1 Introduction

Alluvial aquifers have traditionally been exploited to meet the demand for water. Human development in recent decades has resulted in groundwater being withdrawn from these aquifers and the subsequent capture of nearby stream water resources. These exploitation practices, however, have not considered stream-aquifer interactions (Winter et al. 1998; Tellam and Lerner 2009; Foster and Ait-Kadi 2012). For example, minor streams—those related to thin alluvial formations and with mean discharge rates below a few cubic meters per second—have suffered severe damage, and will continue to as long as the development demand is satisfied. As a result, stored groundwater reserves diminish, in-stream flow declines, ecological dynamics are jeopardized, and riparian biodiversity is lost (Environmental Agency 2005, 2009; Boulton 2007; Kirk and Herbert 2002).

In this scenario, hydrological resources management must be addressed to insure supplies and protect riparian areas from environmental deterioration, as stated in the European Water Framework Directive (WFD; Directive 2000/60/EC). Since groundwater is still abstracted, it must be assumed that sufficient water inputs are available to satisfy human demand, regardless of their impact. Complementary alternative water resources may help to reduce human pressures on the stream-connected alluvial aquifers and on the streams themselves. Regional groundwater flow systems could provide an alternative option for water supply if their occurrence is verified and makes their exploitation feasible (Morgan et al. 2006).

Stream discharge and hydrochemical analyses can determine the origin of both surface and groundwater. Such data can be used to identify the contribution of each source, or end-member, to the existing water resources in the overexploited alluvial reach. In this study, our objective is to use discharge and hydrochemical data to investigate the origin of stream water in an intensively exploited stream-connected alluvial aquifer. Our aim is to identify potential water resources alternatives to ease local human impacts on stream hydrology, based on the mentioned data and methods.

The origin of water resources can thus be determined if potential end-members are identified. From a methodological perspective, knowledge of geological and hydrological factors, and of human pressures, is needed to describe the system, and monitoring the stream reach provides the necessary data to prove whether all end-members have been thoroughly identified. Algebraic end-member mixing evaluates the contribution of each end-member and supports the potential use of alternative water sources to meet the current demand.

The Tordera River valley (NE Spain; Fig. 1) suffers from intense pressures upon the riparian environment. Stream discharge is diverted at several locations to fulfill urban demand, and groundwater from a stream-connected alluvial aquifer is pumped to cope with industrial necessities. Continuous groundwater withdrawal is also responsible for inducing stream recharge and, therefore, for a loss of surface runoff. Because of both pressures, water availability does not always satisfy ecohydrological needs, neither urban demand; and the Tordera River is likely to fail WFD requests for surface water bodies. Under these circumstances and its hydrogeological idiosyncrasy, the Tordera River stands as a potential place where investigating the stream-aquifer relationship and searching for potential alternatives for water resources exploitation.

Fig. 1
figure 1

Study area location and geological setting of the Tordera River basin. Starting and ending points of the surveyed stream reach are indicated as SP and EP, respectively. Geological mapping based on ICC (2011)

Full size image

2 Geographical, Geological and Hydrological Setting

This research was conducted in the Tordera River basin, located in Catalonia (NE Spain; Fig. 1). The area of the basin above the monitored segment is 180 km2. The basin lies mainly on igneous and metamorphic rocks in the Montseny-Guilleries and Montnegre ranges, where tectonic elements, such as regional fault lines, control the drainage pattern. In fact, the Tordera River follows a major tectonic structure, oriented NE-SW, that has defined the main physical features of the regional landscape (a range-and-basin area) since the Neogene (Mayo et al. 2008). In particular, this fault zone defines the Vallès graben, which includes the central Tordera valley. This graben presents an infilling of sedimentary Neogene materials, several tens of meters thick, deposited as alluvial fans along the northern wall in response to the tectonic dynamics. During the Quaternary, alluvial formations developed along the present drainage network, reaching a maximum thickness of 15–20 m. Minor streams from the Montnegre range follow tectonic lines oriented NW-SE.

A thin alluvial formation developed along the middle reach of the Tordera River, with a width of 500 m in its widest sections, and a maximum thickness of about 10–15 m in its central part. This formation is bounded by igneous rocks in the study area, and is therefore considered the main aquifer in that area. To our knowledge, it contains all the abstraction wells for urban and industrial use. Neither the hydrogeological potential of the igneous basement nor the role of faults in groundwater flow has been explored locally, although they have been shown to be of relevance in a nearby area (Menció 2005; Folch 2010; Menció et al. 2010; Folch et al. 2011). Since groundwater withdrawal to meet urban demand is only used to complement surface water resources, industrial uses account for the most continuous and intense exploitation rate, estimated at 4.47 hm3/year (GeoServei 2000).

Upstream from the studied reach, for the period from 1990 to 2010, the annual mean discharge of the Tordera River was about 0.532 ± 0.016 m3/s (Catalan Water Agency gauging station EA15; ACA 2010). Any major variation may be due to intense flooding, and more importantly, to water capture for urban supply during flood recession phases. Moreover, the mean treated wastewater effluent discharge is 0.061 ± 0.001 m3/s, which constitutes 11.5 % of the mean surface runoff. The sampling period (July 2009) was on the recession limb of the hydrograph of a large rise related to late spring rainfall events (Fig. 2).

Fig. 2
figure 2

Tordera River hydrograph at Sant Celoni gauging station located 2 km upreach from the surveyed reach, and mean daily rainfall. Discharge data ata from Catalan Water Agency, station EA15 (ACA 2011), and precipitation data from the Catalan Weather Service (SMC 2011)

Full size image

A stream reach of 5 km was used for this research. The beginning of the reach is located upstream of the confluence of a small creek, the waste water treatment plant (WWTP) and the main stream, the Tordera River (Fig. 3). After the confluence, the alluvial formation widens and the Tordera River flows continuously along this reach, despite the influence of intense groundwater withdrawal. During the sampling period tributaries from the left margin, draining the Montseny-Guilleries Range, contributed to the surface discharge, whereas those from the right margin, originating in the Montnegre Range, ran dry.

Fig. 3
figure 3

Sketch of the sampling point location for surface water Tordera River, tributaries and other inputs, and groundwater (wells) measurements and samples along the studied reach. Starting and ending points are indicated in Fig. 1

Full size image

3 Methodology

3.1 Field and Laboratory Methods

Discharge measurements and water samples for hydrochemical analysis were taken at different points variable spaced to track discharge and chemical variations along the Tordera River, as well as the inputs from tributary streams and industrial effluents. Active wells located in the alluvial aquifer were also sampled for hydrochemical analysis. River sampling was conducted in a single day (July 9, 2009), and groundwater samples were taken during the next week. The final dataset consists of 18 sampling points: 13 along the main stream of the Tordera River; two corresponding to tributary streams (Pertegàs and Gualba Creeks); two representing effluents from industrial areas; and a final sampling point from the WWTP located at the beginning of the investigated stream reach. Seven wells located in the alluvial aquifer were also sampled to determine hydraulic head and groundwater hydrochemistry (Fig. 3).

Discharge was recorded by measuring stream flow velocity using a current meter at intervals of 0.30 to 0.50 m along a cross section of the stream. The maximum stream depth in each cross section varied between 0.2 and 0.5 m.

Hydrochemical parameters for surface and groundwater samples such as temperature, conductivity, pH, and dissolved oxygen were measured in situ using appropriate probes. Samples for analysis were stored at 4 °C, and acidified for cation analysis. Alkalinity, as HCO3 −, was determined by titration; NH4 +, NO3 −, and NO2 − were measured by spectrometry; Cl−, SO4 2 − and NO3 − were determined using capillary electrophoresis (CE) with indirect-UV detection; and Ca2 +, Mg2 +, Na+, and K+ were measured using inductively coupled plasma-atomic emission spectrometry (ICP-AES). Bromide was analyzed by inductively coupled plasma-mass spectrometry (ICP-MS).

3.2 End-Member Mixing Analysis

End-member mixing analysis of conservative components from each pole can be used to estimate their contribution in the samples (EMMA; Christophersen et al. 1990; Hooper et al. 1990; Hooper 2001), and to provide a fair description of the hydrogeological processes that govern the stream-aquifer relationship in this specific reach as well. The mass-balance expressed by EMMA is given by a linear combination of the contribution of each pole according to a proportional factor, in our case the discharge from surface sources or ground water. For a single component,

$$ \left\{ {\begin{array}{*{20}c} {\sum\limits_{j=1}^n {C_j^i{Q_j}={C^i}Q} } \hfill \\ {\sum\limits_{j=1}^n {Q_j =Q} } \hfill \\ \end{array}} \right. $$
(1)

where C is the concentration of the component, in units of mass per unit volume [M/L3], and Q the discharge, in units of volume per unit time [L3/T]. Subindex i refers to the hydrochemical component (in this case i = 1), and subindex j denotes each one of the EM (j = 1,n). The first equation indicates a linear mixing process, while the second one constrains the discharge of each end-member to be equal to the total discharge after mixing.

Nevertheless, for several end-members (n > 2), different components (m) can be used to build a system of equations that considers a lineal combination of all components according to the flow contribution from each end-members, generating an over-determined system of equations where more tracers are used than are necessary to define the end-member contribution; that is,

$$ \left\{ {\begin{array}{*{20}c} {\sum\limits_{j=1}^n {C_j^{i=1 }} {x_j}={C^{i=1 }}} \\ {\sum\limits_{j=1}^n {C_j^{i=2 }} {x_j}={C^{i=2 }}} \\ \vdots \\ {\sum\limits_{j=1}^n {C_j^{i=m }{x_j}={C^{i=m }}} } \\ {\sum\limits_{j=1}^n {x_j =1} } \\ \end{array}} \right. $$
(2)

Its solution provides an estimate of the coefficients x i that denotes the proportion of each end-member, j, contributing to the stream. Recall that the values of x i correspond to the ratio Qi/Q.

Using matrix notation,

$$ A \;x= C $$
(3)

where A is the matrix (m + 1,n) that contains de concentrations of the different components (i) at each one of the EM (j), x is the unknown vector (n,1) that describes the contribution of each cto the mixing, and C is the vector (m + 1,1) with the concentrations of each component in the sample. The system includes the condition that the sum of all contributions (x j ) must be equal to 1.

In the system of equations when m + 1 = n, A is a square matrix, and the solution is given by

$$ x={A^{-1 }}C $$
(4)

Nevertheless, in the case of an over-determined system, where the number of components is larger than that of end-members —(m + 1) > n— the unique least-square solution is given by (Strang 1976):

$$ x={{\left( {A^T A} \right)}^{-1 }}{A^T}\;C $$
(5)

Once the end-members are identified, the components should be: 1) representative so each end-member is adequately identified, 2) conservative so their concentration does not change by any process other than linear homogeneous mixing, and 3) characterized by uniform concentration at the pole. In this study, the components selected are chloride, sulfate, sodium, potassium, and bromide. They are representative of the water-rock interaction under stable conditions in the subsurface and also represent human pressures on surface water.

4 Results

4.1 Hydrological Framework

Rainfall before the sampling period was scarce. Some rainfall events of less than 5 mm were recorded during the week previous to the sampling day, but they did not influence the hydrological behavior of the system (Fig. 2). The mean stream discharge on July 9, 2009, as recorded at the gauging station located about 2 km upreach (gauging station EA15), was 0.126 m3/s (Fig. 2). This value is similar to the mean discharge (0.143 ± 0.06 m3/s) for July, and it corresponds to a 15 % probability of exceedance on the July flow duration curve estimated for the 1990–2009 period.

Most of the sampled wells were active during the field survey, and hydraulic head levels were usually below the stream bed elevation, indicating a losing stream, condition which is usually maintained during exploitation periods. Monthly surveys from 2003 to 2010 of two inactive wells, located at the alluvial aquifer in the final part of the studied reach, at distances of 90 and 210 m from the Tordera River, indicate that a losing stream behavior persists throughout the year, except when it is temporarily inverted to a gaining stream during the flooding events (Mas-Pla et al. 2007; Observatori de la Tordera 2011).

4.2 Stream Discharge Data

Stream flow measurements corresponding to the 18 sampling points along the monitored river reach (Table 1) show a significant increase at the beginning of the segment, between sampling points 1 and 2, because of the contribution of the WWTP and of the Pertegàs Creek. An unanticipated increase, however, was recorded between points 2 and 3. The Gualba Creek also represents a significant increase in the Tordera River discharge, whereas industrial effluents barely account for minor flow increments (Fig. 4).

Table 1 Discharge and hydrochemical data
Full size table
Fig. 4
figure 4

Stream discharge along the studied reach of the Tordera River, and external inputs from tributaries and human downloads

Full size image

After the major increase at sampling point 3, a continuous loss of discharge occurs at a rate of approximately 0.061 m3/s·km between points 3 and 9, and of 0.039 m3/s·km between points 11 and 13. These suppose a loss of 1.26 104 m3/day along the 5 km reach, which corresponds to 48 % of the measured total discharge at point 3.

4.3 Hydrochemical Data

Hydrochemical data along the river is controlled by the inputs from the WWTP and those from the industry effluents that substantially modify the original surface water composition, as recorded at point 1 (Table 1). This information will be used later to track surface and groundwater mixing processes. It should be mentioned that the water quality at point 1 does not correspond to a pristine stream, having already been altered by human pressures some kilometers upstream.

Conductivity values increase significantly when the inputs from the WWTP and from industry C join after points 1 and 4, respectively (Fig. 5). Sampling points after both inputs are combined, i.e., points 2 and 5, do not reflect the expected increase because of incomplete stream dispersion and mixing. Effects, however, are fully noticed at the next sampling location. Conductivity values remain uniform along the river until the pristine input from the Gualba Creek has a dilution effect.

Fig. 5
figure 5

Electrical conductivity of Tordera River surface water along the studied reach, and that of external inputs from tributaries and human downloads

Full size image

The main hydrochemical composition is represented using a Piper-Hill diagram (Fig. 6) that represents the relative concentrations, as percentages, of the major components and that allows hydrochemical facies to be inferred and relationships among samples to be established (Freeze and Cherry 1979). Surface water samples, except the one corresponding to point 1, show similar compositions and are all grouped within a chloride/sulfate-calcium/sodium facies, indicating the influence on the river of human inputs. Groundwater samples show a shift from bicarbonate-calcium facies at the upreach wells (P1 and P2) to hydrochemical facies closer to that of the surface water in the downreach wells. Sources of human inputs, such as WWTP and industries, are clearly located at the chloride/sulfate-sodium area in the plot, whereas the near pristine waters from the Pertegàs and Gualba creeks show a bicarbonate-calcium facies, which in this area is typical of natural waters unaffected by human pressures.

Fig. 6
figure 6

Piper-Hill diagram for the hydrochemical data of surface water (Tordera River and tributaries), groundwater, and human inputs. Samples mentioned in the text are indicated in the central plot

Full size image

Nitrogen compounds vary in the river as a consequence of nitrification and biological assimilation processes in the stream flow. Along the reach the main nitrogen inputs from the WWTP also vary as a result of both processes. In every case, they are below the threshold concentrations for eutrophication and toxicity of surface water (Table 1).

5 Data Analysis and Discussion

5.1 Discharge and Hydrochemical Data

Stream discharge data (Fig. 4) show two specific features. One is a discharge increase along the first kilometer due to a tributary contribution and the WWTP effluent. However, the observed discharge at point 3 was larger than the sum of all the surface contributions, suggesting an additional, local source of water. The second feature, there is a discharge decrease further downstream at a uniform rate due to surface water capture by pumping wells, turning the rest of the studied segment into a losing stream.

Such an unexpected increase in stream discharge is attributed to a significant groundwater contribution to surface discharge, which in turn attributes a gaining stream behavior to the segment between points 1 and 3. A cross section of the alluvial formation does not show special variations in this transect that could deviate groundwater flow to surface discharge, as the river flows between the outcropping igneous rocks of its right margin and the sedimentary Neogene materials on its left. In addition, geological mapping shows that several regional faults in a NW-SE direction intersect the major tectonic structure, which is oriented NE-SW in this particular area (Fig. 1). This last feature is a potential cause for groundwater flux occurrence.

Stream discharge data points out then a contribution from groundwater in the initial transect of the studied reach (from point 1 to point 3). Whether the discharge increase in this segment originates from alluvial aquifer resources or from recharge due to a regional groundwater flow system driven by structural elements (faults), as mentioned, must be ascertained by hydrochemical data. Indeed, a Piper-Hill diagram (Fig. 6) shows a significant hydrochemical difference between groundwater and surface water, especially for the groundwater sampling points located near this initial transect (samples P1 and P2).

Since human pressures on surface and groundwater resources (mainly groundwater withdrawal) are similar upreach and downreach of this initial transect, the occurrence of bicarbonate-calcium facies in groundwater (P1 and P2) indicate a nil contribution from stream water due to induced recharge by pumping. Consequently, a basement groundwater contribution is deduced for this transect. Nevertheless, the Pertegàs Creek does not present a well-developed alluvial aquifer, and the Neogene sediments do not constitute efficient aquifer formations because of the dominant proportion of fine sediments in their uppermost layers. Therefore, the observed discharge increment and the hydrochemical characteristics of groundwater indicate a pertinent contribution of the regional flow system to the alluvial aquifer that, in this particular reach, outcrops and feeds the stream discharge through a fault-driven flux.

Nitrate concentrations in wells P1 and P2 indicate, however, some influence from human activities on this contribution, possibly from agriculture, sewage pipes, or domestic septic tanks.

5.2 End-Member Mixing Analysis of Hydrochemical Data

Geological and hydrochemical data indicate that surface water is a mixture of distinct poles. Conversely, groundwater hydrochemistry at the down reach wells shows the influence of surface water infiltration, an expected result of the induced losing stream condition.

The main distinct poles, or end-members (EM), for this hydrological setting are 1) the upstream contribution of the Tordera River, 2) the effluents from the WWTP, and 3) the regional groundwater contribution. Samples representative of the EM are established as point 1 (Tordera River) for the first EM, as effluents from the WWTP and industry C for the second EM, and as the P1 well for the third EM (Table 1). The groundwater EM assignment assumes that well P1, because of its hydrochemistry, is only slightly influenced by surface water and, therefore, is the closest available data to the unaltered regional groundwater flow system.

The assumption of pole stability is essential in fluvial systems (Kurtenbach et al. 2006). Surface water chemistry varies with the discharge; as dilution processes are enhanced or reduced and human downloads fluctuate. In this case, since the survey was conducted in a single day, during an 8-hour period, in a recession phase of the hydrograph, data are assumed to satisfy the pole stability condition.

For a river that shows gaining and losing stream dynamics along the monitored reach, the EM mixing analysis will estimate the percentage of water originating from each pole at each sampling point if the following assumptions apply:

  • Each sampling point is representative of a river segment that is independent of contiguous segments.

  • The chemical composition of conservative elements/compounds in each sample is a linear combination of the EM and is independent of the composition of the sample upstream. If two adjacent samples have similar chemistry, we conclude that the contribution of each EM has not varied in th is river stretch. If there are some changes, the role of diffuse EM, such as groundwater, or the occurrence of additional point sources should be analyzed.

  • Infiltration through the river bed, or induced surface water capture by wells, does not alter surface water chemistry.

  • Groundwater samples are also a linear mixing between the defined EM. Under forced losing stream conditions, the contribution of the subsurface pole in these samples may be indistinguishable from both the regional groundwater flow and the percentage of groundwater present in the recharging (captured) surface water.

The Pearson correlation between the mentioned components (chloride, sulfate, sodium, potassium, and bromide) reveals that sodium has a coefficient larger than 0.90 for all the other components, except bromide (Table 2). Since sodium could be considered as linearly dependent on the other variables, it has not been used in the EM mixing analysis, resulting in a system of three end-members and four components (Table 3).

Table 2 Pearson correlation coefficients between the components of all water samples
Full size table
Table 3 Contribution of the different EM in the hydrochemical composition of surface water of the Tordera River and groundwater of its alluvial aquifer
Full size table

Bivariate plots of chloride, sulfate and potassium against bromide (Fig. 7) indicate that Tordera River stream water and groundwater both lie within the ternary mixing of the Tordera River upstream EM (represented by sample 1), the WWTP pole (sample A) and the groundwater pole (sample P1), for those samples located upreach from industry input C. Since the industry C download has the largest ion concentration, sampling points for surface water and groundwater located downstream will represent the mixing of four EM: namely, the three already mentioned plus the industry C. From a strictly hydrological perspective, any other input to the Tordera River, such as industry D and the Pertegàs (point B) and Gualba (point E) creeks would actually represent additional EM. Nevertheless, for the sake of simplicity and since their mass contribution is low, they are not included in the performed mixing analysis.

Fig. 7
figure 7

Bivariate plots between different components considered representative of each EM. Lines represent mixing path between EMs. EM samples and legend are indicated only in the first plot, being the same for all the others. EM1: Tordera River sample #1; EM2: WWTP; EM3: Groundwater pole; EM4: Industry C

Full size image

Samples upreach from industry C (samples 1 to 4) show an increase in conductivity; except at point 2 where the conductivity decrease is attributed to non-homogeneous fluvial dispersion of the Pertegàs Creek discharge. Mixing proportions for samples 3 and 4 can therefore be considered representative (Table 3). Following the analysis based on the three end-members mentioned, and chloride, sulfate, and bromide as components, surface water in this first reach shows a contribution of up to 20 % of groundwater attributed to a regional flow system, while the WWTP input is as much as 36 %, and the rest belongs to the upstream discharge of the Tordera River.

Notice that the groundwater contribution at point 3 is 0.60 m3/s, which approximately corresponds to the stream discharge difference recorded between points 2 and 3. This independent observation provides a reasonable corroboration of groundwater as the origin of the extra local water source identified at this location.

Downreach from industry C, the overall contribution of groundwater in this second reach diminishes to approximately 10 % all the way to sampling point 10. At the same time the percentage related to the WWTP is maintained around 30 %, and that of the EM corresponding to the Tordera River increases to 60 %. Because of its low discharge, the contribution from the industry C input is below 3 %.

Plots of sulfate and potassium versus bromide in this second reach already suggest a shift of the surface water samples to a linear combination of the EM related to the Tordera River and the EMs associated with the WWTP and industry C (Fig. 7), with a minimum contribution of the groundwater pole. Since an increase in the first EM is hydrologically unfeasible, we attribute this hydrochemical shift to some unnoticed wastewater dumping from small hamlets or homesteads along the course of the river with a composition close to that of this EM in order to explain the reduction of the groundwater percentage. A similar reasoning is necessary to explain the percentages observed from surface water samples 11 to 13.

Regarding groundwater, except for water from wells P1 and P2, which has been defined as the composition that comes closest to representing the non-altered hydrochemistry of the groundwater, samples from wells P3 to P8 reflect a large contribution from surface water (up to 80 %) as a result of pumping-induced depletion. The percentage of the groundwater pole varies from 22 to 40 %, indicating that upward recharge from the basement through fractures is tangible and spatially variable. Moreover, the supposed predominance of the Tordera River EM is false and hydrologically unrealistic. It should show the contribution of the WWTP which remains relatively unnoticed in percentage after P4. Consequently, a hydrochemically-similar, substitute EM must be found to account for an extra input that masks the effect of the WWTP. In this case, local rainfall infiltration in the alluvial formation and intense urban and industrial development may provide a chemical fingerprint akin to that of surface water sample 1, which also corresponds to water resources altered by human activity.

6 Conclusions

Overexploited stream-related aquifers usually induce a loss of surface discharge that balances groundwater abstraction. The analysis of discharge and hydrochemical surface water data along a reach, and that of its associated groundwater aquifer, provides evidence that the supplementary recharge produced by upward fluxes from the geological basement compensates the water budget. Specific geological structures, as faults, govern such additional water contributions.

This study has revealed that groundwater contributes to surface water at the beginning of the investigated reach of the Tordera River. This contribution is attributed to preferential large-scale groundwater flows through regional fault lines. End-member mixing analysis shows that this specific contribution was as much as 20 % of the total discharge, which also explains the unexpected rise in stream discharge observed at the first monitoring points. Nevertheless, it has been observed under specific flow conditions, and its magnitude may vary depending on the general hydrological conditions of the study area. Moreover, several minor end-members may also contribute to the surface water hydrochemistry. A careful hydrological discussion of the percentages attributed to the main end-members reveals the role of specific secondary contributions, which may also be of interest to stream managers and policy makers.

From a management perspective, this finding highlights basement groundwater as a potential water supply alternative. The wise exploitation of such resources would ease pressures on the alluvial aquifers and the stream flow, and achieve the recommended in-stream discharge more frequently. Moreover, it adds a new aspect to water resources planning that had not been previously considered, simply because industrial demand was already satisfied by capturing stream water despite well-known environmental impacts.

In light of these results, the conjunctive use of surface water and groundwater resources at different scales can now be considered as an opportunity in river basin management plans. Companies that are responsible for impacts must be encouraged to participate in hydrological assessment, and compelled to develop further hydrogeological research through private investment. This research ought to look for supplemental, alternative water sources that may eventually help to reduce environmental costs associated with meeting water needs.