Hydrogeology Journal DOI 10.1007/s10040-014-1170-9

Multi-method assessment of connectivity between surface water and shallow groundwater: the case of Limarí River basin, north-central Chile Ricardo Oyarzún & Felisa Barrera & Pamela Salazar & Hugo Maturana & Jorge Oyarzún & Evelyn Aguirre & Pablo Alvarez & Hervé Jourde & Nicole Kretschmer Abstract A study that tests the applicability and consistency of independent but complementary approaches in the assessment of interactions between surface water and shallow groundwater within a water-stressed basin is described. The mostly agricultural Limarí basin in arid north-central Chile was chosen as a suitable case study. The analyses involved: (1) a connectivity index method, (2) hydrochemistry, and (3) water isotopic geochemistry. Chemical and isotopic data were obtained from two sampling campaigns conducted in April (fall) and December (summer) of 2011 in 22 sampling locations, which included surface water and groundwater. The results obtained by each of the methodologies were mutually consistent and indicate high connectivity conditions. Additionally, the relative contribution by different sources was assessed through end-member mixing analysis, and for reaches of the river that showed gaining conditions,

Received: 18 December 2013 / Accepted: 3 July 2014 * Springer-Verlag Berlin Heidelberg 2014 R. Oyarzún : H. Maturana : J. Oyarzún Departamento Ingeniería de Minas, Universidad de La Serena, Benavente 980, La Serena, Chile R. Oyarzún ()) : N. Kretschmer Centro de Estudios Avanzados en Zonas Aridas, Av. Raúl Bitrán s/n, La Serena, Chile e-mail: [email protected] Tel.: 56-51-204503 F. Barrera : P. Salazar Ingeniería Civil Ambiental, Universidad de La Serena, Benavente 980, La Serena, Chile E. Aguirre Comisión Chilena de Energía Nuclear, Avda. Nueva Bilbao No. 12.501, Las Condes, Chile P. Alvarez Departamento de Agronomía, Universidad de La Serena, Av. La Paz s/n, Ovalle, Chile H. Jourde Université Montpellier 2 - UMR HydroSciences Montpellier, Av. Jeanbrau 300, Montpellier, France

the contribution of groundwater inflow to stream discharge was estimated. It is suggested that this multimethod approach is useful for the characterization of surface-water–groundwater interactions, since it at least represents a suitable starting point for obtaining basic information on these relationships. Thus, it may become the base for further studies in arid and semi-arid basins facing water management challenges. Keywords Stable isotopes . Radon . Over-allocated basin . Arid regions . Chile

Introduction Arid and semi-arid zones cover approximately 40 % of the Earth’s surface (Jolly et al. 2008). These zones are typically found between latitudes 10 and 35 °, immediately north and south of the major tropical convergence zones (Simmers 2003). There is a current increase in water demand in many semi-arid and arid zones worldwide due to factors such as population and industrial (agriculture and mining) growth, at a time when the meager water resources are threatened by climate change (CONAMA 2006; Shi et al. 2001; Souvignet et al. 2010; Vicuña et al. 2012). This is precisely what is happening in northern central Chile where the Limarí watershed, the focus of this study, is located. Additionally, proper understanding of surface-water/groundwater (SW–GW) connectivity, i.e., “the direction and magnitude of flow between water resources located above and below ground” (Geoscience Australia 2013) is a key issue for integrated water resources management (Kalbus et al. 2006). In this regard, SW–GW relationships have recently started to be incorporated into water-related legislation adopted by a number of developed countries such as Australia, where regulation includes the management of surface water and groundwater for rural and urban use as a single resource (Fullagar 2006). At the basin scale, hydrologic exchange between surface water and groundwater is determined by: (1) the spatial pattern and magnitude of hydraulic conductivities,

both in the channel and alluvial-floodplain sediments, (2) the relation of stream stage to the adjacent groundwater level, and (3) the geometry and position of the stream channel relative to the alluvial floodplain (Jolly et al. 2008). Two basic cases of water flow are recognized: (1) an effluent condition or gaining stream, where subsurface water drains into the stream, and (2) an influent condition or losing stream, where surface water contributes to groundwater flow (Bailly-Comte et al. 2009; Dor et al. 2011). Since the behavior of any given SW–GW system is dynamic (Bailly-Comte et al. 2009), a river could be receiving groundwater in certain parts and losing groundwater in other reaches (Kalbus et al. 2006; Newman et al. 2006). Likewise, exchanges between groundwater and surface water are unsteady and seasonally variable (Andersen and Acworth 2009; Jourde et al. 2012; Konrad 2006). There is a large amount of literature on the different methods and techniques available for the characterization of SW–GW interactions at different scales (e.g., Kalbus et al. 2006; Brodie et al. 2007). A very simple approach is one by Ransley et al. (2007), who presented a desktop method to determine SW–GW interactions, designed to be used “as a catchment scale screening tool for identifying potential connection between surface water and groundwater resources”. According to this approach (which will be further addressed and discussed in the methodology section), the SW–GW connectivity potential is determined by means of a rating index approach, based on four data inputs: (1) depth to the water table, (2) stream bed sediments, (3) geology, and (4) geomorphology. These types of indices serve multiple purposes and provide an initial assessment that can be used as a basis for more detailed studies (Parsons et al. 2008). Of course, only the verification of stream–aquifer conditions from field-derived data will allow their scientifically sound evaluation, and make them reliable for water resource planning and management purposes. In this regard, the use of environmental tracers is a commonly used technique for the field assessment of SW–GW relationships, as these have proved to be excellent indicators of spatial and temporal patterns of shallow groundwater/stream exchanges (Constantz et al. 2003). Among the typically used tracers are: (1) common dissolved water constituents such as major ions, (2) water stable isotopes of oxygen (18O) and hydrogen (2H), and (3) radioactive isotopes such as radon (222Rn; Winter et al. 1998). While the use of a single tracer in a given situation could be inadequate to properly characterize hydrologic relationships between surface and groundwater compartments, better results are obtained when complementary tracers are simultaneously used (Andersen and Acworth 2009; Burtnett et al. 2010; Cook et al. 2003; Kalbus et al. 2006; Stellato et al. 2008). Moreover, when this information is analyzed through the use of multivariate statistical techniques such as hierarchical cluster analysis (HCA), a comprehensive understanding may be achieved (Demirel and Güler, 2006; Güler et al. 2002; Tallini et al. 2013). Hydrogeology Journal

Despite harsh aridity conditions, the Coquimbo Region in northern central Chile has significant agricultural activity, with approximately 75,700 ha of irrigated crops (INE 2007). Within the region, the Limarí basin represents the most important agricultural area, where a sustained increase in irrigated land has been taking place during the last 15 years (CIREN 2011). This motivated the official declaration of Limarí as an “over allocated catchment” by the Chilean Water Authority (Dirección General de Aguas: DGA) in 2005 (DGA 2008). As a result of this decision, no new permanent consumptive surface-water rights can be granted within the basin; thus, groundwater resources that are already scarce and under increased demand, face a growing stress. In fact, consumptive groundwater rights granted in the basin increased from 373 l/s in 1994 to 3,982 l/s in 2012 (DGA 2012), which does not necessarily correspond with an increment in groundwater availability. Moreover, the area has recently been experiencing a severe ongoing drought for several years, and evidence points towards the occurrence of a structural change in precipitation regime, and therefore, towards persistent dry conditions in the next decades (Núñez et al. 2013). When groundwater is extracted from shallow wells located in the recent alluvial river floodplain, a very complex situation unfolds. When these “minor stream” systems (i.e., thin alluvial formations, few cubic meters per second of stream discharge rates) are subjected to intense groundwater extractions, important hydrological and ecological disturbances arise, such as loss of riparian biodiversity or salinity increases in wetlands(Jolly et al. 2008; Mas-Pla et al. 2013). Also, in the case of Chile, the liberal water management scheme adds further complexity to the system. Currently, groundwater and surface water are legally two different “assets” in Chile. However, given that water flow does not take into consideration any legal framework, but simple hydrological rules (e.g., SW–GW interactions), a number of disputes have arisen between water users, some even reaching civil suit status (Alvarez and Oyarzún 2006). These conflicts are difficult to settle, because, as recognized by the DGA, “there is very limited information on the degree of interaction between rivers and aquifers in the Limarí basin” (DGA 2008), which is currently a fairly common case in Chile and elsewhere. In this regard, some preliminary efforts have recently been made to improve the existing understanding on these issues, particularly in Limarí (e.g., Strauch et al. 2009; Oyarzún et al. 2013). However, these studies have been limited to one-time sampling campaigns that lack sufficient groundwater data, or to the assessment of the use of specific tracers such as 222Rn. Thus, the need still exists for the application and testing of more comprehensive and integrated approaches to better characterize river– aquifer interactions. The aims of this contribution are three-fold: First, to spatially characterize SW–GW interactions occurring in the Limarí River basin; second, to assess the consistency among several approaches for the identification of SW– GW interactions; and third, to obtain a preliminary quantification of aquifer–river transfer rates. It is expected DOI 10.1007/s10040-014-1170-9

that the methodology and the results presented in this report will be useful for a better assessment of SW–GW relationships and, hence, for the definition of sustainable and integrated water-resource-management strategies in arid and semiarid basins worldwide facing similar challenges to those currently existing in Limarí.

Ovalle, the Limarí River exhibits a wider alluvial plain and Pliocene to Quaternary fluvial, and partly marine terraces cover an important part of the basin.

Methodology Connectivity index

Climatic, hydrological, and geological setting The Limarí basin is located in the mostly arid Coquimbo Region, north-central Chile (Fig. 1). This study focuses on the lower area of the catchment, downstream La Paloma (750 million cubic meters, Mm3) and Recoleta (100 Mm3) water reservoirs. The climate is characterized by mild temperature and humidity, with an average annual rainfall of 130 mm and an annual potential evapotranspiration of over 1,200 mm. Since precipitation is highest between July and September, the dry period lasts 8–9 months (DGA 2004). The Limarí River is formed by the confluence of the Grande and Hurtado rivers, approximately 4 km upstream from Ovalle, the main city in the catchment. From the confluence of the Hurtado and Grande rivers, the watercourse is named Limarí and reaches the Pacific Ocean 60 km downstream. Although variable in space and time, the annual average flows of the rivers Grande, Hurtado and Limarí are approximately 3.9, 1.2, and 2.3 m3/s, respectively (DGA 2004). Given that the flow of the Hurtado and Grande rivers is regulated by the Recoleta and the Paloma dams, respectively, discharge immediately downstream the dams rarely represents the concurrent inflows to the reservoirs. Moreover, delivery from the dams to the rivers, as well as river water diversions (irrigation channels), reach their peak between November and March, during the driest part of the year, the irrigation season. Between Ovalle and the Limarí mouth in the Pacific Ocean, the river receives three minor tributaries: El Ingenio, where a formerly active mining facility exists, contributing to high salinity levels, mainly due to sulfate concentrations; La Placa Creek; and Punitaqui Creek. All three tributaries originate in the Coastal Cordillera zone, and are mainly fed by pluvial precipitations (DGA 2004). The general geological traits of the Limarí basin were established by Thomas (1967). The geological setting of the Limarí River basin expresses the almost continuous generation of calc-alkaline magmas, a consequence of the subduction of Pacific oceanic tectonic plates under the western edge of the continent. Granodioritic composition is dominant in plutonic rocks. Basaltic to andesitic composition is frequent in volcanic flows. Lastly, andesitic to dacitic compositions are common in pyroclasts and in clastic sedimentary beds that intercalate in the volcanic series. As a consequence of the igneous origin of most of these rocks, permeability is largely secondary and restricted to fault zones. Eastward from Ovalle, alluvial sediments are restricted to the narrow plains of the steep tributary rivers; however, west of Hydrogeology Journal

Based on the work of Braaten and Gates (2003), who used depth to the water table as a criterion to characterize a SW–GW system as either connected or disconnected, Ransley et al. (2007) proposed a simple method to determine the river–aquifer potential connectivity. This latter approach considered four sources of information (categories and nomenclature have been kept as originally presented by Ransley et al. 2007): the water-table depth (dw), the river-channel sediments (rs), the dominant geology (ge), and the site geomorphology (gm), each category with specific scores (Table 1). The method determines a weighted connectivity index (CI) as: CI ¼ ð3ÞðdwÞ þ ð5ÞðrsÞ þ ð5ÞðgeÞ þ ð2ÞðgmÞ

ð1Þ

There is no further information in the Ransley et al. (2007) work on how the scores for each factor are defined; however, regarding weights, these authors state that sensitivity analysis was conducted for their determination. In order to use this approach, the area of study was segmented into several reaches based on both location of surface and groundwater sampling points (addressed in the following), as well as on the presence of DGA monitoring wells (Fig. 1). Long-term water-table-data series were obtained from these DGA wells, which in most cases were situated less than 1 km from the river (Fig. 1; Table 2), a suitable distance for these analyses (e.g. Braaten and Gates 2003; Ivkovic 2009). In order to incorporate seasonal variations into the analysis, the data set was divided into three different 4-month periods, referred to as “summer” (November to February), “fall” (March to June), and “spring” (July to October), and an average water-table depth value was obtained for each of these periods. Geology information was obtained from the CHI– 535 report (SERPLAC et al. 1979), which is one of the most complete hydrogeological studies of the Limarí catchment. River channel sediment was characterized from field observations as well as from well stratigraphy data available at DGA. Finally, geomorphology was considered as a depositional environment, based on field observation (i.e. stream channel shape, alluvial sediments size), as well as river sinuosity and slope of the stream; the latter two features were obtained from analysis of Google Earth images and topography maps (1:50.000). Ransley et al. (2007) did not propose any specific range or criteria for CI classification. In fact, they simply state that “the higher the connectivity index, the greater the potential for stream–aquifer connectivity”. In the current work, a value between 53 and 75 is defined as high connectivity condition, a value between −24 and 53 is indicative of medium connectivity, and a value between DOI 10.1007/s10040-014-1170-9

Fig. 1 Surface water (SW) and groundwater (GW) sampling points, Dirección General de Aguas (DGA) monitoring wells, stream reaches (R1–R11), and rainfall collectors (CT Carretera station; LP La Paloma station) in the area of study

−24 and −78.5 is considered as a low connectivity condition. The maximum and minimum scores (75 and −78.5) result from considering the “best” and “worst” factor combination for connectivity (Table 1; i.e., “best”: water-table depth less than 10 m; sand/gravel for river channel sediments; erosion environment for geomorphology; gravel and sand for geology; “worst”: water-table depth greater than 20 m; clay for river channel sediments; hill top for geomorphology; clay for geology). Likewise, a Table 1 Parameters, classes and scores for the connectivity index method (after Ransley et al. 2007) Parameter

Class

Score

Water-table depth

<10 m 10–20 m >20 m Sand/gravel Sandy loam/silty loam Silt/clay loam Clay Gravel/sand Clay/sand Clay Erosional environment Depositional environment Hill top

5 3 0.5 5 3 −1 −4 5 3 −4 5 1 0

Channel bed sediments

Geology Geomorphology

Hydrogeology Journal

mid-range value of 31 is obtained for intermediate conditions (e.g., water table between 10 and 20 m, sandy loam or silt loam channel-bed sediments, clay or sand dominance in the aquifer formation, and depositional environment). From these values (e.g., −78.5, 31, 75), the category thresholds (−24 and 53) were obtained as the mean scores between them (i.e., from −78.5 to 31 and from 31 to 75, respectively).

Field sampling work Sampling campaigns and analytical methods Two sampling campaigns were performed in this study: one in April 2011 (fall, at the end of the irrigation season), and one in December of the same year (early summer, in the peak of the irrigation season). In both campaigns, 13 samples of surface water and 9 samples of groundwater were taken at several sites throughout the study area (Fig. 1). Locations included the area of the rivers Grande (G1, G2, G3), Hurtado (H1, H2, H3), Limarí (L1–L10), and an area of the El Ingenio Creek (E1–E6). Most groundwater samples were obtained from shallow wells (less than 40 m depth) that locally provide drinking water to rural communities (“agua potable rural”, APR). DOI 10.1007/s10040-014-1170-9

67.0 67.0 67.0 67.0 57.0 57.0 67.0 67.0 67.0 57.0 41.0 47.0 47.0 57.0

High High High High High High High High High High Medium Medium Medium Medium

Connectivity category CIa

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

Geomorphology score

5.0 5.0 5.0 5.0 3.0 3.0 5.0 5.0 5.0 5.0 3.0 3.0 3.0 3.0

Geology score

5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 3.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 3.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 3.0 5.0 5.0 5.0 1.67 2.93 2.16 5.00 0.47 1.52 3.43 3.48 1.70 0.64 13.0 1.79 1.34 3.93 1.68 2.94 2.21 4.78 0.47 1.37 3.32 3.31 1.86 0.67 12.59 1.59 1.08 4.23 1.71 2.97 2.10 4.82 0.46 1.72 3.14 2.94 1.65 0.68 13.39 1.70 0.59 4.32

Hydrogeology Journal

CI connectivity index a

R5 R6 R7

R4

R3

R2

1978–1988 1978–2010 1969–2010 1978–1999 1978–2010 1982–2010 1970–2010 1970–2010 1976–1997 1970–2010 1969–2010 1989–2010 1975–2003 1970–2010 R1

BA1 BA2 CT PT SJ1 SJ2 P24 P13 VL FM AS CA EP PP

921 1,184 599 967 863 768 928 1,273 193 393 460 297 80 291

Spring

5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 3.0 3.0 3.0 3.0 5.0

Stream channel sediment score Water-table depth score Summer Fall Average water-table depth [m] Summer Fall Spring Distance, well to river [m] Record (year range) DGA well Reach

Table 2 Degree of connectivity calculated for each stream reach of the studied area

Surface-water samples were obtained as close as possible to groundwater sampling locations, taken as far as possible from the river border, ideally at the center of the stream flow. In addition to the surfacewater sampling, both stream channel characteristics (i.e., width, water depth, alluvial sediments size, stream channel configuration) and stream flow velocity were recorded, the latter using a Hydro-Bios current meter (HB-445510). At each sampling location, two 1-L bottles were filled for subsequent chemical analysis. One bottle was acidified (HNO3, 2 % v/v) and used for cation determination; the second bottle was not acidified and used for anion determination. A Hanna HI982804 multiparameter probe was used for field determinations of total dissolved solids (TDS), electrical conductivity (EC), dissolved oxygen (DO), and pH. Water samples for stable isotope analysis (2H, 18O) were also obtained and stored in 125-ml plastic bottles, sealed with parafilm. Finally, water samples for 222 Rn analyses were collected in 250-ml glass bottles, filled up completely under water to avoid contact with the atmosphere, and hermetically closed. All the samples were kept refrigerated at approximately 4 ° C during field campaigns, and later in the laboratory facilities at the University of La Serena. The samples for chemical analyses were filtered (0.45 μm) the day after sampling and subsequently transported to the Geoquímica Laboratory in Coquimbo. The analyses were performed following the procedures established in Standard methods for the examination of water and wastewater, 20th edition (Clesceri et al. 1999). Measuring techniques (and detection limits in mg/l) were atomic absorption for Ca2+ (0.1), Mg2+ (0.01), K+ (0.05), and Na+ (0.05); volumetry for HCO3− (1) and Cl− (1); and gravimetry for SO42− (10). Isotopic determination of oxygen and deuterium were performed at the Environmental Isotope Laboratory of the Chilean Nuclear Energy Commission (CCHEN) in Santiago, via laser absorption using a Liquid Water Stable Isotope Analyzer (Los Gatos Research, CA, USA). Isotope ratios are expressed in per mil (‰) using the usual δ notation:
dð‰Þ ¼ Rsample =Rstandard – 1 1000 ð2Þ where R refers to either 2H/1H or 18O/16O ratio of a sample and the standard, respectively. The standard for both δ2H and δ18O was VSMOW. Finally, the 222Rn analyses were performed using alpha spectrometry on RadH2O equipment (Durridge Co., MA, USA) at the Environment Laboratory of the University of La Serena, no later than 2 days after sampling. Corrections were made to account for radioactive decay between time of sampling and time of analysis in the laboratory.

Meteoric water line Rainfall water samples were obtained during 2011 on a rainfall event basis, and rainwater was stored in a bottle DOI 10.1007/s10040-014-1170-9

with a mineral oil layer to avoid evaporation and isotopic enrichment (Friedman et al. 1992). Four collectors were installed both in the area of study at Carretera (214 meters above sea level (masl), near the town of Barraza), and La Paloma dam (335 masl, near the water reservoir), as well as upstream the study area at Tulahuén (987 masl), and Las Ramadas (1,380 masl) (both in the Grande River catchment, upstream La Paloma Dam, not shown in Fig. 1). Water samples were collected and isotopic values were weighted by the rainfall amount of each event to obtain a representative isotopic signature, and hence, the local meteoric line. Stable isotopes analyses (18O, 2H) were carried out at CCHEN as previously described.

Data treatment Graphical analysis To characterize the general pattern of major ion distribution, both Piper and Stiff diagrams were used. With respect to stable isotopes, δ2H vs δ18O scatter plots were obtained, including global and local meteoric water lines. Lastly, the spatial distribution of 222Rn activities was assessed.

Comparative SW–GW connectivity analyses In order to characterize connectivity as high, medium or low based on water chemistry and water stable isotope compositions, and to later compare the results with those obtained from the connectivity index approach, some criteria were arbitrarily defined (i.e., expert judgment) based on the chemical composition and the isotopic signature of the water samples. These criteria were:

1. Based on major ions. It was considered that a difference of less than 20 % between a surface-water sample and a groundwater sample located in the same stream reach would indicate a high connectivity case; a difference between 20 and 40 % a medium connectivity condition; and a difference of more than 40 % a low connectivity condition. These differences in percentages were determined from the location of the samples under comparison in the central rhombus of the Piper diagram. 2. Based on water stable isotopes. The range of recorded isotopic signatures was first determined (for both 2H and 18O) and the same criteria (20 and 40 % differences) were used. These thresholds corresponded to ±4 and ±9 ‰ for 2H, and ±0.7 and ±1.4 ‰ for 18O. The rationale behind this analysis is that under connected conditions (either losing or gaining), surface water and groundwater composition (chemistry, isotopes) should resemble each other, whereas under disconnected conditions, it is more likely that Hydrogeology Journal

the compositions will differ, given the different sources of surface water and groundwater as well as changes occurring during the slow vadose zone flow. Thus, three connectivity classifications were available for each stream reach (i.e., major ions, 2H, and 18 O), and the most repeated category was kept for the chemistry- and isotopic-based characterization of the stream. In case a stream reach received a different classification based on hydrochemistry, 2H and 18O (e.g., medium, high, low respectively), a medium connectivity was assigned. For those stream reaches that included more than two sampling locations (R5, R7, R8, and R9), the two closest surface water and groundwater samples were considered for the analysis. These results were compared to those obtained with the CI method. Multivariate analysis As discussed by Guggenmos et al. (2011) and King et al. (2014), a suitable complement to the qualitative assessment of SW–GW interaction based on water composition (chemistry, isotopic signature) corresponds to multivariate statistical analysis such as hierarchical cluster analysis (HCA). This is a statistical multivariate technique that allows the grouping of samples (Q-mode) based on their similarity, and was hence used in this study, integrating hydrochemistry (pH, EC, Na+, K+, Ca2+, Mg2+, Cl, SO42−, HCO3−), stable isotopes (2H, 18O) and radioactive isotopic (222Rn) data. The Ward linkage method and Euclidean distance measurement technique were considered, as it has been shown that they are effective in cluster determination in water studies (Güler et al. 2002; Thyne et al. 2004). Regarding the selection of the number of clusters, some authors (e.g., Astel et al. 2007; Shrestha and Kazama 2007) consider the Sneath index, which is based on the 2/ 3 Dmax threshold (being Dmax the maximum separation distance), whereas other authors use an arbitrary selection criteria (e.g., Thyne et al. 2004). Both approaches were followed in this study.

Groundwater contribution assessment The previous analysis yields information about the degree of connectivity, but not about the type of connection (i.e., losing or gaining stream). This issue was preliminarily addressed through end member mixing analysis (EMMA), and considering the 222Rn activities, as this radioactive isotope has been proven useful for this purpose (e.g., Cook et al. 2003; Oyarzún et al. 2013). An EMMA, assuming a priori three end-members, i.e., surface water that is delivered by dams, local precipitation, and recycled water (irrigation water that infiltrates and recharges shallow groundwater and eventually contributes to river flow), based on the preliminary knowledge of the area (e.g. Strauch et al. 2009; DOI 10.1007/s10040-014-1170-9

Oyarzún et al. 2013), was performed. The contribution ratio (R) from these potential sources, considering as tracers 18O and chloride, was estimated as (following Liu and Yamanaka 2012):

ðδs − δi Þ cp − ci −ðcs − ci Þ δp − δi

Rd ¼ ð3Þ ðδd − δi Þ cp − ci − ðcd − ci Þ δp − δi

Rp ¼

ðδs − δi Þðcd − ci Þ − ðcs − ci Þðδd − δi Þ

δp − δi ðcd − ci Þ − cp − ci ðδd − δi Þ

Ri ¼ 1 – Rd − Rp

ð4Þ

ð5Þ

where δ is the δ18O, c is the Cl− concentration, and s, d, p and i denote sampled water at each location, dam-derived surface water, local precipitation and infiltrated-recycled water, respectively. Regarding the use of 222Rn, four arbitrary 222Rn activity classes were defined, based on a logarithmic distribution of the range of the measured values. For those reaches where gaining stream conditions were inferred based on “elevated” 222Rn levels in surface water, i.e., higher than 1,000 Bq/m3 (Green and Stewart 2008; Oyarzún et al. 2013), and when surface-water samples were available at the beginning and end of the section, the rate of water transfer (aquifer to stream) was estimated following two methods. The first method is based on Stellato et al. (2008), which is done by calculating a theoretical 222 Rn activity (Rncalc) downstream of a given river reach from the upstream 222Rn record (Rnu), i.e., Rncalc ¼ Rnu e

−

1 ðDV Þ =2 x 3= V h 2

− λ Vx

ð6Þ

where D is the molecular diffusivity of 222Rn (cm2/s, calculated from temperature, T [K], by means of –logD= 980/T), V is the stream velocity (m/s), h is the stream water depth (m), x is the distance between upstream and downstream locations (m), and λ is the radioactive decay constant of 222Rn (2.08×10−6l/s). Once Rncalc is obtained, it is compared with the actual (i.e., measured) 222Rn activity downstream the given stream reach (Rnobs). It is assumed that any deviation from the theoretical value will denote incoming water from the aquifer, which can be quantified as Qgw Rnobs − Rncalc ¼ Qr Rngw − Rncalc

ð7Þ

where Qgw/Qr corresponds to the fraction of groundwater discharged to surface water between the two sampling locations, and Rngw is the 222Rn activity determined for groundwater in the reach area. Hydrogeology Journal

The second approach was based on Cook et al. (2003) and is the solution of a flow model that simulates stream longitudinal 222Rn profiles as: Qr


dc ¼ Qgw Rngw − Rnobs þ ωERnobs þ kωRnobs − hωλC dx

ð8Þ

where Qr is the river discharge (m3/day), Qgw* is the groundwater discharge per river unit length (m3/m−1 day); ω is the stream channel width (m); E is the evaporation rate (m/day); k is the gas transfer velocity across the water surface (m/day) for which a value of 1 was assumed (after Cook et al. 2003); and the remaining symbols are as explained in the preceding. Note that nomenclature has been slightly modified in Eq. (8) from the one described by Cook et al. 2003 in order to make it consistent with Eq. (7). As mentioned, these two independent approaches (Stellato et al. 2008, and Cook et al. 2003) were applied to those stream reaches where gaining conditions were previously identified.

Results and discussion Connectivity index The results for the connectivity index method determined for each stream reach, based on the information on watertable depth, geology, geomorphology and stream channel sediments, are shown in Table 2. In most of the reaches, the method yields a high connectivity condition, the only exceptions being reaches R6 and R7 with medium connectivity. Note that the method was not applied for R8 to R11 given the lack of DGA wells and, therefore, water-table depth data. Regarding water-table depth, data were considered separately for summer, fall and spring, with no relevant seasonal variations on this parameter in any of the wells. Thus, it is possible to infer a rather stable behavior of the hydrogeological system in the study area, at least on a long-term time scale, with seasonality not being an important issue in terms of major changes in groundwater depth. With the exception of the “Agua potable Sotaquí 2” (AS) well, all the DGA monitoring wells show shallow water-table depths, and received a score of 5. In AS the water-table depth is greater than 10 m, for which a score of 3 was assigned. Whatever the case, this situation does not greatly affect the overall results since water-table depth has a lower weight than stream channel sediment or local geology when calculating the connectivity index (Eq. 1). In fact, for reaches R6 and R7, the high presence of loam and clay sediments in the alluvial deposits greatly contributes to the determination of their “medium connectivity” condition.

Piper and Stiff graphic analysis Figure 2 presents the Piper diagrams for all the samples and the two sampling campaigns, and Table 3 shows the main water types for each sub-basin according to the disposition of DOI 10.1007/s10040-014-1170-9

Fig. 2 Piper diagrams of surface water (open symbols) and groundwater (filled symbols) samples in the a fall and b summer sampling campaigns

the samples in the Piper diagram. Two main groups can be distinguished (denoted as 1 and 2 in the figure) as well as an outlier that corresponds to a single sample, with a similar pattern during both sampling periods. The first group (1) includes all the samples of the Grande and Hurtado rivers, as well as samples E1 and E2 Hydrogeology Journal

of El Ingenio Creek, and samples L1–L4 on the upper Limarí River. The second group (2) is formed by samples of the downstream section of El Ingenio Creek (E4, E5, and E6) and from the mid and lower Limarí River sections (L5–L10). In the latter group, chloride content increases in the direction of the flow and is generally higher than that DOI 10.1007/s10040-014-1170-9

Table 3 Sub-basin water type classification (n number of samples) Sub-basin Hurtado Grande El Ingenio Limarí

Component Surface water Groundwater Surface water Groundwater Surface water Groundwater Surface water Groundwater

n

Fall

2 1 2 1 4 2 5 5

Ca –SO4 Ca2+–SO42− Ca2+–HCO3− Ca2+–HCO3− Na+–SO42− Na+–Cl− Na+–Cl− Na+–Cl−

of the first group. Finally, the E3 sample (the outlier ‘group 3’ in Fig 2) in the El Ingenio Creek departs from all the other samples given the high sulfate content in the water (over 1,000 mg/l in both campaigns) as a consequence of the long-term effects derived from an abandoned metallurgical facility (Panulcillo), located immediately upstream the sampling location. Two main hydrochemical zones (related to two distinct water types) are recognized in the study area. Given the close location of the neighboring surface water and groundwater in the diagram, a high degree of river– aquifer interaction can be inferred. In fact, surface water and groundwater within each sub-basin area exhibit similar compositions (Table 3). The spatial distribution of these water types is clearly seen on the Stiff diagrams (Fig. 3), where, based on the shape and size of the polygons, it is possible to observe the spatial distribution of the two main groups and the isolated sample previously identified in the Piper diagrams. In summary, the dominant cations are Ca2+ and Na+, with Ca2+ being the most important in the upper reaches of the basin, and Na+ being more relevant in the lower sections of the area (mainly Limarí River). Regarding anions, the most important are Cl−, especially in the Limarí River, and SO42− in the El Ingenio Creek, and to a lesser extent, in the Hurtado River. Also, HCO3− is present in higher proportions (with respect to SO42− and Cl−) in the Grande River, downstream La Paloma Dam.

Stable isotopes Figure 4 presents the isotopic signature of surface water and groundwater considering both sampling campaigns, as well as both the global and local meteoric water lines. All the samples plot below the meteoric water lines, and the slope of the regression lines for surface water and groundwater in both sampling campaigns (6.1–7.0 and 7.0–7.6 ‰ respectively) is lower than the 8 and 8.9 values for the global and local meteoric water lines, which indicate important evaporation processes in the area (Dor et al. 2011; Yin et al. 2011). There is an overall low dispersion in the isotopic signature in the sampled water, which suggests a similar origin, and therefore, an important relationship between the surface water and groundwater components. For the Hydrogeology Journal

2+

Summer 2−

Ca2+–SO42− Ca2+–SO42− Ca2+–HCO3− Ca2+–HCO3− Na+–Cl− Na+–Cl− Na+–Cl− Na+–Cl−

fall campaign, 2H and 18O ranged from −75 to −51 ‰ and from −10 to −6 ‰, respectively. For the summer campaign, the intervals were very similar, i.e., from −71 to −46 ‰ and −9 to −5 ‰ for 2H and 18O. It is also possible to identify three main groups which are rather similar in both campaigns, as shown in Fig. 4. The first group includes two of the three samples collected in the Hurtado River area (H2, H3). Highly depleted values were observed, departing from what is expected from local rain such as that collected at the nearby La Paloma station (around −8 ‰ for 18O and −50 ‰ for 2 H), which may indicate that this water does not correspond to infiltrated water coming from the immediately upstream Recoleta Dam or to local precipitation, but rather to contributions of recharge areas in the upper parts of the basin. In other words, these are high altitude recharged waters that exfiltrate into the sampling area. The second group includes the more enriched samples, i.e., G1 and G2 (Grande River) and L9 and L10 (Limarí River) in the fall sampling campaign, and L8, L9 and L10 (lower part of the Limarí River) in the early summer campaign. Lastly, the third group, situated between the previous two, includes samples from the El Ingenio Creek and those from the upper part of the Limarí River, as well as G1 and G2 (Grande River) in the second campaign. As previously discussed for the chemical composition, surface water and groundwater of neighboring sampling locations exhibit similar isotopic compositions, and therefore are close in the δ18O vs. δ2H diagram. This fact is also indicative of an active interaction between these water components. The isotopic signature also shows that water dynamics (e.g., recharge) are local and highly controlled by the Recoleta and La Paloma reservoirs (with the exceptions of H2 and H3 locations, previously addressed).

Integrated analysis Connectivity index vs. chemically and isotopically based connectivity assessment Table 4 summarizes the connectivity as determined after the connectivity index methodology, as well as the chemical composition and isotopic signatures. Although the CI methodology is representative of long-term conditions and the results from chemistry DOI 10.1007/s10040-014-1170-9

Fig. 3 Stiff diagrams for surface water (light gray shapes) and groundwater (black polygons) for the a fall and b summer sampling campaigns

and isotopes correspond to a “snapshot” at the sampling time, the results from these different approaches show strong agreement in most of the reaches. Thus, the rather simple method proposed by Ransley et al. (2007) arises as a suitable approach, especially in cases where basic information such as water-table depth, well stratigraphy and river configuration is available, or where restrictions (budget, time) Hydrogeology Journal

may exist to carry out more detailed analyses. In fact, even with a lack of hydraulic data, the remaining information required by the method is likely available at the basin scale. Finally, the method may constitute a powerful visual aid, providing a relatively fast assessment that can be used as a starting base for a more detailed conceptualization of the SW–GW relationships in a given system. DOI 10.1007/s10040-014-1170-9

Fig. 4 Plot of δ18O vs. δ2H values of surface water (open symbol) and groundwater (filled symbols) for the a fall and b summer sampling campaigns. The local meteoric water line (LML) and the global meteoric line (GML) are also included, as well as isotope signatures of precipitation samples at Las Ramadas (LR), Tulahuén (TU), La Paloma (LP) and Carretera (CA) stations

Multivariate analysis Figure 5 shows the dendrograms obtained from the cluster analysis and the spatial distribution of the clusters. According to the Sneath criterion, two main groups (C-1 and C-2) are obtained. The first group (C-1) includes samples from the Grande River, the Hurtado River, the initial reaches of the El Ingenio Creek (E1 and E2 samples) and the mid-upper part of the Limarí River (L1–L5). The second major cluster (C-2) is formed by samples from the end section of the El

Ingenio Creek (E3–E5) and the mid-lower section of the Limarí basin (L6–L10). In order to obtain a more detailed analysis, the first cluster was arbitrarily divided into two subgroups. Subgroup C1.1 includes groundwater samples (H3, G3, E1, L2 and L4) and only two surface water samples (H2 and E2). This group presents low pH, δ2H, and δ18O values and the highest 222Rn activities. The C1.2 group includes only surface water samples (H1, G1, G2, L1, L3 and L5), and it is characterized by lower EC levels, the lowest Ca2+, Mg2+, Na+, HCO3−, Cl−, and SO42− concentrations, and the lowest 222Rn activities, as well as by higher pH values. In turn, C2 is a mixture of surface water (E3, E4, E5, L7 and L9) and groundwater (E6, L6, L8 and L10), all from the mid-end reaches of the El Ingenio Creek and the Limarí River. The group presents the highest EC levels and Ca2+, Mg2+, K+, Na+, HCO3−, Cl− and SO42− concentrations. The average composition for each subgroup is detailed in Table 5. There are two important results obtained from this analysis. First, the same samples became grouped within the same clusters in both sampling campaigns, which points towards a stable system with no major seasonal changes in both water chemistry and isotopic signals. Second, the spatial distribution of clusters for the study area is consistent with the previous results obtained from the separate consideration of water chemical data (e.g. Piper and Stiff diagrams) and stable isotopes values. Again, neighboring surface water and groundwater samples are grouped within the same cluster, confirming the active interaction between these water components, as discussed before. Identifying interaction dynamic and estimating transfer rates Given the results obtained from HCA that showed little seasonal changes, an EMMA was carried out for the average chemical (Cl−) and isotopic (18O) composition of surface water and groundwater. Figure 6 shows a plot of the mean δ18O vs. Cl− in surface (river) water and groundwater. It can be observed that, with the exception

Table 4 Comparison of connectivity levels among different approaches considered Reach

R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11

CI

High High High High High Medium Medium NA NA NA NA

Fall sampling campaign Hydrochemistry (Piper)

2

18

High High High High High NA High High High NA High

High High High High Medium NA Medium High High NA Low

High High Medium High Medium NA Low High High NA Low

H

O

Resulta

Summer sampling campaign 2 Hydrochemistry H (Piper)

18

High High High High Medium ND Medium High High NA Low

High High High High High NA High High High NA High

High High High High High NA High High High NA Medium

Medium Medium Medium High Medium NA High High High NA High

NA not available (lack of information and/or samples) a Corresponds to the overall characterization based on hydrochemistry (Piper) and stable isotopes (2 H and Hydrogeology Journal

18

O

Resulta High High High High High NA High High High ND High

O)

DOI 10.1007/s10040-014-1170-9

Fig. 5 Clusters defined for the a fall and b summer sampling campaigns and c their spatial distribution

of H2 and H3 (which, as already discussed, recharge outside the area of study), the surface water and groundwater samples effectively lie within (or very close

to) the limits of a triangle defined by three end-members. These end-members are precipitation (lower vertex of the triangle given by composition of water obtained at La

Table 5 Cluster average compositions Subgroup

pH

EC (μS/cm)

Fall sampling campaign C1.1 7.35 735 C1.2 8.34 546 C2 7.70 2,503 Summer sampling campaign C1.1 7.23 753 C1.2 8.24 507 C2 7.58 2,748 Hydrogeology Journal

Ca2+ (mg/l)

Mg2+ (mg/l)

K+ (mg/l)

Na+ (mg/l)

HCO3− (mg/l)

Cl− (mg/l)

SO42− (mg/l)

δ2H (‰)

δ18O (‰)

222 Rn (Bq/m3)

69.2 50.1 174.0

21.4 12.8 72.3

4.8 5.4 12.2

59.2 51.3 267.1

245 158 269

61.4 48.0 518.7

157 118 462

−64.2 −55.9 −56.8

−7.99 −6.52 −6.96

11,064 603 4,247

50.5 36.0 188.1

18.1 11.0 87.4

17.9 6.8 26.5

60.0 35.7 226.8

247 138 267

78.3 50.7 645.6

151 89 496

−61.4 −56.2 −51.7

−7.78 −7.09 −6.52

10,796 361 3,804

DOI 10.1007/s10040-014-1170-9

Fig. 6 Relationship of mean annual δ18O and Cl− concentration in river water (grey circles) and groundwater (black circles)

Paloma rainfall collector), water from reservoirs (upper left vertex given by H1 and G1), and recycled water (upper right vertex, represented by groundwater sample at L8). Thus, the relative contribution of each source was determined (Fig. 7). It is of interest to note that a group of samples (L1, L2, L3, L4, L5, E1) can be identified with basically two sources: water delivered from dams and local precipitation. These samples are located in the upper parts of the area of study, and therefore, the contribution of recycled water is negligible. On the other hand, samples obtained in the lower part of the study area (L7, L9, and L10) originate mainly from recycled water (surface water that is diverted from the river upstream, conducted through channels and used for irrigation, that infiltrates and returns as groundwater discharge to the surface stream) and from surface water from the river. Again, and as previously discussed from the other analysis, there seems to be an active SW–GW connectivity in general, as

Fig. 7 Estimated contribution ratios by different sources (p local precipitation; d dam-water delivery to the river; i infiltrated water). River water (grey circles), groundwater (black circles) Hydrogeology Journal

neighboring surface water and groundwater lie close to one another in the diagram. Regarding the use of 222Rn, and as expected, surface water and groundwater 222Rn activities were very different from each other in both sampling campaigns. For the fall campaign, the average surface-water activity was on the order of 1,400 Bq/m3, whereas the groundwater average signal reaches 11,100 Bq/m3. These values slightly decreased, reaching 1,000 Bq/m3 and 10,800 Bq/m3 respectively, during the summer campaign. The G1 sampling location, i.e., La Paloma reservoir, registered the lowest 222Rn activity in both campaigns (less than 100 Bq/m3). This behavior is expected, considering the fact that the reservoir accumulates low 222Rn surface water (i.e., incoming upstream surface flow), which is subjected to additional depletion during its residence time in the dam (given the short half-life of this isotope, 3.8 days). On the other hand, the highest activities were found in both sampling campaigns at E1 (El Ingenio Creek area) with a value of approximately 21,000 Bq/m3. This sample was obtained in a location characterized by the existence of granitic rock outcrops, which explains the high 222Rn values. The spatial distribution of measured 222 Rn activities in the study area is presented in Fig. 8. Based on these results, and following Green and Stewart (2008) who state that 222Rn activities higher than 1,000 Bq/m3 in surface water indicate “nearby and recent groundwater outflow”, it may be inferred that exfiltration processes were occurring around L9 (downstream part of the Limarí River) and E2 (upper El Ingenio Creek) during both sampling periods. Also, at least in one of the two campaigns, high activity levels were found in surface water for L3, H2, and L7 sampling locations. Following the approaches of Stellato et al. (2008) and Cook et al. (2003), described in the methodology section, it was possible to estimate the groundwater discharge rate for those reaches where exfiltration (i.e., gaining stream condition) was identified (Table 6). Higher values for summer are seen, which is expected, to some extent, given the inexistence of precipitation and the high water diversions from the river that occur during this period, which should favor an hydraulic gradient oriented towards the river. It must be acknowledged that these results should be taken with caution, given some uncertainties in the methods employed. Regarding the Stellato et al. (2008) approach, it assumes that solute inflow from groundwater dominates other processes, and that parameters such as evaporation, gas transfer velocity and decay constant are negligible (Cook et al. 2003). On the other hand, regarding the Cook et al. (2003) method, Radon gas transfer velocity (k) was not measured but taken from the literature (Cook et al. 2003). However, a simple sensitivity analysis shows that the results are not highly dependent upon the adopted value. In fact, if k is increased or decreased by 50 %, the groundwater discharge rates are modified on the order of 28 % at most (for a complete analysis on model assumptions and limitations on this approach the reader is referred to Cook 2013). Secondly, DOI 10.1007/s10040-014-1170-9

Fig. 8

222

Rn activities in surface water (open circles) and groundwater (black circles) in a fall and b summer sampling campaigns

and perhaps more important, the Cook et al. (2003) approach does not allow to distinguish and separate the contribution of the hyporheic zone, which as shown later by Cook et al. (2006), can be relevant and can yield overestimations of the groundwater inflow rate (up to Table 6 Fraction of surface water coming from groundwater (in %) Campaign Fall Summer

Reach R1 R5 R8 R1 R2

Hydrogeology Journal

Model Stellato et al. (2008)

Cook et al. (2003)

23 23 33 30 57

36 32 40 88 87

70 % in their study area). In order to overcome this difficulty, Cook et al. (2006) and Cook (2013) have proposed the use of a second (ionic) tracer that is not contributed (or minimally affected) by hyporheic flow, such as Cl−. When calculations are made for this ion using Eq. (8)—considering only the first two terms on the right side of the equation that apply for conservative tracers— groundwater contribution rates of 25 and 49 % for R1 and R5 for fall, and 41 and 64 % for R1 and R2 for summer, are determined. Thus, and despite the uncertainties described (which as expressed in Cook 2013 “remain to be resolved to make this method more widely applicable”), it is interesting to note two aspects. First, that the results using 222Rn were in general mutually consistent between both approaches. DOI 10.1007/s10040-014-1170-9

Second, that although normally lower (the difference most likely explained by hyporheic contribution), groundwater inflow rates were also important when Cl− was considered as the tracer for the calculation. In this regard, it is possible to infer that, in general terms, an important fraction of surface discharge (25 % or higher) corresponds to groundwater contribution, and that this amount is particularly high in summer. Sustainable water allocation schemes for the Limarí basin should consider these results in the near future.

Conclusions The connectivity index method is easy to apply, and constitutes a suitable starting point for obtaining awareness regarding SW–GW interactions. In the case study described here, it yielded high connectivity conditions in most of the stream reaches, with the exception of two cases where a medium connectivity situation was identified. Chemical parameters (major ions) presented some differences throughout the area of study. When the analyses were performed specifically for each sub-basin, both surface water and groundwater were classified as the same type, supporting the idea of an active interaction between surface water and groundwater. In addition, stable isotope (2H and 18O) data indicate that the samples from the Hurtado subwatershed, H2 for surface water (located at 255 masl) and H3 for groundwater (located at 288 masl) are not only very similar to each other, but are also the most depleted within the study area. This leads to the conclusion that both waters correspond to recharge occurring at upper locations in the sub-basin (isotopic signature similar to meteoric water signature from Tulahuén station at approximately 1,000 masl) that outflows at the sampling locations. All the other samples, from the Grande and Limarí rivers and El Ingenio Creek presented more enriched isotopic signals than the Hurtado River, and very low dispersion, which seems to be highly controlled by the La Paloma and Recoleta reservoirs and local hydrological conditions. Overall, the results obtained with the different approaches, i.e., connectivity index method and the hydrochemistry and stable isotope data, were mutually consistent. In fact, the identification of zones in the study area with similar characteristics derived from the cluster analysis confirms this idea. Among the major groups, the first (C1) includes the Grande River sub-basin, the Hurtado area, and the initial sectors of the El Ingenio Creek and the Limarí River basin. The second group (C2) includes the middle reaches of the El Ingenio Creek and the rest of the Limarí basin. These clusters (C1 and C2) were identical in both campaigns. Based on end-member analysis, it was possible to generally assess the relative contributions of the dams, local precipitation, and recycling of water for surface water and groundwater throughout the area of study. Also, the use of 222Rn activities as a tool for the assessment of Hydrogeology Journal

groundwater contribution is possible because these are considerably lower in the river than in groundwater. The average surface water does not exceed 1,500 Bq/m3, whereas that in groundwater is on the order of 11,000 Bq/ m3. Based on measured activities at specific locations, gaining stream conditions were inferred in fall in the Barraza zone and the mid-upper part of the Limarí River area (reaches R1, R4, R5), and the initial section of the El Ingenio Creek (R11). In the case of the early summer campaign, this occurs in the final area of the Limarí River (reaches R1, R2) and El Ingenio Creek (R11). The proportion of groundwater entering the surface flow in river sections with gaining conditions was generally high, greater than 25 %, especially in summer. Finally, the results obtained through each of the methodologies used in this study were consistent and allowed the determination of a high connectivity between surface water and groundwater in the basin. Thus, this multi-method approach should prove to be a useful tool when sustainable water-management strategies are required for arid to semi-arid basins facing similar water-management challenges to those of the Limarí basin. Acknowledgements This work was funded by the project Fondecyt 11100040 (CONICYT), and was carried out as part of the Programa de Recursos Hídricos y Medio Ambiente (PRHIMA) of the Departamento Ingeniería de Minas, Universidad de La Serena. The authors are much indebted to the Dirección General de Aguas (DGA, Ministerio de Obras Públicas) for allowing the use of their data files. The report benefited from the comments of two reviewers, Drs. A. Brookfield and J. Gomez-Velez, the associate editor, Dr. T. Gleeson, and the editor, Prof. M. Schafmeister.

References Alvarez P, Oyarzún R (2006) Interacción río–acuífero en zonas áridas: contexto legal y análisis de casos [Aquifer–river interaction in arid zones: legal framework and case study]. VIII Congreso Latino-Americano de Hidrología Subterránea (ALSHUD), Asunción, Paraguay, September 2006 Andersen MS, Acworth RI (2009) Stream-aquifer interactions in the Mules Creek catchment, Namoi Valley, New South Wales, Australia. Hydrogeol J 17:2005–2021 Astel A, Tsakovski S, Barbieri P, Simeonov V (2007) Comparison of self-organizing maps classification approach with cluster and principal components analysis for large environmental data sets. Water Res 41:4566–4578 Bailly-Comte V, Jourde H, Pistre S (2009) Conceptualization and classification of groundwater-surface water hydrodynamic interactions in karst watersheds. J Hydrol 376:456–462 Braaten R, Gates G (2003) Groundwater-surface water interaction in inland New South Wales: a scoping study. Water Sci Technol 48:215–224 Brodie R, Sundaram B, Tottenham R, Hostetler S, Ransley T (2007) An overview of tools for assessing groundwater– surface water connectivity. Bureau of Rural Sciences, Canberra, Australia Burtnett WC, Peterson RN, Santos IR, Hicks RW (2010) Use of automated radon measurements for rapid assessment of groundwater flow into Florida streams. J Hydrol 380:298–304 CIREN (2011) Catastro frutícola: principales resultados—región de Coquimbo [Fruit tree census: main results—Coquimbo region]. Available at http://www.ingenierosagronomos.cl/pcomciac/ DOI 10.1007/s10040-014-1170-9

wp-content/uploads/2011/06/Catastro-Frut%C3%ADcola-IVRegi%C3%B3n-de-Coquimbo.pdf. Accessed 15 Jan 2013 Clesceri LS, Greenberg AE, Eaton AD (1999) Standard methods for the examination of water and wastewater, 20th edn. American Public Health Association, Washington, DC CONAMA (2006) Estudio de la variabilidad climática en Chile para el siglo XXI [Climate variability study for Chile in the XXI century]. Comisión Nacional del Medio Ambiente, Santiago, Chile Constantz J, Cox MH, Su GW (2003) Comparison of heat and bromide as groundwater tracers near streams. Ground Water 41:647–656 Cook P (2013) Estimating groundwater discharge to rivers from river chemistry surveys. Hydrol Process 27:3694–3707 Cook P, Favreau G, Dighton J, Tickell S (2003) Determining natural groundwater influx to a tropical river using radon, chlorofluorocarbons and ionic environmental tracers. J Hydrol 277:74–88 Cook P, Lamontagne S, Berhane D, Clark JF (2006) Quantifying groundwater discharge to Cockburn River, southeastern Australia, using dissolved gas tracers 222Rn and SF6. Water Resour Res 42, W10411. doi:10.1029/2006WR004921 Demirel Z, Güler C (2006) Hydrogeochemical evolution of groundwater in a Mediterranean coastal aquifer, MersinErdemli basin (Turkey). Environ Geol 49:477–487 DGA (2004) Diagnóstico y clasificación de los cursos y cuerpos de agua según objetivos de calidad. Cuenca del Río Limarí [Evaluation and classification of water bodies according quality objectives. Limarí watershed]. Dirección General de Aguas, Ministerio de Obras Públicas, Santiago, Chile DGA (2008) Evaluación de los recursos hídricos subterráneos de la cuenca del río Limarí. Informe Técnico N°268 [Assessment of groundwater resources in the Limarí basin]. Technical report no. 268, Departamento de Administración de Recursos Hídricos, Dirección General de Aguas, Ministerio de Obras Públicas, Santiago, Chile DGA (2012) Catastro de derechos de aprovechamiento de aguas [Water rights information]. Available at www.dga.cl/ administracionrecursoshidricos/derechosconstituidos/paginas/ default.aspx.. Accessed 17 Jan2013 Dor N, Syafalni S, Abustan I, Rahman M, Nazri M, Mostafa R, Mejus L (2011) Verification of surface–groundwater connectivity in an irrigation canal using geophysical, water balance and stable isotope approaches. Water Resour Manag 25:2837–2853 Friedman I, Smith GI, Gleason JD, Warden A, Harris JM (1992) Stable isotope composition of waters in Southeastern California: 1. modern precipitation. J Geophys Res 97:5795–5812 Fullagar I, Brodie R, Sundaram B, Hostetler S, Baker P (2006) Managing connected surface water and groundwater resources. Science for Decision Makers. Australian Government, Bureau of Rural Sciences, Canberra, Australia Geoscience Australia (2013) Groundwater–surface water connectivity. Available at http://www.ga.gov.au/groundwater/ understanding-groundwater-resources/groundwater-surfacewater-connectivity.html. Accessed 11 Mar 2014 Green G, Stewart S (2008) Interactions between groundwater and surface water systems in the Eastern Mount Lofty Ranges, South Australia, DWLBC Report 2008/27. Dept. of Water, Land and Biodiversity Conservation, Gov. of South Australia, Canberra, Australia Guggenmos MR, Daughney CJ, Jackson BM, Morgenstern U (2011) Regional-scale identification of groundwater–surface water interaction using hydrochemistry and multivariate statistical methods, Wairarapa Valley, New Zealand. Hydrol Earth Syst Sci 15:3383–3398 Güler G, Thyne G, McCray J, Turner A (2002) Evaluation of graphical and multivariate statistical methods for classification of water chemistry data. Hydrogeol J 10:455–474 Ivkovic K (2009) A top-down approach to characterize aquifer– river interaction processes. J Hydrol 365:145–155 INE (2007) Censo Agropecuario y Forestal 2007 Resultados por Comuna [Agricultural and forestry census]. Available at http://

Hydrogeology Journal

www.ine.cl/canales/chile_estadistico/censos_agropecuarios/ censo_agropecuario_07_comunas.php. Accessed 25 May 2012 Jolly ID, McEwan KL, Holland KL (2008) A review of groundwater–surface water interactions in arid/semiarid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1:43–58 Jourde H, Rochette R, Blanc M, Brisset N, Ruelland D, Freixas G, Oyarzún R (2012) Characterization of the interactions between subterranean and superficial fluxes within an Andean catchment as a function of the spatio–temporal variability of climate. La Houille Blanche 2:18–25 Kalbus E, Reinstorf F, Schirmer M (2006) Measuring methods for groundwater–surface water interactions: a review. Hydrol Earth Syst Sci 10:873–887 Konrad C (2006) Location and timing of river-aquifer exchanges in six tributaries to the Columbia River in the Pacific Northwest of the United States. J Hydrol 329:444–470 King AC, Raiber M, Cox ME (2014) Multivariate statistical analysis of hydrochemical data to assess alluvial aquifer–stream connectivity during drought and flood: Cresbrook Creek, southeast Queensland, Australia. Hydrogeol J 22:481–500 LiuY Y, Yamanaka T (2012) Tracing groundwater recharge sources in a mountain-plain transitional area using stable isotopes and hydrochemistry. J Hydrol 464–465:116–126 Mas-Pla J, Menció A, Marsiñach A (2013) Basement groundwater as a complementary resource for overexploited streamconnected alluvial aquifers. Water Resour Manag 27:293–303 Núñez J, Rivera D, Oyarzún R, Arumí JL (2013) Influence of Pacific Ocean multidecadal variability on the distributional properties of hydrological variables in north-central Chile. J Hydrol 501:227–240 Newman BD, Vivoni ER, Groffman AR (2006) Surface watergroundwater interactions in semiarid drainages of the American Southwest. Hydrol Process 20:3371–3394 Oyarzún R, Jofré E, Maturana H, Oyarzún J, Aguirre E (2013) Use of 222Rn as a simple tool for surface water–groundwater connectivity assessment: a case study in the arid Limarí basin, north-central Chile. Water Environ J doi:10.1111/wej.12057 Parsons S, Evans R, Hoban M (2008) Surface–groundwater connectivity assessment: a report to the Australian Government from the CSIRO Murray-Darling Basin Sustainable Yields Project. CSIRO, Clayton, Australia Ransley T, Tottenham R, Sundaram B, Brodie R (2007) Development of method to map potential stream aquifer connectivity: a case study in the Borders rivers catchment. Bureau of Rural Sciences, Dept. of Agriculture, Fisheries and Forestry, Australian Gov., Canberra, Australia SERPLAC, DGA, ONU, CORFO (1979) Hidrogeología de la cuenca del río Limarí: investigación de recursos hidráulicos en la IV región [Limarí basin hydrogeology: research on hydraulic resources in the Region IV]. Proyecto CHI-535, SERPLAC, DGA, ONU, CORFO, Santiago, Chile Shrestha S, Kazama F (2007) An assessment of stream water quality of the Río San Juan, Nuevo Leon, Mexico, 1995–1996. J Environ Qual 31:1256–1265 Shi JA, Wang Q, Chen GJ, Wang GY, Zhang ZN (2001) Isotopic geochemistry of the groundwater systems in arid and semiarid areas and its significance: a case study in Shiyang River basin, Gansu province, northwest China. Environ Geol 40:557–565 Simmers I (2003) Hydrological processes and water resources management. In: Simmers I (ed) Understanding water in a dry environment: hydrological processes in arid and semi-arid zones. Balkema, Lisse, The Netherlands Souvignet M, Gaesse H, Ribbe L, Kretschmer N, Oyarzún R (2010) Statistical downscaling of precipitation and temperature in north-central Chile: an assessment of possible climate change impacts in an arid Andean watershed. Hydrol Sci J 55:41–57 Stellato L, Petrella E, Terrasi F, Belloni P, Belli M, Sansone U, Celico F (2008) Some limitations in using 222Rn to assess river– groundwater interactions: the case of Castel di Sangro alluvial plain (central Italy). Hydrogeol J 16:701–712

DOI 10.1007/s10040-014-1170-9

Strauch G, Oyarzún R, ReinstorfF OJ, Schirmer M, Knöller K (2009) Interaction of water components in the semi-arid Huasco and Limarí River basins, north central Chile. Adv Geosci 22:51–57 Tallini M, Parisse B, Petitta M, Spizzico M (2013) Long-term spatio-temporal hydrochemical and 222Rn tracing to investigate groundwater flow and water–rock interaction in the Gran Sasso (central Italy) carbonate aquifer. Hydrogeol J 21:1447–1467 Thomas H (1967) Geología de la Hoja Ovalle, Provincia de Coquimbo [Ovalle sheet geology]. Bull no. 23, Instituto de Investigaciones Geológicas, Santiago, Chile, map 1:250.000

Hydrogeology Journal

Thyne G, Güler C, Poeter E (2004) Sequential analysis of hydrochemical data for watershed characterization. Ground Water 42:711–723 Vicuña S, McPhee J, Garreaud RD (2012) Agricultural vulnerability to climate change in a snowmelt-driven basin in semiarid Chile. J Water Resour Plan Manag 138:431–441 Winter TC, Harvey JW, Franke OL, Alley WM (1998) Groundwater and surface water: a single resource. US Geol Surv Circ 1139 Yin L, Hou G, Su X, Wang D, Dong J, Hao Y, Wang X (2011) Isotopes (δD and δ18O) in precipitation, groundwater and surface water in the Ordos Plateau, China: implications with respect to groundwater recharge and circulation. Hydrogeol J 19:429–443

DOI 10.1007/s10040-014-1170-9