Numerous basin aquifers in arid and semi-arid regions of the world derive a significant portion of their recharge from adjacent mountains. Such recharge can effectively occur through either stream infiltration in the mountain-front zone (mountain-front recharge, MFR) or subsurface flow from the mountain (mountain-block recharge, MBR). While a thorough understanding of recharge mechanisms is critical for conceptualizing and managing groundwater systems, distinguishing between MFR and MBR is difficult. We present an approach that uses hydraulic head, chloride and electrical conductivity (EC) data to distinguish between MFR and MBR. These variables are inexpensive to measure, and may be readily available from hydrogeological databases in many cases. Hydraulic heads can provide information on groundwater flow directions and stream–aquifer interactions, while chloride concentrations and EC values can be used to distinguish between different water sources if these have a distinct signature. Such information can provide evidence for the occurrence or absence of MFR and MBR. This approach is tested through application to the Adelaide Plains basin, South Australia. The recharge mechanisms of this basin have long been debated, in part due to difficulties in understanding the hydraulic role of faults. Both hydraulic head and chloride (equivalently, EC) data consistently suggest that streams are gaining in the adjacent Mount Lofty Ranges and losing when entering the basin. Moreover, the data indicate that not only the Quaternary aquifers but also the deeper Tertiary aquifers are recharged through MFR and not MBR. It is expected that this finding will have a significant impact on the management of water resources in the region. This study demonstrates the relevance of using hydraulic head, chloride and EC data to distinguish between MFR and MBR.
Numerous basin aquifers in arid and semi-arid regions receive a significant portion of their recharge from adjacent mountains, largely because the latter typically benefit from higher precipitation and lower evapotranspiration (Winograd et al., 1998; Wilson and Guan, 2004; Earman et al., 2006). Two recharge mechanisms can be recognized (Wahi et al., 2008): mountain-front recharge (MFR), which predominantly consists of stream infiltration in the mountain-front zone, and mountain-block recharge (MBR), which consists of subsurface flow from the mountain towards the basin. Here the mountain-front zone is defined after Wilson and Guan (2004) as the upper zone of the basin, between the basin floor and the mountain block (Fig. 1a). The term MFR has traditionally been used to encompass the two recharge mechanisms described above, but it may be more appropriate to use it for the first one only. Following Wahi et al. (2008), the collective process of MFR and MBR is referred to as mountain system recharge (MSR).
Conceptual models of the transition between mountain and basin:
The distinction between MFR and MBR is important. The conceptualization of a basin groundwater system critically depends on whether recharge occurs through MFR or MBR, as each of these mechanisms implies different groundwater flow paths, groundwater age and geochemical characteristics. MFR and MBR can also imply different responses to land and water resource management practices (both in the basin and the mountain) as well as to climate change. A good understanding of these mechanisms is thus essential for an effective coordinated management approach of water resources in basins and adjacent mountains (Manning and Solomon, 2003; Wilson and Guan, 2004).
While various methods exist to estimate MSR as a bulk, characterizing the individual contributions of MFR and MBR is difficult. For instance, Darcy's law calculations and inverse groundwater flow modelling typically provide bulk MSR estimates (e.g. Hely et al., 1971; Anderson, 1972; Maurer and Berger, 1997; Siade et al., 2015). It is possible to consider MFR and MBR independently in a groundwater flow model, but the solution to the inverse problem is more likely to be non-unique (e.g. Bresciani et al., 2015b). The water balance and chloride mass balance methods also provide bulk MSR estimates when the analysis is performed at the base of the mountain-front zone or further downstream in the basin (e.g. Maxey and Eakin, 1949; Dettinger, 1989). Environmental tracers such as noble gases (e.g. Manning and Solomon, 2003), stable isotopes (e.g. Liu and Yamanaka, 2012) and radioactive isotopes (e.g. Plummer et al., 2004) can help to determine which of MFR or MBR is the dominant mechanism, but their analysis remains expensive and their interpretation can be difficult. The most robust approach for characterizing MFR and MBR might be the integrated analysis of all available hydraulic, temperature and concentration data through the coupled modelling of groundwater flow, heat and solute transport in the combined basin–mountain system (e.g. Manning and Solomon, 2005) – but it is also arguably the most complex approach.
In this study, we explore alternatives to expensive and complex methods to
investigate whether MSR to basin aquifers is dominated by MFR
(Fig. 1b) or MBR (Fig. 1c), or if both recharge mechanisms are significant
(Fig. 1d). We focus on the use of hydraulic head,
chloride (Cl
Situation map showing elevation (in metres according to the AHD, i.e. Australian Height Datum) and relevant features of this study.
Schematic diagram showing triangular facets at the base of the mountain block (after Welch and Allen (2012)).
Schematic diagram showing hydraulic head contours and groundwater
flow directions in the horizontal plane for
After presenting a general rationale for the use of hydraulic head, Cl
Impact of a difference in basement elevation induced as a result
of faulting on hydraulic head in a hypothetical setting.
In this section, a generic rationale is presented for the use of hydraulic
head and Cl
Hydraulic heads directly relate to groundwater dynamics. Consequently,
hydraulic head patterns could theoretically enable the identification of
groundwater flow paths, both in mountains and basins. Specifically, four
types of analysis are suggested that could inform the likely occurrence or
absence of MFR and MBR:
In cases where faults run between the basin and the mountain, it may be
tempting to study the difference in hydraulic head between the two sides of
the fault zones. Intuitively, a large head difference would indicate that a
fault zone constitutes a barrier to flow in the direction perpendicular to
it (e.g. Bense et al., 2013), and consequently that
MBR would be low. However, a large head difference across a fault zone may
not always imply that the fault zone constitutes a hydraulic barrier. Let us
consider the hypothetical case of a sedimentary layer overlying a basement
of relatively low hydraulic conductivity and that features a sharp
transition in elevation as a consequence of faulting
(Fig. 5a). The hydraulic conductivity of the
fault zone itself is assumed to be no different from that of the embedding
materials (i.e. the fault only implies a difference in basement elevation).
In this simple configuration, if the groundwater level right below the fault
is lower than the basement elevation above the fault (as a result of
downstream hydraulic controls), the groundwater level above the fault is
essentially “disconnected” from the lower part
(Fig. 5b). This is because in all cases, the
groundwater level above the fault has to satisfy a minimum height (i.e.
transmissivity has to be large enough) for groundwater to flow there. Hence,
this example demonstrates that a large difference in head can exist across
the fault zone despite the fault zone itself having no particularly low
hydraulic conductivity. It should also be noted that regardless of the
cause, the implications of a large difference in head, in terms of the amount
of flow eventually crossing the fault zone, is far from obvious, as it
depends on the hydraulic conductivity of the fault or the basement (which is
in either case difficult to determine). A more relevant analysis may be to
investigate whether or not the hydraulic head above of the fault is so high
(relative to topography) as to imply that groundwater discharges locally to
mountain streams instead of flowing across the fault towards the basin. In
other words, what matters is the partitioning of the mountain groundwater
between these two pathways. This is precisely what the first three types of
analysis presented above should contribute to determining.
Chloride (Cl
Total thickness of the Quaternary (Q) sediments
In many cases, Cl
Schematic hydrogeological cross sections in
In this study, the proposed strategy consists of analysing three types of
water for Cl
Cl
Electrical conductivity values are known to be strongly correlated to
Cl
The Adelaide Plains basin is a coastal sedimentary embayment of 1700 km
Hydraulic head and topographic contours in the NAP
Precipitation is relatively low and potential evapotranspiration is high in
this semi-arid area. The average rainfall is 445 mm yr
The basin comprises complex, spatially dependent sequences of Quaternary and Tertiary sedimentary deposits (Gerges, 1999). The Quaternary sediments are dominated by fluvio-lacustrine clay interbedded with sand and gravel. The Tertiary sediments are dominated by sand, sandstone, limestone, chert, marl and shell remains interbedded with clay. A number of faults dissect the basin. Among these, the Eden–Burnside Fault and the Para Fault are of primary interest in this study since these faults run along the foothill, almost at the margin of the CAP and the NAP sub-basins, respectively (Fig. 2). The total thickness of the sedimentary units increases sharply on the downthrown side of the major faults (up to 400 m in places). The thickness of the Quaternary sediments ranges from 0 to about 140 m across the basin (Fig. 6a), while that of the Tertiary sediments ranges from 0 to about 500 m (Fig. 6b). The Tertiary sediments are directly outcropping in the northeast part of the CAP. The basement of the basin and the Mount Lofty Ranges are mostly comprised of Proterozoic fractured rocks of various lithologies including slate, phyllite, quartzite, limestone and dolomite. Superficial sedimentary deposits (typically less than 20 m in thickness) also exist locally in the Mount Lofty Ranges.
Head difference in the NAP
Up to six semi-confined aquifers (named Q1 to Q6) are recognized within the
Quaternary sediments from the central to western side of the basin
(Gerges, 1999) (i.e. downstream of the mountain-front zone). These
aquifers contain water of variable salinity with a median value of around
1300 mg L
Cl
Early investigations suggested that the natural (i.e. pre-development)
recharge to the Tertiary aquifers of the basin was dominated by stream
infiltration along the mountain front (i.e. MFR) (Miles, 1952;
Shepherd, 1975). In contrast, subsequent investigations suggested that the
natural recharge of the Tertiary aquifers was dominated by subsurface flow
from the Mount Lofty Ranges (i.e. MBR) (Gerges, 1999, 2006). The
latter conceptual model has formed the basis of most investigations of the
Tertiary aquifers since its presentation and has underpinned the development
of a number of groundwater flow and transport models of the basin aquifers
(Jeuken, 2006a, b; Zulfic et al., 2008; Baird, 2010; Georgiou et al.,
2011; Bresciani et al., 2015b). However, studies from Green et al. (2010) and
Bresciani et al. (2015a)
produced results supporting the hypothesis that MFR could also be
significant. To further investigate this question, the present study
provides a re-appraisal of available hydraulic head, Cl
Hydraulic head data in the AP catchment (i.e. the area including both the
basin and contributing mountain areas based on surface topography) were
retrieved from the WaterConnect database (
The data were subsequently split according to three aquifer groups: the AP Quaternary aquifers, the AP Tertiary aquifers (“AP” in these expressions will be omitted in the remaining text) and the Mount Lofty Ranges aquifers. This grouping is relevant in view of the hydrogeological characteristics of the system and the objective of the study. In particular, we did not distinguish between the T1 and T2 aquifers (i.e. the two main aquifers of the AP basin) because, as mentioned earlier, they are undifferentiated along most of the mountain front. In the Mount Lofty Ranges, the presence of complex fracture networks and high relief can induce the blurring of otherwise depth-dependent hydraulic signals, and so splitting the data according to depth in this area may not be very meaningful and it would reduce data density. The name of the aquifer into which the wells were screened was informed in the database for about two-thirds of the wells (6209). This allowed for assignment of these wells to one of the above aquifer groups. For the remaining one-third of wells, the aquifer group for the wells located in the basin was determined by comparing the well mid-screen elevation to the bottom elevation of the Quaternary sediments and to the top elevation of the basement. The largest number of wells was from the Quaternary aquifers (3964), followed by the Mount Lofty Ranges aquifers (3589) and the Tertiary aquifers (1768). Wells screened into the basement of the basin were disregarded (240 wells).
Groundwater-level fluctuations can be an issue for data interpretation. In particular, as this study focuses on natural recharge mechanisms, the impact of pumping constitutes a potentially important bias. It should be noted that the density of hydraulic head data is higher in areas of lower groundwater salinity, which coincides with areas that have experienced greater changes due to pumping. The measurements made before the main development period (i.e. before 1950) may have been less affected by pumping than more recent measurements, but limiting the analysis only to these measurements would dramatically reduce the data density. In addition, even the earliest measurements may not be free of pumping influence, since it is likely that these were precisely taken to monitor the impact of pumping. Hence, instead of subjectively fixing an arbitrary date beyond which the data would be excluded, all data were retained regardless of the measurement date. For each of the wells that had multiple measurements, the temporal mean hydraulic head was calculated in an effort to smooth out the measurement errors and temporal fluctuations. The analysis focuses on these mean values.
Groundwater Cl
The Cl
Pumping may also have impacted Cl
Surface water flow rate, EC values and their conversion into
Cl
Flow rate and EC data from streams running down from the Mount Lofty Ranges
into the AP basin were also retrieved from the WaterConnect database. Six
gauging stations were located close enough to the mountain front to be
relevant to the current study. Details on this dataset are given in
Table 1. The reported EC values of surface water
were converted into Cl
Given the relatively large area investigated, the analysis presented below concentrates on two “focus areas” that cover the transition between the Mount Lofty Ranges and the AP basin: one at the margin of the NAP sub-basin and one at the margin of the CAP sub-basin (locations indicated in Fig. 2). Figures for the entire study area are also available in the Supplement (Figs. S1–S12). These do not call for a different interpretation.
Hydraulic head maps were constructed for the three aquifer groups (Quaternary aquifers, Tertiary aquifers and Mount Lofty Ranges aquifers) (Figs. 9, 10). In constructing these maps, the choice of the interpolation method and associated parameters revealed to be critical. The classical inverse distance weighting method would produce the well-known “bull's eye” effect around individual data points. This could severely compromise the interpretation of head contours. Instead, the diffusion kernel interpolation method from the Geostatistical Analyst extension of ArcGIS 10.4.1 was used. This method allows for a more realistic interpolation when the underlying phenomenon governing the data is diffusive, as is the case for hydraulic heads. The most important parameter in this method is the bandwidth, which is used to specify the maximum distance within which data points are used for prediction. Making this parameter too small would undermine the prediction capability as many areas would remain uncovered by the interpolation, while making it too large would produce overly smoothed results. A good compromise was found by setting this parameter to 1200 m for the NAP focus area and to 800 m for the CAP focus area – reflecting a higher density of streams and data in the latter case. Topographic contours were also constructed. To facilitate the comparison with head contours, these were created after application of a circular moving-average window to the topography using a radius that matches the bandwidth used in the interpolation method for hydraulic head (i.e. 1200 m in the NAP focus area and 800 m in the CAP focus area).
Figure 9a displays hydraulic head and topographic contours in the NAP focus area, showing the Quaternary aquifers on the basin side and the Mount Lofty Ranges aquifers on the mountain side. The results are quite contrasted between the mountain and the basin. In the mountain, head contours follow the topographic contours relatively closely, and their shape is most often indicative of gaining stream conditions. Note that one should not expect to see sharp “V” shapes where head contours cross streams (i.e. as in Fig. 4) due to the limited data density. Instead, head contours are smoothly curved. In the basin, head contours do not closely follow the topographic contours, and their shape is generally indicative of losing stream conditions (especially close to the basin margin). Figure 9b also displays hydraulic head and topographic contours in the NAP focus area, but showing the Tertiary aquifers on the basin side instead of the Quaternary aquifers (on the mountain side, this figure is identical to Fig. 9a). Head contours in the Tertiary aquifers are generally indicative of focused recharge along streams, but on a somewhat larger scale, i.e. showing wider curvatures than in the Quaternary aquifers (mostly around Gawler River and Little Para River). Figure 9c and d display analogous results for the CAP focus area. Similar to above, in the mountain, head contours are relatively well correlated with topographic contours and their shape is generally indicative of gaining conditions. In the basin, head contours in the Quaternary aquifers are not very distinct from topographic contours, but nevertheless tend to indicate losing rather than gaining stream conditions close to the basin margin (Fig. 9c). In the Tertiary aquifers, head contours are quite distinct from topographic contours and are quite clearly indicative of focused recharge along a majority of streams (Fig. 9d). Near Glen Osmond Creek and Brownhill Creek, groundwater flow predominantly appears oriented towards the southwest, which may result from the bedrock sloping in this direction (the Tertiary sediment thickness can be seen to increase in Fig. 6b).
Figure 10a–d display the difference, in every point, between river head (approximated by the topographic elevation of the nearest river) and groundwater head. Figure 10a shows the NAP focus area, with the Quaternary aquifers on the basin side and the Mount Lofty Ranges aquifers on the mountain side. In the mountain, the results generally reveal a potential for gaining stream conditions along large portions of the main rivers (i.e. North Para River, South Para River and Little Para River). Potential losing stream conditions are indicated around the upper reaches of streams, suggesting that these are not supported by groundwater discharge, but are rather initiated by overland flow or interflow. This observation is consistent with the fact that most of the stream headwaters in the Mount Lofty Ranges are ephemeral. Potential losing stream conditions are also observed locally around a few streams in the lowest part of the Mount Lofty Ranges (e.g. South Para River and Smith Creek). This observation is not in line with the interpretation of head contours made from Fig. 9a, and this inconsistency might be an artefact of the temporal averaging of hydraulic heads, i.e. the hydraulic heads might be on average lower than the river head but the stream might still be gaining due to important groundwater discharge in some periods (but this explanation remains a hypothesis). In the basin, the Quaternary aquifers are revealed as potentially receiving water from streams everywhere, and especially close to the basin margin where the head difference is the largest. Figure 10b shows the head difference between the Quaternary aquifers and the Tertiary aquifers on the basin side, such as to investigate the vertical connection between these aquifers (on the mountain side, this figure is identical to Fig. 10a). The hydraulic head appears larger in the Quaternary aquifers than in the Tertiary aquifers over most of the area. This indicates a potential for downward groundwater leakage from the Quaternary aquifers to the Tertiary aquifers. The rate at which this leakage occurs is nonetheless difficult to estimate, since it is also a function of the effective vertical hydraulic conductivity and vertical distance between these units, which are largely unknown. Similar observations and interpretations can be made of the CAP focus area (Fig. 10c, d).
Most observations from Figs. 9 and 10 suggest that groundwater flow is dominated by local flow systems in the Mount Lofty Ranges. This indicates that only a small proportion of the recharge occurring over the mountain may make its way towards the basin. Hence, if MBR occurs, it would be probably limited to the routing of the recharge occurring over triangular facets in between stream catchments at the base of the mountain (see Sect. 2.1). Moreover, the results generally suggest that MFR through stream leakage is an important recharge mechanism for both the Quaternary aquifers and the Tertiary aquifers of the AP basin.
Cl
Figure 11a shows the Cl
Scatter plot of chloride concentration versus streamflow rate for six streams running down from the Mount Lofty Ranges into the AP basin.
The Cl
One of the main strengths of hydraulic head data is that they can indicate the contemporary flow direction (if hydraulic conductivity can be considered to be isotropic). In this study, the analysis of head contours suggested that groundwater in the Mount Lofty Ranges discharges to local streams for a large part. It also allowed for the identification of losing stream conditions in the mountain-front zone, where leakage from streams appears to recharge not only the Quaternary aquifers but also the deeper Tertiary aquifers. Studying the head variation with depth in the basin gave further evidence that groundwater flows from the Quaternary aquifers to the underlying Tertiary aquifers, even though the rate of this flow is unknown.
The main limitation of hydraulic head data is probably that these data are quite sensitive to pumping. This is problematic when the objective is to study the natural (i.e. pre-development) recharge mechanisms. Pumping in the AP basin mostly affects groundwater levels in the western part of basin, where large cones of depression exist in the Tertiary aquifers due to extensive historical and ongoing pumping (Bresciani et al., 2015a). Therefore, for the purpose of this study, which focuses on the eastern part of the basin (where the mountain-front zone is located), this issue may not be as critical. However, smaller-scale pumping wells surely also exist in the mountain-front zone and in the mountain, and may affect the results to an unknown degree. This represents a non-negligible source of uncertainty.
Cl
Stream water chloride concentration and flow rate over selected
periods for
As for hydraulic heads, pumping can potentially distort the Cl
The interpretation of Cl
Conceptual model of the recharge mechanisms for the aquifers of the AP basin, as seen in a cross section perpendicular to a stream in the mountain-front zone, and how these mechanisms can explain the observed groundwater salinity. Blue to red colours indicate relatively low to high salinity.
Finally, the assumption of conservative Cl
Compared to other methods that use noble gas, radioactive or isotopic
tracers, the above approach appears simpler, more cost effective and more
reliable due to the much higher data density generally achievable (i.e.
given current technologies and budget constraints). Nevertheless, in
contrast with some other methods (e.g. noble gases), it should be noted
that this approach is only qualitative. I.e. it does not allow for the
quantification of the relative proportion of MFR and MBR and their absolute
rate. One way to extend the ideas presented in this study to gain more
quantitative insight would be to use the data as calibration targets in a
groundwater flow and Cl
In an early study, Miles (1952) noted that the pre-development groundwater levels along the mountain front of the AP basin were reflective of unconfined conditions, and that the subsurface materials in this zone were favourable to stream infiltration. In addition, Miles (1952) already analysed the groundwater salinity distribution. He observed that salinity contours were forming fan-shaped zones of low-salinity “mushrooming” outwards from streams, with such patterns being visible up to more than 100 m below the ground surface. He concluded that stream infiltration in the mountain-front zone was a major recharge mechanism for the basin aquifers. Later, in a study of the NAP aquifers, Shepherd (1975) arrived at the same conclusion, partly using similar arguments and further noting that (i) groundwater hydrographs in the Quaternary aquifers were each year showing a rapid rise in water level shortly after Gawler River and Little Para River started to flow, and (ii) the vertical head gradient and vertical hydraulic conductivity were indicative of significant downward flow from the Quaternary to the Tertiary aquifers. Additionally, a number of studies directly measured groundwater gains and losses using differential flow-gauging along streams entering the AP basin (Hutton, 1977; Green et al., 2010; Cranswick and Cook, 2015). All found that several streams were losing a significant amount of water in the mountain-front zone. Finally, Zulfic et al. (2010) found that bore yield based on air-lift testing conducted at the time of drilling (e.g. Williams et al., 2004) in the Mount Lofty Ranges did not increase beyond 100 m depth for most geology types. This finding can be interpreted as showing that hydraulic conductivity is relatively low beyond that depth. This would promote local groundwater flow systems in the mountain instead of deep flow towards the basin, in line with the current analysis.
In contrast, Gerges (1999) and Batlle-Aguilar et al. (2017)
proposed that MBR is the dominant recharge mechanism for the Tertiary
aquifers of the AP basin. A major argument in these studies was based on the
observation that salinity is generally higher in the Quaternary aquifers
than in the T1 and T2 aquifers. From this observation, the authors suggested
that the relatively fresh water found in the T1 and T2 aquifers could not be
the result of downward leakage. Instead, they proposed that this water
should come from the Mount Lofty Ranges (where the salinity is lower)
through subsurface flow. However, along streams, the Cl
Hence, on the basis of robust consistent evidence given in this work
(including consistent findings from hydraulic head and Cl
We presented and demonstrated through a regional-scale example the use of
hydraulic head, Cl
In the above case study, both hydraulic head and Cl
Difficulties in the interpretation of hydraulic head and Cl
The relevance of the presented approach lies in that the variables used
(hydraulic heads, Cl
The AP basin serves as an example of a region where the recharge mechanisms have long been debated in the context of groundwater resources management. Application of the proposed rationale was revealed to be effective in resolving this debate. It is expected that the findings of this study will have important consequences for future hydrogeological investigations and the management of water resources in the Adelaide region.
Groundwater hydraulic head, Cl
The supplement related to this article is available online at:
The authors declare that they have no conflict of interest.
This study was supported by the Goyder Institute for Water Research through the project I.1.6 “Assessment of Adelaide Plains Groundwater Resources” (2013–2015), and by the Korea Research Fellowship Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (project number 2016H1D3A1908042). The authors thank two anonymous reviewers for their constructive comments. Edited by: Graham Fogg Reviewed by: two anonymous referees