Introduction

The massive Great Artesian Basin (GAB) aquifer extends across ~20% of Australia and provides critical water supplies to economically important mines, ranches, and towns in an otherwise arid landscape (Fig. 1; Ransley et al. 2015). Over 600 springs in 11 major groups discharge from the GAB, with Dalhousie Springs (Dalhousie) being the largest spring complex in the western GAB (Fig. 2; Habermehl 1982). Dalhousie, as with other springs in South Australia (Keppel et al. 2011, 2012), is composed of travertine mound-form springs (Fig. 3b; following nomenclature of Springer and Stevens 2009) with over 145 individual vents (White et al. 2013). Dalhousie is thought to account for ~40% of GAB spring discharge (Habermehl 1982) and supports groundwater-dependent ecosystems covering ~200 km2 (White et al. 2013), with vegetation from the largest spring extending ~10 km (White et al. 2016). The immense size and geographic isolation of aquatic and terrestrial habitats of the Dalhousie ecosystem permitted evolution of high taxonomic richness and diversity, with taxa quite dissimilar to other GAB springs (Kodric-Brown and Brown 2007; Rossini et al. 2018; Zeider and Ponder 1989).

Fig. 1
figure 1

Dalhousie Springs and western margin of the Great Artesian Basin, Australia. Dalhousie Springs is located at the South Australia-Northern Territory border (see inset; black square indicates extent of Fig. 3). Surface geology is composed primarily of sedimentary rocks of Jurassic–Quaternary age (Geoscience Australia 2012). To the west of the GAB boundary where Jurassic–Cretaceous rocks crop out (red line), surface geology is primarily composed of Precambrian–Paleozoic rocks. Line A–A′ shows the location of the hydrogeologic conceptual cross section (Fig. 9). Legend: Petroleum exploration wells: a McDills 1. b Mount Crispe 1. c Purni 1. d Mokari 1. e Macumba. Water wells: (1) Alberga; (2) Pedirka Siding; (3) Hamilton Creek; (4) Crans Grave; (5) Bloods Creek; (6) Opossum; (7) 3 O’Clock Creek; (8) Mount Crispe 1; (9) Macumba HS Bore 2; (10) Stevensons; (11) Apperinna Bore 2; (12) Onqueedinna 1; (13) Horseshoe; (14) Witcherrie 1; (15) Purni; (16) Junction; (17) Charlotte Waters; (18) Anacoora; (19) McDills (completed as water well); (20) Town Bore Apatula; (21) Apatula Community; (22) Andado Station; (23) New Crown Station. Table S1 of the electronic supplementary material (ESM) lists detailed hydrochemical data

Full size image
Fig. 2
figure 2

Dalhousie Springs temperature measurements. Dalhousie Springs is situated in a breached anticline where the Cretaceous Bulldog Shale confining bed and overlying sediments have eroded. Only spring temperatures >30 °C (more likely to be actual spring vent temperatures) are shown (Table 1). Groundwater-dependent vegetation, called “spring tails”, appears as dark green areas. Long-term discharge measurement locations at weirs are shown as black rectangles (Mark Keppel, Department of Environment, Water and Natural Resources, South Australian Government, Adelaide, Australia, unpublished data, 2019; see also Fig. 5). Background is a high-resolution multispectral aerial photo (Lewis and White 2013; White and Lewis 2011)

Full size image
Fig. 3
figure 3

Dalhousie Springs, aquifer outcrops, and Finke River ephemeral flow. a Main Spring at Dalhousie Springs. b Typical spring discharge vegetation and older travertine at Dalhousie Springs (mound at right of photo). c Fractured outcrop of Cadna-owie Formation in Dalhousie Anticline. d Algebuckina Formation in recharge zone along western margin of GAB near Finke, Northern Territory; e Crown Point Formation in recharge zone ~25 km west of Finke, Northern Territory; f Ephemeral flow of Finke River ~30 km west of Finke, Northern Territory in November 2008 (for Finke River historical discharge, see Fulton et al. 2013)

Full size image

Dalhousie and other desert spring systems globally provide important ecological benefits, as they are often the only perennial aquatic environments maintaining biodiversity in otherwise arid areas (Davis et al. 2017; Hunter 2017). However, aridland springs are threatened by groundwater development and other anthropogenic threats because of their relatively small size and isolation. Species that inhabit aridland spring ecosystems are particularly at risk because they are found in only a few locations (i.e., low redundancy), and if a spring dries completely, the species cannot bounce back, even if water returns (i.e., low resiliency, following nomenclature of Smith et al. 2018); however, compared to larger aquatic ecosystems, conservation of desert springs may be a more tractable challenge (Davis et al. 2017). Dalhousie is culturally important to the Lower Southern Arrente peoples, who refer to the area as Irrwanyere, or “healing water.” Determining the source of groundwater to the springs is essential to its conservation and is the main objective of this study.

Site physiography

Dalhousie is located in central Australia near the South Australia and Northern Territory border, ~350 km south of Alice Springs, ~1,000 km north of Adelaide, and near the western margin of the GAB (Fig. 1). Mean annual precipitation is ~175 mm/year (BOM 2010) and maximum annual temperature is 36–39 °C during summer and 18–24 °C in winter (Allan 1990), resulting in a semi-arid to arid climate. Land surface elevation ranges from ~400 to ~10 m below sea level at Lake Eyre. Ephemeral streams flow following summer monsoonal precipitation or winter frontal precipitation (Fig. 3f; Costelloe et al. 2007; McMahon et al. 2005); however, low gradients and high evaporation typically limit stream flow to headwater and middle reaches (Habermehl 1980).

Dalhousie is sited in a topographic depression along the axis of the breached Dalhousie Anticline (45 km by 35 km; Fig. 2). Pools and lakes (Fig. 3a) are associated with the ovoid-shaped central spring complex (20 km × 10 km; Krieg 1985, 1989), which is drained by Spring Creek and flows ephemerally for ~50 km before terminating in the Simpson Desert.

Geology

Dalhousie regional geology includes several stacked sedimentary basins that range in age from Neoproterozoic to Cenozoic, with basin names corresponding to different ages of stratigraphic units (i.e., Adelaide Geosyncline, Western Warburton Basin, Southeast Amadeus Basin, Pedirka Basin, Simpson Basin, Eromanga Basin, Lake Eyre Basin) (Figs. 1 and 4; Drexel and Preiss 1995; Drexel et al. 1993; Geoscience Australia 2012; Krieg 1985, 1989). This study focuses on Carboniferous–Cretaceous strata of the Pedirka and Eromanga basins because of their hydrogeologic importance (discussed in the next section). Surface geology generally includes Cretaceous and Cenozoic rocks east of the GAB margin, with Proterozoic and Paleozoic rocks exposed west of the GAB margin (Geoscience Australia 2012).

Fig. 4
figure 4

Stratigraphic column of western margin of Great Artesian Basin. The Algebuckina Sandstone and Cadna-owie Formation (Fig. 3d,e, respectively) are considered primary aquifers providing flows to Dalhousie Springs and confined by the Bulldog Shale; however, this study suggests that the Purni and Crown Point formations (Fig. 3e) also discharge to the springs (after Carne and Alexander 1997 using information from Drexel and Preiss 1995; Drexel et al. 1993; Krieg 1985, 1986). In the Northern Territory, the Algebuckina Sandstone is known as the “De Souza Sandstone” and the Bulldog Shale as the “Rumbalara Shale”

Full size image

Two formations compose the shallow, foreland, intracratonic Pedirka Basin (Figs. 1 and 4; Middleton et al. 2005): The Carboniferous–Permian Crown Point Formation Sandstone is made up of siltstone and sandy–pebbly diamictite (Fig. 3e; Wells et al. 1966; Youngs 1975). The Permian Purni Formation (Youngs 1975) is divided into three facies with differing amounts of sandstone, shale, and coal (Wells et al. 1966). The lowest contains carbonaceous shale; the middle is a fining-upward, cross-bedded sandstone with minor shale and coal; and the upper is a fine-grained, silty sandstone.

The Pedirka Basin is overlain by terrigenous and marine sediments of the Eromanga Basin. The Early Jurassic–Early Cretaceous Algebuckina Sandstone (known as the De Souza Sandstone in the Northern Territory) is a fine- to coarse-grained, cross-bedded, fluvial, poorly consolidated, quartzose sandstone with pebble and coal layers, and minor shale and siltstone lenses (Fig. 3d; Krieg 1985; Wopfner et al. 1970). The Early Cretaceous Cadna-owie Formation includes a lower unit of fine- to medium-grained, well-sorted, porous, quartzose, cross-bedded/massive sandstone (Fig. 3c; Krieg 1985; Moore and Pitt 1982). The Early Cretaceous Bulldog Shale (Rumbalara Shale in the Northern Territory) is a bioturbated, fossiliferous, shaley, marine mudstone with thin parallel bedding and micaceous silt to very fine-grained sandstone interbeds (Freytag 1966). The Bulldog Shale is regionally extensive throughout the Eromanga Basin; however, it is eroded from the Dalhousie Anticline (Fig. 2).

Many springs are associated with Cretaceous Cadna-owie Formation exposures (Fig. 2). The rim of the breached anticline lies ~60 m above modern and relict spring-related sedimentation modified by fluvial activity (Fig. 3b). Within ~5 km of and higher than present-day springs are at least two levels of Quaternary travertines (Krieg 1985, 1986). Sedimentary morphologies of Dalhousie platform travertines resemble those formed in lakes, whereas mound-shaped deposits likely formed from single pools and vents similar to mound structures of Lake Eyre (Fig. 1; Keppel et al. 2011). Travertine dating using U-Th series suggests long-lived, episodic spring discharge since at least ~465 ka (close to the method’s limit; Priestley et al. 2018), and stratigraphic relationships place the oldest Dalhousie spring activity at 2–1 Ma (Krieg 1985).

Hydrogeology

Aquifers and aquitards

Jurassic–Cretaceous sandstone units of the Algebuckina Sandstone and the Cadna-owie Formation (generally known as the J Aquifer or J-K Aquifer) are considered the main source of springs at Dalhousie (Habermehl 1980). Erosion of the Bulldog Shale—the primary GAB confining bed—along the crest of the Dalhousie Anticline permits discharge from the J-K Aquifer via the Cadna-owie Formation. A detailed schematic of hydrostratigraphy for the western margin of the GAB is provided in Keppel et al. (2013) and in Fig. 4. The deeper Purni Formation and Crown Point Formation were not historically considered aquifers (Habermehl 1980 and later works); however, this study evaluates their input to Dalhousie discharge.

Potentiometric surface, recharge, and discharge

A density-corrected regional potentiometric surface map (Fig. 1; from Ransley et al. 2015) reveals generalized regional groundwater flow from postulated recharge zones where Jurassic–Cretaceous-age aquifer units crop out along the western GAB margin to discharge areas at topographic lows of Lake Eyre and other salt lakes (Fig. 1). However, monitoring wells drilled in close proximity to ephemeral channels suggest that modern-day groundwater recharge only occurs along the Finke and Plenty rivers (Figs. 1 and 3f; Fulton et al. 2013).

Spring discharge rates and temperature

Dalhousie is the largest western GAB spring complex, with discharge ranging from <1 L/s to ~160 L/s (~5 GL/year) at Main Spring (Fig. 3a; Habermehl 1982; Radke et al. 2000). Erosion of the Cretaceous Bulldog Shale confining bed and overlying sediments (red-colored land surface in Fig. 2) permits discharge through fractures in the exposed Cadna-owie Formation (Fig. 3c), forming dozens of springs (Figs. 2 and 3a,b). Spring discharge measured at weirs was relatively stable from late 1997 to mid-2012 (Fig. 5). Spring-vent temperature measurements range from ~29 to 46 °C (Fig. 2; Table 1; Radke et al. 2000; Smith 1989). Warmer springs are found to the north along a northeast trend that may reflect springs emerging along faults in the core of the anticline. Springs get cooler to the south (Department for Water 2010; Radke et al. 2000; Smith 1989) and approach mean annual air temperature (approximately 20 °C), suggesting cooling from vent to sample location. A high-resolution multispectral image collected as part of efforts to develop remote-sensing techniques to monitor temporal changes in spring-dependent ecosystems (Lewis and White 2013; White and Lewis 2011) reveals groundwater-dependent vegetation (dark green in Fig. 2) in spring tails up to 5 km long.

Fig. 5
figure 5

Selected discharge at Dalhousie Springs. Long-term discharge measured with weirs at selected larger springs (locations shown on Fig. 2; South Australia Department of Environment and Water, “Dalhousie Springs weir gauging”). Unpublished dataset provided by Mark Keppel (2019). Bars indicate ±10% to account for field sampling uncertainty. Overall trend of Dalhousie Main Spring (Weir 2) shows a possible decrease in discharge since 2000

Full size image
Table 1 Thermal spring source depth input data and results
Full size table

Structural controls on Dalhousie Springs

Dalhousie is situated within a region of complex crustal and lithospheric weakness along a rifted passive continental margin (Kennet et al. 2004; Shaw et al. 1991). Intersecting northwest- and northeast-trending structures in the study area are representative of normal faults inverted during contractional events. Such structures have been documented to produce complex bending monoclines and intersecting fault-and-fold networks in similar structural settings (Martel et al. 1989). The Pedirka Basin (Fig. 1) is bisected into the western Eringa and eastern Madigan troughs by the McDills-Dalhousie Ridge, which is a major north–northeast trending, folded and faulted structural ridge composed of Devonian and older rocks (Drexel and Preiss 1995). Seismic surveys through Dalhousie confirmed the presence of faults in close proximity to springs (Aldam and Kuang 1989). Normal faults were reactivated in the middle–late Carboniferous to create compressional, high-angle, north-northeast striking reverse faults (Drexel and Preiss 1995; Youngs 1976). During the Oligocene–Miocene, reverse faults were further reactivated, with fold amplitudes increasing in the Pliocene–Pleistocene (Krieg 1985).

Using spring temperature to estimate aquifer source depth

Anderson (2005) highlighted the importance of heat as a groundwater tracer, including the evaluation of spring systems. Thermal springs are found around the world in a variety of geologic settings and require a heating mechanism, fracture permeability, groundwater source, and pressure gradients (Waring et al. 1965). Groundwater-heating mechanisms include recent volcanism, magmatic intrusion, and deep circulation of groundwater heated by Earth’s geothermal gradient—the only mechanism considered here (Waring et al. 1965). Fracture permeability caused by recent faulting is an important influence on thermal spring locations (Curewitz and Karson 1997; Hancock et al. 1999; Waring et al. 1965).

Objectives

This study elucidates aquifer sources to Dalhousie Springs by integrating current hydrogeologic and hydrochemical data with historic, previously unpublished test data from petroleum exploration wells. Specifically, the study:

  1. 1.

    Compiles spring locations, discharge rates and temperatures, surface and subsurface geologic characterization, and mapping of potentiometric surfaces

  2. 2.

    Evaluates hydrochemistry and hydrogeology of springs, groundwater, and data collected from petroleum exploration wells, including formation water samples, drill-stem tests, sidewall cores, and geophysical logs

  3. 3.

    Assesses potential aquifer source depths to springs using a thermal model

  4. 4.

    Updates the hydrogeologic conceptual model of Dalhousie to inform potential groundwater management strategies

Although this approach is employed to identify sources to Dalhousie Springs, the methods developed by this study may be applied to improve conservation outcomes for similar aridland spring systems by identifying linked aquifers.

Methods

Oil and gas exploration wells: geophysical logs, side-wall cores, and drill-stem tests

This study presents a new conceptual model for the hydrogeology of the Dalhousie spring region by evaluating previously overlooked data obtained from 1960s and 1970s-era petroleum exploration wells (Fig. 1), including Witcherrie 1 (Magnier 1964b), Purni 1 (Magnier 1964a), McDills 1 (Amerada 1965), Mokari 1 (Jacque 1966a), Mount Crispe 1 (Jacque 1966b), and Macumba 1 (Wiltshire 1978). Geologic data considered include interpretation of subsurface stratigraphy from lithologic and geophysical logs, permeability tests of side-wall cores, and estimates of formation water salinity from geophysical logs. Geophysical estimates of formation porosity were calculated from gamma and sonic logs during drilling; however, of greatest value were hydrochemical analyses of groundwater samples collected during drill-stem tests (DSTs) or artesian flow that tested sediments of the Pedirka and Eromanga Basins. These data were used to identify potential aquifers in Pedirka Basin sediments that previously had not been characterized by existing groundwater wells.

Hydrochemistry

Previously unpublished hydrochemical analyses of groundwater samples collected from petroleum exploration (n = 6) wells were compiled (Amerada 1965; Jacque 1966a, b; Magnier 1964a, b; Wiltshire 1978), as were publicly available data from Dalhousie springs (n = 17; only springs with temperature > 30° to omit springs with excessive cooling), groundwater wells (n = 23), and Lake Eyre springs (for regional context, n = 11; Radke et al. 2000). Parameters assessed included total dissolved solids (TDS) and concentrations of major anions and cations–see Table S1 of electronic supplementary material (ESM). Potential hydrogeologic connections between springs and groundwater were evaluated by comparing hydrochemical facies of Dalhousie-area DSTs, wells and springs with hydrochemical facies of wells from the larger region, and springs of the Lake Eyre complex. Refer to Priestley et al. (2013) and (2019, this issue) for a detailed evaluation of the use of hydrochemical variations of springs and groundwater to understand the regional groundwater flow system of the western GAB.

Thermal model of spring aquifer source depth

Spring discharge temperatures (Fig. 1; Table 1; Radke et al. 2000; Smith 1989) were used to estimate the depths of aquifers sourcing thermal springs, defined here as a point of naturally occurring groundwater discharge at greater than mean annual surface air temperature (but not warm enough for two-phase fluid conditions).

The traditional adiabatic approach to estimating spring source depth in an aquifer considers spring discharge temperature, mean annual air temperature, and an assumed geothermal gradient (Fig. 6). It also ignores heat loss by assuming that source temperature is the same as spring outlet temperature and that the ground surface temperature is the same as the atmospheric air temperature. The earliest known reference to the adiabatic approach is McCallie (1913), and the method has persisted in the literature with, for example, Alfoldi et al. (1985) and Sanz and Yélamos (1998), as noted by Manga (2001). Whether or not heat losses in the spring conduit are considered, all models estimating spring source depth require geothermal gradients, which is estimated by subtracting temperatures measured in a well at two depths and dividing by the measurement depth difference (Polak and Horsfall 1979). Ground surface temperature (Te) is estimated by extrapolating the well-depth-temperature slope to ground surface and is often 1–4 °C above the mean annual surface air temperature (T0) (McCutchin 1930).

Fig. 6
figure 6

Spring geothermal schematic. Steady-state geothermal conditions of a linear rock matrix temperature T are noted where L (e.g., 270 m) is the depth from the top of an aquifer to ground surface where a thermal spring discharges with temperature Ts (e.g., 30.00). Temperature at the top of the aquifer Ta is located at vertical distance z = L. No heat losses are assumed to occur in the spring conduit; thus, Ta =  Ts. Matrix temperature T at 0 < z < L is controlled by the geothermal gradient γ (3.0 °C/100 m). Mean annual air temperature is T0 (i.e., 21.9 °C). Refer to Table 1 for thermal spring source depth model input data

Full size image

Aquifer source depth was estimated by comparing measured mean annual spring discharge temperature (Ts) to T0 and assuming a geothermal gradient, γ, defined by:

$$ T={T}_{\mathrm{a}}\hbox{--} \gamma z $$
(1)

where T is the linear aquifer rock matrix temperature at a given depth; Ta is the aquifer top surface temperature; and z is the distance from aquifer top in a vertical direction. At the top of an aquifer where depth z = 0, the temperature in the matrix surrounding the bottom of the spring conduit is the same as at the top of the aquifer, T = Ta. The distance from aquifer top to ground surface in the z direction is L. The temperature at the ground surface where z = L is assumed to be the same as ground surface temperature, T = T0. If it is assumed that there is no heat loss in the spring conduit, then the spring temperature measured at the ground surface is the same as the temperature at the top of the aquifer, and Ts = Ta. At the top of the aquifer (spring source depth z = L), the geothermal gradient is T = Ta–γL. Aquifer source depth (L0) can be found by setting Ta → Ts, T → T0, and L → L0 (after Alfoldi et al. 1985):

$$ {L}_0=\frac{\left({T}_s-{T}_0\right)}{\gamma}\kern0.5em \mathrm{i}.\mathrm{e}.,\kern0.5em {T}_s={T}_0+\gamma {L}_0 $$
(2)

This study uses measurements of thermal spring temperatures obtained during this study and values available in the open literature (Table 1; Radke et al. 2000; Smith 1989). The geothermal gradient for each spring location is interpolated from Cull and Conley (1983). Mean annual air temperature was interpolated from stations near Dalhousie (BOM 2010). Calculated aquifer source depths were compared with stratigraphic depths inferred from surface horizons in seismic surveys (refer to maps in Appendix 1 of Keppel et al. 2013). Because seismic surveys were acquired by the oil industry during 1960s-era exploration under difficult field conditions using now-antiquated equipment, some uncertainty is to be expected in interpreted stratigraphic depths. The top of the Cadna-owie Formation corresponds to the C surface horizon, and the bottom of the Crown Point Formation corresponds to the Z surface horizon. The Cadna-owie Formation only crops out in the highest portion of the Dalhousie Anticline; otherwise, the C surface horizon is located stratigraphically below the Bulldog Shale and younger spring deposits.

Results

Oil and gas exploration wells: geophysical logs, lithologic logs, and drill-stem tests

Previously unpublished petroleum exploration well-testing data allowed for evaluation of the hydrogeologic properties of deeper formations than those available in Dalhousie-area groundwater wells (Fig. 1 and Table 2; Amerada 1965; Jacque 1966a, b; Magnier 1964a, b; Wiltshire 1978). Well-completion reports for Witcherrie 1, Purni 1, and McDills 1 revealed formation water quality (650–2,400 mg/L TDS), porosity (up to ~30%), and lithology that indicate that the Permian-age Crown Point Formation is a high-quality aquifer. Witcherrie 1, Purni 1, and McDills 1 were ultimately completed as groundwater supply wells in the Algebuckina Sandstone, producing groundwater TDS of 750, 2,750, and 2,750 mg/L, respectively (however, McDills 1 now has a corroded casing and discharges mixed Jurassic–Permian groundwater; Matthews 1997). Testing of deeper sediments in the southeast Amadeus Basin, western Warburton Basin, and Adelaide Geosyncline did not reveal fresh formation water or high permeability. Refer to Appendix 1 of Keppel et al. (2013) for full results of water-quality analyses of samples collected at petroleum exploration wells, springs, and groundwater wells.

Table 2 Petroleum exploration well-test data
Full size table

Hydrochemistry

Hydrochemistry of springs at Dalhousie, selected nearby groundwater wells, formation fluids produced during petroleum exploration well testing, and Lake Eyre springs (provided for regional context) are presented in a Piper plot (Fig. 7; well locations on Fig. 1; full data set in Table S1 of the ESM). Spring water at Dalhousie primarily belongs to Na-Cl and Na-Cl-SO4 hydrochemical facies and are distinct from Lake Eyre springs, which are predominantly Na-Cl-HCO3. Witcherrie Bore (screened in the J-K Aquifer) and artesian flow collected at the McDills 1 petroleum exploration well during testing of the Permian Crown Point Formation (Na-Mg-Cl-SO4 hydrochemical facies) exhibit similarities to spring waters at Dalhousie. Upgradient to Dalhousie (see Fig. 1 for potentiometric surface map), wells screened in Jurassic–Cretaceous formations at Charlotte Waters and Andado Station also show hydrochemical similarities to Dalhousie; however, other J-K Aquifer wells upgradient to Dalhousie and located near the Finke River at Apatula (Finke) Community and at New Crown Station belong to Ca-Mg-Cl-SO4 hydrochemical facies. Approximately 10 km east of Dalhousie, a groundwater sample from Cambrian–Ordovician strata at Mount Crispe also belongs to the Na-Cl-SO4 hydrochemical facies (DST No. 1; Jacque 1966b). However, TDS approaches 10,000 mg/L and the unit is “nonporous”; thus, Cambrian–Ordovician formations are not likely to be major sources of Dalhousie water. Water samples collected during DSTs of other petroleum exploration wells have more Na+ and K+ and much more Cl−, indicative of basinal brines.

Fig. 7
figure 7

Piper plot of spring water and groundwater samples. Dalhousie Springs, Lake Eyre springs, and selected Dalhousie Springs-area bores (Radke et al. 2000) are synthesized with Dalhousie Springs–area petroleum exploration boreholes with drill-stem tests or artesian flow encountered during drilling (data provided in Table S1 of the ESM; Amerada 1965; Jacque 1966a, b; Magnier 1964a, b; Wiltshire 1978)

Full size image

Thermal model of spring aquifer source depth

In general terms, springs in the northern part of the Dalhousie Spring Complex have higher temperatures and lower salinities than those observed in the south of the region (Smith 1989). Estimated thermal-spring aquifer source depths range from 270 to 802 m below ground surface (bgs; Table 1), a greater depth than that of the J-K Aquifer. In addition, the top of the Cadna-owie Formation (i.e., C surface horizon) is found at shallower depths ranging from 82 to 200 m bgs (refer to maps in Appendix 1 of Keppel et al. 2013 for C and Z surface horizon depths). At spring locations, depth to the base of the Crown Point Formation (i.e., Z surface horizon) ranges from 356 to 471 m bgs. Thus, thermal-modeling results suggest that spring discharge also originates from the deeper, warmer Crown Point Formation at depths up to ~800 m.

Discussion

Hydrochemistry and petroleum exploration well-test data

This study evaluated the hydrochemistry of samples collected at Dalhousie-area springs and groundwater wells, as well as water quality and hydraulic properties from petroleum exploration well testing. The salinity and generalized water type of Permian Crown Point Formation waters at McDills 1 petroleum well and J-K Aquifer groundwater from the Witcherrie groundwater well (completed from the petroleum exploration well) are similar to those at Dalhousie, suggesting shared water sources. In addition, similarities in the hydrochemistry of groundwater in upgradient wells located to the north and northwest of Dalhousie are consistent with providing flows to the springs.

Groundwater samples collected from wells screened in the J-K Aquifer in the postulated recharge area along the Finke River at Apatula (Finke) Community and New Crown Station belong to the Ca-Mg-Cl-SO4 hydrochemical facies, and wells at Andado Station and Charlotte Waters are also similar to Dalhousie, suggesting the wells belong to the same regional flow system that discharges at the springs. Groundwater movement from the Finke River area to Dalhousie, along with mixing of Permian Crown Point Formation groundwater, may qualitatively explain observed Dalhousie hydrochemistry. This interpretation is also consistent with prior studies (Abu Risha et al. 2009; Abu Risha 2016) that used chlorine, carbon, oxygen, hydrogen, and strontium isotopic analyses to infer flow paths from recharge zones along the western margin of the GAB to Dalhousie. Focused ephemeral stream recharge to the northeast and west of spring discharge at Dalhousie is also suggested by a persistent groundwater mound along >100 km of the Finke River, as well as mounds along Stevenson and Lindsay creeks (Ransley et al. 2015). Thus, results of hydrochemical analyses of spring water, groundwater, and Permian-age formation water from the Dalhousie Springs region indicate that Dalhousie Springs hydrochemistry is consistent with a Jurassic-Cretaceous-age or Permian Crown Point Formation aquifer source. This is the first study to suggest that the previously unidentified Permian Crown Point Formation is a regionally extensive aquifer that, in addition to the generally accepted J-K Aquifer, may also provide a significant component of the groundwater discharge to springs at Dalhousie.

Thermal model

A thermal model was used to test the hypothesis that groundwater from the Permian Crown Point Formation, as well as that from the J-K Aquifer, flows to Dalhousie Springs. All previous studies of Dalhousie (Habermehl 1982; Krieg 1985; Radke et al. 2000; Smith 1989) suggest that groundwater originates solely from the J-K Aquifer. Using spring temperature as a proxy for source depth, results of the model suggest that Dalhousie discharges a mixture of shallower, cooler groundwater from the Algebuckina Sandstone and deeper, warmer groundwater from the Crown Point Formation. This is a novel interpretation of the source of Dalhousie groundwater.

The thermal model used by this study has some limitations because of necessary assumptions—for example, Alfoldi et al. (1985) noted typical model input measurement errors of ±5–10% for geothermal gradient and ±0.1 °C for spring temperature measured using a handheld field water-quality meter. At Dalhousie, spring temperatures generally increase from south to north, suggesting a deeper source of northern springs or thermal diffusion cooling caused by slower water velocity in southern spring vents. Spring temperatures below mean annual air temperature should be treated with caution because they may have been caused by sampling during cool air temperatures when heat loss is too high or by the occurrence of near-surface lateral groundwater mixing. Thus, it is important that spring temperature measurements be taken at the greatest depth and as close to the vent as possible to minimize such errors. The model assumption that aquifer temperature and spring temperature are the same is reasonable, given relatively high discharge rates of springs at Dalhousie.

Additional model limitations include the assumption that groundwater flows from the source aquifer to the spring vent along a cylindrical vertical conduit (Figs. 6 and 8a,b) without contributions from other water sources (e.g., Fig. 8c–e). As such, this model assumes a thermal spring conceptual model composed of a vertical fault conduit cutting through an aquifer and its low-permeability upper confining bed with negligible displacement (Fig. 8a). It is also reasonable to assume that the thermal model presented here may also be applied when the fault associated with a thermal spring has a relatively small displacement of ≤0.1 L (Fig. 8b). When fault displacement is large (e.g., ≥ 0.1 L), estimated spring source depths have higher uncertainty but are still useful to constrain the range of possible source depths (Fig. 8c). Similarly, when a fault intersects aquifers with multiple source depths and temperatures (Fig. 8d,e), spring temperatures reflect mixing of different aquifers and results have higher uncertainty.

Fig. 8
figure 8

Spring orifice conceptual models. a Vertical fault conduit cutting through an aquifer (dappled pattern) and its low-permeability upper confining bed (no fill) with negligible displacement. Groundwater flow direction is indicated by arrow, and distance from top of aquifer to ground surface is L. b Fault associated with a thermal spring with a relatively small displacement of ≤0.1 L. (c–e) Fault intersects aquifers with multiple source depths and temperatures. Diagrams after Van der Pluijm and Marshak (2004)

Full size image

The model also assumes homogeneity and isotropy of the rock matrix composing the spring conduit. Other modeling assumptions include that the spring conduit, spring vent, and place where geothermal gradient is measured are located away from other heat sources or sinks and that values for all input parameters are accurate. However, despite imperfect input data typical of hydrogeologic studies (e.g., the possibility of more than one aquifer source depth at Dalhousie), a reasonable first-order estimate of aquifer source depths may be generated with this approach.

Updated hydrogeologic conceptual model of spring discharge

Hydrogeologic data evaluated in this report reveal that springs at Dalhousie are sourced by a mixture of groundwater from the Carboniferous–Permian Crown Point Formation and J-K Aquifer. Fractures in the McDills-Dalhousie Ridge anticline, as well as erosion of the Bulldog Shale confining unit at this structural high, enabled formation of this spring system. A hydrogeologic cross section summarizing this conceptual model (Fig. 9) was constructed using published hydrogeologic data (Drexel and Preiss 1995; Krieg 1985; Questa 1990), seismic surveys (refer to maps in Appendix 1 of Keppel et al. 2013), and previously unpublished petroleum exploration well-testing data (Amerada 1965; Jacque 1966a, b; Magnier 1964a, b; Wiltshire 1978).

Fig. 9
figure 9

Hydrogeologic conceptual model of groundwater flow to Dalhousie Springs. Cross section A–A′ location is shown on Fig. 1. Dalhousie Springs likely discharges a mixture of groundwater from the Crown Point Formation and Algebuckina Sandstone along fractures in McDills-Dalhousie Ridge. Groundwater recharge occurs where ephemeral streams flow over outcrops of the Crown Point Formation and J-K Aquifer along the western margin of the GAB (Fig. 3f). The Bulldog Shale and lower-permeability units of the Crown Point Formation confine groundwater flows down gradient from northwest to southeast until ascending along fractures of the western anticline limb of the McDills-Dalhousie Ridge. Cross section is after Figure 8.27 of Drexel and Preiss (1995)

Full size image

Aquifers of the Crown Point Formation and J-K Aquifer crop out along the Finke River northwest of Dalhousie (Figs. 1 and 3d,e). Groundwater recharge occurs during ephemeral stream flow of the Finke and neighboring streams (Fig. 3f; Abu Risha 2016; Fulton et al. 2013; Ransley et al. 2015). Groundwater flows down-gradient (i.e., northwest to southeast) and is confined by the Bulldog Shale and lower-permeability units of the Crown Point Formation (Fig. 9). Groundwater moves through the synclinal Eringa Trough, ascends along the western anticline limb of the McDills-Dalhousie Ridge, and flows to the surface through fractures associated with faults in the McDills-Dalhousie Ridge. An interpretation of faults cross-cutting multiple aquifers is consistent with 87Sr/86Sr > 0.715, which suggests interaction with radiogenic basement rocks (Ring et al. 2016) and high total helium, high 3He/4He ratios, and excess CO2 signifying deeply derived fluids (Crossey et al. 2013). This conceptual model of Dalhousie differs from those in previous studies, which only considered the J-K Aquifer as a Dalhousie source (Habermehl 1980). Thus, an important finding of this study is that groundwater management planning needs to consider the Permo-Carboniferous Crown Point Formation, as well as Jurassic–Cretaceous sediments.

Implications for groundwater management: mitigating potential threats to springs discharge

Limited understanding of groundwater flow systems and hydrologic connections between aquifers and springs complicates successful groundwater management. In the eastern GAB, groundwater development since the late 1800s resulted in pressure declines >100 m and caused springs to dry (Habermehl 1980), in part because early groundwater users poorly understood the potential effects of excessive groundwater development. Thus, groundwater management decisions that maintain natural discharge of linked aquifer-spring systems such as Dalhousie must be based upon an accurate hydrogeologic conceptual model.

As suggested by Harrington et al. (2017), sustainable groundwater development may be informed by the construction of monitoring wells located between well fields and a groundwater discharge zone to provide an early warning of possible pumping effects on springs. Groundwater models may also be used to identify groundwater trigger levels at specific monitoring locations upgradient from springs below which pumping should be reduced or curtailed. Thus, the preservation of Dalhousie requires that any future groundwater development in the region be operated so that hydraulic heads around the complex and upgradient recharge areas are maintained in both Permian and Cretaceous aquifers. To this end, long-term monitoring, including hydrochemical mixing studies, in both aquifers is needed at a regional scale. In addition, numerical groundwater modeling could be used to assist in forecasting and mitigating the potential impacts of new wells on spring discharge. Thus, conservation of spring discharge at Dalhousie requires identifying and possibly mitigating potential future groundwater development threats, particularly groundwater pumping for mining and energy development.

Mines: dewatering and/or groundwater supply

Mines along the western margin of the GAB currently rely on groundwater for industrial processes—for example, the massive Olympic Dam mine maintains extensive GAB supply wells (Geoscience Australia 2013). BHP, the Olympic Dam mine owner, has also proposed nearly doubling current copper production to 350 kt/year, which would require an additional 50 ML of groundwater to be pumped from wells in the GAB (BHP 2019; DEM 2019a). While this new well field would be sited far from Dalhousie (>400 km), springs of the Lake Eyre group are closer (~100 km; e.g., springs listed in Keppel et al. 2011); thus, long-term pumping may affect artesian heads in the vicinity of these springs, potentially adversely affecting spring flow. Several other mines have been proposed along the western margin of the GAB, which could potentially increase GAB groundwater extraction (e.g., DEM 2019b). Thus, as part of a comprehensive risk assessment process (e.g., Green et al. 2013), groundwater wellfields for future mines should be sited and operated to minimize potential effects on GAB springs.

Coal seam gas

Most of Australia’s coalbed methane industry is located in eastern Australia (Geoscience Australia 2018); however, production potential in the Pedirka Basin from the upper member of the Purni Formation is estimated at ~47 trillion cubic feet (TCF) (Goldstein et al. 2012; Meaney 2007). Development is also being investigated near the Hale River, ~225 km north of Dalhousie (Scott et al. 2012), and ~100 km north of Dalhousie and ~50 km to the north of the Finke River (Ebony Energy Ltd. 2017). Of interest to maintaining Dalhousie spring flows are the potential effects of groundwater pumping to dewater coal beds during gas production and the effects on the regional groundwater flow system. While development of Permian coal may be unlikely in the near-term, potential coalbed methane dewatering may still pose a risk to springs.

Unconventional shale and tight gas

In the US, development of unconventional oil and gas resources using directional drilling and hydraulic fracturing (HF) has been rapid and intense (Allred et al. 2015). Of concern are water demands of this process (e.g., Permian Basin median ~23 ML/HF well; Scanlon et al. 2017). In Australia, unconventional gas resources remain essentially undeveloped (Geoscience Australia 2018; IEA 2018). In the Pedirka Basin, shales are not currently exploited because they are generally thin, laterally discontinuous, and of low thermal maturity (Goldstein et al. 2012). In deeper parts of the Pedirka Basin (~100 km east of Dalhousie), thermal maturity is sufficient, but thermogenic gas shows were limited to thin coal seams with shale partings (i.e., Oolarina 1 well; Goldstein et al. 2012; Menpes et al. 2012). During the drilling of McDills 1 (~80 km northeast of Dalhousie), gas was detected in the lower Rumbalara Shale (Amerada 1965) but was not subsequently developed. While Australian energy operators do not currently target prospective shale and tight gas resources (Geoscience Australia 2018), future commodity prices may incentivize their development. Thus, forecasts of where future development may occur may be warranted to mitigate potential impacts of HF water sourcing on groundwater resources and linked spring systems (e.g., Ikonnikova et al. 2017; Wolaver et al. 2018a).

Future work

In light of current and potential groundwater development in the Dalhousie region, conducting source aquifer assessments to “define the connection of springs to source aquifers” remains an important research topic required to manage groundwater development to maintain spring flows along the GAB western margin (Lai et al. 2016). To this end, continued monitoring of the existing well network should be continued and data openly shared with the public. Multiple geochemical tracers can be used to evaluate water quality and changes in source mixing. In addition, running groundwater models to quantify potential impacts of groundwater development on Dalhousie could be included as a condition of state well permitting, so that operations could be adjusted prior to construction to mitigate potential adverse effects on spring systems. In addition, groundwater modeling could be used to identify monitoring gaps to be filled by additional wells and triggers developed to reduce pumping prior to impacting springs. Conservation outcomes of aridland spring systems like Dalhousie could also be improved by developing science to inform decision-making before any groundwater development occurs as part of a public, transparent stakeholder-driven process (i.e., including state and federal agencies, private industry, nongovernmental organizations, ranchers, indigenous peoples, etc.). Facilitated science development sessions similar to the approach of Wolaver et al. (2018b) and Gulley (2015) could be used to gain early project buy-in and assure that the right science is developed to inform on-the-ground conservation actions that result in preservation of springs and associated groundwater-dependent ecosystems.

Conclusions

Historic, previously unpublished drill-stem test data are integrated with current hydrogeologic and hydrochemical data to identify the Permian Crown Point Formation as an unrecognized and important contributor to spring discharge at the Dalhousie Spring Complex. The revised hydrogeologic conceptual model presented here is important because it identifies the overlooked Permian aquifer and other deep-fluid sources as providing groundwater discharge to Dalhousie, in addition to the previously accepted J-K Aquifer (Habermehl 1980).

The most plausible mechanism for Permian and Jurassic–Cretaceous aquifer discharge at Dalhousie is through fault-associated fractures in the McDills-Dalhousie Ridge anticline. Uplift in the anticline formed a topographic high and eroded the Bulldog Shale confining bed to expose the Cadna-owie Formation. Fracturing associated with interformational faulting permits both mixing of groundwater and discharge of groundwater at the surface to form the spring complex.

Preservation of Dalhousie requires that the potentiometric surface in the vicinity of Dalhousie be maintained at or near present levels in both Permian and Jurassic–Cretaceous aquifers. Consequently, any future groundwater development in the Dalhousie region should consider these two aquifer systems as contributors with respect to the maintenance of head in the vicinity of Dalhousie. While this approach is used to elucidate sources to springs at Dalhousie to inform groundwater management strategies resulting in their conservation, this technique may be applied to preserve other aridland spring systems globally.