Darling basin, and has been for decades — for well over a century in places. Before we get to the numbers, it’s worthwhile understanding how.

Most streamgauges do not measure flow; they measure water level. Traditionally that was done by a lackey walking down to the river each morning and reading a staff gauge. That still happens in many places, but the water level is usually also recorded by some device: originally a submerged gas bubbler tube, pressure gauge and clockwork chart recorder; often now by similar with a digital recorder, or by a submerged pressure sensor, or by an overhead radar, or … there are many ways.

Darling River at Wilcannia, staff gauges — WaterNSW pic

Water level is converted to flow using a rating curve, established over many years by making physical flow measurements at the gauge. Again, traditionally that was done using a mechanical propeller meter to measure flow velocity across a surveyed river cross-section, working from a small boat (scary in a big flood), or from a flying fox, or at low flows just by wading. You multiply each velocity measurement by its grid square of cross-sectional area, then sum all of those to get the total flow. These days there are better ways (‘acoustic Doppler current profilers‘), but the old manual methods are still used.

To be useful, these physical gaugings need to be done many times at many different flow levels, so that a suitable average curve can be fitted (often aided by hydraulic calculation). Some fits are better than others:

Darling River at Wilcannia — rating curve

Obviously, if the river channel changes (through scour or accretion) then the rating curve will change too, and the hydrographer will need to take that into account when computing flows. Ideally one chooses a site for a gauge just upstream of some feature that won’t change, like a rock bar or weir. That’s called the control.

Accurate river flow measurement requires knowledge and skill.


Flow data

The Bureau of Meteorology provides a useful national streamflow and water quality data aggregation service, but it only ingests daily and shorter-duration flow data contained in the source databases (mostly run by the states and territories). There is often lots more older data that hasn’t made it to the front of those system.

The Murray-Darling Basin Authority also runs a useful data portal for River Murray data, with links to flow records.

We really need to venture further up the data chain. There are some local government gauges, but most are run by the states, which all maintain hydrological databases with web portals³:

Here’s WaterNSW:

WaterNSW river flow database web portal

You can navigate to major gauges via the summary map. Failing that, you need to browse the list under ‘Real Time Data — Rivers and Streams’. Here’s the Darling River at Wilcannia data page and flow data plot:

WaterNSW Darling River at Wilcannia gauge 425008

Wilcannia ‘Plot of Continuous Height and Flow Data’

The important thing to notice there is that’s just the plot of the available daily (and shorter) interval data, meaning the data that has been digitally recorded or digitised from chart recorder plots or hand-written record sheets. There is often more data.

Here is the ‘Period of Record Daily Flow Summary Report’ for Wilcannia, with monthly flow data back to 1913:

Wilcannia ‘Period of Record Daily Flow Summary Report’

It turns out there’s lots of old data. The Menindee town gauge has data back to 1881 (Burke and Wills left there on 19 October 1860!). Of course data quality is a concern. Care is needed, but it’s a mistake to assume that the quality of old data is necessarily poor. People had more time a century ago; knew well how to do this; may often have done it well.

That’s enough for now. In the next installment we’ll look at some actual data, and simple analysis.



  1. Investigation of the causes of mass fish kills in the Menindee Region NSW over the summer of 2018–2019‘, Australian Academy of Science, 18 February 2019.
    Lead author: Prof Craig Moritz, Director, Centre for Biodiversity Analysis, Australian National University.
  2. Independent assessment of the 2018-19 fish deaths in the lower Darling — Final Report‘, Vertessy et al, 29 March 2019.
  3. They’re all using the industry-standard Hydstra database, but South Australia seems to have written its own web portal … which might be better, or not.

2 comments to Murray-Darling Basin flows: #1 The data

  • David Laughlin

    Is it possible to determine average velocity at the Barham gauging station.
    There is a cross section as well as flow in ML/day , so velocity could be estimated if we could plot the rive height on cross-section.

    Has this been done or do I need to do it manually some how.

    I expect that velocity varies throughout the cross-section , but I am trying to work out some sort of idea of current velocity.

    Thank you