Introduction

The water resources benefits of managed aquifer recharge (MAR) are compelling. There are now considerable long-term operational data that demonstrate the feasibility of using aquifers to store and improve the quality of waters. Managed aquifer recharge systems may be significantly less expensive than other storage and treatment options. Managed aquifer recharge also includes “green” and “right-sized” technologies that are appropriate for the technical and financial resources available in developing countries. However, MAR does not work everywhere. Some MAR systems have failed to meet performance objectives and expectations. The unavoidable reality is that the performance of MAR systems depends primarily upon local hydrogeological conditions, which may not be favorable for the planned type of MAR system. As was reviewed with respect to aquifer storage and recovery (ASR), some failed systems had hydrogeological conditions (e.g., extreme aquifer heterogeneity) unfavorable for the recovery of recharged water that were evident during the initial exploratory well programs (Maliva and Missimer 2010).

An important goal for the implementation of MAR is to improve the understanding of local hydrogeological conditions so that it is possible to better predict the performance of various system types and design options and to optimize the performance of systems given any local hydrogeological constraints. Design and operational variables that influence system performance include the location and spacing of wells, well completion (e.g., design and depths of well screens), design of infiltration trench systems (size, shape, and depth), pretreatment of injected water, and recharge and recovery rates. These variables can be evaluated through groundwater flow, solute-transport, and geochemical modeling to determine the optimal solution. The objective is to more accurately determine both the likelihood that a system will meet performance objectives and optimal system design before the time and financial investment is made to construct a full-scale system. If conditions unfavorable for MAR exist at a site, then it is important to identify them was early as possible in a project.

Prediction of MAR system performance requires groundwater modeling, which requires an accurate conceptual model and data on aquifer hydraulic and transport parameters. The primary reason for inaccurate model predictions, in general, commonly lies in incorrect conceptual models (Poeter and Anderson 2005; Bredehoeft 2003, 2005; Poeter 2007). In the case of underperforming ASR systems, conceptual models failed to recognize and quantitatively incorporate aquifer heterogeneity (Maliva and Missimer 2010). Opportunities for improved implementation of MAR lie in more detailed and accurate aquifer characterization, which can be achieved by the targeted application of conventional and advanced technologies developed for both the oil and gas industry and specifically for water resources investigations. Another avenue for improved implementation of MAR lies in workflows and modeling tools that allow for the efficient processing and interpretation of all collected hydrogeological data. The data are then incorporated into models that more accurately predict groundwater flow and mixing as well as the hydraulic and geochemical responses of aquifers to recharged water. The objective is to extract the maximum value out of data collected as part of aquifer characterization programs.

Surface geophysics

The principal value of surface geophysical technologies is that they permit a denser data coverage at a lower cost and in a shorter time period than borehole-based methods. However, surface geophysics compliments, but does not replace, more traditional borehole-based field measurements. Borehole geological, water quality, and hydraulic data are needed to calibrate and validate surface geophysical data. Surface geophysical methods require the presence of contrasts in subsurface geological and hydrogeological properties that are detectable at land surface such as a sharp transition from clay-rich to clean sediments, boundaries between waters with pronounced differences in salinity (e.g., interface between fresh and saline groundwater), the water table (change in fluid saturation), and the contact between rock and sediments with different hardness and densities.

Surface geophysical data are typically interpreted by inversion solutions that essentially involve finding a best fit solution between theoretical geophysical responses generated for various earth models (e.g., configurations of layers and their properties) and the actual field (measured) data. Inversion solutions are not unique and ambiguity applies to the interpretations from all geophysical methods. Surface geophysical interpretations are more accurate if the data processing is based on a sound conceptual knowledge of the strata being evaluated. Surface geophysical methods are coarse techniques with relatively low vertical resolution compared to borehole geophysics.

Failure to adequately evaluate whether or not the target of a planned geophysical survey is detectable at a project site using various proposed techniques can result in costly and embarrassing failed surveys. Forward modeling of the geophysical response to various hydrogeological systems can be used to determine in advance if the target is detectable using a given technique and to select and design the appropriate testing procedures for an investigation (Fitterman and Stewart 1986; Mills et al. 1988). For example, Minsley et al. (2011) used forward modeling to evaluate the sensitivity of hydrogeophysical methods to map a subsurface plume of injected freshwater and changes in the effective stress in the storage zone resulting from increased pore water pressure in a proposed ASR system in Kuwait. The sensitivity analyses indicted that resistivity methods could differentiate between stored freshwater and native groundwater, but that the seismic data could not provide useful data on effective stress.

Surface geophysics has been used in MAR projects as part of general hydrogeological investigations. DC resistivity and time-domain electromagnetic induction (TDEM) profiles and soundings have been used to determine the base of aquifers (contact with hard rock and sediments) and to locate depths to salinity interfaces. For example, TDEM sounding data were used to locate zones of moderately brackish groundwater that are suitable for use as ASR storage zones in central Florida (Maliva and Missimer 2010). A seismoelectric survey was performed as part of the investigation of an ASR site in Abu Dhabi (Dean et al. 2012). Resistivity-based methods are particularly useful for the evaluation and design of salinity barrier systems because of the large contrast in salinity associated with the saline-water interface.

Airborne geophysical methods allow for the efficient coverage of large geographic areas and can be performed in areas poorly accessible to ground travel. Large areas can be covered in a short time with low cost compared to ground-based investigations, but at the expense of lesser lateral and vertical resolution. Airborne electromagnetics (AEM) have been used to obtain data on the geology and salinity of shallow aquifers. Helicopter-based AEM was used, for example, as part of the Broken Hill Managed Aquifer Recharge (MAR) Project, in western New South Wales, Australia, to map the three-dimensional hydrostratigraphy, structure and groundwater salinity of the 7,541.5 km2 study area (Lawrie et al. 2012). The AEM and ground-based data were used to map and assess potential MAR wellfield locations. A general limitation of AEM is that it is vulnerable to anthropogenic interference and is thus not effective in developed areas.

Surface NMR, also referred to as magnetic resonance sounding, (MRS), measures the free water and thus provides a measure of porosity. Magnetic resonance sounding has the potential to provide quantitative data on pore size distribution, which can be processed to provide an estimate of hydraulic conductivity. Magnetic resonance sounding has already passed the experimental stage and is evolving into a useful tool for hydrogeophysics (Yaramanci and Müller-Petke 2009). A promising application of MRS is as a non-invasive screening method for cost-effectively obtaining a large data set of transmissivity and specific yields for shallow aquifers, which can aid in aquifer model parameterization (Boucher et al. 2009). However, there are still limited data available on the accuracy of the MRS for quantitative estimation of aquifer hydraulic properties (Mueller-Petke et al. 2011).

Surface geophysical methods have much to offer in terms of providing operational monitoring data for MAR systems in which recharge results in a substantial change in subsurface properties. Time series of measurements have been used to map local changes in water levels and salinity. Time series of resistivity measurements allow for the mapping of the movement of the saline-water interface induced by aquifer recharge. For example, a combination of vertical electrical soundings (VES), seismic refraction surveys, and TDEM were used to map the position of the saline-water interface in Wadi Al Hawasinah, located along the Batinal coast of northern Oman (Abdalla et al. 2010). The position of the saline-water/freshwater interface was found to have migrated seawards by about 600 m between 2002 and 2007, which was attributed to increased recharge induced by a wadi dam and regulation of groundwater pumping (Abdalla et al. 2010).

Microgravity methods provide information on changes in the mass of water in aquifers, which has been demonstrated to be of value for monitoring water levels in unconfined aquifers. Pool and Schmidt (1997) documented the use of time series of relative microgravity surveys to monitor water levels changes in an aquifer recharge project in Tucson, Arizona (USA). The gravitational survey data allowed for the quantification of the increase in the volume of stored water due to recharge and subsequent decreases caused by groundwater withdrawal and a net outflow of groundwater from the study area. Howle et al. (2002) similarly demonstrated the use of microgravity data to map the mounding of the water table at the Lancaster, California, ASR site. Microgravity was successfully used to map the location of stored water at the City of Arvada ASR project in Colorado (Davis et al. 2005, 2008). Microgravity is not a substitute for groundwater monitoring wells, but can be used to cost effectively increase the spatial density of monitoring points.

Borehole geophysics

Borehole geophysical logging has long been a critical tool for reservoir characterization in the oil and gas industry and is widely used on a lower technical level in the water industry. The applications of conventional and advanced borehole geophysical logging techniques to MAR projects was reviewed by Maliva et al. (2009a), Herrmann (2009, 2010), Maliva and Missimer (2010). Conventional borehole geophysical logs, such as caliper, natural gamma ray, resistivity, and sonic, are commonly used to provide information on lithology, porosity, and groundwater salinity. Resistivity logs are used to provide qualitative information on relative permeability. For example, flow zones are often identifiable by the invasion of drilling fluids into a formation, as indicated by a difference in shallow versus deep resistivity.

Flowmeter logging, when combined with aquifer hydraulic data from pumping tests, is used to provide information on the location and transmissivity of flow zones within strata being investigated for use as a storage or recharge zone. Flowmeter logging and its applications to hydrogeologic investigations has been discussed by Javandel and Witherspoon (1969), Keys (1989), Molz et al. (1989, 1990, 1994), Paillet (1998), Paillet and Crowder (1996), Young et al. (1998), Paillet and Reese (2002), and Maliva and Missimer (2010). Conventional spinner (impeller) and electromagnetic flowmeter logging provides a means to evaluate large-scale, layered aquifer heterogeneity. The contribution of flow from various depth intervals is used to divide an aquifer into multiple zones (and in turn, model layers) for which a transmissivity value is calculated. High-resolution heat pulse flowmeter logs have been used in general hydrogeologic investigations to identify hydraulically active fractures (e.g., Morin et al. 1997).

Pavelic et al. (2006) provide an excellent example of the application of flowmeter logging to a MAR project (Bolivar ASR system in South Australia). Electromagnetic flowmeter data revealed a two order of magnitude variation in hydraulic conductivity in the approximately 50 m thick aquifer. Solute-transport modeling results indicated that the subdivision of the aquifer into four layers based on the flowmeter log data was sufficient to allow for satisfactory model calibration. Another key observation was that the hydraulic conductivity values obtained from core samples resulted in an underestimation of the transmissivity by a factor of almost 50. The bias toward low values was attributed to the preferential recovery of core from well-cemented sections of the aquifer.

Advanced borehole geophysical logs, such as nuclear magnetic resonance (NMR) and microresistivity imaging logs, allow for the fine-scale characterizations of aquifer heterogeneity. Borehole NMR logging was reviewed by Kenyon et al. (1995), Coates et al. (1999), Allen et al. (2000), Henderson (2004), Freedman (2006), and Serra (2008). Borehole NMR logs provide a measurement of both the total porosity and pore size distribution of the logged strata. These data are further processed to obtain profiles of hydraulic conductivity versus depth. The logs can identify immobile water that could be sources of solutes that diffuse into (and impact the quality of) water stored in MAR systems. NMR logs were obtained for ASR projects in South Florida to obtain a detailed profile of hydraulic conductivity versus depth (Fig. 1), which was used to identify potential storage and confining zone strata (Maliva et al. 2009a, 2011).

Fig. 1
figure 1

NMR and FMI logs from an ASR exploratory well in Daytona Beach, Florida. Photograph of a core from an adjoining borehole illustrates actual features observed on FMI log. The FMI provides a clear image of secondary pores formed by the dissolution of benthic foraminifera. The pore-size distribution obtained from the NMR log and macroporosity data from the FMI log were processed to obtain a detailed profile of hydraulic conductivity versus depth

Full size image

The most important applications in aquifer characterization is the identification of secondary porosity features, such as fractures and cavities, which may have a high permeability and, as a result, may be preferential loci for groundwater flow. Borehole imaging logs vary in their image type, with the most commonly used being the optical televiewer (OTV), acoustic televiewer (ATV), and microresistivity imaging logs. Prensky (1999), Hurley (2004), and Serra (2008) provide good overviews of the history and applications of borehole imaging technology including references to key papers. All three imaging log types provide continuous or near continuous and orientated 360° views of the borehole wall. The OTV uses a ring of lights to illuminate the borehole, a conical or hyperbolic reflector, and camera to record the images. The OTV was used, for example, to evaluate porosity types in the Biscyane Aquifer of southeastern Florida (Cunningham et al. 2004), which is the primary water source for the region and the subject of numerous existing and planned aquifer recharge projects.

The ATV log, which is also referred to as the borehole televiewer log (BTV), provides a continuous high-resolution image of the borehole wall, even in mudded holes. The ATV tool consists of a rotating piezoelectric ultrasonic transducer that operates in a pulse echo mode. The transducer, which acts as both a transmitter and receiver, measures the travel time and amplitude of an acoustic pulse that is reflected off the borehole wall (Prensky 1999). The signal amplitude is a function of the acoustic impedance (density and acoustic velocity) of the formation. As is the case for microresistivity images, the ATV images are typically presented as unwrapped 360º diagrams. Acoustic televiewer logs work best where there are large contrasts in acoustic impedance, with a prime example being between solid rock and open factures or cavities. The ATV is not commonly used in groundwater investigations because of the high cost of the equipment and the need for experienced operators. It has been applied in investigations where information is required on the abundance and extent of fractures and other secondary flow features. In Florida, for example, the ATV log has become a standard tool for confinement analyses of deep injection well systems.

Williams and Johnson (2004) reviewed and compared the ATV and OTV. Turbidity in the water and coatings on the borehole walls impact the quality of OTV images. The OTV will also not be able to detect features if there is not an associated difference in color. An advantage of the ATV is that it does not require clear water in the borehole (can be run on mud-filled holes). However, the ATV may not show bedding planes or other features if there is no associated change in borehole relief or acoustic properties.

Microresistivity imaging logs provide high-resolution images of the borehole wall. A 360-degree oriented image of the borehole wall is generated, which is typically presented as a flat “unwrapped” colored diagram. As is the case for conventional resistivity logs, the measured microresistivity values are a function of porosity, pore fluid resistivity (salinity), cementation, and clay content through the Archie (1942) relationship (Prensky 1999). Microresistivity imaging logs are used to identify and quantify the abundance and distribution of secondary macroporosity types, such as fractures and vugs, that could potentially dominate groundwater flow and solute transport. Microresistivity imaging logs can also assist in the identification and characterization of relatively low permeability beds that could act as confining strata and allow for the visualization of sedimentary and structural features such as bedding, lamination, brecciation, and slumping. Structural strike and dip are determined from the orientation of bedding. Fullbore Formation Imager (FMI™) logs run on dolomitic strata being considered for an ASR storage zone in South Florida clearly imaged features as small as the molds of foraminifera that are less than one millimeter in diameter (Fig. 1). The image was processed to quantify the abundance (percentage) of secondary macroporosity. Gamma ray spectroscopy logs (e.g., Elemental Capture Spectroscopy; ECS™ log) provide information on the mineralogy of the logged formation. The ECS sonde has a neutron source and measures the full spectrum of gamma rays generated from neutron-element interactions. The measured gamma ray energy spectrums are processed to determine the contributions from specific elements, which data are further processed to determine the abundance of common sedimentary rock types and constituents including calcite, dolomite, total clay (and some clay mineral types), QFM (quartz, feldspar, and mica), siderite, and pyrite.

Reservoir simulation platforms and modeling

Considerable amounts of data are collected during hydrogeologic evaluations, often at great expense, but much of the information is often not fully utilized in the investigations. Often, borehole geophysical logs are relegated to an appendix in reports. Software are available, such as the Petrel™ reservoir simulation platform, that allow for the incorporation and processing of a wide variety of different data types to create geological models, which are then used to develop groundwater flow and solute-transport models. Workflows are available that enable the quantitative extraction of data from geophysical logs, upscaling of the log data, and then incorporation of the log and other data into hydrogeological models, and then numerical models. The objective is to maximize the value obtained from the borehole geophysical log and other data and use those data to create models that more accurately simulate the groundwater systems of interest. For the Shwaib ASR system in Abu Dhabi (UAE), for example, advanced borehole geophysical and other well data were processed to obtain a facies map the captures much of the aquifer heterogeneity that would influence system performance (Fig. 2).

Fig. 2
figure 2

Vertical (top) and horizontal (bottom) high-resolution facies profiles for the Shwaib ASR system (Abu Dhabi, UAE), which were generated from well lithological and advanced borehole geophysical log data. The data were processed to generate a groundwater model that captures much of the aquifer heterogeneity. Cells are 10 m by 10 m

Full size image

The type of modeling performed for MAR systems varies with system type and complexity. For systems that recharge freshwater into similar quality freshwater aquifers, basic groundwater flow models are sufficient to evaluate processes such as mounding during recharge and drawdown during recovery. In ASR systems in which freshwater is injected into brackish aquifers, the mixing and movement of the injected water is of concern and solute-transport modeling is required. Solute-transport modeling is also necessary for MAR systems that store impaired water or utilize aquifer storage for treatment. Aquifer heterogeneity needs to be incorporated into the models as the presence of preferential flow paths could allow for much more rapid transport than would occur under more homogenous conditions. Density-dependent solute-transport modeling may be required in MAR systems in which there are large differences in salinity between the waters involved.

A review of the performance of ASR systems the use brackish storage zones indicates that the degree of aquifer heterogeneity is a very important factor controlling the recoverability of recharged water (Maliva and Missimer 2010). In particular, well-developed dual-porosity conditions were found to be highly adverse for recovery efficiency. Advanced modeling software, such as Eclipse®, allows for the simulation of density-dependent dual-porosity conditions. Theoretical model results demonstrated how dual-porosity conditions in ASR systems can result in high levels of dispersive mixing and buoyancy-induced movement of stored water, which would result in relatively low recovery efficiencies (Guo et al. 2013).

Geochemical processes may be both beneficial and problematic for MAR systems. Water quality improvement is a specific objective of some MAR systems such as soil aquifer treatment (SAT) and aquifer storage transport and recovery (ASTR) systems. Geochemical processes, such as the leaching of arsenic and metals, has adversely impacted the quality of recharged and stored waters. It is clearly advantageous if the geochemistry of MAR systems is accurately simulated in advance of system construction and operation. Reactive solute transport codes are available that combine a solute-transport code with a geochemical modeling code. Two codes that have been used for reactive multicomponent modeling of ASR and MAR systems are PHAST (Parkhurst and Kipp 2002) and PHT3D (Prommer et al. 2003). The PHT3D code (Prommer et al. 2003) is a general purpose reactive 3D multi-component model that combines the widely established MODFLOW/MT3DMS codes with the batch-type geochemical model PHREEQC-2 to compute a wide range of biogeochemical reactions.

The ultimate modeling goal for MAR systems is the development of a reactive multi-component transport model that accurately predicts water quality changes, but such a goal has been elusive (Maliva and Missimer 2010). The fundamental challenge is the non-uniqueness of model solutions. Generally, the benefits of multi-component reactive transport models have been in their capacity to constrain or reject existing or new hypotheses on interactions of physical, chemical, and biological processes and, to a lesser extent, in their predictive capabilities (Saaltink et al. 2003; Greskowiak et al. 2005, 2006; Prommer and Stuyfzand 2005, 2006; Dillon et al. 2006).

Groundwater modeling for MAR projects is usually performed during the design and permitting phase of projects and is often subsequently never revisited. Only rarely are post-audits performed on the accuracy of model predictions (Bredehoeft 2005). A groundwater model based on sound aquifer characterization and calibrated against initial operational data can serve as an invaluable tool for system expansion design and the optimization of performance. Various design and operational scenarios can be simulated to determine, for example, the configuration of a wellfield for an ASR or salinity barrier system that most effectively achieves the established system goals. Similarly, different options for recharge and recovery volumes and rates can be simulated to determine the operational protocols that maximize recovery efficiency or recharge rates.

Conclusions

A wide variety of technologies are now available that allow for improved aquifer characterization and simulation of the performance of MAR systems. Improved aquifer characterization may allow for the optimization of wellfield design and operational management. This may subsequently result in a reduced number of wells to be drilled, minimized total screen length, greater recovered volume, better recovered water quality, and hence, significant cost savings for the system owner and operator. As with the case for any technology, it is important that the technology be applied with the goal of obtaining specific data of real value for a project. It is also important to recognize that projects have finite budgets, which constrain the amount of hydrogeological investigative work that can be performed. Project managers thus face the challenge of determining which package of tools will provide the most useful data for a given amount of available financial resources. MAR includes a variety of proven water management technologies. The challenge moving forward is the target application of technology to improve the performance of MAR systems and thus increase their water management benefits.