Abstract

In the Loa water system of the Atacama Desert in northern Chile, careful management of groundwater is vital and data are sparse. Several key management questions focus on aquifers that occur in the Calama sedimentary basin, through which groundwater and Loa surface water flow to the west. The complexity of the two major aquifers and their discharge to wetlands and rivers are governed by primary facies variations of the sedimentary rocks as well as by faults and folds that create discontinuities in the strata. This study integrates geological studies with groundwater hydrology data to document how the aquifers overlay the formations and facies. Neither the phreatic aquifer nor the confined or semiconfined aquifer, each of which is identified in most basin sectors, corresponds to a laterally persistent geological unit. The variable properties of low-permeability units sandwiched between units of moderate to high permeability cause a patchwork pattern of areas in which water is exchanged between the two aquifers and areas where the lower aquifer is confined. The westward termination of most of the sedimentary rocks against a north-trending basement uplift at an old fault zone terminates the principal aquitard and the lower aquifer. That termination causes lower aquifer water to flow into the upper aquifer or discharge to the rivers. The regionally important West fault juxtaposes formations with differing lithological and hydraulic properties, resulting in some exchange of water between the upper and lower aquifers across the fault.

INTRODUCTION

The Loa River water system of northern Chile’s Atacama Desert (Fig. 1), in the Antofagasta region, exemplifies the high stakes involved in sustainable management of scarce water resources. The Loa surface and groundwater system supplies the great majority of water used in Antofagasta Region, and meets much of the municipal and agricultural demands. It is vital to Antofagasta Region copper mining, which constitutes ∼50% of Chile’s copper production (Servicio Nacional de Geología y Minería, 2011), which in turn supplies one-third of the world’s copper needs. However, a key property of the Loa system is the scarcity of surface water.

The Loa groundwater and surface water are inseparable resources; the Loa River is the discharge channel of the groundwater of a >34,000 km2 area basin. The coincidence of modern climate, topography, and geology leads to a geography in which recharge occurs far from where humans use the water resources, and groundwater aquifers supply the vast majority of stream water. The aridity of the region sharply restricts the number of human inhabitants and extent of native plants or animals. However, under different climate states during the past few millennia the water flux was greater than now (Rech et al., 2002; Latorre et al., 2006); this leads to great uncertainty in estimations of how much of the current water flow is renewable versus fossil (Houston and Hart, 2004).

Under the Chilean water code, water is treated as a property independent of land, to which a permanent right is granted for a given production rate from a given extraction site. Although the water code treats groundwater rights and surface-water rights separately, for the Loa system the regulatory body (Dirección General de Aguas, hereafter DGA) treats them as strictly coupled. In current practice, all requests for additional rights or for changes in the points of extraction are subject to an environmental impact review by an independent agency, the Ministerio del Medio Ambiente.

Initial management efforts distributed rights to exploit surface and groundwater when only a limited understanding of the natural system was available. Now management practices strive to minimize both economical and ecological problems, yet to do so requires improved information about the natural system. The desire to better inform the management of this coupled natural-human resource system is a central motivator for this study. Furthermore, because the Loa system is at the junction of a natural extreme (e.g., an arid to hyperarid climate) and an atypical water governance approach (e.g., private water rights), lessons from the Loa system may be broadly useful to those who are considering regional-scale groundwater management strategies.

Because of extreme aridity at low elevations, precipitation that is likely to lead to direct recharge of aquifers occurs only in the eastern mountainous fringe of the hydrologic basin (Fig. 1B) (CORFO 1977, see Table 1; Houston, 2009). At lower elevations there is extensive exchange of water between aquifers and rivers (DGA 2001, see Table 1; Houston, 2006), but the locations and net outcomes of those exchanges are not well documented and constitute major themes for ongoing research.

This paper focuses on the rocks through which the groundwater flows in the central sector of the Loa system, within and adjacent to the Calama Valley (Fig. 1B), in which an upper phreatic aquifer and a lower confined aquifer are routinely described (Figs. 2 and 3) (CORFO 1977, see Table 1; Houston, 2004). There are three primary purposes of this paper: to clarify the spatial distribution of the rocks with hydraulic conductivity favorable to function as aquifers; to identify where the lower aquifers discharge to the surface water system; and to identify the most likely sectors in which water is exchanged between upper and lower aquifers. Although the data available for hydraulic properties are sparse, the combined use of knowledge of the sedimentary architecture of the Calama sedimentary basin and of piezometric head enables informed extrapolation of the hydraulic data laterally and vertically. For the first time for the Loa system, we analyze the controls on the spatial variability of major aquifers imposed by the complex stratigraphic architecture, and the uncertainties that remain. Examination of the state of knowledge reveals sectors of the groundwater basin for which it is most critical to obtain hydrochemical, geophysical, and hydrological data with which to monitor the impacts of water extraction or to constrain parameters in a numerical model.

A broader objective is to advance appreciation that knowledge of sedimentary basin architecture is valuable for regional hydrogeology research. Barthel (2014) drew attention to the need for improved fundamental regional hydrogeological approaches. The hydrocarbon resource industry routinely utilizes the architecture of entire sedimentary basins as a tool to predict reservoir flow properties. Likewise, for groundwater systems within sedimentary basins, the large-scale architecture of the strata likely plays a major role in determining the continuity of hydraulic properties. This paper demonstrates an application of knowledge of sedimentary basin architecture to a major sector of a coupled groundwater–surface-water basin.

After describing the existing management premises and the physical context of the study area, we describe the available piezometric framework of the aquifers, then illuminate the spatial distribution of the geological units in which the aquifers and aquitards occur across the middle part of the Loa groundwater basin. The paper concludes with identification of hydrogeological trends and unresolved problems.

Water Management Hydrological Premises and Uncertainties

The Loa hydrologic system is located on the western flank of the Andes Mountains in northern Chile and extends westward to the Pacific Ocean coast (Fig. 1). The generally accepted premise is that the natural water basin is at steady state such that recharge equals combined flows out of the basin plus extraction plus evapotranspiration. An alternative conceptual model holds that some flow results from head decay established during times of wetter climate (Houston and Hart, 2004).

Based on empirical relationships between precipitation and elevation as well as temperature and elevation, combined with the topography of the basin, the DGA (2003) estimated that the total annual available recharge of the Loa surface and groundwater hydrologic basin is 6.4 m3/s. However, the Loa River discharges only 0.6 m3/s to the Pacific Ocean (Salazar, 2003) (Table 2). The difference, 5.8 m3/s, is attributed to evapotranspiration and consumptive water use. Large uncertainties exist with this steady-state model. A trend is found for water to flow predominantly in the subsurface in the upper parts of the basin, in a combination of surface channels and subsurface flow in the middle Loa basin, and in surface channels in the lower Loa basin. It is thought that the final significant transfer from subsurface to surface flow occurs just west of Calama city (Salazar, 2003). Consequently, the official measure of the water in the system available for ecosystem use, and potentially for additional human use, is given by the discharge in the Loa and San Salvador Rivers west of this assumed final location of transfer from aquifers to surface streams.

The water balance model that underpins water management decisions is informed by a set of long-term stream gauging stations that exist within the central Loa basin as well as by a small set of monitoring wells (Fig. 2; Table 2). However, there are long reaches of the Loa River where flow is not measured, or where only single-year gauging campaigns have been reported (Matraz 2012, see Table 1). There has not been a previous analysis of the potential for hydraulic interconnections among the rocks that contain the aquifers, although extensive monitoring plans have been developed to try to demonstrate the presence or absence of hydraulic connections. Although recently published geological and stratigraphic studies illuminate the stratigraphic and spatial positions of units that may function as aquitards or aquifers, the resulting insight into the likely complexity of the groundwater system has not been integrated into basin-scale water management assessments that are important to the integrated management of the groundwater and surface water system.

The lack of understanding of the architecture of the sedimentary-hosted aquifers within the Calama Basin contributes to a lack of understanding of where groundwater exits the middle sector of the Loa groundwater basin. This gap in knowledge is particularly relevant to deriving, let alone monitoring, a water budget. Absent data regarding the western distribution or terminations of the aquifers, an arbitrary location of where discharge to the Loa is measured could produce misleading information, especially for monitoring of the impacts of operating well fields. An outcome from this paper, a data-based hypothesis for the locations of aquifer discharge west of Calama city, should be considered when planning monitoring stations.

The extraction of water has had significant impact on stream flow. Stream gauge measurements from the early twentieth century provide data least affected by extraction (Table 2; 1916 data). Some water would have been diverted then for agricultural use, and the first sluice, built in 1915, supplied the early copper industry. By the 1960s there was important extraction of water for mining purposes, and in 1979 the high Conchi Dam was built to reduce the impacts of the rare floods. The result (Table 2) was a decrease between the early 1960s to the 1990s by >50% in stream flow. By the 2000s, the regulatory agency DGA began to tighten evaluations of petitions for additional extractions, in response to the assessment that the water system was in deficit.

Basic Components: Surface Water, Groundwater, and Sedimentary Rocks

Three distinct types of basin are important to the Loa system hydrology. From a hydrological perspective, the Loa basin is tightly coupled to continental-scale landforms and is extensive, whether one considers its surface catchment area (first type of basin, the surface-water basin) or its groundwater recharge-discharge footprint (second type of basin, the groundwater basin) (Fig. 1). The third basin type is geological, and pertains directly to the physical properties of the aquifer rocks. The Calama sedimentary basin forms the subsurface rocks of a central sector of the Loa hydrological basin (∼2400 km2; sedimentary basin) and constitutes one of two major regions for groundwater exploitation (Fig. 2). All three classes of basin are important, and it is necessary to clarify which basin (surface catchment, groundwater, or sedimentary) is the focus of various parts of the analysis.

Surface Catchment Basin

The surface catchment basin (33,570 km2; Tables 2 and 3) is limited at the topographic crestline in the Andes Mountains on the east, the Pacific coast on the west, and ∼lat 21°S and 23°S on the north and south, respectively. The Loa River main stem measures 440 km in length, with a series of four orthogonal reaches, each 50–150 km long (Fig. 1). The first broad valley through which the surface drainage system passes is the 50 km by 50 km Calama Valley, the focus here, whose low-relief floor is ∼2200–2800 m above sea level. The two main tributaries, the upper Loa River and the Salado River, join within the Calama Valley.

The general attributes of the topography, characterized by mountains with elevations >4000 m in the east and lowlands toward the west, set the boundary conditions for groundwater recharge and water flow. In the Atacama Desert precipitation increases with elevation (Table 3) and evaporation is intense. Precipitation currently capable of recharging the aquifers only occurs above 3500 m above sea level (asl) but, after consideration of evapotranspiration, recharge is most likely above 4000 m elevation (Fig. 1) (DGA, 2003; Houston, 2009). The area in the Loa catchment with widespread potential for modern recharge is the eastern mountains (Western Cordillera and Altiplano; Fig. 1), although there may have been recharge at elevations below 4000 m during wetter times of the Pleistocene or Holocene (Latorre et al., 2002; e.g., Rech et al., 2002). Outside of the eastern highlands, the Loa system surface water bodies are fed by aquifers (DGA, 2003).

The San Salvador River rises at springs just north and west of Calama city (Fig. 2). The Loa and San Salvador Rivers are ∼5 km apart and are parallel as the south and north canyon boundaries of a 65-km-long east-trending valley that has a table-like planar surface. With the exception of ∼20 km distance of the Loa main stem, the Loa and San Salvador Rivers pass through narrow canyons incised 20–200 m into either consolidated sedimentary rocks, volcanic rocks, or crystalline basement.

River flow data from the early austral spring season in a year prior to extensive consumptive use of the Loa River system (September 1916) indicate that natural flow of the middle Loa River where it entered the Calama Valley (∼2200 L/s) was increased substantially by influx from the Salado River (∼1000 L/s) (Table 2). The San Salvador River carried ∼600 L/s. Between the junctions of the Salado and the San Salvador, the Loa River flows through the hyperarid Calama Valley and the Calama sedimentary basin. Exchanges between surface water and groundwater are suggested by both downstream increases and decreases in Loa River flow in reaches where there are no tributary streams (Table 2; e.g., gains between stations 2–4; losses between stations 4–5).

Groundwater Basin

The boundaries of the groundwater basin are not well known (Table 4), both in the western region (Coastal Cordillera, Fig. 1) and in the eastern highlands, where the border may coincide with the surface-water basin or may extend east of the surface catchment, beneath the Altiplano Plateau. The areal extent, 34,000–65,000 km2, is very uncertain (Fig. 1).

The Loa groundwater basin can be considered to include three geographical sectors, upper, middle, and lower; the lower groundwater basin is not discussed herein. The eastern limit of the upper, or eastern, groundwater basin likely occurs among the volcanic centers that form the Western Cordillera and that cover broadly the southwestern Altiplano Plateau (Fig. 1). Those volcanic peaks overlie laterally extensive Miocene and Pliocene pyroclastic volcanic deposits and interbedded epiclastic sands and gravels (de Silva, 1989; Montgomery et al., 2003; Houston, 2007) that compose aquifers in some of the upland basins (Mardones Perez, 1998; Montgomery et al., 2003; Houston, 2007; DGA, 2003). A reasonable but unproven extrapolation is that some water recharged east of the surface catchment divide flows as groundwater westward into the surface water Loa catchment (Pourrut and Covarrubias, 1995; Houston, 2007).

The middle sector of the groundwater basin (Table 4) is strongly linked to the Calama sedimentary basin. Across the Calama topographic valley the water table drops nearly 500 m, from ∼2700 m asl to 2240 m asl at Calama city, with a further drop of >100 m to the western limit of the aquifers (Fig. 2). Groundwater exploration, monitoring, and production in some parts of the Calama Valley have demonstrated that aquifers extend to several hundred meters depth. A phreatic aquifer extends locally to a depth as great as 100 m (Houston, 2006), and lower aquifers occur in a depth range of 100–300 m (Fig. 3A) (inclusive of DGA, 2003; EIA 2011, Matraz 2012, and Mayco 2013 in Table 1). These depths greatly exceed the thickness of unconsolidated sediment (Blanco and Tomlinson, 2009; Tomlinson et al., 2010), and are within the compacted Cenozoic sedimentary rock. To date, no aquifers are known within the Paleozoic or Mesozoic highly indurated strata, and therefore these units plus plutonic rocks, metamorphic rocks, and lava flows are treated as the hydraulic basement. Fracture flow may be possible within the rock units treated as hydraulic basement, but is not discussed here.

Calama Sedimentary Basin

The Cenozoic sedimentary rocks of the Calama Basin compose the third category of basin (Table 5) and host the major aquifers of the middle Loa groundwater basin. Both the Calama Valley and the mesa-like surface between the San Salvador and Loa Rivers (Fig. 4) are the modern expressions of this long-lived sedimentary basin. The Calama Basin is composed of moderately consolidated sedimentary rocks of Eocene–Pliocene age (May, 1997; May et al., 1999, 2005; Blanco et al., 2003; Jordan et al., 2006; Blanco, 2008; Blanco and Tomlinson, 2009; Tomlinson et al., 2010) (Fig. 5). The depocenter of the Cenozoic sedimentary basin has shifted in location at least four times (Fig. 6). First, Eocene-age conglomerates and volcanic-associated strata reach ∼1000 m thickness and occur mostly in the southern third of the valley (Fig. 6D). This first stratigraphic unit encompasses the confined aquifers in some sectors of the groundwater basin. Second, Oligocene to lowermost Miocene strata reach ∼2000 m in thickness beneath the north-central part of the valley (Fig. 6C). Third, the overlying lower and middle Miocene strata spread widely across what is now the Calama Valley (Fig. 6B). In the northern part of the basin some facies of these units host both the lower and upper aquifers, but in the southern and western part of the basin the fine-grained facies that corresponds to this time slice forms an important confining layer. Fourth, the upper Miocene and Pliocene Opache and Chiquinaputo Formations extend across most of the Calama Valley as well as northward along the lower reach of the surface water upper Loa basin and westward across the mesa-like surface between the San Salvador and Loa Rivers (Figs. 4 and 6A). Parts of the middle Miocene strata and much of the upper Miocene–lower Pliocene strata constitute the phreatic aquifers.

Geohydrological Framework

Water in the Loa catchment north and east of the focus region enters permeable units of the subsurface (Houston, 2007), but those units are not laterally continuous with possible host rocks of the Calama Valley focus area. The possible aquifer rock units within the Calama sedimentary basin are not tabular or continuous sheets. Thus water must pass from one set of stratigraphic units to another set, in order to exit the middle section of the Loa system. A comparison of the large amount of mapping data added by recent surface and subsurface geological studies (e.g., Jordan et al., 2006; Blanco, 2008; Blanco and Tomlinson, 2009; Tomlinson et al., 2010) to the hydrogeological reports, both published (Houston, 2004) and unpublished (EIA 2005 in Table 1), reveals that the stated stratigraphic position of the gravels that serve as confined aquifers is commonly inconsistent with the recent geological mapping.

Although the Calama sedimentary basin strata are little deformed compared to other Cenozoic basins of northern Chile, deformation by faults and folds are important to the groundwater and surface-water systems. A set of faults and associated folds (Mesozoic and Eocene faults of Cerros de Guacate and Sierra de San Lorenzo; the Cere fault; Table 6) controlled the topographic margins of the Calama Basin while sedimentary rocks accumulated. Four major and one minor tectonically controlled fault sets traverse the sedimentary basin (Table 6; Fig. 7) and bound hydraulically conducting stratigraphic units. An additional set of nontectonic folds and fractures creates a large-scale and laterally continuous zone of likely hydraulic discontinuity within the Calama Valley (Fig. 7; Table 6). Some of the faults (e.g., West fault, Milagro fault; Table 6) displace by kilometers the continuity of rocks that are the hosts for the lower aquifer. Only the Chiu Chiu monocline (Table 6; Fig. 7) offsets significantly the strata that host the upper aquifer. The result of the primary stratigraphic patterns and secondary deformation is a complex architecture of the strata that are plausible aquifers and aquitards beneath the surface of the Calama Valley and San Salvador–Loa Valley.

The geological units herein are the lithological units mapped by Marinovic and Lahsen (1984), Marinovic et al. (1995), Blanco and Tomlinson (2009), and Tomlinson et al. (2010). These mappable lithological units are not strictly the same as the aquifers and aquitards defined by hydraulic conductivity. Rather, this paper documents the positions in space of the geological units whose facies are both anticipated, based on textures observed in outcrop, as well as demonstrated in boreholes, to have appropriate porosity and permeability to serve as aquifers or as aquitards. Thus the maps and geological cross sections place bounds on the distributions of the host rocks of the aquifers and aquitards.

MATERIALS AND METHODS

Data for groundwater are from reports that companies file with the Chilean agencies Ministerio del Medio Ambiente (Ministry of the Environment) and/or DGA. Those reports systematically record the static water level and, in some cases, useful aquifer parameters (transmissivity and storativity). The maps of piezometric surfaces (Figs. 2 and 3A) were generated using both a compilation of piezometric contours from previous reports (EIA 2005, EIA 2011, Matraz 2012, Mayco 2013, and Minera Leonor 2007, see Table 1) and data from 118 wells (Tables 1 and 7). Time series of water levels are available for more than 50% of these wells (Table 8; Figs. 2 and 3A). Wells reported to be production wells were not used unless a time series was available from which to identify the impacts of pumping, and therefore to select data prior to that impact. Likewise, monthly time series enabled recognition of the impacts of pumping at nearby wells (e.g., Fig. 8), and exclusion of those data. Ideally the maps would represent a single month in a single year, for a time prior to human intervention in the hydrological system. In reality, the data on which the maps are based represent either the oldest reported water levels for each well or data for 2003–2005, which were the earliest years of widespread well monitoring (Table 8). Overall, the oldest measurements used were recorded in 1993 and for a few sectors the earliest monitoring wells reported are as recent as 2011. For the phreatic aquifer, the surface of the water in the Loa and San Salvador Rivers was included in the data set. Although the data sources routinely indicate whether each well measures an upper or lower aquifer, the depths of screened intervals are reported for only 39 of the wells, and the year of construction is known for fewer than half of the wells. Well integrity problems may affect the segregation of water in these wells, allowing upper aquifer water to affect the recordings of lower aquifer head. Nevertheless, the clear distinctions between upper and lower aquifer heights of most near-neighbor wells (Table 9) indicates that many of these wells successfully restrict water entry to desired intervals in a single aquifer. In the primary data sets (Table 1) there are a few examples of wells whose data suggest the mixing of the two aquifers, and we avoided use of those wells. Further evidence that these monitoring wells successfully isolate the waters of the two aquifers is provided by hydrochemistry studies. For example, Matraz 2012 (see Table 1) examined water chemistry for 134 wells that overlap with the set listed in Table 7 (19 wells in common for upper aquifer; 24 well in common in lower aquifer) and interpreted from the sulfate concentrations that the upper aquifer and lower aquifer waters are distinctive. For those hydrochemistry monitoring wells, the well owners are required to maintain official certification of the well integrity.

Subsurface data for rock properties come from a small fraction of the numerous boreholes that have been drilled in the study area to explore for minerals within the rocks underlying the sediments of the Calama Basin. In addition to a very small number of published analyses of exploration boreholes (May, 1997; Blanco, 2008), this study used reports of lithologies from 44 exploration boreholes (Table 7). For 15 of those well reports, the driller or mudlog records include mention of depths at which water or wet rock was encountered. This study utilizes geological information regarding aquifer and aquitard lithologies from 131 groundwater wells and logged mineral exploration borehole records that appear in reports prepared for the DGA (e.g., Matraz 2012, see Table 1) or to comply with environmental impact and mitigation regulations (e.g., EIA 2005, see Table 1). These were put in the public domain through the website of Chile’s Environmental Evaluation Service (Ministerio del Medio Ambiente, http://sea.gob.cl). In those reports the lithological data appear either in detailed borehole-specific illustrations (e.g., EIA 2011, see Table 1) or embedded within geological cross sections (e.g., EIA 2005, see Table 1).

Geophysical profiles collected for minerals exploration, groundwater studies, and petroleum exploration exist in the study area, although only a small fraction of the results is published (e.g., interpretations of seismic reflection profiles collected for petroleum exploration: Jordan et al., 2006; Blanco, 2008; gravity survey: Matraz 2012 and Mayco 2013, see Table 1). Additional examples consulted for this study that were embedded as supporting documents within environmental impact analyses include NanoTEM™ (http://zonge.com.au/capability/method/nano-tem) profiles (GAC 2012, see Table 1) and TEM (Transient Electromagnetic) profiles (EIA 2011 and Mayco 2013, see Table 1).

The lithological information from the boreholes and geophysical profiles was combined with data from geological maps (Marinovic and Lahsen, 1984; Blanco and Tomlinson, 2009; Tomlinson et al., 2010) and stratigraphic studies (May, 1997; May et al., 1999, 2005; Jordan et al., 2006; Blanco, 2008; Blanco and Tomlinson, 2009) to map the distributions of sedimentary units that are independently reported to hold the aquifers and form the aquitards. Our geological field work and satellite image analysis using Google Earth led to creation of a new geological map for the study area west of Calama city (Fig. 9). Combining surface information, geophysical data, and borehole data, geological cross sections (Figs. 10–14) were created to illuminate major attributes of the three-dimensional distribution of the mapped units of major importance to aquifer and aquitard architecture.

A subset of both the mineral exploration boreholes and the groundwater well reports document the depths to water-bearing rocks or indicate the depth and lithology at which groundwater wells are screened for water entry (e.g., EIA 2005, see Table 1). Those reports are of special value in this analysis and were used to determine the direct connections between a rock unit and an aquifer. The few TEM profiles provided data for the depth to aquifers at locations between wells. The positions of the piezometric surfaces (from Figs. 2 and 3A) and of the corresponding borehole-specific aquifers were superimposed on the geological cross sections (Figs. 10, 11, 13, and 14).

RESULTS

Aquifers and Hydraulic Parameters

In the Llalqui area (Figs. 4 and 13) in the eastern part of the Calama Valley, the Opache and Chiquinaputo Formations contain the upper aquifer (Fig. 15A–15C) (Houston, 2004, 2006, 2007). A few wells document a lower aquifer, which is locally artesian. The aquifer corresponds to rocks at >200 m depth that are capped by the Jalquinche Formation mudstone and/or an ignimbrite, with lateral variability in the thicknesses of those low-hydraulic-conductivity units. Houston (2004) presented evidence that the Sifón Ignimbrite is an effective confining layer between the two aquifers.

In the central sector of the Calama Valley, most wells encounter an upper phreatic aquifer. In south-central areas the phreatic aquifer occurs in a limestone-dominated rock (EIA 2005 and EIA 2011, see Table 1), the Opache Formation. In the north-central area, the upper aquifer is variably in limestone mapped as the Opache Formation or in sandstone of the upper part of the Lasana Formation. A lower set of aquifers also occurs widely; at some locations the wells are artesian (e.g., northeast of location T in Fig. 3B). In the north-central region the lower aquifer occurs in conglomerate below 80–130 m depth (Figs. 10, 13, and 14); current geological mapping places this conglomerate in the Lasana Formation. A mudstone is considered to be the confining unit and attributed commonly to the Jalquinche Formation. However, Blanco and Tomlinson (2009) reported that the Jalquinche mudstone facies is only a few meters thick in the north-central area, and therefore is probably not an effective confining layer. Instead, mudstone intervals within the Lasana Formation (Blanco and Tomlinson, 2009) are more likely candidates for the low-transmissivity layers in that area.

South of Talabre (T in Fig. 2) but north of the Loa River, there is a region where there is no phreatic aquifer even though there appears to be lateral continuity of the limestone-rich unit. In that area, several wells prove a confined lower aquifer in a conglomerate unit (EIA 2005, see Table 1).

Both upper and lower aquifers are found near and west of Calama Hill (sector between H and O in Figs. 2 and 3) (Mayco 2013, see Table 1). An upper phreatic aquifer exists in unconsolidated alluvium, karstic limestone, calcareous sandstone, sandstone, and conglomerate. The calcareous upper part is referred to as the Opache Formation (Fig. 15D) and the lower detrital strata has informal unit names (e.g., black sandstone). A lower aquifer occurs in conglomerate referred to in most reports as the Calama Formation (Fig. 16C), although evidence presented in the following indicates that the aquifer is in multiple sedimentary and volcanic units. Above the lower aquifer is a thick clay-rich siltstone, commonly attributed to the Jalquinche Formation.

West of the central Calama Valley and north of the San Salvador River and Calama city, where the surface elevations rise toward the northwestern mountain range, the confining layer is an ignimbrite (EIA 2011, see Table 1). However, that ignimbrite pinches out northward so that the two aquifers become one phreatic aquifer (EIA 2011, see Table 1).

The sparse data for the hydraulic properties of the rocks are summarized in Table 10. Typical permeability for the upper aquifer in the central Calama Valley (L and T in Fig. 4) is 0.7–1.2 m/day. In the San Salvador–Loa Valley area west of Calama Hill (sector between H and O, Fig. 4), productive well fields in the upper aquifer report higher permeability (3.9 m/day average) (Fuentes Carrasco, 2009). Houston (2004) reported that some upper aquifer horizons in the eastern part of the study area have much higher permeability (120 m/day). Heterogeneity was also emphasized by Fuentes Carrasco (2009), who considered the upper aquifer west of Calama Hill to contain elongate channels of exceptionally high permeability.

The permeability of the conglomeratic lower aquifer in the eastern region ranges from 1 to 4 m/day (Table 10), with localized horizons of much higher fissure permeability (40–100 m/day) (Houston, 2004). In the central Calama Valley as well as in the area west of Calama Hill, well tests for the lower aquifer display generally higher permeability, 2–21 m/day (Table 10).

Data for the basement rocks upon which the Eocene–Quaternary sedimentary rocks accumulated and for units reported to serve as confining layers confirm that they are significantly less permeable. The few well tests conducted for the mudstone or ignimbrite that overlies the lower aquifer reveal permeability of ∼10–3 m/day (Table 10). Araya Torres (2010) documented that most permeability values for basement rocks in the adjacent mountain range on the west side of the basin, where hundreds of measurements exist because of mining-related geotechnical studies, are 10–3 to 10–6 m/day.

Groundwater Flow

Across the eastern, central and northern sectors of the Calama Valley, the piezometric surface of a phreatic aquifer (Fig. 2) declines from 2660 m in the northeast to ∼2300 m at the West fault, driving flow toward the southwest. West of Calama city, the flow direction is west-southwest (Fig. 2) where the water table declines 90 m to the springs of Ojos de Opache.

The aquifers discharge to surface water bodies in several areas. In the central Calama Valley, short-term stream gauging campaigns document significant transfer of water (e.g., several hundred liters/second at multiple locations) from the aquifers into the south-trending reach of the Loa River (between blue diamonds 2 and 3 in Fig. 2) (EIA 2005 and Matraz 2012, see Table 1). West of Calama city (Fig. 2), examples of discharge into the Loa River occur near the location of the La Cascada waterfalls (LC in Fig. 2) and through diffuse zones of springs that produce extensive wetlands in two tributary canyons on the Loa’s north bank (blue diamond 5 in Fig. 2). Although no gauging stations document the flow of the Loa above and below those springs, at times of low flow there is a visually pronounced downstream increase in the flow of the Loa River. Similarly, the San Salvador River is entirely spring fed, primarily from a set of springs at Ojos de Opache (O in Fig. 2; CORFO 1973, see Table 1). Klohn (1972) reported that the water chemistry of surface streams, springs, and wetlands in the area between Calama city and Ojos de Opache reveals a connection between the upper aquifer and the surface waters.

For a set of wells with roughly monthly measurements during the interval of time corresponding to most of the hydrological data, 2004–2005, the time-variable groundwater levels (Fig. 8) reveal a mixture of natural variability and human management. An annual cycle of increased river flow during austral summer (December–March) is perceptible in the Loa River flow (Fig. 8A) after it exits the Conchi Reservoir to pass into the Calama Valley, although the seasonal variations are managed and spikes in discharge are likely related to managed flow through dam drains. The Chiu Chiu well (Fig. 8B) offers a history of an upper aquifer well that is tightly coupled to the other major river that is a conveyance from the high-elevation parts of the catchment, as it is located <500 m from the Salado River near its juncture with the Loa. This well displayed a small-magnitude increase in the phreatic water table late in 2003, and a subsequent slow decline. That September increase predated by three months the anticipated annual precipitation cycle in the catchment highlands (Chaffaut, 1998). For wells located farther downflow and at considerable distance from the rivers (Fig. 3B), neither the upper nor lower aquifer heads display a seasonal variability that is tightly related to the upper Loa and Salado inflow to the Calama Valley, even with a phase shift. For those wells, rapid fluctuations in water levels are more likely the result of pumping of nearby wells (e.g., Figs. 8C, 8I) than natural causes. For the wells that are distant from the rivers, the range of variability of the head during 2003–2005 is 1–13 m, inclusive of large values interpreted to reflect human management (Table 9). With the available data, identification of natural variability is subjective, but during 2003–2005 the natural range is interpreted to be <2 m for both the upper and lower aquifers (Table 9).

The elevations of the aquifer surface in the north-central Calama Valley and of the Loa River bed constrain the plausible geology of the aquifer in the area north of the well data. The elevation of the phreatic aquifer in the two northernmost control wells (Fig. 2) is ∼2610 m asl in the valley center and ∼2660 m asl near the western margin of the valley. Following the piezometric gradient to the northeast, the landscape surface elevation rises to 2720–2800 m asl near the Loa River, before dropping 40–60 m to the river bed. Although the Opache Formation is exposed widely in the northern part of the Calama Valley, it was removed by erosion at the Loa canyon. Instead, the bedrock exposed at the depth of the river bed is within the Lasana Formation (Figs. 2 and 4). If a phreatic aquifer is recharged by the Loa River south of the Conchi Dam (Fig. 2), the most likely aquifer host is the Lasana Formation. Groundwater must then flow southwestward into what become both the lower and the upper aquifers.

Through the central and northern sectors of the Calama Valley, the head of the lower aquifer declines ∼400 m and the form of its piezometric surface is similar to that of the upper aquifer. The flow direction is from northeast to southwest (Fig. 3A). For the lower aquifer in the western part of the Calama Valley and San Salvador–Loa Valley, the pattern of the piezometric surface is more irregular. Immediately west of Calama Hill, the piezometric gradient is low (∼10 m/km) over a 5-km-wide zone before transitioning westward to a steep slope (∼20 m/km) in the region of the Ojos de Opache springs (Fig. 3A). Although sparse data south of Calama Hill (H in Fig. 3) suggest that lower aquifer water may flow toward the Loa canyon south of Calama city, there are no control wells near the Loa River southwest of Calama city or river gauges west of station 5 (Fig. 3) with which to verify possible lower aquifer conditions. Station 5 (Fig. 3A) marks the western point at which the Opache Formation, the typical upper aquifer host rock throughout the western part of the basin, crops out at the base of the Loa canyon. West of station 5, extensive volumes of carbonate deposits occur along some sectors of the river bed north of Chintoraste hills (tr in Fig. 3A), and spring-like tufa carbonate deposits occur along the traces of some east-trending faults on the walls of the Loa canyon. These deposits suggest spring drainage that is not tied to the upper aquifer.

The relative heights of the piezometric surfaces of the upper and lower aquifers vary across broad regions (Fig. 3B). In most of the northern Calama Valley west of the Loa River and in a zone along the west-central margin of the valley, the lower aquifer head is above that of the upper aquifer, producing localized artesian or near-artesian conditions (Fig. 3B). For well pairs 1 and 6 (Fig. 3B; Table 9), the variability of head over the reported months is sufficiently large to plausibly reverse these relative piezometric heights in some months. Nevertheless, persistently flowing artesian wells near well pair 1 demonstrate the robustness of the relative pressures within some parts of the northern region. In other broad areas, especially one in the south-central region and another near and west of Calama city, the head in the upper aquifer is higher than that in the lower aquifer (Fig. 3B). There are two smaller regions in which the lower aquifer head is similar to that of the upper aquifer (Fig. 3B). In most of the subareas these relative heights are robust over the 2003–2005 data years (Table 9).

Geology

Geohydrological Consequences of Faults and Folds

The long-lived, north-trending, oblique-slip West fault system occurs near the boundary between the Calama Valley and the San Salvador–Loa Valley (Reutter et al., 1996). Due to lateral offset during the mid-Oligocene–early Miocene, the Eocene and lower Oligocene strata in the southwestern Calama Valley would not have formed in continuity with similar-aged deposits in the San Salvador–Loa Valley (Figs. 7, 10, 11, and 14). Within the Calama Basin, vertical offset across the West fault apparently displaces the contact of the crystalline basement with strata by <200 m (Figs. 11 and 14).

Where the West fault zone cuts crystalline basement rock in the >800-m-deep Chuquicamata open pit mine, the zone of fault gouge and breccia is at least 3 m wide (Tomlinson and Blanco, 1997). Araya Torres (2010) documented that the fault within the mine region is a barrier to groundwater flow; however, that data set examined the hydraulic conductivity of fracture zones related to basement rock and did not evaluate hydraulic properties where the faults cut the moderately lithified sedimentary units that are the aquifers.

The east-trending Milagro fault system (between locations H and T in Fig. 7) formed contemporaneously with accumulation of the Eocene Calama Formation, and is buried by post-Eocene strata along most of its trace (Blanco, 2008; Tomlinson et al., 2010). Available outcrop and subsurface data imply that it is a north-dipping reverse fault in its central sector (Fig. 10) and that it declines in offset in the western and eastern sectors, where folds dominate the structure (Fig. 14). In general, lower aquifer rocks north of the Milagro deformation zone occur in the Lasana Formation, whereas lower aquifer rocks near and south of this fault occur in the Calama Formation (Figs. 10 and 14).

The fault within the basin with the largest vertical displacement, the northeast-trending Loa fault (Fig. 7; Table 6), displaces the pre–middle Miocene units and is a major discontinuity in the deeply buried Yalqui Formation (Figs. 10 and 13). Well data in the northern part of the Calama Valley are insufficient to test whether the fault impacts the aquifers (Figs. 10 and 13).

In the western sector, in the San Salvador–Loa Valley, near-vertical, small-displacement (tens to hundreds of meters), east-trending faults that are exposed in the canyon of the Loa River (Figs. 7 and 9) have an important local impact on the groundwater system. Because these faults juxtapose rocks of markedly different physical properties, such as coarse Eocene conglomerate against Jurassic metasediments, hydraulic conductivity changes abruptly. Preliminary data suggest that groundwater is forced to the surface along one of these faults, to form extensive calcium carbonate mineral deposits along the Loa River bed (tr in Fig. 3) and at paleosprings located on the canyon walls. There are no stream gauge data for the Loa River at suitable locations to test this discharge hypothesis. With many fewer constraints, it is inferred that parallel faults of similar magnitude occur 3–4 km to the north. These inferred faults would control the east-trending walls of the Quebrada de Opache canyon, and might control vertical displacement by tens of meters of some of the lithologic units described in the water monitoring wells (Fig. 14, southern extreme of cross section E-E′). To date, the potential affects by these faults on the hydrology of the Ojos de Opache springs area have not been considered, and no wells monitor the region down-gradient (southwest) of this set of inferred faults.

The final pair of structures known to cause major displacement of the strata that serve as aquifers are the Chiu Chiu monocline and accompanying Salado syncline (Figs. 7 and 17). At gross scale, the monocline is a gentle fold with down-to-the-east sense of stratigraphic offset of 100–200 m (Blanco and Tomlinson, 2009). The synclinal axis is a broad, shallow, and complexly folded zone, within which the Loa and Salado Rivers flow. The Chiu Chiu monocline is younger than the Opache Formation and older than the Quaternary Chiu Chiu Formation. The serpentine form of the monocline and evident subsidence of the area encircled by the monocline (Fig. 17) led Blanco and Tomlinson (2009) to interpret it to be the result of the subsurface dissolution and removal of an evaporite-rich unit located at many hundred meters depth (Fig. 13A). The 2–3-km-wide syncline would thus represent the topographic low formed above the area of maximum subsurface material loss and subsidence. The vertical position of the strata that contain the upper aquifer rises >100 m from east to west across the monocline (Fig. 13). South of the Loa River the monocline diminishes progressively in relief. Fractures related to the original dissolution and subsidence along the axis of the syncline, as well as fractures related to strain within overlying units (Blanco and Tomlinson, 2009), may have enhanced permeability through some rock units.

Distribution of Aquifer Host Rocks East and North of Calama Hill

Three cross sections (Figs. 10, 13, and 14) illustrate the geometry of the strata that contain the aquifers in the Calama Valley. Cross sections A-A′ (Fig. 10) and E-E′ (Fig. 14) are approximately parallel to the groundwater flow direction; cross section D-D′ (Fig. 13) is essentially perpendicular to groundwater flow.

The regionally extensive limestone of the upper Miocene and Pliocene Opache Formation, the host for the upper aquifer in many areas, displays a variety of facies (Figs. 15A, 15C), some with centimeter-scale vugs (Fig. 15C) and microkarst (May et al., 1999; Houston, 2004). Both bedding-parallel and nearly vertical fractures are common (Fig. 15A) and likely contribute to the hydraulic conductivity of the Opache. Laterally toward all the basin boundaries, the Opache grades to conglomerates (Figs. 6A and 17). In the northeastern Calama Valley, the Chiquinaputo Formation, which is also a host to the phreatic aquifer (Houston, 2004), interfingers with the Opache limestone (Figs. 5 and 13) as well as locally underlying the limestone (Blanco, 2008). The Chiquinaputo consists of well-sorted fluvial gravels (Fig. 15B). Zones of siltstone within the Chiquinaputo Formation, reported by Blanco (2008), likely have poor hydraulic conductivity and diminish the continuity of flow within the upper aquifer. Elsewhere, the basin-margin conglomerates are alluvial fan gravels (May, 1997; Blanco, 2008) that are not well sorted and likely have low hydraulic conductivity.

The Eocene Calama conglomerate (Figs. 6D and 16C) is the host to the lower aquifer in most of the southern Calama Valley. Blanco et al. (2003) and Blanco (2008) documented that, in outcrop, the lower 100 m of the Calama Formation contains several interbeds of andesite lava overlain by ∼450 m of alluvial and fluvial conglomerate cemented by either gypsum or clays. The Calama Formation pinches out northward near the Eocene-age Milagro fault and fold system (Figs. 7, 10, and 14) (Blanco, 2008; Tomlinson et al., 2010). In the eastern sector of the Calama Valley, depths to a local lower aquifer coincide with the interpreted depth to the Yalqui Formation (Figs. 6C and 13). Although facies of some exposures of the Oligocene–lowest Miocene Yalqui Formation are suitable to serve as a lower aquifer (Fig. 16A), much of the Yalqui Formation is a matrix-supported conglomerate (Fig. 16B) that is not likely to have adequate permeability. The kilometer-scale lateral variations from clast-supported to matrix-supported conglomerate texture (Figs. 16A, 16B) seen in outcrop suggest that aquifers in the Yalqui Formation in the eastern sector of the basin must be laterally complex. There is inadequate information in the zone between the Talabre, Llalqui, and Calama Hill areas (T, L, and H in Fig. 3A) to deduce how the gravel-dominated lower aquifer host rocks change across the Chiu Chiu monocline.

In the northern and eastern Calama Valley the middle Miocene Lasana Formation (Fig. 5) is the most likely candidate to be the gravel reported to host both the lower and upper aquifers (Figs. 10, 13, and 14). The Lasana Formation is at least 100 m thick where its lower member crops out along the Loa canyon north of Chiu Chiu (Fig. 4) (Blanco, 2008). Blanco (2008) and Blanco and Tomlinson (2009) described fluvial conglomerate, sandstone, and siltstone in the lower member, in repeated complex fining-upward series several meters thick, that become progressively dominated by siltstone to the south and west. The upper member has a higher percentage of mudstone (Blanco, 2008). Aquifers composed of the Lasana Formation are likely internally complex and laterally limited by facies changes. Blanco (2008) described preferred orientations of the sedimentological features of the lower member that may cause a favored south-southwest alignment of conductive aquifer properties. The Lasana Formation is age equivalent to and interfingers laterally with the Jalquinche Formation clay-rich siltstone unit (Figs. 5, 13, and 14), which functions as an aquitard (Table 10).

Widespread ignimbrites in the eastern and northern extremes of the Calama Valley (Ramírez and Gardeweg, 1982; Marinovic and Lahsen, 1984; de Silva, 1989) may be aquitards. The distribution of the Sifón Ignimbrite is well established in the east-central and southern parts of the Loa catchment basin. In the northern and easternmost sector, published maps and reports do not clarify the locations of the boundaries between the Sifón Ignimbrite and other very thick ignimbrites (e.g., Cupo, Divísico, Rio Salado, and lower San Pedro Ignimbrites; e.g., de Silva, 1989) in similar stratigraphic positions. Herein, these five ignimbrites are treated as a single map unit (Figs. 4, 10, and 13). Although locally at least part of the 1–100-m-thick ignimbrites is welded tuff, more widely the ignimbrites are not welded. The effectiveness of these ignimbrites to impede water flow is unresolved: common vertical fractures may serve as flow paths (Montgomery et al., 2003), yet water-pressure data in the Llalqui area demonstrate that the Sifón Ignimbrite locally confines an artesian aquifer (Houston, 2004, 2007).

In the southern part of the Calama Valley a major aquitard or confining unit is created by the fine sandstone and mudstone of the middle Miocene Jalquinche Formation, which is laterally extensive (Figs. 5 and 9–14) and as much as 200 m thick (May, 1997; May et al., 2005; Blanco, 2008; Blanco and Tomlinson, 2009; Tomlinson et al., 2010). Red mudstone is an important component of the Jalquinche Formation (Blanco, 2008), and the reddish color is suggestive of a diagenetic Fe-rich clay. Fine-grained reddish sandstone is both common and rich in gypsum (May, 1997; Blanco, 2008). Its color and gypsum content together suggest that even in the sandstone facies the primary porosity may have been occluded. North of the Talabre area (T in Fig. 4) the Jalquinche Formation thins to only 10 m, and is replaced by coarse-grained facies of the Lasana Formation (Fig. 6B) (Blanco and Tomlinson, 2009).

Not only are the geological units that serve as aquitards laterally variable from mudstone to ignimbrite, in addition their effectiveness as aquitards is heterogeneous within a single geological unit. Both the Jalquinche Formation and the ignimbrites change markedly in thickness as well as pinch out entirely. In some parts of the central Calama Valley the coarse facies of the lower member of the Lasana Formation either underlies the Sifón Ignimbrite or underlies a thick mudstone interpreted as the Jalquinche Formation or as a facies variation within the Lasana Formation (Figs. 10, 13, and 14); however, elsewhere available well data are not conclusive that any of these units is the aquitard (see especially Fig. 10, where boundaries are dashed lines). As a consequence of the heterogeneity of the overlying set of aquitards, down-gradient flow in deep aquifer horizons is likely to pass laterally from phreatic zones to confined zones (Fig. 3B).

Distribution of Units West of Calama Hill

Three cross sections, B-B′ (highly oblique to groundwater flow), C-C′, and the southern 15 km of E-E′ (the latter two subparallel to groundwater flow; Figs. 12, 13, and 14), illustrate major changes in the distribution of Calama Basin strata near the West fault and in the region where the Calama Valley meets the San Salvador–Loa Valley. Cross section C-C′, drawn south of the Loa River, is isolated by the deep Loa canyon from the regional groundwater flow. Nevertheless, the exposures in this southern zone permit detailed understanding of the architecture of the sedimentary and volcanic units (Figs. 2 and 9), unlike the area between the Loa and San Salvador Rivers where only the Opache Formation is exposed. It is assumed that cross section C-C′ (Fig. 12) approximates the geometries and depths of formations that are appropriate for the aquifer-hosting rocks between the two rivers.

Rocks related to a local Eocene volcanic center (Mpodozis et al., 1993; Trumbull et al., 2006) at the Chintoraste hills are an important part of the geohydrological setting west of the West fault (Figs. 4 and 9). Pyroclastic deposits are abundant, as well as epiclastic conglomerates, minor lava flows, and subvolcanic intrusives. The Chintoraste hills are a circular feature ∼4 km in diameter with an outer ring of outward-tilting (∼40°–50°) ignimbrites and a center that includes contact-metamorphosed Mesozoic strata. Reports from mineral exploration boreholes in the 10-km-wide valley between the West fault and Chintoraste hills indicate widespread pyroclastic deposits, primarily tuffs, and interbedded coarse fluvial siliciclastics. The lithologic distribution suggests that pyroclastic deposits near the Chintoraste center interfinger northward and eastward with conglomerates. In some of the mineral boreholes, water was reported within these deposits.

Capping the localized Chintoraste volcanic center is an Oligocene(?) or Miocene conglomerate that constitutes a more widespread sheet, ∼10 m thick (Fig. 6). From a depositional history perspective, the Eocene Chintoraste strata and the overlying conglomerate sheet are very different. However, few subsurface data exist that enable differentiation of this capping conglomerate from Chintoraste lithologies.

Borehole reports and paleospring locations suggest that the lower aquifer west of the West fault is dominated by pyroclastic deposits and interbedded conglomerates. This study attributes most of these water-bearing horizons to the Chintoraste complex (Figs. 11 and 14) and the overlying thin conglomerate sheet. A consequence of this interpretation of the lower aquifer host rocks is that, at the West fault, down-gradient flow in the lower aquifer (Fig. 3) transfers from an eastern host rock dominated by conglomerate (Calama Formation) to a western host rock dominated by pyroclastic facies and volcaniclastic facies (Chintoraste unit).

In the area west of Calama Hill (H in Fig. 4), the Opache Formation limestone and immediately underlying sands and gravels serve as the regionally extensive phreatic aquifer (Figs. 11, 14 and 15D; Table 10). The sub-Opache medium-grained sandstone to well-sorted cobble conglomerate ranges in thickness between 0 and 4 m (May et al., 1999) to 20–30 m (our mapping), too thin to distinguish in Figure 9 as a separate map unit. The outcrop belt of the sandstone and conglomerate at the base of the Opache Formation is associated with important springs.

The uppermost major unit with poor capacity to transmit water is a part of the Jalquinche Formation. It is composed mostly of gypsum-rich and clay-rich mudstones and very fine to fine sandstones. Heterogeneities in the Jalquinche Formation encompass local horizons of coarser sandstone and conglomerate. Overall, the thickness of the Jalquinche Formation varies markedly, from 0 to 200 m (May, 1997), while it thins westward, toward a basement paleoridgeline near which it pinches out (brown line, Fig. 9).

Rocks with properties suitable to act as aquifers are heterogeneous in this western area. A persistence throughout the Neogene of topography that funneled rivers through the narrow San Salvador–Loa Valley (Figs. 6A–6C) would likely have created preferred elongations of sedimentary facies in an east-west direction. Near what is today the canyon of the Loa River exists evidence of paleo–Loa River positions, expressed both in cross sections of channel forms and in landforms. One stratigraphically deep example is a paleochannel with an apparent width of 250 m that cuts the Chintoraste unit and has well-bedded siliciclastic fill that underlies the Jalquinche Formation. Shallow examples may either control zones of karst in limestone of the Opache limestone or control the distribution of Quaternary gravels.

To the west of Calama Hill, bedrock of the Precordillera constricts the sedimentary units that form the aquifers. This is especially true of the upper aquifer, where the horizontal distribution and vertical relationships of rocks with relatively high permeability are reduced from an eastern wide area (∼25 km north-south distance) to a narrow western area (∼5 km wide, measured north-south) (Fig. 18). Although the width of the lower aquifer also narrows westward near Calama Hill, the Chintoraste volcanic and volcaniclastic unit may continue at a similar elevation for more than 10 km southward, underlying a broad valley for which there are few data (Fig. 4). Given that the Jalquinche Formation coarsens southward from the Loa River, and given that the scant borehole reports in the southern valley reveal no evidence of a Jalquinche-like mudstone, an aquifer within that southern valley is likely to be phreatic.

In addition to the narrow passage between the north and south bedrock boundaries of the San Salvador–Loa Valley, water that passes north of Chintoraste hills in the lower aquifer encounters a second major bedrock constriction. A north-trending bedrock ridge separates the eastern sedimentary basin domain from a western domain of high-standing folded Mesozoic metasedimentary rocks and Cretaceous–Eocene intrusive bodies (Figs. 9 and 12), with a north-trending fault at the boundary. South of the Loa River the western bedrock domain and faulted boundary have a thin cover of Miocene–Pliocene gravel; between the San Salvador and Loa Rivers there is a thicker cover of Opache Formation, which obscures the details of the bedrock ridge (Fig. 9). On the east side of that bedrock ridge (west end of section C-C′, Fig. 12), both the Jalquinche mudstones and the Eocene Chintoraste unit with its permeable interbeds thin westward. In the intercanyon plain the host rocks for the regional lower aquifer shallow westward until they are in direct contact with the base of the thin sandstone and gravel that underlies the Opache Formation. In Figure 9, the brown contour marked 0 m traces where the Jalquinche Formation pinches out, and the green contour labeled 0 m traces where the Chintoraste unit pinches out. Between the brown and green lines and for some distance to the east of the brown line, where the Jalquinche is thin, there is no effective aquitard and therefore there is a single aquifer that is phreatic. Thus within a span of 10–15 km west of the West fault, it is inferred that the lower aquifer connects to the upper aquifer (intercanyon region) and can discharge directly to both rivers.

At a smaller scale near the Loa River, two intersecting sets of faults control the position of the western boundary of the rock unit that hosts the lower aquifer. The first fault set is north trending (F1 and F3, Fig. 9) and forms the western boundary of the Eocene Chintoraste unit. The second set, east trending and of small displacement, juxtaposes the Chintoraste unit against the Mesozoic impermeable basement (F2 in Fig. 9). That set also juxtaposes multiple lithologies within the Eocene volcano-sedimentary package against one another, some likely permeable and some of low permeability. Locally, the result is that the lower aquifer host rocks terminate at a corner.

DISCUSSION

Overview

For the Loa system, the existing subsurface and surface flow data are inadequate to quantify numerous parts of the hydrological system. It would be convenient to assume that aquifer properties are laterally homogeneous, so that sparse data can be widely applied. However, the geological properties of the sedimentary basin that hosts the major aquifers of the Loa system point toward considerable heterogeneity (Fig. 18). This paper contributes an improved understanding of the architecture of the rocks of suitable hydraulic conductivities to serve as aquifers or as aquitards within and adjacent to the Calama Valley, and their influences on the preferential flow pathways and possible discharge regions.

Within the Calama Valley the groundwater supplied from the eastern mountains and from the northern mountains join at multiple levels. At the surface, the baseflow-fed Salado River and baseflow-fed Loa River merge. In the subsurface, groundwater enters from the highlands and encounters sedimentary rocks of the Calama Basin. Figure 18 illustrates in a simple geohydrological sketch the vertical changes in hydraulic conductivity but displays only the most rudimentary aspects of the horizontal variability. The steep regional topographic gradient imposes a strong piezometric gradient that directs groundwater to the southwest and then west across the study area.

The aquifers of the north sector of the Calama Valley are filled at least in part by infiltration of water from the Loa River into the conglomerate and sandstone of the Lasana Formation along the sector of the river south of the Conchi Dam (Figs. 2 and 18). Additional groundwater may enter the Calama Valley from the peaks above 4000 m elevation beyond the northeast end of Figure 18 section X-X’, passing beneath the bed of the Loa River in the lower parts of the Lasana Formation. Once in the Lasana Formation, the groundwater migrates southwest and encounters lenses and formations of variable permeability, with the outcome that the groundwater is progressively split into multiple aquifers (Fig. 18B).

Groundwater in the east-central Calama Valley, east of the Loa River, is sourced by water that precipitated in the highlands to the east and northeast, by recharge that may be both direct (Houston, 2007) as well as indirect, e.g., by infiltration of Salado River water into its alluvial bed. The extent of or locations where the groundwater from the eastern aquifers (L, Llalqui area, Fig. 18A) mixes with the water in the central Calama Basin aquifers are not yet documented (Fig. 18B; note question marks where the eastern and northern cross sections should connect).

From the perspective of the broad Loa water system, within the Calama Valley there is no net increase in water because there is essentially no direct precipitation. However, within the study area the rivers exchange with the aquifers at various locations. A key river-groundwater exchange may occur near the northern limit of the Calama sedimentary basin, where the Loa River likely loses water into the Lasana Formation, which down-gradient hosts both a lower confined aquifer and the upper phreatic aquifer (Figs. 10 and 14). Elsewhere in the basin, east of the West fault, the available data suggest that only the upper aquifer exchanges with the rivers. However, at the West fault and within 20 km distance to its west, most of the water in the lower aquifer must discharge to the upper aquifer or to the rivers directly.

Calama Valley and West Fault System Hydrogeology

Across the western and central Calama Valley the piezometric surfaces of the two aquifers are of similar elevations in some sectors, but elsewhere the lower aquifer piezometric surface is higher than that of the upper aquifer (Fig. 3B). Within those broad areas (Fig. 3B) the aquitard likely permits a small degree of slow flux upward and the lower aquifer may recharge the upper aquifer. The sedimentary rocks that serve as the aquifer are highly heterogeneous across this region. Near the north-trending reach of the upper Loa River and the Chiu Chiu monocline, southward and westward thinning of the Sifón (Blanco, 2008; Blanco and Tomlinson, 2009; Tomlinson et al., 2010) and Cupo Ignimbrites (Figs. 10 and 13) lead to their loss of effectiveness as confining units. Similarly, lateral variations in the thickness of the Jalquinche Formation and mudstones in the Lasana Formation lead to transitions in their capacity to be effective confining layers (Figs. 10, 14, and 18B).

In areas in the western and central Calama Valley where the head in the upper aquifer exceeds the piezometric surface of the lower aquifer (red in Fig. 3B), we expect that slow downward flow from the upper aquifer may recharge the lower aquifer. The narrow strip north and east of the Talabre region (Fig. 3B) where the pressure in the two aquifers is similar corresponds to a transition zone from excess head in the lower aquifer to excess head in the upper aquifer.

The piezometric gradient of the lower aquifer in the 5 km east of the West fault is markedly steeper than in the central sector of the Calama Valley (broad region around T in Figs. 3A and 14). A less pronounced increase in gradient occurs also within the upper aquifer east of the West fault (Fig. 2). Changes in the piezometric gradient of the lower aquifer may reflect the Milagro deformation zone and related changes in the thickness of suitable aquifer host rocks (Fig. 14), but lithological data for the lower aquifer are not sufficient to test this hypothesis.

The heads of the two aquifers come into equilibrium near the West fault zone, which comprises the West fault and a subsidiary parallel fault, located 1 km to the east (Fig. 3B) (Tomlinson et al., 2010). The Loa River turns sharply to the south near the trace of the eastern branch of the fault, continues ∼6 km in the zone between the two faults, and then resumes its westward direction (Fig. 7). In the subsurface, rocks that host the lower aquifer, the Eocene Calama Formation conglomerate to the east and Eocene Chintoraste pyroclastic and sedimentary rocks to the west, meet at the West fault (Figs. 10, 11, and 14). Lower aquifer water flow from east of the West fault to west of the West fault system navigates through permeability pathways that are not stratigraphically continuous. Although those deeper units are discontinuous, the Jalquinche Formation and the upper aquifer-bearing Opache Formation accumulated after most of the displacement across the West fault zone, and likely underwent much less disruption (Tomlinson et al., 2010). Nevertheless, the nearly equal heads (blue zone north of H in Fig. 3B) likely indicate an active flow between the lower and upper aquifers through a less effective aquitard. The faults may increase the heterogeneity of rock units that would otherwise act as aquitards, and they may place low-conductivity rocks adjacent to high-conductivity rocks. Recent research shows that large faults can effectively produce a greater hydraulic connection between shallow and deep aquifers (Bense et al., 2013).

San Salvador–Loa Valley Hydrogeology

The interpretation that the lower aquifer in the western region is hosted by the Chintoraste complex and a thin overlying conglomerate is a departure from prior interpretations, which ascribed the aquifer to the Calama Formation. The borehole geology and the outcrops at Chintoraste hills reveal that the host rocks west of the West fault are much more pyroclastic and volcaniclastic than is the Calama Formation (Blanco et al., 2003; Blanco, 2008). Furthermore, considering that tens of kilometers of left-lateral displacement along the West fault are interpreted to have postdated accumulation of the Eocene Calama Formation (Tomlinson and Blanco, 1997), it is unlikely that the Calama Formation continues to the west of that major fault.

Although the Loa River turns abruptly southward where it intersects the east branch of the West fault and parallels the fault set for ∼2 km (Fig. 7), potential exchanges between the river and aquifers cannot be quantified with the sparse stream gauge data (Table 2), especially because several irrigation channels tap the river in this reach. Only a few kilometers farther west, widespread springs discharge from the upper aquifer to the Loa River channel and to the main tributary to the San Salvador River. Much of the discharge occurs because the upper aquifer is intersected by the canyon walls (e.g., springs near LC, Fig. 9).

Whereas at the West fault the piezometric height of both aquifers is ∼2240 m (Figs. 2 and 3A), throughout the area with data west of the fault the piezometric height of the upper aquifer is 20–40 m higher than that of the lower aquifer. This relative loss of head in the lower aquifer might occur for either of two reasons. First, the rocks with high permeability below the aquitard layer might thicken westward, allowing more vertical dispersion in the lower aquifer of water that infiltrated across the West fault. Second, part of the water from the lower aquifer east of the fault might have transferred upward within the fault zone into the upper aquifer. Data are not available to test which explanation is more viable. Whatever the cause, the result is a zone with relative pressures that create a hydraulic gradient favoring downward seepage of upper aquifer water into the lower aquifer, across the intermediate Jalquinche Formation (Fig. 18).

Unlike prior studies, this study concludes that within 20 km west of the West fault the lower aquifer discharges significantly, if not completely, to the surface water system, because the host rocks for the lower aquifer unit thin between impermeable basement and a thinning confining unit. This lower aquifer discharge is in part direct, because the host rocks crop out in the San Salvador, Ojos de Opache, and Loa canyon walls (Fig. 9). Although hydrological evidence of groundwater discharges into the Loa River canyon where the Chintoraste unit crops out is lacking, extensive carbonate deposits in the bed of the Loa River north of Chintoraste hills (Fig. 9) and paleospring carbonates located above the modern water level on the canyon walls are hypothesized to be byproducts of lower aquifer springs. The lower aquifer discharge is also indirect in part, through the upper aquifer between the San Salvador and Loa canyons. This indirect discharge is inferred from the westward termination against a basement ridge of a major aquitard, the Jalquinche Formation (brown dashed line in Fig. 9), and of the subjacent units with lithologies that are suitable for moderately high permeability (green line in Fig. 9). Borehole data imply that little mudstone separates the Chintoraste unit from the sub-Opache sandstone of the upper aquifer (westernmost 4 km E-E′, Fig. 14), and hence water could migrate easily from the lower aquifer to the upper aquifer.

Important improvements in knowledge of the hydrogeology of the critical region west of Calama city will require geological mapping at high spatial resolution. In addition to providing greater precision on positions of formation boundaries, mapping is needed to specify facies variations in the water-bearing rocks, to establish the positions of channelized strata of high hydraulic conductivity, and to relate paleospring deposits to the modern groundwater hydrology. Given the widespread cover by the Opache Formation of underlying complex lateral changes in aquifer-host units and aquitard units, and the vertical canyon walls along the Loa River, novel observation techniques and high-resolution geophysical surveys may be needed.

Intersection of the Aquifers of the Eastern and Central Sectors

A lack of publicly available piezometric data for the eastern sector of the basin (from Llalqui to the Salado River; Figs. 2 and 3A) results in very little documentation of what happens to either aquifer near the Chiu Chiu monocline (Figs. 13B and 18). In that sector, where the Loa River flows parallel to and immediately east of the monocline, a short-term stream gauge campaign conducted late in a summer season (March) indicated that the upper aquifer discharges ∼700 L/s to the river (Matraz 2012, see Table 1), accounting for about a quarter of the surface water flow. Another discharge from the upper aquifer to a spring on the north bank of the Loa River, ∼100 L/s, occurs at the southern crossing of the Loa River canyon over the monocline (Fig. 4) (Matraz 2012, see Table 1). This spring location suggests a likely structural control on upper aquifer groundwater flow, which is combined with insight into aquifer host rocks developed elsewhere in the Calama Valley to put forth a hypothesis for groundwater flow near the Chiu Chiu monocline. The lower aquifer occurs at a depth exceeding 200 m (hosted in the Yalqui Formation and a conglomerate that may be age equivalent with the Lasana Formation; Fig. 13; Blanco and Tomlinson, 2009), with a considerable thicknesses of two overlying low-permeability units (Jalquinche Formation and Sifón Ignimbrite). Given the westward dip of the aquifer host strata and aquitards east of the Loa River and their higher elevations across the monocline west of the river (Fig. 13B), it seems likely that the Salado-Llalqui region groundwater does not cross the monocline to mix with the northern source region groundwater of the Talabre area. Blocked by the rise in elevation of the aquifer hosts at the monocline, lower aquifer water may flow south into the region where the thick (hundreds of meters) Eocene Calama Formation is expected to occur (Figs. 4 and 6D). Within the Calama conglomerates, water may continue southward until the monocline tip is reached, south of which it then flows westward between, and paralleling, the Loa River and the southern basin boundary. Some lower aquifer water would thus flow south of Calama Hill (H in Fig. 4). An absence of data near and south of the west-flowing reach of the Loa River (Fig. 3A) precludes any further evaluation of this hypothesis.

CONCLUSIONS

The integration of data describing spatial variations in sedimentary facies and their thicknesses has the potential to improve understanding of any complex groundwater system located within the fill of a sedimentary basin. This Calama Basin study integrated information about lateral variations in the potential to store and transmit water with an assessment of the primary aquifers, and thereby clarified the spatial distribution of the units with which the confined or semiconfined aquifer system is associated. The results are data-based hypotheses for recharge of the aquifers of the northern Calama Basin and discharge west of Calama city.

The results suggest that neither the upper nor lower aquifer corresponds to a laterally persistent geological unit. A comparison of piezometric maps for the two major aquifers implies that there is a patchwork pattern of areas in which water exchanges between the two aquifers, areas where the lower aquifer is confined, and areas where a phreatic aquifer is absent (Figs. 3B and 18). Across the central Calama Basin this pattern results from the lateral variability in thickness and in hydraulic properties of the aquitards, which are both ignimbrites and mudstones. Much of the lateral variability in aquifer properties results from facies changes in the middle Miocene Lasana and Jalquinche Formations, as well as in the upper Miocene to Pliocene Opache and Chiquinaputo Formations.

Folds and faults add to the architectural complexity of the aquifers. A principal example occurs west of Calama city, where a fault-controlled, north-trending basement ridge against which most of the Calama Basin sedimentary rocks terminate controls discharge from a lower, semiconfined aquifer to springs and rivers. A second important example is the West fault, across which piezometric gradients change and groundwater flow navigates major changes in the host rocks.

Two general conclusions of this analysis should be useful in the design of studies that will improve understanding of the coupled surface water and groundwater system. First, the spatial variations of the aquitards exert a key control on the exchanges of water between the upper and lower aquifers. Additional research focused on the sedimentary architecture of the middle and late Miocene sedimentary basin, with a focus on the aquitard facies, would likely lead to better understanding both of preferential flow pathways and the locations where there are exchanges between an upper and lower aquifer. Second, faults and folds near which specific sedimentary units terminate, change in thickness, or change in elevation likely have groundwater flow consequences. The hypothetical consequences can be tested by monitoring the groundwater and surface water at locations near those faults and folds, in both upflow and downflow directions, or with new geophysical and geochemical studies.

We thank Nicolás Blanco Pavez and Andrew J. Tomlinson of the Servicio Nacional de Geología y Minería (Chile), Luís Baeza Assis, Jorge Jemio Figueroa, and Manuel Bucci Ramirez of CODELCO (Corporación Nacional del Cobre de Chile), and Luís Rojas B. and Arturo Beltrán Schwartz of the Dirección General de Aguas (Chile) for initial access to locations, reports, and data, as well as for in-depth discussions of the Loa system. We appreciate the early encouragement by and discussions with John Houston. Alex Covarrubias Aranda and Rodrigo Riquelme Salazar of Universidad Católica del Norte, Oscar Cristi of the Universidad del Desarrollo, Gary Libecap and Eric Edwards of University of California Santa Barbara, and Lovell Jarvis of University of California Davis contributed greatly to our understanding of the management of the Loa water. U.S. National Science Foundation grant OISE-1037929 enabled Jordan, Godfrey, and Kirk-Lawlor to gain an understanding of the multifaceted challenges for water management in the study area. A Fulbright Fellowship for Jordan in 2012 provided partial support of this project. We also thank the makers of Google Earth and the agencies and companies who acquire satellite images displayed on Google Earth for the free availability of this information and tool, which were fundamental for this study. A critique of an earlier manuscript by Maria-Theresia Schafmeister was very helpful. We are grateful for reviews by Andrew Tomlinson and John Houston that led to improvements in the manuscript.