Understanding the surface hydrology of the Lake Eyre Basin: Part 2—Streamflow

https://doi.org/10.1016/j.jaridenv.2008.06.001Get rights and content

Abstract

This is the second of two papers examining the surface hydrology of the Lake Eyre Basin (LEB) (1,140,000 km2) in Australia. The streams are unregulated and are characterised by extreme discharge variation. The analyses reported cover only surface hydrology and include comparisons with arid zone catchments globally. The paper discusses spatial runoff and annual streamflow characteristics, flow duration and baseflow index (BFI) analyses, annual flood frequency analysis and flood transmission losses, a water balance study, wet and dry run length analysis and, finally, yield from hypothetical reservoirs located across the LEB. There are 12 conclusions listed at the end of the paper. We identify two highlights as follows:

  • The coefficient of variation of annual flows, Cv, varies from 0.98 to 2.62. Compared with 45 arid zone rivers world-wide excluding Australian rivers, the annual Cv of the LEB streams are approximately double the average variability found world-wide.

  • Large transmission losses occur as flood flows move down the middle reaches of the major river systems. The transmission losses vary non-linearly with flood size as a result of differing transmission efficiencies between primary channels and the floodplain, and varying floodplain utilisation.

Introduction

This is the second of two complementary papers examining how different the surface hydrology of the Lake Eyre Basin (LEB) (Fig. 1) is from other arid regions world-wide. In Part 1 (McMahon et al., 2007b), we outline the background to this study and discuss the annual rainfall features that make the LEB different to elsewhere. In this paper, we build on the rainfall analyses and describe and interpret the special features of the surface hydrology of the LEB.

The catchments of the LEB together are 1,140,000 km2 in area; the major rivers are long (>1000 km) and have very low gradients. The area has a high conservation value as the aquatic ecosystems are largely rare and remain relatively pristine (Morton et al., 1995). Ramsar-listed wetlands (Coongie Lakes), grasslands, and arid regions are located in the LEB, as well as rare and endangered species of plants and animals (Ramsar Convention Secretariat, 2004). The fluvial ecology of the LEB is characterised by vast booms and busts in animal populations driven by the interannual variation in the flooding cycle (Kingsford et al., 1999; Puckridge et al., 2000).

Streams in the LEB are unregulated and are characterised by extreme variation in discharge and are amongst the most variable rivers in the world (Puckridge et al., 1998). Lake Eyre is primarily fed by the eastern and northern tributaries that originate in more humid areas: Cooper Creek, Diamantina River and Georgina River systems, with the Diamantina River contributing the most runoff to Lake Eyre in terms of volume and frequency of flow events (Kotwicki, 1986). These rivers flow seasonally in the middle to upper catchment in response to summer rainfall. Many reaches of LEB rivers have complex flow paths with extensive anastomosing channel systems (that is, the channels bifurcate and then rejoin irregularly) with greatly varying widths of active channel and floodplain (Fagan and Nanson, 2004; Nanson et al., 1986). Only a small proportion of the flows reach Lake Eyre, due to high transmission losses (Knighton and Nanson, 1994a; Kotwicki, 1986).

In combination with the relatively long river lengths for the major rivers, the subdued elevations (headwater elevations mostly less than 400 m) result in low gradients. For example, the Diamantina River has an average gradient of only 2.7×10−4 m/m and Cooper Creek has an average gradient of 1.7×10−4 m/m from the junction of the Barcoo and Thomson Rivers to Lake Eyre North (Kotwicki, 1986). The lowest point within the LEB is located within Lake Eyre, at a depth of 16 m below sea level (Bye et al., 1978). Runoff also originates from the west of the LEB but the catchments that drain into Lake Eyre North (Neales and Macumba Rivers) are ungauged. These rivers generally have some flow each year but only large floods enter Lake Eyre and these occur approximately one in every 10 years (Kotwicki, 1986).

A feature of the LEB is that it contains large areas that do not contribute any streamflow to Lake Eyre. Areas of arheic drainage include the major dunefields of the Simpson and Strzelecki Deserts where drainage is generally disconnected and accumulates in interdunal swales. The catchments of the north-western sector of the LEB (e.g. Finke, Todd, Hay Rivers) drain into the Simpson Desert and do not contribute any streamflow to Lake Eyre. Even during the largest flood on record in 1974, Landsat satellite images indicated that no flow from the Finke River entered Lake Eyre via the Macumba River catchment. In addition, there are a number of large endorheic “sub-basins” where drainage flows into terminal salt lakes (e.g. Lakes Frome, Callabonna, Cadibarrawirricanna) or clay-pans (e.g. Bilpea Morea) with no connection to Lake Eyre under current climatic conditions. At smaller scales, even within integrated catchments, some areas may also contain disconnected drainage where flow terminates in ephemeral low-lying wetlands that are not generally connected with the main stream network. A number of rivers show evidence of interaction between aeolian and fluvial processes, including fluvial reworking of sand dunes and encroachment of sand dunes onto floodplains (Hollands et al., 2006). These geomorphic interactions are not a focus of this paper but we note their importance in long- and short-term influence on fluvial flow paths (and hence on transmission losses) and influence on the location of waterholes (Hollands et al., 2006; Knighton and Nanson, 1994b).

Lake Eyre, the terminus of the LEB, is located in north-east South Australia. The lake is the fifth largest terminal lake in the world, consisting of two sections: Lake Eyre North (∼88% of the total Lake area) and Lake Eyre South (∼12%). The total surface area of the lake is approximately 9690 km2, supporting a volume of 30.1 km3 (3.01×104 GL) at −9.5 AHD (International Lake Environment Committee, undated). Originally, it was believed by European settlers that Lake Eyre North was permanently dry, however, this was disproved in 1949, the first scientifically recorded filling of the lake. Since that time, numerous inflow events into Lake Eyre have been recorded, the most significant being 1949–1950, 1973–1977, 1989–1991 (Hope et al., 2004; Kotwicki, 1986). Numerous smaller inflow events have also been recorded in the past 50 years and it is relatively rare that Lake Eyre North is completely dry (Hope et al., 2004).

Lake Eyre South is known to have filled in 1938, 1955, 1963, 1968, 1973, 1974, 1975, 1976 and 1984. In 1984 Lake Eyre South overflowed to Lake Eyre North (Allan et al., 1986). In 1974 water flowed from Lake Eyre North to Lake Eyre South between March and October when an equilibrium level was obtained.

The geology of the LEB is dominated by sedimentary rocks of Cainozoic and Mesozoic age. In general, the headwaters of the Cooper Creek, Diamantina River and south-western catchments are formed over Mesozoic sedimentary rocks, which also contain significant volumes of volcanogenic sediments (Bryan et al., 1997). The major anastomosing rivers (Cooper Creek, Diamantina River, Georgina River) predominantly transport a clay-rich mud load (4–20% sand, 25–60% silt, 35–65% clay), commonly as sand-sized pellets (Fagan and Nanson, 2004; Nanson et al., 1986) but also with a high suspended sediment concentration (Knighton and Nanson, 1994a, Knighton and Nanson, 1994b). The anastomosing rivers have areas of significant floodplain expansion and wetland development that are covered by these muds, deposited under the current fluvial regime (Nanson et al., 1988). A consequence of the volcanogenic sediments in the LEB is that one of their weathering products, smectite, is the dominant clay present in the floodplain sediments of the major rivers (Fagan and Nanson, 2004) and results in the development of highly cracked soils when desiccated (Nanson et al., 1986). Some areas in the north-western quadrant of the LEB are comprised of metamorphic and igneous basement rocks and some Palaeozoic sedimentary rocks. The rivers draining these areas, particularly the Finke, Todd and Hay Rivers, contain modern alluvial deposits ranging from boulders to sands with minor amounts of silt and clay (Williams, 1971).

Following this introduction we outline the streamflow data used herein. In Section 3, we characterise the annual streamflow data including an analysis of an appropriate probability density function (PDF) describing the annual data, and discuss monthly flows. Next, the spatial variation of runoff across the LEB based on a water balance analysis is described in Section 4. A discussion of flow duration curves follows in Section 5. Section 6 examines the baseflow index (BFI). A flood frequency analysis is described in Section 7 which is followed by a discussion of flood flow transmission losses. In Section 9 an analysis of the wet and dry run lengths of annual streamflow is carried out. The penultimate section deals with yield from hypothetical regulated systems. In the final section a list of conclusions is presented.

Section snippets

Streamflow data

Data from a number of stream gauges are available in the LEB, however, we restrict the analyses to stations with adequate record lengths. There are 17 stations with at least 10 years of complete records, and 12 have at least 20 years (Fig. 2).

Over time, two stations have been located on the Diamantina River at Birdsville. The first station (002101A) operated by the Government of Queensland recorded between 1949 and 1966. A new station (002101B), in close proximity to the old one, was opened in

Annual streamflow characteristics

To explore the annual streamflow features for the LEB rivers, the statistical characteristics at each gauging station with sufficient data were computed and are summarised in Table 1.

In general, stations at the downstream end of a catchment tend to have higher annual flow volumes than those upstream. This is illustrated in Fig. 3, which shows a plot of the mean annual flow and catchment area. Knighton and Nanson (2001) carried out a similar analysis (but with less data) showing similar results.

Spatial distribution of mean annual runoff

To explore the spatial difference of mean annual runoff across the LEB, the mean annual runoff (mean annual flow divided by catchment area expressed in mm) is computed for catchments with 9 or more years of concurrent annual rainfall and runoff data as listed in Table 3. This table shows that the runoff coefficients defined as the ratio of mean annual runoff to mean annual rainfall vary from a low 1% for the Alice River at Barcaldine to 15% for Diamantina River at the Diamantina Lakes. This

Flow duration curve

Flow duration curves show the percentage of time discharges at a given site are equal to or greater than a given value. Curves based on historical daily data for four rivers (Diamantina at Birdsville, Thomson at Longreach, Cooper Creek at Callamurra Water Hole (often spelt Cullyamurra) and Todd at Wigley Gorge) are presented in Fig. 9. Runoffs are expressed as flows in terms of mm/day (calculated by dividing the daily streamflows by the associated catchment area), allowing a direct comparison

Baseflow index

The BFI, which is the ratio of baseflow to total flow in a streamflow time series, was computed for each set of daily flows using the Lyne and Hollick filter (Nathan and McMahon, 1990) adopting 0.925 as the filter parameter. The filter partitions the daily total flows into surface runoff and baseflow. BFI values listed in Table 3 range from 0.03 (Todd River at Wigley Gorge) to 0.29 (002103a/AW003501-Cooper Creek near Innamincka) and are plotted against catchment area in Fig. 10. The strong

Annual flood frequency analysis

Annual flood frequency analysis was carried out using daily mean flows based on the annual flood series. (Instantaneous peaks were not available.) Fig. 11 displays for the same four rivers used in the flow duration analysis the annual flood frequency values along with a theoretical PDF. The adopted PDF was the generalised extreme value distribution (GEV) fitted by LH-moments (with a shift of two, Wang 1997). Overall, the fits are satisfactory although the upper parts of the Thomson and Cooper

Flood flow transmission losses

A major feature of the large rivers of the LEB is their large transmission losses that vary non-linearly with discharge (Knighton and Nanson, 1994a). This feature, observed during flood events, is masked by the variability in the general relationship observed in Fig. 3 where mean annual discharge increases with catchment area. The change from a positive relationship between flood discharge and catchment area to a negative relationship in these rivers coincides with the junctions of the major

Wet/dry spell length and cumulative surplus/deficit analysis for annual streamflow

Explanations of run length, run length skewness and run magnitude vulnerability were provided in Section 4, Part 1 (see also Peel et al., 2004b, Peel et al., 2005). In this section, the wet and dry spell analysis is applied to the selected LEB streamflow records. The median was used to differentiate wet years from dry years. The results of the run length and run magnitude analysis for annual streamflow data are provided in Table 4. Wet (gwet) and dry (gdry) run length skewness and vulnerability

Yield from regulated systems

Any discussion regarding the surface hydrology would be incomplete without some comments on the yield available for hypothetically regulated systems within the LEB. We have done this by assuming that reservoirs of capacity 1× and 2×mean annual flow (μ) are located at each stream gauging station in the LEB. This exercise is based on annual streamflows and assumes, initially, that there are no evaporation or other losses from the reservoir, and yield (or draft) from the reservoir is constant.

Conclusions

This and the complementary paper (McMahon et al., 2008) examine the surface hydrology of the Lake Eyre Basin (LEB) mainly in terms of annual rainfall and streamflow characteristics. The catchments of the LEB together are 1,140,000 km2 in area and drain to Lake Eyre which is located in the south of the LEB. The Lake is the fifth largest terminal lake in the world with an area of 9690 km2. The tributaries feeding Lake Eyre are primarily from the east and north and originate in humid areas. They are

Acknowledgements

This paper is based on a report commissioned by the Australian Department for the Environment and Heritage on behalf of the Lake Eyre Basin Scientific Advisory Panel. In addition to the present authors, Dr. D.-D. Kandel from Murray Darling Basin Commission and Dr. Rory Nathan, Pat Little and Georgina Race from Sinclair Knight Merz assisted in the preparation of that report. We are grateful for their input. Professor Geoff Pegram offered some helpful comments on an early draft. We thank

References (51)

  • T.A. McMahon et al.

    Revisiting reservoir storage–yield relationships using a global streamflow database

    Advances in Water Resources

    (2007)
  • M.C. Peel et al.

    Identification and explanation of continental differences in the variability of annual runoff

    Journal of Hydrology

    (2001)
  • M.C. Peel et al.

    Continental differences in the variability of annual runoff—update and reassessment

    Journal of Hydrology

    (2004)
  • R.J. Allan et al.

    The 1984 flooding of Lake Eyre South

    Transactions of the Royal Society of South Australia

    (1986)
  • J.A.T. Bye et al.

    Bathymetry of Lake Eyre

    Transactions of the Royal Society of South Australia

    (1978)
  • J.F. Costelloe et al.

    Determining loss characteristics of arid zone river waterbodies

    River Research and Applications

    (2007)
  • B.L. Finlayson et al.

    Australia vs. the World: a comparative analysis of streamflow characteristics

  • Hardison, C.H., 1971. Prediction error of regression estimates of streamflow characteristics at ungaged sites. USGS...
  • P.K. Hope et al.

    The rainfall response to permanent inland water in Australia

    Australian Meteorological Magazine

    (2004)
  • J.R.M. Hosking et al.

    Regional Frequency Analysis: an approach based on L-moments

    (1997)
  • International Lake Environment Committee, undated. World Lakes Database—Lake Eyre....
  • Kellett, J.R., Ransley, T.R., Coram, J., Jaycock, J., Barclay, D.F., McMahon, G.A., Foster, L.M., Hillier, J.R., 2003....
  • A.D. Knighton et al.

    Flow transmission along an arid zone anastomosing river, Cooper Creek, Australia

    Hydrological Processes

    (1994)
  • A.D. Knighton et al.

    Inbank and overbank velocity conditions in an arid zone anastomosing river

    Hydrological Processes

    (2002)
  • W. Köppen

    Das geographisca System der Klimate

  • Cited by (48)

    • A shifting ‘river of sand’: The profound response of Australia's Warrego River to Holocene hydroclimatic change

      2020, Geomorphology
      Citation Excerpt :

      Fluctuations between La Niña and El Niño conditions (Fig. 5A) correlate with some of the largest rainfall and flood events and with some of the driest spells, respectively, particularly during the early 1970s and the 2010–2012 La Niña episodes (Fig. 5B-F). In conjunction with significant downstream flow attenuation and peak discharge decreases, the passage of flood waves across the fan takes several months, providing a further illustration of the slow moving, long duration floods that are characteristic of many large, low gradient Australian dryland rivers (cf. Knighton and Nanson, 1994; Costelloe et al., 2003; McMahon et al., 2008). For example, the 2010 and 2012 floods had peak discharges of ~3000 to 3500 m3 s−1 at Wyandra, but peak discharge declined substantially downstream to ~1500 m3 s−1 at Cunnamulla (Fig. 5F).

    View all citing articles on Scopus
    View full text