Prospects for Managed Underground Storage of Recoverable Water (2008)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

3 Hydrogeological Considerations Development of an aquifer conceptual model through appropriate charac- terization of the physical underground storage system is a critical step in the development of a sustainable managed underground storage (MUS) system. In addition, analytical and/or numerical models can also be developed to evaluate water flow and solute transport in the aquifer and assess its potential as an MUS reservoir. To design a storage reservoir, engineers and hydrogeologists must have a good understanding of the hydrological properties of the aquifers to be used for storage and of the associated hydraulics. In particular, a successful MUS system design is predicated on answers to the following questions about the aquifer physical system and its hydraulics (including factors affecting suc- cess as listed by ASCE, 2001; Bouwer, 2002): • What are the spatial constraints of the aquifer (basin extent, basin depth, aquifer thickness, interlenses, other boundary conditions)? • What geological units are available for storage, and what are the hy- draulic properties of these units (hydraulic conductivity, porosity, stor- age coefficient) (e.g., confined or unconfined aquifer, specific yield or storativity, hydraulic conductivities/transmissivities and hydraulic gra- dients, degree of homogeneity and isotropy, hydrocompaction, interaquifer hydraulic connection)? • What temporal variations will affect the system (seasonal, climatic)? • What are the short- and long-term impacts of the MUS system on the aquifer matrix, groundwater flow, or surface waters? Additional decisions about the MUS system that significantly influence, or are influenced by, hydraulic characterization or aquifer attributes include the following: • Will the water be recharged through spreading basins, wells, or other methods? • Will the stored water be recovered by neighboring production wells (single function), recharge wells (i.e., aquifer storage and recovery [ASR] wells), or through gains in stream baseflow? • How much of the stored water is intended to be recovered? Successful design also requires identification of the source of water to be recharged and the anticipated uses of recovered water, which are discussed in other chapters. Hydrochemical and biological processes critical to MUS system 47 

48 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER success are described in Chapter 4. Factors that can preclude MUS development include low available aquifer storage; low hydraulic conductivity; high probability of clogging during re- charge; anticipated loss of recharge water; anticipated degradation of water qual- ity due to physical, chemical, or biological processes, and anticipated changes in patterns of potentiometric gradients that would adversely affect existing water supplies. The significance of these factors must be considered on a case-by-case ba- sis. Depending on the operational goals of the MUS system, some of these negative factors may be acceptable provided regulatory requirements are met. Addressed briefly in Chapter 6 and not covered here are operational issues that affect MUS viability. This chapter reviews the status of knowledge on the hydrogeology of re- charge, storage and recovery processes as they relate to MUS. The chapter in- cludes discussion of the hydrological properties of the geological formation to be used for storage, the aquifer boundary conditions, recharge and recovery methods to be used, and potential impacts of the MUS system on the groundwa- ter flow and aquifer integrity. In addition, knowledge gaps and research needs related to the hydrogeology of MUS systems are identified. AQUIFER TYPES AND CHARACTERISTICS IN THE CONTEXT OF MUS SYSTEMS A requirement for the success of an MUS system is a comprehensive under- standing of the hydrogeological properties of the aquifer to be used for storage. An aquifer is a layer, formation, or group of formations of permeable rock or sediment saturated with water and with a degree of permeability that allows wa- ter to be withdrawn or injected (Fetter, 2001; Marsily, 1986; Lohman et al., 1972). Sand and gravel layers, sandstone, and carbonate rocks usually form aquifers. This section describes hydraulic and hydrogeologic properties of aqui- fers, including flow and storage characteristics, and discusses aquifer classifica- tion with emphasis on considerations that are important to MUS. Aquifer Classifications Aquifer classification is generally based on composition, degree of con- finement, and geometry at local and regional scales. Each of these is described below. 

HYDROGEOLOGICAL CONSIDERATIONS 49 Lithology (Composition) There are 66 principal aquifers—that is, regionally extensive aquifers or aq- uifer systems that have the potential to be used as a source of potable water—in the United States (Maupin and Barber, 2005). Each principal aquifer is classi- fied into one of five lithologic types: unconsolidated and semiconsolidated sand and gravel aquifers; sandstone aquifers; interbedded sandstone and carbonate rock aquifers; carbonate rock aquifers; and igneous and metamorphic-rock aqui- fers. The total withdrawals of fresh water from these aquifers were estimated at 93.3 million acre-feet (83,300 million gallon per day [Mgal/d]) for the year 2000 (Maupin and Barber, 2005). About 92 percent of the total fresh groundwater withdrawals were used for irrigation, public supply, and self-supplied industrial applications. Withdrawals from the unconsolidated and semiconsolidated sand and gravel aquifers, including the High Plains aquifer, Central Valley aquifer system, Mississippi River Valley alluvial aquifer, and Basin and Range basin-fill aquifers, accounted for 80 percent (or 62,400 Mgal/d) of total fresh groundwater withdrawal for the above listed uses. In 2000, carbonate rock aquifers, primarily from the Floridian aquifer system, igneous and metamorphic rock aquifers (pri- marily the Snake-River Plain aquifer), and sandstone aquifers (primarily from the Cambrian-Ordovician aquifer system) provided 8 percent, 6 percent, and 2percent of total fresh groundwater withdrawal, respectively, from all aquifers in 2000. In the western United States, MUS activities have been conducted primarily within unconsolidated alluvial fan, floodplain, coastal plain, and inland valley deposits. However, in other regions, consolidated aquifers are also used for MUS, such as carbonate aquifers in Florida and fractured igneous-metamorphic rocks in the northwestern United States. All types of aquifers have been used for ASR, but in general ASR is easier to manage in consolidated aquifers where the formation provides a competent well without the requirement for screen and gravel pack (Dillon and Molloy, 2006). Carbonate aquifers show offsetting effects of carbonate dissolution on well clogging (Herczeg et al., 2004), but as discussed later in the chapter may have problems with mixing of injected and native waters. Fractured rock aqui- fers, even low-yielding ones, have been used successfully for ASR (Murray and Tredoux, 2002) with injection rates in some wells exceeding airlift yields. Coarse-grained sand and gravel are also very suitable for ASR storage targets, but care needs to be taken with well construction and completion, to reduce as much as possible the concentrations of organic and colloidal material introduced into the well. Storage in fine-grained unconsolidated media is more problematic and requires water with very low nutrient and colloidal concentrations in order to avoid chronic and irrecoverable depletion of the specific capacity of the ASR well. Table 3-1 summarizes properties of major types of aquifers. The shape and extent of these aquifer types is governed by the geological history of the region, including the depositional environment and subsequent deformation (if any). 

50 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER TABLE 3-1 Properties of Major Types of Aquifers Matrix Composition Confinement Porosity Type Carbonate C, S, U Dual porosity— intergranular & joints, fractures, solution conduits Unconsolidated and consolidated C, S, U Intergranular siliciclastic sediments Fractured or jointed igneous, metamor- C, S, U Joints, fractures phic Fractured sedimentary rocks C, S, U Dual porosity— intergranular and frac- ture NOTES: Confined (C), semiconfined (S), and unconfined (U) including water table and may or may not be perched. Degree of Confinement There are three aquifer conditions with respect to confinement: unconfined, semiconfined, and confined. Aquifer confinement affects or limits methods of recharge, storage, and recovery. Therefore, MUS system performance varies for these different aquifer conditions. Importantly, confined and semiconfined aqui- fers can be recharged only by wells. Unconfined aquifers can generally be re- charged by either wells or by surface spreading methods. .Unconfined aquifers allow flow of water from the land surface into the aq- uifer (i.e., recharge). Therefore, unconfined aquifers are naturally unprotected from contamination due to a lack of intervening low-hydraulic-conductivity units, known as confining layers between the land surface and the aquifer. Un- confined aquifers are also referred to as water table aquifers because the upper surface of the saturated zone is at equilibrium with the atmospheric pressure. This surface is called the water table, which often follows the land surface to- pography with variations due to recharge and boundary conditions. As a result, the water table may reflect hills, valleys, and plains. Localized recharge may also cause mounding. In very highly permeable aquifers the water table is more controlled by the presence of boundary conditions, such as lakes and rivers. In general, unconfined aquifers receive more recharge in upland areas where precipitation infiltrates into the ground, as well as near water bodies where seepage occurs. Discharge from an unconfined aquifer to the ground sur- face in low-lying areas usually occurs at springs or the bottom of surface waters (Fitts, 2002). Therefore, groundwater in unconfined aquifers interacts with sur- face water via several points or areas of connection, (e.g. rivers, lakes, wetlands, springs, and along coastal zones). By observing the hydraulic gradient, one can determine if a water body is “gaining” or “losing.” For example, a gaining 

HYDROGEOLOGICAL CONSIDERATIONS 51 stream is recharged by the aquifer, whereas a losing stream discharges to the aquifer. Unlike unconfined aquifers, confined aquifers are recognized by being iso- lated by a saturated or partially saturated low-hydraulic-conductivity, or “confin- ing,” layer on top of the aquifer. Rock or clay can form low-permeability barri- ers that impede or constrain the flow of water into and out of the aquifer. These confining layers allow pressure to build up in the aquifer system. An artesian well results when the pressure in a confined aquifer is sufficiently high that the groundwater in a well rises above the land surface. The water elevation in a well open to a particular point in a confined aquifer is known as the piezometric head at that point, which is the sum of the pressure head and the elevation head (Bear, 1988). The two-dimensional surface that is defined by mapping the head across the extent of a confined aquifer is the potentiometric surface or pressure surface. Natural recharge zones where a confined aquifer becomes unconfined are important aquifer characteristics. In confined aquifers, these areas are created when the geological confining layers are absent, exposing the aquifer to infiltra- tion. If a well is drilled in a confined aquifer, the water in this well will rise to the elevation of the recharge area. Last, semiconfined or leaky aquifers are saturated aquifers underlying a low-permeability layer, or aquitard. The low permeability of the confining unit allows for limited recharge into and discharge out of this aquifer. The degree of confinement can vary with natural variability of the confining unit: composition (i.e., clay content), pinchouts, or localized discontinuities (i.e., breaches due to sinkholes or fractures). Geometry and Scale Conceptual knowledge of aquifer geometry at both regional and local scales is required in order to identify boundary conditions, which are important con- straints on an MUS application. Aquifers within the hydrogeologic framework of a given region occur either closed or open basins. An aquifer at the margin between the land and the ocean exemplifies an open basin condition. Open basins that reflect a broad shallow paleocoastal margin depositional environment for sediment deposition may contain sheet-like strata comprising the storage zones; hence, the lateral boundary conditions can often be considered infinite. On the other hand, vertical boundary conditions exert an important control on the behavior of the system in this hydrogeological setting, especially with regard to ASR. If the anticipated storage formation is located in a closed basin, almost all of the recharged water can be retained within the basin except water lost through evapotranspiration in discharge areas. Most alluvial aquifers in the southwest United States, for example, are located in closed basins. These aquifers are sur- rounded by bedrocks and receive limited recharge from the mountain fronts or 

52 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER captured flow from the surface water system. Under natural conditions, water table slopes and groundwater movement will tend to conform to the surface to- pography. In many inland basins, this results in drainage from the basin at its lower end. Under such conditions, depths to groundwater will tend to decrease toward the downstream portions of these basins, particularly if there are geo- logic constrictions to reduce the rate of movement. If the water table intercepts the surface, discharge will occur either directly to surface water or as evapotran- spiration via phreatophytes. This results in a loss of water from the basin. Should groundwater levels in these areas be drawn down as a result of artificial extraction, there will be a saving in the water that would otherwise be consump- tively used by the phreatophytes. The value of water supply gained will need to be compared to the environmental values of the phreatophytes lost. With artifi- cial recharge, water levels will typically rise, which can lead to increased dis- charge. As a result, the recoverable water may diminish as the length of storage time increases. The storage zone geometry is also affected by local scale features and local variability (heterogeneity) in the hydrophysical properties of the aquifer. In sedimentary aquifers, the paleoenvironment in which the sediments were depos- ited affects the geometry of the storage zone. For example, if the storage zone is located with a paleofluvial (riverine) system, the geometry of the more perme- able zones may be ribbon-like (Prothero and Schwab, 2004). In a mixed clastic- carbonate aquifer, storage zones may be more isolated both vertically and later- ally than they are in a more homogeneous sandy alluvial aquifer. Hydrogeological Properties The hydrogeological aquifer properties that are most significant with re- spect to underground storage are the hydraulic conductivity (or transmissivity for a confined aquifer) and storage coefficient (either specific yield or storativ- ity) (see text below and Glossary for definitions). Leakage from adjacent water- bearing zones (quantified through the leakance) also affects an underground storage reservoir. The geological processes that create the aquifer control the hydrogeologic properties that the aquifer possesses. For example, in aquifers comprising sedimentary rocks, the environment of deposition, depositional processes, and lithology (types of grains) affect hydraulic conductivity and stor- age properties through the spatial arrangements of and variations in the grain size and sorting, packing, roundness, and so on. Postdepositional processes such as compaction and cementation can reduce hydraulic conductivity while dissolu- tion and fracturing tend to increase hydraulic conductivity. 

HYDROGEOLOGICAL CONSIDERATIONS 53 Storage The capacity of an aquifer to store water is described or quantified by the storage coefficient; specific storage and specific yield are the terms used for con- fined and unconfined aquifers, respectively. The aquifer properties that affect the specific storage are the total porosity and compressibility of the aquifer matrix. Specific storage ranges from less than 3 ×10-6 m-1 in rocks to 2 ×10-2 m-1 in plas- tic clays (Anderson and Woessner, 1992). Storativity, which is equal to the product of specific storage and aquifer thickness, defines the volume of water released from storage per unit decline in hydraulic head in the aquifer per unit surface area of the aquifer (Table 3-2). The relationship between fluid pressure, effective stress, and flow is essen- tial to understanding the mechanism of aquifer storage (Charbonneau, 2000; Fitts, 2002). Storage capacity is modified by compression or expansion in the soil or rock matrix as a response to effective stress. Effective stress is defined as the difference between the total stress and the stress supported by the fluid. The total stress is the weight supported by the surface divided by the surface area (Charbonneau, 2000). In other words, when pressures are lowered by removal of water during pumping, stress is transferred to the solid matrix and the solid matrix compacts as a result of the increased effective stress. When pumping ceases, water flows toward the area of reduced head, causing an increase in fluid pressure and a transfer of stress to the fluid phase. The reduced effective stress on the solid matrix causes an expansion of the matrix. The specific yield quantifies the pore space that is drainable by gravity. In other words, it expresses the difference between the total water filled porosity and the water held by surface tension (i.e., undrainable water). Values of specific yield range from close to 0 for clays to more than 0.25 for coarse gravel (see Table 3-2). There are two types of storage space used most commonly for MUS. One is the drained pore space within a geological unit; this space may have been cre- ated by historical groundwater withdrawal (i.e., groundwater overdraft or min- ing). In general, the available storage spaces in such depleted aquifers are later- ally extensive and may have experienced a reduction in storage capacity as a consequence of consolidation or compaction of the aquifer matrix during his- toric pumping. The second type of storage space is created by displacement of native water with recharge water creating a zone of freshwater around the recharge well (Fig- ure 3-1). In other words, injecting freshwater into a confined aquifer will create an increase in the piezometric head commonly known as the “mounding effect” (e.g., Bouwer, 2002). An example of this type of storage would be an ASR well in a saline or brine aquifer. This type of storage space may be limited by avail- able recharge area and/or by allowable pressures in the aquifer. Porosity in an aquifer system changes throughout the geologic history of the media. The primary porosity, comprising, primarily intergranular space, is cre- ated during deposition in sedimentary rocks. It can be reduced by subsequent 

54 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER compaction and lithification. Secondary porosity is created through marked al- teration of the original aquifer media. Examples include conduits formed by carbonate dissolution, partings along bedding planes, or fractures. The term “dual porosity” characterizes an aquifer that contains both primary and secon- dary porosity. Hydraulic Conductivity and Transmissivity Hydraulic conductivity describes the ability of the aquifer or any unit or volume within it to allow water flow. Hydraulic conductivity is dependent on the fluid (viscosity and density) and the geological medium (Viessman & Lewis, 2003). The dimensions of the connected water- filled pore spaces are the physi- cal attributes of the medium that control the hydraulic conductivity. Hydraulic conductivity values can range over 12 orders of magnitude (Domenico & Schwartz, 1990). Low–hydraulic-conductivity values are indicative of a less permeable matrix such as clay or shale (confining units), while high values are indicative of a highly permeable matrix such as sand and gravel (Schwartz & Zhang, 2003). Transmissivity is equal to the product of the hydraulic conductiv- ity and the aquifer thickness and is most often used in the context of confined aquifers. It thus quantifies the capability of the entire thickness of the aquifer to conduct water flow. Water also moves from one aquifer to another through a semiconfined or confined layer. Leakance, which is defined as the ratio of verti- cal hydraulic conductivity to the thickness of the confining unit or aquitard, was generally used to denote how fast or slow the confining unit may allow water pass through it. Table 3-2 summarizes ranges of these hydrogeological parame- ters, as well as storage parameters, from known MUS projects within common aquifer storage media. The hydraulic conductivity of an aquifer can vary with location in the aqui- fer—termed heterogeneity—and/or with the direction of groundwater flow— termed anisotropy. The Heterogeneity and anisotropy of aquifer hydraulic prop- erties must be known in order to plan an MUS system and develop accurate groundwater flow or solute transport models for such a system. The aquifer created in a fluvial sedimentary deposit provides an example of one that has heterogeneous hydraulic conductivity with lower conductivity in the finer- grained overbank or floodplain-generated units and higher values in the channel features. Heterogeneity in the hydraulic conductivity of aquifer storage units is the norm, rather than the exception. As discussed later in the chapter, heteroge- neity often leads to a highly nonuniform distribution of water recharged by wells (Vacher et al., 2006)—not the subsurface ”bubble” of stored water employed in simpler conceptual models. Whereas heterogeneity indicates that hydraulic conductivity differs between points in an aquifer, anisotropy is the term that characterizes differences in hy- draulic conductivity with direction of flow. Anisotropy can result in observations 

HYDROGEOLOGICAL CONSIDERATIONS 55 TABLE 3-2 Approximate Hydrogeological Parameters in Aquifers Used for Underground Storage Hydraulic Specific Matrix conductivity Transmissivity Specific Capacity1 Leakance Composition (ft/day) (ft2/day) Yield Storativity (ft3/day/ft) (per day) Carbonate 10-1 to 103 102 to 105 0.01 to 0.1 10-3 to 10-5 103 to 105 10-2 to 10-5 Unconsolidated 10-1 to 102 102 to 104 0.1 to 0.3 10-3 to 10-6 103 10-3 to 10-5 and consolidated siliciclastic sediments Fractured 100 to 10-4 102 0.05 to 10-2 to 10-5 103 to 105 - igneous, 0.1 metamorphic, and sedimentary rocks SOURCES: Brown et al. (2005); Driscoll (1995); Leonard (1992); Pyne (2005); Reese (2003); Reese and Alvarez-Zarikian (2007); and Ward et al. (2003). 1 An expression of the productivity of a well. It is defined as the ratio of discharge of water from the well to the drawdown of the water level in the well. It should be described on the basis of the number of hours of pumping prior to the time the drawdown measurement is made. of order-of-magnitude differences in the vertical and horizontal hydraulic con- ductivities in a single core sample of aquifer material. Anisotropy contrasts are generally greater when vertical and horizontal flow directions are compared. Within a layered sedimentary system, for example, flow in the vertical direction is impeded by the presence of any low-hydraulic-conductivity layers, whereas flow in the horizontal direction may travel in laterally continuous, more perme- able zones unimpeded by the low-hydraulic-conductivity layers. A massive (i.e., unbedded), very well sorted quartz sand or carbonate grainstone aquifer (i.e., nearly free of a clay-sized fraction) would be characterized as homogeneous and isotropic. On the other hand, a mixed siliciclastic-carbonate aquifer typical of the southeastern U.S. Coastal Plain would be considered heterogeneous and ani- sotropic. In the context of aquifer storage, a dual porosity aquifer system can be con- sidered a dual reservoir. While most of the water may exist within connected primary pore spaces through which water moves relatively slowly, water resid- ing in the secondary porosity may travel at greater velocities (e.g., conduit flow in a carbonate aquifer). A prominent example of a dual- porosity unit that is frequently considered for MUS systems is the “Chalk” of England, which has up to 40 percent primary porosity, yet most of the flow is through fractures (Gale et al., 2002). The scale of measurement strongly influences the resulting observations in dual-porosity aquifers. Because only the permeability of the matrix or primary porosity is captured in laboratory sample-sized measurements, much greater hydraulic conductivities are observed at the well-field scale where the volume of 

56 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER aquifer measured includes flow through the more permeable secondary porosity features. Fluid flow within secondary porosity can be non-Darcian including turbulent flow (high Reynolds number), and velocities may range from 102 to 103 feet per day, where these gravel seams or fractures are not continuous over large distances. Hydraulic conductivity values generally range from 10-3 to 101 feet per day in the less permeable (primary) counterpart of the dual-porosity system (Brown et al., 2005; Driscoll, 1995). Open basins and coastal plain aqui- fers that are comprised dominantly of dual-porosity carbonates are especially susceptible to issues of scale with regard to hydrogeological parameters. Igneous and metamorphic rocks are generally not considered to have dual porosity because fracture porosity comprises nearly all of the open volume in which water can flow or be stored. Primary porosity in these comparatively brittle rocks is extremely low and rarely interconnected, unless the rocks have been significantly weathered. In a basaltic aquifer, zones of greatest hydraulic conductivity occur along lava flow boundaries; lava tubes comprise a unique type of secondary porosity. Both groundwater modeling and effective monitoring design are facilitated by understanding the physical characteristics of the secondary porosity such as the size, orientation, and distribution of fractures or partings. The orientation of fractures and joints is generally related to present or paleo-stress fields; widen- ing of these features may occur due to rock dissolution and mechanical break- down. Conduit size is more dependent on the aquifer lithology (e.g., carbonate rocks dissolve more readily than silicic rocks) and history of exposure to chemi- cally aggressive water. Additional influences on the distribution of secondary porosity in carbonate rocks include changes in the position of the freshwater-seawater interface, sea- level fluctuations, climate change, and extensive pumping. Variations in lithol- ogy, depositional environment, and position of bedding planes also contribute to evolution of conduits that may yield complex flow systems. Water Movement Between Aquifers or Between Aquifers and Surface Water Aquifer Interaction In an aquifer system, it is possible for water to move from a semiconfined aquifer of higher hydraulic pressure into an unconfined one or vice versa when the semiconfined aquifer hydraulic head is reduced by pumping. Water move- ment may also occur through windows or lenses between confined aquifers due to potentiometric head differences. Adding water to a confined aquifer can be accomplished only by increasing the pressure of water in already saturated pores (contrasted with the ability to add water to partially saturated pores above the water table in an unconfined aquifer) Interaction among aquifers at different physical elevations depends on the piezometric head between them and on the 

HYDROGEOLOGICAL CONSIDERATIONS 57 thickness, hydraulic conductivity, and integrity of the confining unit. Water from different aquifers may also be transferred through uncased wells or aban- doned wells. Leakage between unconfined aquifers and semiconfined aquifers can be enhanced by increased head difference or reduced by decreased head difference as a result of recharge of one aquifer. Surface Water and Groundwater Interaction Groundwater commonly is connected hydraulically to surface water (Alley et al., 1999). In the natural system, the interaction takes place in three basic ways: a water body gains water from inflow of groundwater through its bed, through its margins, or via a spring or seep; loses water to groundwater by out- flow in the same manner (seepage or sinkholes); or does both, gaining in some places and losing in others depending on local and temporal changes in hydrau- lics (seasonal or climatic changes affecting relative pressures). Groundwater- surface water interactions occur between aquifers and rivers, lakes, wetlands, retention ponds, infiltration trenches, and spreader canals. If the vertical gradient or the hydraulic conductivity is low, the flow rate between the water body and the aquifer is lower. Wells located closer to water bodies may have strong im- pacts on surface water flow, whereas distant wells tend to have lesser impacts. Pumpage of wells in close proximity to water bodies may greatly increase seep- age, especially from coarse-grained stream channels or unlined canals and later- als. These types of interactions are relevant to MUS projects because surface water bodies serve as boundaries that recharge or drain the aquifer. For example, water reuse projects could be implemented in coastal aquifers, where water de- livered to canal systems that recharge the aquifer prevents saltwater intrusion from wellfield drawdown. HYDRAULICS OF RECHARGE As noted in the previous chapter, managed underground storage of recover- able water can be achieved using three different methods, namely surface spreading (e.g., recharge basins, modified stream beds, pits and shafts), vadose zone wells, and recharge or ASR wells, plus others including watershed man- agement (water harvesting or enhancement of natural recharge). Each method is governed by its own hydraulics (ASCE, 2001; Bouwer, 2002; Pyne, 2005). 

58 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER Surface Spreading This method consists of releasing water from the source to a recharge basin, pond, pit, or channel for infiltration. This method of aquifer recharge can be used only for unconfined aquifers (see Box 3-1). It also requires large (land) surface areas to accommodate the recharge scheme that can also allow signifi- cant evaporation if infiltration is slow. Surface spreading usually requires both a diversion structure and an infiltration structure (ASCE, 2001). Evaluation of infiltration capacity is critical for MUS because it dictates the method and size of the recharge site. The factors that affect infiltration capacity of artificial recharge projects include the composition of surface soils, the geol- ogy, subsurface hydrologic conditions, source water quality, and procedures used in the construction, operation, and maintenance of the recharge structure. The operational factors can ordinarily be managed to maintain favorable infiltration capacity. Therefore, the most important attributes to characterize the suitability of a recharge location for MUS are the soils, geology, and hydrogeol- ogy of the recharge location. Of particular importance are geologic structures or low-hydraulic-conductivity units that might form a barrier to groundwater movement and the position and hydraulic gradient of the existing water table or potentiometric surface. Under certain geologic and hydrologic conditions, the groundwater mound developed as a result of spreading intersects the land surface. This can occur (1) when subsurface lenses with sufficiently low permeability exist that restrict the downward movement of the recharged water, creating localized mounds, and (2) when the water table is sufficiently close to the surface to cause a similar effect. In both cases, the infiltration capacity is essentially limited to the quantity of lateral flow from the mound, although under the first of these conditions there is probably a small amount of movement through the less permeable lens. The lateral movement of water away from the mound is generally found to be in sub- stantial conformity to Darcy's law. Considerable literature exists presenting methods to estimate the shape and the rates of buildup and recession of ground- water mounds beneath recharge areas (Bouwer et al., 1999). Artificial recharge by injection consists of using a conduit access, such as a tube well, shaft, or connector well, to convey the water to the aquifer (Figures 2- 2 and 3-1). It is the only method to artificially recharge confined aquifers or aq- uifers with low-hydraulic-conductivity overburden. The water is recharged di- rectly into the storage zone, and there are no transit or evaporation losses. This method can be particularly effective in highly fractured hard rocks and karstic limestone, but it is also used in unconsolidated or alluvial sediments. 

HYDROGEOLOGICAL CONSIDERATIONS 59 BOX 3-1 Case Study: Capturing Water from the Santa Ana (California) River The Orange County Water District (OCWD) in coastal Southern California is responsi- ble for managing the underground water reserves that supply approximately 270,000 acre- feet per year from about 500 wells within OCWD’s boundary. That quantity grows steadily, and projections indicate the demand may reach 450,000 acre-feet a year in the next quar- ter century (OCWD, 2006; http://www.ocwd.com/_html/recharge.htm). Groundwater re- serves are maintained by a recharge system, which replaces water that is pumped from wells. OCWD facilities have a recharge capacity of approximately 300,000 acre-feet per year. About 2 million people depend on this source for about three-quarters of their water. Groundwater producers pump water from the groundwater basin and deliver it by pipeline to consumers. Along a 6-mile long section of the Santa Ana River that belongs to OCWD, a system of diversion structures and recharge basins captures most of the river water that would otherwise flow into the Pacific Ocean. The district has 1,500 acres of land for use in its recharge program. The current average annual baseflow of the Santa Ana River is ap- proximately 140,000 acre- feet. Storm flows add an average of 60,000 acre-feet per year, ranging from 10,000 to 500,000 acre-feet. The baseflow may increase by 100,000 acre feet over the next 20 years due to urban development in upstream areas. Increased urbaniza- tion creates more buildings and paved areas, which re- sults in greater quantities of storm runoff. Population growth also causes a propor- tional increase in wastewater discharges to the river chan- nel. Water flows down the Santa Ana River from River- side and San Bernardino Counties, together with Anaheim Lake, one of OCWD's recharge basins. Photo supplies imported from the courtesy of Orange County Water District. Colorado River and from the California State Water Project. In the Cities of Anaheim and Orange in Orange County, a pattern of interlaced levees built of sand helps to slow the river’s flow to maximize the amount of water that can percolate through the bottom of the river channel. Water is also diverted from the river into a series of recharge basins. These basins, with depths ranging from 50 to 150 feet, were formed in years past by sand and gravel mining operations. The soil along this stretch of the Santa Ana River is coarse-grained and sandy. Therefore, water readily seeps into sand and gravel layers below the ground surface. Groundwater is stored in the underground sand and gravel aquifers. Certain aquifers reach the surface in this recharge area of Orange County and can easily be recharged, while in other areas closer to the coast, a layer of dense clay overlies the aquifer and prevents efficient percolation of significant quantities of surface water. continued next page 

60 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER BOX 3-1 Continued The district’s deep recharge basins, such as Anaheim Lake, Warner Basin, and Kraemer Basin, gradually accumulate a thin layer of fine sediments and biological material that slows and can even stop percolation. Although the percolation rate in a newly cleaned deep basin can reach 10 feet per day, the rate can drop to nearly zero after six to eight months. Each of these deep basins is periodically emptied by means of submersible pumps, and the clogging layer is removed by scrapers or by a sand-washing device. Clog- ging affects only the upper 2 to 3 inches of soil. A twice-yearly cleaning cycle, which has replaced a single annual cleaning, increases percolation by as much as 40 percent. Prevention plays an important role in solving the problem of clogging. A flocculation system at the Imperial headgates is being considered that could coagulate suspended solid particles so that they will settle out of the water as it passes through a series of desilting ponds. Three of these ponds help reduce the sediment load in water that is diverted to the recharge basins. This slows the formation of a clogging layer and thereby helps to maintain efficient percolation. More than 100 species of wildlife are found on district lands, and OCWD cooperates with environmental organizations to preserve the natural habitat of these animals. Recrea- tional opportunities include river trails for horseback riding, bicycling, and jogging; two re- charge basins are also stocked for sport fishing. Injection Vadose Zone Wells Vadose zone wells are boreholes (usually 10 to 50 m deep and about 1 to 1.5 m in diameter) in the unsaturated zone completed with a center pipe and the annual space between the pipe and the wall of the borehole filled with sand (ASCE, 2001). They are often used to dispose of storm runoff and to reduce flooding (also called drainage or recharge wells) most commonly in areas of relatively low rainfall.. A negative aspect of vadose zone wells is the introduc- tion of contaminants that comes from recharge of untreated urban runoff (petro- leum byproducts, metals, nutrients, pesticides, surface microbes). Pretreatment strategies are needed, including first-flush bypass, screens, filters, and disinfec- tion systems. Such systems require assessment and monitoring of contaminant fate and transport during wet and dry periods. An important limitation of vadose wells is that there are no effective and reliable methods to reverse clogging. Recharge Wells Recharge of water into abandoned wells and wells specifically designed for artificial recharge has been practiced for many years with varying degrees of success. The use of recharge wells is confined largely to those areas where sur- face spreading is not feasible owing to the presence of low-permeability layers overlying the principal water-bearing deposits. They may also be more eco- 

HYDROGEOLOGICAL CONSIDERATIONS 61 nomical in metropolitan areas where land values are too high to utilize the more common basin, flooding, and ditch and furrow methods (ASCE, 2001). Many attempts to recharge groundwater through wells in alluvial and sedi- mentary aquifers have yielded disappointing results. Difficulties encountered in maintaining adequate recharge rates have been attributed to silting, bacterial and algae growths, air entrainment, release of dissolved gases, rearrangement of soil particles, deflocculation caused by reaction of high-sodium water with soil par- ticles, and chemical reactions between recharged waters and native groundwa- ters resulting in precipitates in the aquifer or well-casing perforations (Bouwer, 2002). However, the Los Angeles County Flood Control District, in California, has successfully operated recharge wells for many years, creating and maintain- ing a freshwater ridge to halt seawater intrusion in the Manhattan-Redondo Beach area in Los Angeles County. Favorable recharge rates have been main- tained by chlorination and deaeration of the water supply and by conducting a comprehensive maintenance program on the wells. The spacing of the recharge wells depends on the range of influence of a well, which in turn depends on the rate of water recharge, and on aquifer and well hydraulic properties, including the aquifer hydraulic conductivity, hydraulic gradient, length and diameter of perforated casing or screen penetrating the aqui- fer, and the open area of casing perforations or screen. In general, it has been found that gravel-packed wells operate more efficiently and require less mainte- nance than non-gravel-packed wells in alluvial aquifers. In addition, where wa- ter is being injected under pressure, it has been found that a concrete seal should be provided on the outside of the casing at a point where it passes through the relatively impermeable confining bed, to prevent the upward movement of water outside the casing. For reasons covered in Chapters 4 and 6, long-term use of recharge wells in alluvial aquifers requires treatment of the injected water. Sediments must be removed completely. The clear water should be treated with chlorine, calcium hypochlorite, or copper sulfate to prevent the growth of bacterial slime and al- gae. The water must also be free of dissolved gases that may be released into the formation and cause air binding or air entrainment, which reduces permeability. In addition, care should be taken to ensure that in formations containing appre- ciable proportions of base-exchangeable clay, water containing a high percent- age of sodium will not be used, since this will cause deflocculation of the aqui- fer sediments and rapid decrease in transmissivity. These treatment require- ments may not be necessary in highly permeable limestone and volcanic aqui- fers. Gravity feed of stormwaters and treated wastewaters into the carbonates of the Floridan Aquifer System in Florida has been effective for many years with no evidence of significant plugging. 

62 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER ASR Wells An aquifer storage recovery well (Pyne, 2005) is designed in such a way that water can be injected through a single well and recovered through the same well at a later time. Recharge for ASR is similar to that for other recharge well systems except for the cycling of injection and pumping. With an ASR well or system, it is assumed that most of the recharged water will be pumped back be- cause a relatively constrained zone of recharge water (i.e., a so-called bubble) will be retained until the stored water is recovered. However, aquifer heteroge- neity and anisotropy, as well as density differences (if any) between the source water and native groundwater, tend to produce relatively amorphous shapes that describe the three-dimensional limits of the recharge water (e.g., bottle brush [Vacher et al., 2006], upside-down Christmas tree [Missimer et al., 2002]), or an “octopus” shape) that reflect preferential flowpaths in dual-porosity settings. In some of these cases, especially within a dual-porosity setting, injected water often cannot be fully recovered by the ASR well. Therefore, the term “bubble” in reference to the shape of the recharge water body could be misleading. Casing, screen design, and storage interval should be determined by hydro- logical properties of the aquifer. Hydrogeologic constraints that are not assessed at the time of design and/or that change over time, such as the collapse of unsta- ble geologic layers into the well borehole, may cause plugging and fouling. Naturally occurring and/or artificial fracturing, sinkholes, and karst terrain fea- tures may dictate where recharge water can flow and how much recharged water can be recovered. Poor well design and/or construction practices, including in- sufficient placement of grout; improper design of pumps, valves, and fittings; and excessive drawdown allowance can lead to low recharge rate, low storage capacity, and low recovery efficiency (see “Recovery Efficiency and Target Storage Volume” later in this chapter) (Bloetcher et al., 2005; Pyne, 2005). High recharge rates in wells can result in turbulent (also termed non-Darcian) flow around the well casing, which will impact well performance and water move- ment in the recharge zone. Incorrect injection pressures can also alter fracture networks by reopening existing fractures or generating new ones. In most cases, it is easier to get a steady pumping rate through a production well than a steady recharge rate. Therefore, well designers tend to use pumping rates instead of injection rates to design recharge wells. Such assumption tends to overestimate the capacity of recharge wells, resulting in an underestimate of overall costs of the system. It was found, for example, that the recharge rate of an ASR well was approximately 50 percent to 70 percent of the pumping rate in El Paso, Texas (Boyle Engineering, 1999). The ratio of recharge to recovery specific capacity for comparable flows and durations typically ranges from 25 to 100 percent, with 50 to 80 percent being a reasonable range for unconsolidated aquifers (Pyne, 2005). Despite the many possible complications, ASR has worked successfully in many locations. An example is given in Box 3-2. There are also cases, albeit many fewer, in which ASR has been attempted but abandoned (Box 3-3). 

HYDROGEOLOGICAL CONSIDERATIONS 63 BOX 3-2 Case Study: ASR System at Boynton Beach, Southeast Florida Among the 21 ASR systems in southern Florida (Brown, 2005), one of the most suc- cessful is the Boynton Beach East Water Treatment Plant located on the east coast in Palm Beach County. Compared to other ASR systems, this system has the highest recovery efficiency achieved per cycle (Reese, 2002) during its operation since 1992. Treated drink- ing water from the Wastewater Treatment Plant (WTP) is the source of recharged water. The ASR was constructed to recharge into the Hawthorn formation (i.e., sandy phosphate limestone), which is located within the upper Floridan Aquifer at a depth of 804 to 1,200 feet below land surface (Reese, 2002). The thickness of the storage zone’s open interval at the Boynton Beach site is 105 feet, and transmissivity is reported to be about 9,400 square feet per day (CH2M Hill, 1993). The zone of influence is at least 800 feet as reflected by observation from the monitoring well. The equilibrium pressure of the system is 10 pounds per square inch (psi) and there are no upward leaks. The maximum pressure range for recharge is 55-60 psi. Recharge occurs during the wet season: June through December. After recharge has been completed, the pressure of the aquifer system drops from 60 psi to the natural value of 10 psi. One full cycle is defined as one wet and one dry season. Approximately 100 Mgal are recovered per cycle, while the pump itself moves 2.5 Mgal/d. A total of 24 sepa- rate recharge and recovery cycles have been completed with recovery efficiencies (i.e., the percentage of the total amount of potable water recharged for each cycle that is recovered) varying from 40 percent to 100 percent (Reese, 2002). Recharge-recovery cycles had been conducted for an average of about two cycles per year. Recovery efficiency seems to be linked to the length of the storage periods (Reese, 2002). Water was recovered until the chloride concentration in the recovered water slightly exceeded 300 mg/L during the dry season. This is related to the fact that native groundwa- ter had a chloride concentration of 1,900 mg/L. The goal of the water treatment plant is to have a chloride range around 70-80 mg/L. Recovered water has 250 mg/L of CaCO3 due to the geology. Injected water has 40-50 µg/L of trihalomethanes, but when the water is extracted, these levels fall 3-5 µg/L, which is below the safe drinking water standard of 80 µg/L. Economic evaluation shows that the ASR alternative is considerably less expensive than other seasonal water supply options (Brown, 2005; CH2M Hill, 1993; Muniz and Ziegler, 1994). The main reason that the quality and amount of storage water are well maintained is the fact that the aquifer has a relatively low permeability zone located just underneath the confining unit. Aquifer Storage, Transfer, and Recovery Wells In aquifer storage, transfer, and recovery (ASTR), water is pumped from a different well than the alternate from the recharge well. This has been tested in Australia (Pavelic et al. 2004). In fact, it is very similar to the reclaimed water recharge systems in Orange County, California (Dillon et al., 2004) and El Paso, Texas (Sheng, 2005). In large projects, such as the Everglades Restoration, hun- dreds of what are nominally called ASR wells are planned. In fact, it seems unlikely that each well will only recover the water that it recharged. In such a case, ASTR would probably occur—whether planned or unplanned—and com- bined with ASR this might add to the overall efficiency of the system. In some 

64 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER BOX 3-3 Case Study: ASR System at Taylor Creek (Lake Okeechobee), South Florida The Taylor Creek ASR project site is located along the northeastern portion of Lake Okeechobee, in south Florida. It has generally been considered an unsuccessful ASR pro- ject due to its recovery efficiency (see “Recovery Efficiency and Target Storage Volume” later in this chapter) of only 15 to 36 percent on four tests in 1989 and 1991 (Reese and Alvarez-Zarikian, 2006). The site was a South Florida Water Management District demonstration project to test the feasibility of storing large volumes of phosphorus-rich stormwater from Taylor Creek underground to prevent it from reaching nutrient-enriched Lake Okeechobee (CH2M Hill, 1989). The project was conceived as a single test well with an on-site groundwater monitor- ing well. The proposed storage zone in the upper Floridan Aquifer System is highly trans- missive and porous fossiliferous limestone, and based upon test data and water quality sampling data, the ASR well was completed with an open-hole interval from 1,268 to 1,700 feet below land surface. The storage zone represented a confined leaky aquifer with a transmissivity of about 570,000 feet per day (CH2M Hill, 1989). The source water from Taylor Creek was highly variable in composition, with total dis- solved solids (TDS) ranging from 268 to 996 mg/L and total coliforms ranging from nonde- tectable to 7,500 per 100 mL. The ASR storage zone also had variable groundwater quality (TDS from 4,000 to 6,900 mg/L; CH2M Hill, 1989). The source water was treated prior to recharge. Brown (2005) did an extensive analysis of the site’s potential based on application of a newly proposed ASR planning decision framework. He concluded that the project seemed to be infeasible both technically and economically for each of three alternatives evaluated and recommended that the Everglades restoration program consider a more suitable pro- ject location if it is to do future testing of ASR in the area. cases, recharge water that moves beyond the capture zone of an ASR well could be captured by a single-purpose recovery well. Other Recharge Methods Enhancement of natural recharge or watershed management offers an effec- tive method to intercept dispersed runoff. Many water conservation techniques have been developed for hillslopes with the intention of preventing soil erosion and reducing surface runoff. These increase the infiltration and aquifer recharge. Traditional terraced agriculture is certainly one of the most common water har- vesting methods in arid areas, particularly in the Near East such as Jordan (http://www.ruralpoverty portal.org/english/learn/water/harvesting.htm). Where the terraces are well maintained, they effectively control runoff and improve aquifer recharge, but once allowed to fall into disuse, they progressively lead to gully erosion, collapse of the retaining walls, destruction of the whole system, and severe modification of the hydrological regime. Therefore, whatever the economic virtues of such terraces, it should be recognized that their abandon- ment on a large scale can upset the hydrological conditions within a basin for a considerable period of time. 

HYDROGEOLOGICAL CONSIDERATIONS 65 RECOVERY OF STORED WATER Fate of Recharged Water Will the recharged water remain in the aquifer until recovery occurs? In many cases, recharged water will migrate or mix with native groundwater due to hydrological and boundary conditions of the aquifer. If water is stored in shal- low aquifers, stored water may flow back into a neighboring stream as baseflow or flow downward to recharge deep aquifers as the hydraulic head increases. As water levels increase in the shallow aquifer, evapotranspiration from the shallow aquifer may also increase. This section describes how processes in the aquifer affect recovery. Differences exist between recharge basin systems and recharge wells with regard to these processes. In recharge basins (Figure 3-1(a)), the source water infiltrates into an unconfined storage zone through a recharge zone. Recharged water is depicted as a “mound” on native groundwater. The base of the lens is a mixture of recharge water and native groundwater, termed transitional water. The entire lens is the storage zone, whereas the vadose zone located above the water table is the recharge zone. The groundwater level will still fluctuate in response to changes in natural recharge, which is affected by changes in land use, seasons, and climates. Other factors include pumping of water wells and the influence of nearby recharge projects. In vadose zone wells (Figure 3-1(b)), the source water is injected into an unconfined storage zone through a recharge zone. As in Figure 3-1(a), recharged water is shown as a mound on native groundwater with transitional water be- tween the two. In ASR wells (Figures 3-1(c) and (d)), the storage and recharge zones over- lap. Recharge water has an irregular shape to reflect aquifer heterogeneity. Only in nonbuoyant, isotropic, homogeneous aquifer conditions would the term bubble appropriately describe the shape taken on by the recharge water. Proxi- mal to the recharge water is the transitional water, in what is referred to as a buffer or mixing zone. In any of the scenarios shown in Figure 3-1, the zones vary in size and ge- ometry as a function of hydraulic gradient, dispersivity, presence of dual poros- ity, and relative density of the native and recharge water. If significant chemical differences exist between recharge water and native groundwater, the degree to which this mixing occurs depends on the dispersivity of the aquifer. Mechanical dispersion is a scale-dependent process that pertains to fluid mixing due to flow through heterogeneous media. Diffusion reflects the movement of dissolved species from higher to lower areas of concentration and does not require flow. Dispersive mixing will tend to increase with time and distance from the recharge well due to both molecular diffusion and “mechanical dispersion,” which results from the heterogeneous nature of aquifers. In a dual-porosity storage zone, the role of diffusion is significant, whereas in a more homogeneous aquifer, the ef- fects of diffusion are masked by dispersion. 

66 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER SW SW RW RW TW TW NGW (a) Recharge Basin (b) Vadose zone well SW SW RW NGW TW RW TW NGW (c) Recharge/ASR well in a confined aquifer (d) Recharge/ASR well in an unconfined aquifer FIGURE 3-1 Mixing of waters with different recharge methods: source water (SW), either surface water, groundwater, or reclaimed wastewater; recharge water (RW); transitional water (TW, mixed recharge water and native groundwater); native groundwater (NGW). Recovery Efficiency and Target Storage Volume Recovery efficiency (RE) broadly reflects the proportion of recovered water in an ASR system and is an important concept in terms of MUS system perform- ance. Variable applications of the term underscore the need to clarify its mean- ing. For example, RE has been defined as a fraction or a percentage and has been applied to describe usable water from individual cycles as well as the total or cumulative performance of a system. By some definitions, recovery effi- ciency may exceed 100 percent; however, many prefer to calculate RE in a man- ner that reflects recovery of the actual water injected into the aquifer through an ASR well. In this context, RE would not exceed 100 percent. From an MUS system management perspective, RE needs to be defined in terms of individual cycle tests and overall system performance. Therefore, the terms “cumulative 

HYDROGEOLOGICAL CONSIDERATIONS 67 recovery efficiency” (CRE) and “operational recovery efficiency” (ORE) are adopted herein. Kimbler et al. (1975) define CRE as the ratio of the cumulative volume of fresh water injected minus the volume of unrecovered fresh water divided by the cumulative volume of fresh water injected. This definition yields a fraction and requires use of threshold value of a water quality parameter to determine the volume of recovered water that originated as artificial recharge. This limiting parameter should allow clear distinction between recharge and native water; examples include salinity, total dissolved solids, electrical conductivity, or a nonreactive tracer. The asterisk in Equation (3-1) reflects this limiting parame- ter. As an example, if salinity is used, the U.S. Environmental Protection Agency (EPA) drinking water standard for chloride (250 mg/L) is often the con- straint (Reese and Alvarez-Zarikian, 2007). Pavelic et al. (2002) quantify this concept in terms of a “recovered mass” fraction. Modifying the Kimbler et al. (1975) definition to a percentage is more widely used in the industry: cumulative volume of recharge water - cumulative volume of unrecovered recharge water CRE = 100 × cumulative volume of recharge water (3-1) As defined, the CRE reflects the overall recovery efficiency of the MUS system. For a given recharge and recovery cycle, ORE is applied in a similar form: volume of recharge water during cycle - cumulative volume of recharge water not recovered ORE = 100 × volume of recharge water (3-2) Recharge and recovery of fresh water into a freshwater aquifer requires careful selection of the water quality parameter used to distinguish between the recharge water and the native groundwater, given that CRE and ORE should not exceed 100 percent. Operational recovery efficiency (Equation 3-2) is consistent with Pyne’s (2005) definition of RE: “the percentage of the water volume stored in an operating cycle that is subsequently recovered in the same cycle while meeting a target water-quality criterion in recovered water.” Bear (1979) noted that a certain volume of recharge water—namely, that portion of the injected water body extending beyond the water divide for pump- ing—can never be recovered by the recharge well itself. However, this portion of recharge water can be partially recovered by pumping from downgradient neighboring wells (e.g., ASTR, Pavelic et al. 2004) or by simply discharging back into the surface water system. In addition, native brackish water, which otherwise may not be usable, can be recovered for beneficial uses after treatment or mixing with recharged water. 

68 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER In an ASR feasibility study, Merritt (1985) investigated the effect of hydro- geological parameters on RE. He noted that “recovery efficiency [ORE] im- proves considerably with successive cycles providing that each recovery phase ends when the chloride concentration of withdrawn water exceeds established criteria for potability (usually 250 milligrams per liter), and that freshwater in- jected into highly permeable or highly saline aquifers (such as the boulder zone) would buoy rapidly.” In this manner, recharge water mixed with native ground- water is left in the transition zone with successive cycles. Recovery efficiency varies with recharge method, hydrological and hydro- chemical properties of the aquifer, and recovery methods. Brown (2005) has done an extensive analysis of these factors. The dispersivity, thickness of the storage zone, preexisting groundwater gradient, recharge volume, rock type, presence of high-permeability zones, length of storage time, density of ambient groundwater relative to recharge water, ambient groundwater quality, and num- ber of recharge and recovery cycles can all be important. The roles of transmis- sivity and anisotropy are still unclear; porosity does not appear to be a major factor (Brown, 2005). In general, surface recharge compared to deep injection in the same aquifer may have a lower recovery efficiency due to evapotranspira- tion and other losses related to surface spreading. Maximized ORE is a long-term operational objective for any ASR system. To this end, Pyne (2005) recommends a “target storage volume” (TSV) ap- proach to meet a predetermined recovery volume goal. The TSV is defined as “the sum of the stored water volume and the buffer zone volume in an ASR well” (Pyne, 2005). In the context of the physical hydrogeologic setting (Figure 3-1), the TSV is the sum of recharge and transitional water volumes. Implemen- tation of the TSV concept involves one or more high-volume recharge phases intended to displace native groundwater early in the development of the system. This initial large-volume buildup is designed to develop a transition zone suffi- ciently far from the ASR well such that subsequent smaller cycle test volumes would minimally recover transition water. Ideally, successive cycles would yield increasing OREs approaching 100 percent to meet the targeted operational objective. The rate at which the TSV is developed depends on the water availability, cost of water, aquifer hydraulic parameters, and regulatory issues (Pyne, 2005). For example, the rapid development of the TSV will likely maximize operational recovery efficiency and can be timed to minimize the cost of water or maximize water availability (off-peak demand periods). Characterization of the hydrogeochemical system (Chapter 4) at a given ASR site prior to TSV development may facilitate progress along the path of regulatory authorization. For example, if water-rock-microbial interactions that may affect water quality are anticipated during recharge or storage based on experience from similar hydrogeochemical settings, the rapid development of a large storage and transition zone may inhibit the ability of the regulatory com- munity to assess water quality changes at the ASR monitor well(s). Bench-scale studies, geochemical modeling, or initial smaller-volume cycle tests may be 

HYDROGEOLOGICAL CONSIDERATIONS 69 warranted. Sufficient time between each cycle test would allow completion of water quality analyses and data interpretation. Results of such an assessment could then be used to modify the next cycle test design (duration, pumping rate, volume, etc.) to optimize data collection and improve characterization of the hydrogeochemical process with the goal of mitigating its effects. Ideally an adaptive characterization method can move forward in parallel with operational strategies that improve ORE, such as the TSV concept. A high hydraulic conductivity not only increases the potential for water to travel beyond the zone of recovery, but also promotes mixing with native groundwater. Conversely, low transmissivity may require extreme wellhead in- jection pressures during recharge and may cause excessive drawdown during recovery. Different aquifer types result in marked differences in recovered water qual- ity with time, as demonstrated by preliminary field data for ASR systems in Fig- ure 3-2. Even at the earliest observation, the pumped water in the dual-porosity aquifers (fractured chalk and fractured sandstone) and, to a lesser extent, basalt and heterogeneous sandstone has the chemical signature of a mixture of the in- jected and native waters. When water is injected into a dual-porosity system, it flows through the secondary porosity much more rapidly than through the pri- mary porosity. Water in some fraction of the primary pores is not replaced by recharged water. During storage, solutes in the primary and secondary pore spaces will move toward equilibrium through diffusion. The figure shows that when native groundwater quality is poor, MUS that employs wells for recharge in dual porosity aquifers faces greater challenges in recovery efficiency because of degraded water quality than systems in more homogeneous aquifers, such as sands and gravels. METHODS FOR CHARACTERIZATION OF AQUIFERS AND MUS SYSTEMS To evaluate the conditions for MUS, geoscientists and engineers must de- termine the aquifer hydrogeologic and hydraulic properties. This section dis- cusses methods to determine these properties, as well as knowledge gained by cycle testing, monitoring, development of a conceptual hydrogeologic frame- work, and groundwater flow modeling. Determination of Aquifer Properties Using Laboratory Tests, Pumping Tests, or Slug Tests Estimating the hydrological properties of water-bearing layers is an essen- tial part of aquifer characterization. Conventional aquifer tests include draw- down (pumping), recovery, interference, and step-drawdown tests (ASCE, 1985). During the test, a well is pumped at a constant rate or stepped rates and 

70 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER 1.0 Native Groundwater 0.8 Fractured Sandstone 0.6 Fractured Chalk C/Co 0.4 Sandstone/ Claystone Basalt 0.2 Sand & Gravel 0.0 Recharged 0.0 0.1 1.0 10.0 Water Volume Recovered/Volume Injected (-) Figure 3-2 Chemical signature of water recovered from ASR wells in various kinds of aqui- fers, normalized to represent recharged water (conventionally C = 0) and native groundwa- ter (C = 1). Mixture of the two waters is common in high-dispersivity environments (e.g., fractured chalk or sandstone, and to a lesser extent, basalt and heterogeneous sandstone) even at an early time. This suggests that the application of ASR in aquifers with poor water quality faces more challenges in high-dispersivity environments than in lower-dispersivity environments, such as homogeneous sands and gravels. SOURCE: Chris Pitre, Golder Associates, personal communication, March 2006. Reprinted, with permission, from Pitre (2006). Copyright 2006 by Chris Pitre. variations of water levels with time are observed in the well and/or in one or more observation wells in its vicinity. For a confined aquifer, transmissivity and storativity can be determined by aquifer tests. For an unconfined aquifer, the hydraulic conductivity and specific yield can be determined. For a semiconfined aquifer, the leakance factor and the storativity of the semipervious (leaky) for- mation can also be determined in addition to the transmissivity and storativity of the aquifer itself (Bear, 1979; Fetter, 2001). As an alternative to an aquifer test, a slug test or bail-down test can be con- ducted in a small-diameter monitoring well. In this test, the water level in the well is raised quickly (or lowered), often by lowering (raising) into it a solid piece of pipe (slug). The rate at which the water in the well returns to ambient is measured. Slug tests are used to determine the hydraulic conductivity of the formation in the immediate vicinity of the well (Bouwer and Rice, 1978). A slug test provides an estimate of storativity with low accuracy. Can conventional aq- 

HYDROGEOLOGICAL CONSIDERATIONS 71 uifer tests provide hydraulic parameters needed for evaluation of behavior of the MUS system? To some degree, they can, especially if the duration of the pump- ing test is long enough for the aquifer response to be representative of a broad area around the well. However, some limitations exist. First, during pumping, wells tend to re- move fine particles from the aquifer, which can improve performance of the well and formation. However, during recharge, wells tend to bring fine particles into the formation, which in turn clogs the formation and well screen. More infor- mation on clogging by chemical and biological reactions can be found in Chap- ters 4 and 6. Second, for an unconfined aquifer, pumping tests cannot character- ize the behavior of the vadose zone located immediately above the current groundwater surface, through which the injected water will pass. Therefore, pumping tests cannot be used to replace the injection-pumping cycle test for performance evaluation of injection wells. Third, the heterogeneity and anisot- ropy of the formation affect the results of the pumping test. Effects of fractured aquifers or carbonate conduits on the pumping tests cannot be identified by a single well pumping test. Therefore, additional pumping tests should be done or a new methodology should be developed to avoid mischaracterization of the formation. Primary porosity may be determined in laboratory tests from data collected for rock material properties (Moore, 2002). • Very low primary porosity if the pores are not interconnected or free draining • Low primary porosity if pores are visible under a l0x hand lens • Highly permeable if pores are visible to naked eye Secondary porosity (at least in the form of fractures or solution openings) is not very amenable to laboratory analysis since lab samples are usually too small to include representative amounts of secondary porosity. It is usually inferred by lab analysis from data collected for rock mass properties. Moore (2002) uses the following characteristics to determine the permeability of the material, as based on secondary porosity features: number of joint sets (including the bedding plane), joint aperture, and type of infilling (plastic compared to cohesionless materials). Moreover, the presence of major voids and solution features (cav- erns, sinkholes, enlarged joints), the occurrence of depositional features (lava tubes or interbedded gravels and lava beds), and the structural setup (faults, stress relief joints) are also indicators of secondary porosity. A study by Nastev et al. (2004) of hydraulic conductivity measurements ob- tained by different methods indicates scale-dependent issues. The study, con- ducted in southwestern Quebec, focused on measured hydraulic conductivity differences between constant-head injection tests, specific capacity tests, pump- ing tests, and single and multiwell pumping tests. Nastev et al. (2004) demon- strate that at the local scale, groundwater flow is influenced by fractures as op- posed to regional flow, which because of the closing of fractures at depth is gen- 

72 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER erally influenced more by the porosity and permeability of the matrix. The availability of fracture zones within an aquifer would increase the potential of preferential pathways allowing higher hydraulic conductivities. Cycle Injection and Pumping Test and Monitoring for Recovery Efficiency A cycle injection and pumping test includes recharge of water to be stored and pumping of part or all of the stored water. A comprehensive cycle testing plan can accomplish multiple objectives pertaining to MUS site hydrogeological characterization, operations, and regulatory requirements. For example, tracer tests and analyses of selected physical and chemical parameters can help deter- mine degrees of mixing between source and native groundwater during a cycle test, as well as characterize hydraulic properties of the aquifer. Monitoring and evaluation of injection and pumping pressures can help optimize system per- formance, identify the presence of dual porosity, and assess reduction in perme- ability. Historically, the design of ASR cycle test plans (i.e., recharge volumes and rates, pH adjustment) has focused predominantly on operational issues such as recovery efficiency. Water-quality monitoring for regulatory purposes occurred during cycle testing, but only recently have cycle test plans been designed with more emphasis on scientific issues (e.g., water quality). The same change in emphasis is seen in the decision matrix for placement of monitoring wells. Rather than placing a single monitoring well a standard distance downgradient of the recharge well, more consideration is now given to aquifer anisotropy, ad- verse hydrogeochemical reactions, and changes in hydraulic gradients through- out all phases of cycle testing. For example, a Wisconsin Department of Natural Resources (WDNR) ASR Technical Advisory Group (2002) states: “The fact that ASR operations will cause frequent changes in directions of groundwater flow in the vicinity of the ASR well means that there may not be a single ‘downgradient’ direction along which to install a monitoring well. … Thus, strict demonstration of compliance could require numerous monitoring wells… .” With regard to the number of cycle tests needed for a particular ASR well, two primary factors are generally considered: (1) operational testing and optimiza- tion, and (2) regulatory requirements. Again, adaptive management effectively guides the decision, while effective communication with regulatory authorities helps streamline the process as detailed in Chapter 6. In an assessment of the Comprehensive Everglades Restoration Plan (CERP), the National Research Council (NRC) (2001) recognized that limiting factors with regard to recoverability of recharged water include mixing between recharge (source) water and poorer-quality native groundwater and the effects of water-rock interactions. If a cycle test monitoring plan is to fully assess poten- tial water quality changes and system performance that may occur at the scale of full operation, long-term monitoring is required (NRC, 2001), especially if the kinetics of geochemical reactions are not well established. 

HYDROGEOLOGICAL CONSIDERATIONS 73 Although CERP (U.S. Army Corps of Engineers and the South Florida Wa- ter Management District, 1999) is orders of magnitude larger in scale and scope than typical ASR operations, most of the issues are the same; therefore similar goals exist for monitoring during cycle testing. These goals can be subdivided relative to which component of a cycle test is involved (i.e., recharge, storage, or recovery) as well as whether the testing should be accomplished during earlier or later cycle tests (U.S. Army Corps of Engineers and the South Florida Water Management District, 2004). Using the project-specific goals of CERP as a foundation, the remainder of this section summarizes what cycle tests can ac- complish and what behavior of the aquifer should be monitored: Before the cycle test, baseline hydraulic and geochemical analyses are con- ducted on both native groundwater and source water for all parameters to be measured during the cycle test. Borehole geophysical logs completed at this time also provide site hydrogeological characterization, which can be compared to post-cycle test logs to assess potentially adverse changes in borehole charac- teristics, such as dissolution-related widening of fractures. A standard step- drawdown pumping test is conducted to establish well and formation loss coeffi- cients and well efficiency. Following water level recovery, a long duration pumping test can then be conducted to estimate hydraulic characteristics in the vicinity of the ASR well. Upon completion of the long-term pumping test and associated recovery of water levels to background, a step-injection test is usually conducted to characterize water level response in the ASR well under reverse conditions from the previous step-drawdown test (Pyne, 2005). During recharge (early cycles), wellhead pressure is monitored and com- pared to the recharge rate to measure the potential effects of well plugging and estimate the “steady-state” pressure of system. Water samples are collected to evaluate geochemical changes as the source water moves through the aquifer. Tracer tests1 (tracer added to the ASR well) can be conducted using a monitoring well to guide duration of recharge. There is a need to assess the fate and trans- port of microorganisms, for which microsphere and/or microphage tracer test can be used. If water quality differences between stored and native water are small and there are no significant concerns regarding geochemical reactions, then a small number of long cycles is appropriate to focus on plugging rates and backflushing frequency required to maintain recharge rates. If there are signifi- cant water quality differences between stored and native water, a larger number of cycles is required. After the first cycle, the next three cycles have the same recharge volume and storage period in order to determine the improvement in recovery efficiency with successive identical cycles. During storage (early cycles), one major concern is the recovery of the stored water. How does the storage time affect recoverability? Samples can be collected from all wells to assess hydrogeochemical reactions with slower rates. If a concern exists with respect to the fate of microorganisms, down-hole diffu- 1 These could be completed during any cycle test; with increased cycle testing, preferential pathways will become more developed. 

74 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER sion chambers may be employed to assess survivability. The storage duration selection process should consider the following factors: anticipated water-rock interactions; the effect of storage on coliform bacteria and other microbiological parameters (if present in the source water), nutrients, metals, radionuclides, and mercury species; and estimated time required for detection of various tracers. If there is a real concern regarding potential geochemical reactions, care should be taken to avoid shocking the formation with a sudden change in quality. Fur- thermore, storage time should be built into the test program since some reactions such as manganese dissolution require several days or weeks to occur. During recovery (early cycles), the recovery efficiency can be estimated for comparison to other sites or subsequent cycles. If waters recovered are for the purpose of ecosystem restoration, bioassays may be performed on the recovered water to determine its toxicity to living organisms near the discharge area. This assessment includes the potential effects of mercury bioaccumulation, referring to details in Chapter 4. The water levels or wellhead pressure will be measured along with the pumping rate to assess the effects of water withdrawal on the well and aquifer. Water quality changes in the recovered water and as it moves past monitor wells will be evaluated. Tracer test (tracer added to the monitor well) can be used to provide hydrogeological characterization of the aquifer, and travel time will guide the duration of recovery. The effect of decreased recovery rates on recovery efficiency is also evaluated. During recharge (later cycles), besides monitoring the same parameters as in early cycles, more parameters for long-term performance of the MUS system are collected. The system will be operated based on projected conditions envi- sioned for an anticipated larger-scale system. The subsurface storage volume will be built up. The effect of buoyancy on system efficiency is evaluated using longer recharge duration. Upward migration of recharge water will also be evaluated by monitoring units above the confining units. During storage (later cycles), the system will be operated based on pro- jected conditions at full scale. The effect of buoyancy on system efficiency is evaluated using longer storage duration. During recovery (later cycles), the system will be operated based on pro- jected conditions at full scale. The potential for upconing due to longer periods of recovery is assessed. The water quality changes will also be assessed to bet- ter define recovery of the stored water. If multiple ASR and monitoring wells are being tested, the characteristics of the storage volume (shape, thickness, expansion rate, etc.) can also be assessed. The maximum pressure buildup ASR well field can be evaluated. Post-cycle test borehole logging can be used to assess physical changes in an aquifer due to repeated cycle testing. Hydrogeological Framework of an MUS System Geologic and geophysical data collection from airborne surveys, land sur- 

HYDROGEOLOGICAL CONSIDERATIONS 75 face, and boreholes, coupled with hydrologic data from various tests at the labo- ratory and field scale, comprise the foundation of knowledge required to develop a robust conceptual hydrogeological framework. This framework should be developed at multiple scales to accommodate site-specific MUS system plan- ning needs as well as regional water supply needs. Surface geologic maps, cross sections, and subsurface maps (e.g., surfaces and thicknesses of underground units) of lithostratigraphic and hydrostratigraphic units (including aquifers and relative confining units) characterize the framework. Integration of geologic, hydrogeologic, and hydraulic data with the hydrogeological framework facili- tates modeling and assessment in support of MUS. Three-Dimensional Models for Aquifer Characterization The amount and types of data required for aquifer characterization depend on the heterogeneity of the aquifer and the optimum resolution required to de- velop models on which the MUS system design will be based. Such data may originate from multiple sources (i.e., consultants; local, state, and federal agen- cies) and multiple projects. Amassing multiple data types from disparate sources comes with challenges, including quality assurance and quality control (QA/QC), data standardization (e.g., units, methodologies), metadata, and reso- lution issues. While development of such databases requires significant finan- cial and human resources, organization of these data into a seamless application facilitates effective use of time and funds over the duration of long-term MUS projects. Continuing advancements in geographic information systems (GIS) and hy- drologic computer models facilitate integration of complex databases with three- dimensional (3D) applications. Storage of geologic or hydrogeologic data in three dimensions allows interpolation of 3D hydrogeologic units, designation of measured or interpreted properties to the units, volume calculations, morphol- ogy analysis, representation of complex fault systems, parameter flux (i.e., groundwater flow, chemical diffusion) between units, and interpolation of hy- drologic properties within the unit volumes. Once the 3D framework is estab- lished, “virtual” cross sections (e.g., geologic, hydrologic, geophysical, hydro- chemical) and borehole stratigraphy can be predicted and represented graphi- cally. Moreover, 3D visualization of aquifer properties, which is available through commercial software packages, can be rendered. Ross et al. (2005) state that “3D geomodeling is expected to become a stan- dard in the near future.” Numerous advantages exist regarding integration of complex database solutions with 3D hydrogeologic framework mapping (e.g., Artimo et al., 2003; Faunt et al., 2005; Ross et al., 2005; Soller et al., 1998; Thorleifson et al., 2005): (1) the 3D framework model is comprised of relational discrete surfaces or volumes representing best-available data in a common and internally consistent framework that does not require high-end computer power; (2) the degree of complexity of the framework can be modified to fit the needs 

76 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER of individual projects and the development of derivative products (e.g., flow simulations, cross sections, aquifer vulnerability mapping);(3) the 3D frame- work model not only facilitates development of groundwater flow models, but also provides a feedback loop (adaptive modeling) between the framework and groundwater model, allowing for refinement of both models to reduce uncer- tainty; (4) data redundancy is minimized and data standardization is maximized; (5) data gaps can be efficiently identified, such as optimizing the location, con- struction, and depth of wells to address specific data needs; (6) automated data entry and semiautomated QA/QC can be streamlined (7) this facilitates GIS and data analyses using complex data sets, including graphic logs; (8) once imple- mented, the time required for data entry and model development is significantly reduced; and (9) 3D visualization (static or animated) helps scientists, engineers, and environmental managers more fully understand the dynamics and complexi- ties of the hydrogeologic system. These advantages, however, are accompanied by caveats, especially with regard to spatial resolution and interpolation uncertainty. For example, the cor- relation length of hydraulic conductivity measured in a core sample may be on the order of a few meters. In such a case, interpolative maps drawn by GIS, 3D mapping applications, or other methods using kriging or similar algorithms are to be interpreted with care if the data set is separated by more than this distance. Potential misuse of 3D maps or models is minimized through assessments of interpolated surfaces or volumes that consider factors such as map prediction error and effects of data quality, gaps, and clusters. Box 3-4 shows a case study of 3D geologic modeling and database solutions in an MUS system. Additional MUS applications of aquifer framework characterization include development of models or tools that identify (1) cultural impediments to MUS, (2) hydrogeologic settings and land use applications that increase the contamina- tion potential of an MUS storage zone, and (3) storage zones suitable for spe- cific MUS activities. Examples of these tools include ASR suitability scoring (Brown, 2005) or ASR potential mapping (Dudding et al., 2006), and aquifer vulnerability as- sessments (e.g., Doerfliger et al., 1999; Huaming and Wang, 2004; Arthur et al., 2007). Tracers, Geophysics, and Other Aquifer Characterization Methods Tracers. In MUS systems, there are four general applications for tracers: (1) assess the fate and transport of microorganisms; (2) determine aquifer prop- erties such as porosity and hydraulic conductivity; (3) determine movement of the recharge water, including the degree of mixing between recharge and native waters, as well as dispersion and diffusion; and (4) evaluate in situ reaction rates 

HYDROGEOLOGICAL CONSIDERATIONS 77 BOX 3-4 Case Study: The Role of 3D Geologic Modeling and Database Solutions in the Virt- taankangas Aquifer Artificial Recharge Project, Southwestern Finland The purpose of the Virttaankangas Aquifer artificial recharge project is to provide the 285,000 inhabitants of the Turku area, southwestern Finland, with good-quality potable water by 2010. The total budget of this project will be about 100 million euros. Pretreated river water from the Kokemäenjoki River will be conducted by pipeline to the Virttaankan- gas Aquifer for infiltration. To provide acceptable water quality, the residence time of the water in the aquifer is designed to be at least three months. The water will then be pumped from the aquifer to the Turku region for consumption. The costs of planning and building the 100-km pipeline and associated infrastructure are the largest items of expenditure in the project. However, research on the geology and hydrogeology of the Virttaankangas Aquifer will be critical to its success. In this project, geological, geochemical, and geophysical data are organized within an integrated database solution that also accepts manually and automatically measured groundwater field data. Among the wide array of data types are sedimentological, ground- penetrating radar, gravimetric, isotopic, and physical water quality parameters; results of pump and infiltration tests; and hydraulic heads. Through implementation of this database, all hydrologic and hydrogeologic data are accessible “on demand” for development of semiautomated, internally consistent 3D model units, from which hydrogeologic framework models, and subsequently, groundwater flow models are generated. This dynamic and flexible database will be used and expanded through the development and production phase of this MUS project. SOURCE: Artimo et al. (2005). (see Chapter 4). Tracer studies are also needed to address regulatory compliance and hydrogeological and hydrogeochemical characterization. Some states, for example, require reporting of residence times and travel distances for recharge water (Shrier, 2002), as well as demonstration of microbial die-off (EPA, 2006). Dispersivity and the degree of mixing can be the single most important factors in recovery efficiency, which is often a measure of MUS system performance (Brown, 2005). Chemical tracers allow distinction between the two waters of interest. These tracers include basic water quality parameters, stable and radiogenic iso- topes, and constituents added to the recharge water to give it a unique “signa- ture.” To provide optimal results, the tracer should have the following charac- teristics: • impart or represent a unique physical or chemical characteristic of the recharge water, • be nonreactive with the aquifer matrix, • be nonreactive with any combination of the two waters, • be photochemically stable, • be unaffected by microbial activity, • be easily measured, and • not be readily sorbed. 

78 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER For example, Cl− can serve as a natural conservative tracer if no evaporite minerals are present in the aquifer and if source water concentrations are consis- tent (i.e., no evaporative concentration of Cl− occurs within reservoirs of re- charge water). In a similar manner, F− is generally a useful tracer as long as fluorine-bearing minerals are not present in the aquifer or predicted to precipi- tate based on water-rock geochemical (equilibrium) models. Br− is another tracer in the same family. Visible fluorescent dyes such as rhodamine WT and fluorescein, which are not sorbed by the aquifer materials, can also be useful; their presence can be detected visually or with the use of fluorometers or ultra- violet light. Organic contaminants such as ethylenediaminetetraacetic acid (EDTA), which may be present within recharge basins, can also serve as indica- tors of the recharge water. Additional water quality parameters commonly used as tracers include, but are not limited to, NO3−, SO42−, and boron. Under certain conditions, including a sufficient contrast between the two waters, parameters such as electrical conductivity, total dissolved solids (TDS) and even tempera- ture can be used as tracers, at least to a qualitative degree. Stable isotopes of the water molecule itself (e.g., 18O, 2H [deuterium]) can be used to determine ratio of mixing when waters of different origins are mixed. “Emerging” chemical tracers may also be used. The rare-earth element gadolinium (Gd) is an example. Magnetic resonance imaging (MRI) technology uses a gadolinium-based acid to improve contrast of the image. This stable acid compound is passed through the human body into wastewater and is not re- moved during effluent treatment. Knappe et al. (2005) found that the gadolinium compound has many of the characteristics outlined above, and as a result, it was a useful tracer in their study of bank filtration. Additional emerging tracers in- clude endocrine disrupting compounds (EDCs) (e.g., Verstraeten et al., 2005). The use of microbial tracers for groundwater movement has been around since the late nineteenth century (Harvey and Ryan, 2004). However, biological tracers have been used extensively in studying groundwater movement only since the 1970s (Wimpenny et al., 1972). Viruses were first used to study the hydrology of aquifers in the early 1970s (Martin and Thomas, 1974). The viruses of choice were bacteriophages; these viruses infect bacterial hosts. Under certain circumstances, bacteriophages are preferred over colored dyes as tracers for water movement because they can easily be obtained in high titer in a relatively small volume (e.g. 1014 pfu in 1-10 L versus 30-40 L for Rhodamine WT dye). In addition, because of the host specificity of the phages, different bacteriophages can be injected at the same time or at different time in the same water system (Rossi et al., 1998). Biological tracers (in this case, bacteriophages) may represent a better model for the trans- port behavior of waterborne pathogens in water because they are present as col- loids whereas chemical tracers are generally water soluble. Bacteriophages that have been used extensively in hydrology studies as surrogates for viral patho- gens are PRD1(P22) and MS2 (Yahya et al., 1993). Salmonella typhimurium, strain LT-2, is one of the most commonly used hosts for propagation and enumeration of P22 in environmental studies (Harvey 

HYDROGEOLOGICAL CONSIDERATIONS 79 and Ryan, 2004). Because of its small size and double-stranded DNA nature, P22 has been found to be very stable in natural environments compared to other viruses and can be used as the worst-case scenario model. An inactivation rate of 0.2 log per day in river water was observed at 21-25°C (John and Rose, 2005b). Overall, tracer tests can be highly useful, but they do have limitations. The expense may limit the tests to a few available monitoring wells whose location and spacing may not have been designed with tracer tests in mind. Also, during recovery tests, regulatory limitations may pose a severe constraint on tracer re- covery due to other parameters (e.g., chloride, TDS) that must be disposed of or discharged into nearby waterways. These constraints may affect pumping tests as well. Hydrogeophysical Methods. Hydrogeophysical technology developed in recent years provides qualitative and quantitative information about subsurface hydrological parameters or processes (Hubbard and Rubin, 2005). Various plat- forms can be used to collect data to characterize hydrological parameters and processes at different stages of testing and operation of an MUS system. Hy- drogeophysical surveys for an MUS system can range from laboratory (or point) scales (10-4 to 1 m), local scales (10-1 to 102 m), to regional scales (101 to 105 m). Several factors are considered when selecting a characterization or data acquisition approach for MUS system applications: the objective of the investi- gation relative to the sensitivity of different geophysical methods; the desired level of resolution; site conditions (e.g., power lines or other cultural impedi- ments); available time, funds, and computational resources; experience of the investigator; and availability of geologic or hydrologic data for calibration or verification of the geophysical data. The most common land surface or airborne (fly-over) hydrogeophysical studies are electromagnetic (EM) induction (both frequency and time domain), gamma-ray spectrometry (radiometrics), and magnetics (Paine and Minty, 2005). Electromagnetic induction, including both frequency-domain EM (FDEM) and time-domain EM (TDEM) approaches, measures the apparent electrical conduc- tivity of the ground to depths ranging from a few meters to a few hundred me- ters, depending on the instrument resolution and the ground conductivity (Everett and Meju, 2005). Bulk conductivity of the ground is a function of wa- ter content, water chemistry, pore volume and structure, and electrical properties of the host mineral grains (McNeill, 1980). Gamma-ray surveys map the distri- bution of radioactive elements—potassium (K), uranium (U), and thorium (Th), at the earth’s surface, which varies with source rock mineralogy and surface processes such as erosion, pedogenesis, and sediment deposition as well as hu- man activities (development of an MUS system). Surface geophysical methods such as seismic refraction and reflection (Pride, 2005), microgravity, controlled source audiofrequency magnetotelluric (CSMAT) profiling, TDEM, and resistivity surveys may help identify heteroge- neities in permeability, including preferential flow paths. For example, Dobecki et al. (2007) employed CSMAT before, during, and after an ASR recharge event 

80 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER to characterize the distribution of recharge water within the storage zone. Some of these methods require certain site characteristics including an order-of- magnitude resistivity contrast between recharged source water and native groundwater. Target depth and cultural noise (e.g., pipelines, large grounded metal structures) also affect the ability of EM methods to resolve features. Re- lated emerging technologies involving an induced magnetic field via groundwa- ter low-voltage charging (Rollins, 2006) are also of potential application for MUS. Ground-penetrating radar (GPR) can be employed to address numerous hy- drogeological questions, ranging from geological structure to material properties (Annan, 2005). It can delineate fine-scale depositional stratigraphy and its spa- tial variations, which are important in terms of understanding correlation lengths and the scale of heterogeneity for hydraulic conductivity (Annan, 2005; Bristow and Jol, 2002; van Overmeeren, 1998). GPR sensitivity to water content also provides a technique for mapping groundwater surfaces including perched water tables. During 2005, the U.S. Geological Survey (USGS) continued its work on a web-enabled earth resistivity tomography (ERT) monitoring system that will be used to assess and monitor hydrologic processes including ASR and saltwater intrusion into coastal aquifers. Monitoring in a dual-porosity system can be very problematic. Due to the high ratio of intragranular porosity to open spaces in the aquifer (conduits, fissures, fractures, etc.), storage zone monitoring wells will likely not represent the chemical and physical conditions in the open storage zone network. Moreover, if geologic processes that led to the development of dual porosity have affected zones above or below the storage zone, confinement of the storage zone, if needed (as in ASR), is compromised and recovery effi- ciency will decline. Borehole geophysics includes all methods for making continuous profiles or point measurements at discrete depths in a borehole using different types of probes (Kobr et al., 2005). One of the most important attributes of geophysical log tools is the ability to make several different physical or chemical measure- ments in a borehole. Examples include spontaneous potential; normal and/or lateral resistivity logs; conductively focused current logs and micro-focused logs; gamma-ray logs; gamma-gamma logs; neutron logs; elastic wave propaga- tion logs; acoustic televiewer; and temperature logs. These methods can be employed to delineate stratigraphy; determine bulk density, porosity, and mois- ture; and characterize the structure of aquifer materials (orientation of fractures, fracture openings and bedding), water movement (vertical and horizontal), and water quality. Each method has its own limitations; therefore, multiple methods are employed to have a better understanding of the aquifer system. Dual or secondary porosity in an MUS storage zone can be identified using caliper logs, which provide borehole diameter. Poorly consolidated materials, washout zones, and possible fractures or conduits are among the features that can be identified. Loss of drilling fluids or circulation may indicate influence of secondary porosity as well. Borehole video logs also help identify fractures, 

HYDROGEOLOGICAL CONSIDERATIONS 81 cavities, and conduits intersecting the borehole. These video logs, as well as borehole flowmeters, allow measurement of groundwater flow rates and direc- tions. These methods may be complemented by microgravity surveys to detect conduits. Cross-borehole ERT is a method that applies what has historically been a surface geophysical survey, EM. Cross-borehole ERT provides a vertical profile of resistivity (e.g., a resistivity cross section) between boreholes. Results of multiple cross-borehole ERT surveys can be combined to construct 3D distribu- tions of groundwater quality, which can be validated by hydrochemical sam- pling. In a 2D application, Johnson et al. (2004) designed an ERT survey to monitor the injection of relatively low-resistivity water into a brackish-water fractured limestone aquifer utilized as an ASR storage zone. Table 3-3 lists several common geophysical characterization methods, which are classified according to their acquisition category (Hubbard and Rubin, 2005). The attribute that is typically obtained from each method is given, along with some examples of hydrogeological objectives for which each method is particularly well suited. These objectives can be broadly categorized into three key areas: hydrogeological mapping, hydrogeological parameter estimation, and monitoring of hydrological processes. Modeling of Groundwater Flow During Recharge, Storage, and Recovery A groundwater model is a simplification of a real-world aquifer system that provides a cost-effective instrument for planning, design, or operation of MUS. Models are used largely to understand the behavior of a flow system and to pre- dict how the system will behave in the future (Fetter, 2001), they can be built as analytical or numerical models. Analytical models are derived from differential equations that describe the distribution of hydraulic heads in space (x, y, z) and time. Storage properties and hydraulic conductivities in a groundwater system can sometimes be solved analytically when simplified. Among those assump- tions may be homogeneity, flow exclusively in one or two dimensions, and sim- ple boundary conditions (e.g., time and space invariant). Therefore, analytical models are elegant and useful in wellhead protection, and dewatering, among other studies. However, they are less versatile than numerical solutions since the problem has often been simplified. Numerical models simulate groundwater flow using algebraic equations. Heterogeneity, three-dimensional flow, and complex boundary conditions are more easily incorporated into numerical equations. These models require more data, conceptualization, design, and expertise. Equations can be solved via fi- nite differences or finite elements, which are among the most used methods. Most often, a computer code solves the flow equations by applying approxima- tion techniques. In selecting the type of model for use, it is necessary to deter- mine whether the model equations account for the key processes occurring at the site. Each model, whether it is a simple analytical model or a complex numerical 

82 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER TABLE 3-3 Common Geophysical Characterization Methods that are Used to Assist in Hydrogeological Investigations Acquisition Characterization Attributes Typically Examples of Approaches Methods Obtained Hydrogeological Objectives Airborne Remote Sensing Electrical resistivity, Mapping of bedrock, freshwater- gamma radiation, mag- saltwater interfaces and faults, netic and gravitational assessment of regional water field, thermal radiation, quality electromagnetic reflec- tivity Surface Seismic refraction P-wave velocity Mapping of top of bedrock, groundwater surface, and faults Seismic reflection P-wave reflectivity and Mapping of stratigraphy, top of velocity bedrock, and delineation of faults or fracture zones Electrical resistivity Electrical resistivity Mapping aquifer zonation, groundwater surface, top of bedrock, freshwater-saltwater interfaces and plume boundaries estimation of hydraulic anisot- ropy, and estimation or monitor- ing of water content and quality Electromagnetic Electrical resistivity Mapping aquifer zonation, groundwater surface, freshwater saltwater interfaces, and estima- tion or monitoring of water con- tent and quality Ground- Dielectric constant val- Mapping of stratigraphy and penetrating radar ues and dielectric con- groundwater surface, estimation trasts and monitoring of water content Crosshole Seismic P-wave velocity Estimation of lithology and frac- ture zone detection Electrical resistivity Electrical resistivity Mapping aquifer zonation and estimation or monitoring of water content and quality Radar Dielectric constant Estimation or monitoring of wate content and quality, mapping aquifer zonation Wellbore Geophysical well Electrical resistivity, Lithology, water content, water log seismic velocity, and quality, and fracture imaging gamma activity Laboratory/ Electrical, seismic, Electrical resistivity, Development of petrophysical Point dielectric, and x- seismic velocity and relationships, model validation, ray methods attenuation, dielectric investigation of processes and constant, and x-ray instrumentation sensitivity attenuation SOURCE: Modified from Rubin and Hubbard (2005). Reprinted, with permission, from Rubin and Hubbard (2005). Copyright 2005 by Rubin and Hubbard, with kind permission of Springer Science and Business Media. 

HYDROGEOLOGICAL CONSIDERATIONS 83 model, may have utility in hydrogeological and remedial investigations. The key is to make certain that the problem is clearly defined and that the selected model is the best choice to answer the posed questions. Analyses using groundwater and solute transport numerical modeling may help to • Evaluate the performance of a regional set of ASR wells to aid in estab- lishing the design, spacing, orientation, and capacity of those wells; • Evaluate regional changes in hydraulic head and flow patterns; • Evaluate the impact on the environment, including neighboring surface water flow, and existing users; • Evaluate the critical pressure for rock fracturing or widening of existing fractures; • Analyze the relationship between storage interval recovery rates and recharge volume so that the recovery of water is optimized; • Visualize the movement of stored water throughout the wellfield, which is of special interest where the storage zone contains water of lesser quality or where dual porosity is present; and • Evaluate the extent of potential water quality changes in the aquifer during storage and movement. Modeling Protocol A protocol should be followed when developing a model, as documented in relevant ASTM (American Society for Testing and Materials) guidelines and other literature, such as Anderson and Woessner (1992), Spitz and Moreno (1996), and Mercer and Faust (1981). There are several important steps to take during model development. First, the purpose of the model should be identified clearly, and it should be determined whether modeling is the appropriate type of analysis. The next step requires building the conceptual model, which can be a pictorial and/or a written description of the real-world aquifer system and the simplifying assumptions that describe the primary hydrogeologic processes. The process of building a conceptual model starts by identifying the boundaries of the model—any physi- cal (e.g., faults) and/or hydraulic (e.g., groundwater divide, large water body, ocean) boundaries within the area of interest, formulating the general water budget (evapotranspiration, baseflow, pumping, etc.), and defining the flow sys- tem from water levels or contour maps in terms of the locations of recharge and discharge areas and the connection between surface and groundwater systems. Field data are collected for hydraulic properties of the aquifer (e.g., hydraulic conductivity, storage). Estimates of the vertical aquifer properties are especially important for an MUS model, since vertical flow through an aquifer and/or con- fining unit can play a large role in determining flow behavior when the system is 

84 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER stressed hydraulically (i.e., ASR recharge or recovery). Typically, building the conceptual model is an iterative process, which requires reevaluating the con- ceptual model throughout the model development process. The steps described in the conceptual model must be formulated mathe- matically using appropriate governing equations. In the case of MUS, governing equations for groundwater flow, variable density (if this is the case) and trans- port should be solved accurately by the numerical method applied in the selected computer code. Standard flow models assume that the density of groundwater is constant, which is reasonable if salinity and temperature show little temporal or spatial variance. However, differences in the density of miscible fluids, associ- ated with saltwater intrusion, or freshwater recharge into brine aquifers, among other processes, require the use of models that solve for solute transport and/or temperature in addition to flow (Brown et al., 2006). It should be noted that solving for groundwater flow with variable density and transport may require large run times. If the model run time becomes unmanageable, it may be neces- sary to re-visit the model design (e.g., using a coarser discretization or aggregat- ing the stress periods). The next step is MUS model design, which sets the spatial and temporal discretization of the grid. A refined grid improves model stability, and accuracy may be improved while numerical dispersion in the solute transport components is reduced although this may have a negative effect on the model run time. A difficult modeling problem exists when a combination of steep head gradients and sharp concentration fronts is present and could result in model instability and inaccuracy. Brown et al. (2006) found that the concentration changes around an ASR well are focused within 250 feet of the well (pumped at a rate of 5 Mgal/d) when simulating ASR in the Floridan Aquifer System; therefore, a constant horizontal grid or mesh resolution within that radius can be established with an increase in grid size at a reasonable ratio beyond 250 feet. ASR wells with large recharge or recovery rates may need additional horizontal and vertical grid refinement (Brown et al., 2006). Excessive refinement is not advised so as to maintain reasonable run times for calibration and prediction. In effect, finer resolution should be employed at the areas of interest, while areas of less interest may be modeled with a coarser resolution. Meeting specified calibration criteria or targets by reproducing measured water levels or flow rates is the goal of model calibration. The MUS model builds on calibration against seasonal water levels and water quality, as well as performing transient calibration at existing ASR and well sites where aquifer test or ASR cycle testing data are available. Once the model is calibrated, a sensitiv- ity analysis is recommended to quantify and show the effects of uncertainty in the calibrated model. Modeling Software Modeling an MUS project can be a daunting challenge. Fortunately, there 

HYDROGEOLOGICAL CONSIDERATIONS 85 are several, available model codes for both finite-element and finite-difference methods that could be used for MUS projects. The several models are available in both the public and the private domain. A description of the model codes, which are available in the public domain, is presented herein including: MOD- FLOW-MT3D, SEAWAT, HST3D, SUTRA, and WASH123D. Several other established codes could also have been selected, including FEFLOW (a proprie- tary code from Europe that may be more difficult to procure for U.S. govern- ment work efforts; Pavelic et al., 2004, 2006). Each code exhibits both strengths and weaknesses. All of the codes provide much of the model functionality de- sired for the MUS Regional Study for saturated flow. WASH123 codes provide the best overall functionality for any type of MUS due to the coupling of surface and groundwater components. MODFLOW is a three-dimensional finite-difference groundwater model that was first published by the USGS in 1984 (McDonald and Harbaugh, 1988) and has been updated periodically. The groundwater flow equation is solved using the finite-difference approximation. The flow region is subdivided into a grid; within each cell, properties are assumed uniform. The model layers can have varying thickness. A flow equation is written for each cell. Several solvers are provided for solving the resulting matrix problem. MODFLOW is consid- ered to be the most widely used program for constant-density groundwater flow problems. Key factors for MODFLOW’s popularity in the modeling community are its thorough documentation, its modular structure, and the public availability of the software and source code. The major limitations of MODFLOW are that the model cannot provide a water budget for the full hydrologic cycle because overland flow and the unsaturated zone are not simulated. However, MOD- FLOW 2005 has a new package with the capability to simulate unsaturated flow. MODFLOW-MT3DMS is a suitable tool for a mass-balance approach to evaluating storage and recovery, but is not suitable if there are significant den- sity issues in the study area. MT3DMS (Zheng and Wang, 1999) is based on a modular structure to permit simulation of solute transport and particle tracking (dispersion or advection). MT3DMS interfaces directly with MODFLOW. MT3DMS is a 3-D transport model, where MS denotes the multispecies struc- ture for accommodating add-on reaction packages. MT3DMS has a comprehen- sive set of options and capabilities for simulating advection, dispersion or diffu- sion, and chemical reactions of contaminants in groundwater flow systems under general hydrogeological conditions (Zheng and Wang, 1999). SEAWAT (Guo and Langevin, 2002) is a computer program for the simula- tion of three-dimensional, variable-density, transient groundwater flow with sol- ute transport in porous media. The program combines MODFLOW and MT3DMS (Zheng and Wang, 1999) into a single computer program that solves the coupled flow and solute transport equations. A disadvantage in using this code is the long model run times due to small time-step requirements for con- taminant transport models. HST3D (Kipp, 1997) is a heat and solute transport program that simulates groundwater flow and related heat and solute transport in three dimensions. This 

86 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER program may be used to analyze subsurface waste injection, saltwater intrusion, and freshwater recharge and recovery (Kipp, 1997). This code solves three- dimensional, saturated groundwater flow with heat and solute transport. These three equations are coupled through the dependence of advective transport on the interstitial fluid velocity field, the dependence of fluid viscosity on tempera- ture and solute concentration, and the dependence of fluid density on pressure, temperature, and solute concentration. SUTRA (Voss and Provost, 2002) is a model for saturated-unsaturated, vari- able-density groundwater flow with solute or energy transport. SUTRA (satu- rated-unsaturated transport) can simulate fluid movement and transport of either energy or dissolved substances in a subsurface environment. The code employs a two- or three-dimensional finite-element and finite-difference method to ap- proximate the governing equations that describe the two interdependent proc- esses that are simulated: (1) fluid density-dependent saturated or unsaturated groundwater flow; and (2) either transport of a solute in the groundwater or transport of thermal energy in the ground water and solid matrix of the aquifer. SUTRA energy transport simulation may be employed to model thermal regimes in aquifers, subsurface heat conduction, aquifer thermal energy storage systems, geothermal reservoirs, thermal pollution of aquifers, and natural hydrogeologic convection systems. SUTRA has been used for past ASR simulation studies. Voss (1999) provides a review of SUTRA applications. WASH123D (Watershed Systems of 1D Stream-River Network, 2D Over- land Regime, and 3D Subsurface Media) is a public domain model developed by the Waterways Experiment Station for the U.S. Environmental Protection Agency (Yeh, 1998). The EPA and the U.S. Army Corps of Engineers endorse WASH123D for modeling of comprehensive watershed management plans. WASH123D is a finite-element numerical model designed to simulate variably saturated, variable-density water flow and reactive chemical and sediment trans- port in watershed systems. It is capable of representing a watershed system as a combination of 1D river or stream, 2D overland, and 3D subsurface subdomains. WASH123D is a physically based, spatially distributed, finite-element, inte- grated surface water and groundwater model. WASH123D is applicable to a variety of problems, including flood control, water supply, water quality, struc- tures, weirs, gates, junctions, evapotranspiration, and sediment transport for both event and continuous simulations. WASH123D can provide a water budget for the full hydrologic cycle. The groundwater flow portion of the code utilizes an adaptation of the FEMWATER code (Lin et al., 1997). A disadvantage of using this code is the long model run times due to small time-step requirements for contaminant transport models. WASH123D has been applied in south Florida (Brown et al., 2006). 

HYDROGEOLOGICAL CONSIDERATIONS 87 Data Needs Data needs can be extensive for modeling MUS projects. Developing a groundwater flow model is always the first step before adding the variable- density, solute transport, and/or heat components. As discussed in the modeling protocol, it is important to define the hydrogeologic system in terms of aquifer characterization and hydraulic properties. The physical framework is defined by the characteristics of the hydrostratigraphic layers and boundary conditions. Geological maps and cross sections should be shown and should identify the hydrostratigraphic units. Standard groundwater flow model properties, such as hydraulic conductivity or transmissivity and storativity, must be defined spa- tially and in the context of the physical framework. In addition, a topographic map, of basins and surface water bodies should be compiled. The extent and thickness of stream and lake sediments are important to show the connection between surface and groundwater systems. Box 3-5 describes a case study of applications of WASH123D in an MUS system in a brackish aquifer in south Florida. Other data needs include properties that describe water quality. These prop- erties are of special importance to MUS projects, especially when waters of dif- ferent qualities may be combined. In MUS, the ambient (native) water quality refers to water measured upstream (or outside) of the influence of a pollutant or contaminant during average flow conditions. Water that is no longer fit for use is said to be polluted or contaminated but legal standards change according to use, such as environmental, human consumption, irrigation, or others. One measure of water quality is total dissolved solids, which is defined by the weight of the solids that remain after the water sample is completely evapo- rated (milligram per liter). Among the major constituents (natural constituents) are calcium, magnesium, sodium and potassium, chlorine, sulfite, carbonate, bicarbonate, and silicon. Minor constituents are iron, manganese, fluorine, ni- trate, strontium, and boron. TDS can also be classified as fresh, brackish, saline, and brine (Fetter, 2001). Other quality standards are associated with dissolved oxygen, the presence of radioactive constituents, and bacterial content (see Chapter 4 for details). Regional- Versus Local-Scale Modeling Models can be applied to address both regional- and local-scale issues. Model applications can assess system-wide impacts as well as other impacts from the local, complex geometry (e.g., dual porosity, changes in the storage zone) of the MUS System. Models of different spatial, and sometimes temporal, scales are needed to address regional and local issues. Regional models are tools used for evaluating the cumulative effects of multiple, and potentially competing, uses of water resources in a region. These models are valuable tools for (1) determining regional changes in aquifer heads 

88 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER BOX 3-5 Case Study: ASR Modeling for a Brackish Aquifer in South Florida Mixing of fresh water and native water from a brackish aquifer can be modeled with the WASH123D numerical model, which computes solute transport and density-dependent flow. A hypothetical case for the Comprehensive Everglades Restoration Plan is presented to show a typical model design and to depict results from an ASR site. The WASH123D finite-element code is applied to simulate injection, storage and recovery using a box model with a domain of approximately 40 miles x 40 miles x 2,340 feet. Flow and concentration gradients may be high in the neighborhood of an ASR well. Refinement of the vertical and horizontal resolution of the 3D mesh in the vicinity of the ASR well may be necessary. In the hypothetical case, the horizontal mesh resolution at the ASR well is 10 feet and expands gradually to 5,000 feet along the model perimeter as illus- trated in part A of the figure below. Vertical mesh resolution, described in part B, varies among the different conceptual geologic units, which represent the Surficial Aquifer System (SAS), Hawthorn Group (HG) confining unit, and Florida Aquifer System (FAS) including the Upper (UFA), Middle (MF) and Lower Floridan (LF). Vertical resolution is increased in the confining units directly above and below the recharge zone (i.e., UFA). This increased resolution allows the model to depict the large head and concentration gradient at the inter- faces of these confining units. The results of a sensitivity analysis indicate that the change of the computational re- sult becomes insignificant as the time step size is reduced to less than 0.5 day. Therefore, a time-step size of 0.5 day is used. The boundary conditions are applied to the element faces representing the well screen within the UFA. Constant boundary conditions were used to assign the total head along the eastern and western model boundaries. No-flow boundary conditions were used along the northern and southern model boundaries. Con- stant boundary conditions were also used to assign the concentration along the model perimeter. The model simulated a 30-day recharge period, followed by a 305-day storage period and a 30-day recovery period. The following sections discuss the results of the WASH123D model during each simulation phase. • Initial Condition. The nodes in the geologic units above the UFA were assigned a constant concentration of fresh water (150 mg/L), while the nodes in and below the UFA were assigned a constant concentration of seawater (35,000 mg/L). Re- fer to part B of the figure for an understanding of the assignment of the initial conditions and location of ASR well prior to recharge. • Recharge Phase. Starting with the initial conditions, the ASR pumping well in- jects fresh water (with a concentration of 150 mg/L) into the UFA at a rate of 5 Mgal/d for 30 days. The hydraulic head at and immediately surrounding the ASR well increases nearly doubling in magnitude. Part A of the figure shows a cross- sectional view of the concentration profile in the vicinity of the ASR well at the end of the injection cycle, it shows that the injected fresh water has displaced the ambient saline water forming a “spheroid” of lower concentration water in the vi- cinity of the ASR well. • Storage Phase. After the recharge phase, the ASR well is turned off for 305 days. During this storage period, the hydraulic conditions stabilize close to steady-state conditions. Part B of the figure shows a cross-sectional view of the concentration profile in the vicinity of the ASR well at the end of the storage pe- riod. Although the concentration at the ASR well remains relatively constant, the effects of buoyancy stratification are noticeable. During the storage period, the density effect is the dominant factor in the flow fields. The concentrations at the lower portion of the UFA increase substantially faster than at the top of the aqui- fer. 

HYDROGEOLOGICAL CONSIDERATIONS 89 • Extraction Phase. After the storage period, the ASR recovery begins at a rate of 5 Mgal/d for 30 days. During this extraction cycle the hydraulic head at and im- mediately surrounding the ASR well decreases substantially. Part C of the figure shows a cross-sectional view of the concentration profile in the vicinity of the ASR well at the end of the storage period. Up-coning of the higher-concentration water below the ASR well is computed during extraction. A B (A) Horizontal mesh resolution and (B) conceptual geology and vertical mesh resolution of numerical model for an ASR well. 

90 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER BOX 3-5 Continued (a) (b) (c) 35000 30000 25000 15000 10000 1000 500 (C) Cross sectional views of the concentration profile in the vicinity of the ASR well: (a) time = 30 days, at the end of the recharge phase; (b) time = 335 days, at the end of the storage period; and (c) time = 365 days, also at the end of the storage period. SOURCE: Brown et al. (2006). and flows; (2) determining regional changes in aquifer water quality TDS, sul- fate, and chloride; (3) estimating groundwater recharge, discharge, and storage at larger spatial scales; (4) assessing the cumulative effects of existing and pro- posed water resource uses and developments; (5) assessing regional impacts on existing well users of the aquifer; and (6) evaluating the cumulative effects of various water management scenarios on water resources (Mack, 2003). Devel- opment of a regional model for the MUS system can be used to assess potential impacts related to a full-scale implementation of MUS, specifically ASR, and the feasibility of its development on such a scale. In the case of ASR, these models are valuable tools for (1) increased potential for saltwater intrusion caused by ASR pumping; (2) ASR well cluster site selection; (3) ASR well clus- ter design, layout performance including estimating recovery efficiency; (4) ASR well site evaluation of pressure- induced changes; and (5) localized ASR well pump design (dependent upon the appropriate model resolution). When conducting a regional study, it is advised that model development be- gin early in the process. Model development can begin as soon as the prelimi- 

HYDROGEOLOGICAL CONSIDERATIONS 91 nary hydrogeologic framework has been assembled. In this way, the model can help define a framework to study the system dynamics and organize field data as they become available. This type of analysis can occur before the calibration and sensitivity have been completed (Anderson and Woessner, 1992). Although the objective of the final model is to simulate density-dependent flow, the initial regional-scale modeling effort employing available codes—even those that are only capable of simulating constant density flow—can be useful. This effort may assist in guiding the test-drilling program (e.g., APTs) and associated data acqui- sition (e.g., monitoring). The described adaptations to the model development process can be thought of as an iterative process in which the modeler is con- tinually updating and improving the model throughout the duration of the re- gional study. As additional data are integrated in space and time, the regional modeling tool becomes more valued as a tool that has potential to extrapolate findings over large areas and long periods of time. Furthermore, the regional model can begin to identify areas where local scale modeling may be preferred or required. Local models are required when the spatial resolution of a regional model is not sufficient to capture changes in the storage zone. Unlike regional models, local models provide an in-depth analysis of local hydrogeologic properties for specific areas of interest within the regional model. Generally speaking, a re- gional model may have a resolution of 250 to 5,000 feet, while a local model may have a resolution of 50 to 250 feet. In local models, more emphasis would be placed on small groups of ASR wells, which would allow for a better under- standing of the performance and impact of these wells within the MUS systems. The feasibility of injecting, storing, and recovering specified volumes of water at individual ASR wells and in local clusters of ASR wells is the issue to which the local models are most specifically directed. A horizontal variable resolution of a regional model should be developed at a resolution of 100 to 250 feet around the proposed ASR wells, increasing to 10,000 feet in the far reaches of the study area. Use of finite-element code would benefit from its geometry ad- vantages such that the model mesh would have varying resolution across the model domain. Overall, it is common practice to use both regional and local modeling tools because of the need to assess the system-wide impacts as well as other impacts from the local, complex geometry (e.g., dual porosity, changes in the storage zone). Regional models can assess potential impacts when the MUS system is being implemented. Local models can assess potential impacts of fracturing and changes in the storage zone. Monitoring with modeling can complement and strengthen the regional study as a whole. Depending on the project, modeling tools can be implemented to help define the storage zone, buffer zone, and na- tive groundwater area in conjunction with technical experience, especially since deep monitoring is expensive. As in any modeling effort, it is advisable to start collecting data early, start model development early (including running sensitiv- ity analyses to determine data gaps), and practice the iterative process of adapt- ing the modeling tool to newly acquired data. 

92 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER IMPACTS AND CONSTRAINTS OF THE MUS SYSTEM Local, Intermediate, and Regional Flow Local, intermediate, and regional flows can affect groundwater movement through an aquifer. Factors affecting the amount of local versus regional flow include topography, climate, nature and extent of unconfined and confined aqui- fers, hydraulic conductivity of the aquifer and confining layers (Johnston, 1997), and recharge patterns. Local flow systems are dominated by a topographically high recharge area and a topographically low discharge area (Schwartz and Zhang, 2003). Groundwater flow through unconsolidated sandy aquifers is usu- ally the result of local flow systems (http://capp.water.usgs.gov). Johnston (1997) argues that under natural conditions, the amount of local flow can be much greater than regional flow. In addition, the flow velocity in local flow systems is usually greater than that in regional systems due to nearby streams (Schwartz and Zhang, 2003). As such, local flow systems are highly dependent on topography and recharge patterns. A system with various topographically low sections between recharge and discharge zones is recognized as having intermediate flow (Schwartz and Zhang, 2003). Basin-filled aquifers, such as the California Central Valley, have interme- diate flow that is representative of regional flow systems (http://capp.water. usgs.gov). In contrast, regional flows are dominant in topographically extensive low areas under natural conditions. In a regional flow system, changes in hy- draulic conductivity can impact the vertical and horizontal hydraulic head of the aquifer. For example, in an aquifer with increasing hydraulic conductivity in a lower layer and a consistently lower hydraulic conductivity in a higher layer, the vertical hydraulic gradient would be greater than the horizontal hydraulic gradi- ent in much of the system (Schwartz and Zhang, 2003). Karst aquifers can be described by high heterogeneity resulting from groundwater flow, large voids, and high flow velocities (Bakalowicz, 2005). Groundwater flow through karst aquifers can be affected by extensive fracture systems, which increase total porosity and permeability (Herrera, 2002). In kar- stic environments, Bakalowicz (2005) makes a distinction between local and regional flows, indicating that conduit patterns depend on porosity and recharge and direction of the hydraulic gradient and the drainage planes, respectively. It must be noted, however, that secondary porosity can affect regional flows by increasing the hydraulic conductivity (Herrera, 2002). Karstic aquifers gener- ally favor rapid flow and transport of solutes. With a recharge basin, vertical flow beneath the basin becomes dominant. A groundwater mound will be developed during the recharge, but it can also re- verse local flow direction against regional flow direction. Vertical flow can po- tentially increase baseflow in neighboring streams. With a vadose zone well, a groundwater mound will be developed beneath the well. However, its effect on the regional flow is minimal due to limited recharge capacity. Vertical flow is a 

HYDROGEOLOGICAL CONSIDERATIONS 93 primary component beneath the well, while horizontal flow becomes dominant after water reaches the water table. With well recharge, a groundwater mound will be developed surrounding the well. It can reverse the local flow direction relative to the regional flow, depending on the recharge capacity (Sheng, 2005). In general, horizontal flow is the dominating component, especially within a confined aquifer, even though a mound may be developed around the well. For stressed aquifers in an arid or semiarid region, the MUS system can also restore local water levels created by historic groundwater mining or overwith- drawal within a wellfield and, in turn may, restore natural flow pattern of the regional aquifer. MUS in shallow aquifers can thereby have major effects on groundwater-surface water interaction. If surface water is well connected hy- draulically to the shallow aquifer, a rise in the water table will likely increase groundwater flow into local streams, lakes, and wetlands via seeps and springs. In the case of wetlands, this could potentially have a major effect on water budgets (and, therefore, water depth) and nutrient budgets. The combination of changing depth and water chemistry might alter the aquatic ecosystems substan- tially. The loss of groundwater to the surface environment might also raise legal questions as to the ownership of the discharged water (Chapter 5). Conversely, if surface water upstream is diverted from streams and lakes for recharge, it may not only affect downstream flow, but also cause water quality deterioration. Such complications underscore the importance of both having a clear under- standing of the hydrogeologic system and keeping MUS in the context of other water management activities and tools (Chapter 7). Water Density: Uniform Density Versus Variable Density Groundwater flow in an aquifer can be caused by density differences (Cserepes & Lenkey, 2004). Water with dissolved solids such as seawater is denser than fresh water, and as such, density calculations are imperative in esti- mating flow directions (Boulding and Ginn, 2003). Water’s fluid properties vary with temperature, and hydraulic conductivity is affected as a result (Driscoll, 1995). To compensate for density effects on flow in modeling, flow equations can be adjusted to account for a variable-density fluid by including measure- ments of fluid pressure, intrinsic permeability, dynamic viscosity, and elevation (Ingebritsen & Sanford, 1998). In an aquifer, dispersion and diffusion of solutes are critical to the effective- ness of MUS, especially for ASR. Buoyancy stratification allows injected water in a high-permeability zone to swell under the overlying confining unit floating atop the native, more saline groundwater due to its higher density (Merritt, 1985; Vacher et al., 2006). Multiple “wedges” can result from injection into low- permeability beds within heterogeneous storage zones (Vacher et al., 2006). Maliva et al. (2006), recognizing recovery issues associated with density-driven fluids, suggest a dual-zone approach to ASR whereby the open interval includes the storage zone; however, a second well in the upper part of the storage zone is 

94 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER used for recovery, taking advantage of the stratification effect. Alternatively, they suggest use of a “flapper valve” that would allow recharge through the en- tire open interval, but only recover within the upper part of the storage zone. The effects of diffusion are highly evident in zones of lower permeabilities (Kasteel et al., 2000). Aquifers with high clay contents and native water salinity can greatly decrease their recovery efficiency as illustrated by Konikow (2001). Recovery efficiency is dependent on formation properties. Salinity, permeabil- ity, and thickness of the aquifer can affect recovery efficiency. Konikow (2001) indicated that the potential for clay dispersion is greatest in aquifers with swell- ing-type (2:1 clay lattice) clays. A successful ASR in coastal systems with brack- ish water is possible if flow patterns during recharge and withdrawal are con- stant (Konikow, 2001). Changes in the pattern can negatively impact an ASR. Solute concentration can create buoyancy forces greater than the fluid velocities created by hydraulic forces (Ingebritsen & Sanford, 1998). Temperature Increases in fluid temperature can result in a decrease in density, which in turn causes molecules to move faster (Fitts, 2002). Driscoll (1995) defines vis- cosity as the degree of resistance of a liquid to an applied force. A temperature increase causes a fluid’s viscosity to decrease (Fitts, 2002). As a result, hydrau- lic conductivity increases by 50 percent between 10 and 26°C (Zijl and Nawalany, 1993). As groundwater flows through an aquifer, its temperature changes despite its high specific heat capacity (Boulding & Ginn, 2003). The natural geothermal gradient leads to an increase in temperature with depth of about 1°C per 20-40 meters (Bear, 1979). If a significant temperature difference exists between native groundwater and recharge water, the MUS system will affect the hydraulic gradient between the native and recharge water. The size and shape of the storage zone could then be affected depending on the direction and magnitude of the temperature gradi- ent. In addition, the MUS system could also cause changes in temperature for spring flow, which may be a critical parameter for sustaining a neighboring eco- logical system. Aquifer Matrix Physical Impacts Land subsidence is a gradual settling or sudden sinking of the earth’s sur- face owing to subsurface movement of earth materials (Galloway et al., 1999). More than 80 percent of the identified subsidence in the United States is a con- sequence of our exploitation of underground water, and the increasing develop- ment of land and water resources threatens to exacerbate existing land subsi- 

HYDROGEOLOGICAL CONSIDERATIONS 95 dence problems and initiate new ones. Subsidence is virtually an irreversible process, and cracks and fissures may coexist at the land surface as results of aquifer movement (Helm, 1994; Holzer, 1984; Sheng, et al., 2003). Decrease in artesian head in compressible confined aquifer systems results in increased effective stress (grain-to-grain load) on the confined sediments. The magnitude of subsidence depends on the magnitude of change in head and on the compaction characteristics and thickness of the sediments. The greater the number of clayey beds in the aquifer system, the greater may be the compaction. Continuous measurement of compaction of materials in deep holes indicates rapid response to head change at most places in the subsiding areas. Subsidence can be slowed down or stopped by raising the level of the potentiometric surface or the artesian head sufficiently. One of the methods for controlling land subsi- dence is artificial recharge, which injects water into the stressed aquifer to raise the hydraulic head and curtail ongoing subsidence resulting from overwith- drawal of groundwater (ASCE, 2001). Observed subsidence and uplift after recharge in Santa Clara, California, demonstrates deformation of aquifer materi- als as results of groundwater pumping and artificial recharge (Schmidt and Burgmann, 2003). During cycles of recharge-storage-recovery of an MUS sys- tem, resulting subsidence and uplift may impact system operations and cause possible damages of infrastructure in the vicinity of the system. Li (2000) and Li and Sheng (2002) developed conceptual models to assess impacts of different scenarios of cycle loading of the ASR system on aquifer materials and con- cluded that confining units, especially clay layers or interlenses, will deform and result in additional subsidence during recovery of the stored water, and partial recovery of subsidence or seasonal uplift is also expected under a favorable con- dition even with an ASR system to retain groundwater levels. The magnitude of subsidence associated with water-level decline appears to be related in large part to geologic factors such as (1) differences in mineral composition, (2) particle size, (3) sorting, (4) degree of consolidation, (5) degree of cementation, and (6) degree of confinement of the deposits in the groundwa- ter reservoir. Thus, the ratio of subsidence to head decline will vary between groundwater reservoirs and even within a single groundwater reservoir. For ex- ample, measured ratios of subsidence to head decline vary from 0.008 to 0.1. Annual measured rates of land-surface subsidence in groundwater reservoirs range from a fractional value to about 0.5 m (1.5 feet) (ASCE, 1987). Allowed Change in Hydraulic Head Development and operation of a groundwater basin or aquifer in an area subject to alternating periods of drought and surplus suggests utilization of the groundwater storage during periods of deficient supply and the subsequent re- plenishment of the storage during periods of surplus (ASCE, 2001). In this con- text, these operations will occur in much the same manner as a surface reservoir would be operated. This will result in an artificial lowering of the water table or 

96 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER potentiometric surface during periods of deficient supply and a consequent re- turn to former conditions during periods of surplus. The actual limit or allowable range of fluctuation of the water table aquifer is primarily a matter of economics and aquifer characteristics. During wet peri- ods, levels should not be permitted to rise so high as to cause waterlogging or property damage. On the other hand, levels cannot be drawn down to the point where it is economically impossible to extract and utilize the supply. Moreover, excessive drawdown can deteriorate the storage capacity of the aquifer by com- paction, reduction of pore connections, or in extreme cases even the collapsing of structures. When excessive drawdown occurs, the costs of deepening existing wells, resetting pumps, or drilling new wells, in addition to the cost of obtaining new pumping equipment, are particularly important. In addition, environmental factors such as reduced baseflow to surface water bodies, spring discharge, and increased potential for sinkhole formation could also result from excessive drawdown. Excessive replenishment of the aquifer during surplus periods can also be undesirable because excessive recharge may result in hydraulic fractur- ing. Hydraulic fracturing is a result of increased fluid pressure over stress and rock strength (Domenico and Schwartz, 1998; Ingebritsen and Sanford, 1998). Brown et al. (2005) identified four failure mechanisms related to hydraulic fracturing. These mechanisms include regional-scale shear failure of the rock matrix, hydraulic failure of the rock matrix, pore volume increase due to micro- fracture formation, and localized stress concentrations around the wellhead. The failure mechanisms—whether on a regional scale or localized—may limit or prevent implementation of an MUS project, since any one of the mechanisms could lead to the formation of preferential flow paths in the confining unit (Brown et al., 2005). In April 1999, the Comprehensive Everglades Restoration Plan (proposed large-scale development of ASR facilities as the preferred method of providing additional freshwater storage required for overall restoration success (USACE/SFWMD, 1999). The proposed CERP system includes a total of 333 ASR wells and related surface facilities at the general locations. All proposed ASR wells have a target capacity of 18,927 m3/d (5 Mgal/d) with water treat- ment facilities included in the conceptual CERP ASR components. The total cost of the proposed CERP ASR system is approximately $1.7 billion, about one-fifth of the total estimated cost of the CERP. In cooperation with a multiagency project delivery team, the U.S. Army Corps of Engineers and the South Florida Water Management District were tasked with evaluating the feasibility of the proposed CERP ASR projects indi- vidually and through the development of a regional feasibility study. A compo- nent of the ASR Regional Study, outlined in Brown et al. (2005), was to deter- mine the pressure induced effects of an anticipated daily ASR recharge volume of 1.67 billion gallons, to the upper Floridan Aquifer System (FAS) and overly- ing Hawthorn Group sediments, specifically, the effects on piezometric pressure and hydraulic fracturing potential. The magnitude of the increase or decrease in piezometric pressure within 

HYDROGEOLOGICAL CONSIDERATIONS 97 the upper FAS during recharge and recovery cycles is highly dependent on nu- merous factors such as aquifer transmissivity, well spacing, and aquifer porosity. During ASR recharge, increases in static piezometric (hydraulic) head of 30.48 to 60.96 m (100 to 200 feet) near the pumping wells are certainly possible based on both analytical and numerical models. Conversely, during ASR recovery, decreases in static head of similar magnitudes are possible. These anticipated pressure changes may present planning and engineering constraints that some- what limit ASR development. The most important of these constraints is the potential for hydraulic frac- turing of the limestone rock of the upper FAS or overlying Hawthorn Group sediments. Brown et al. (2005) concluded that microfracturing of the FAS lime- stone due to dilatancy may occur at a total hydraulic head greater than 183 feet. In addition, during recovery operations, settlement or subsidence of the overly- ing Hawthorn Group clays is a possibility that requires further examination. Conversely, during ASR recharge cycles, expansion or “lengthening” of the Hawthorn Group clay is a remote possibility. For ASR design purposes, the pressure changes will also likely constrain wellhead design or pump selection. In addition, continually injecting at high pressures could result in high costs for electricity. Brown et al. (2005) noted that the estimated electricity cost for the 333 planned ASR wells would surpass 50 percent of the entire operation and maintenance budget for the CERP ASR pro- gram. FAS heads substantially higher than the current regional flow system could also lead to changes in flow direction or velocity. Slow regional subsidence poses serious problems in the operations of many types of engineering structures, particularly those involved in the storage, trans- port, and pumping of water. For example, tilting of the land surface can appre- ciably reduce the flow of water in low-gradient gravity canals. This has occurred in the United States Bureau of Reclamation's Delta-Mendota Canal along the west side of the San Joaquin Valley in California. In addition, smaller structures such as drains and sewers can be affected, and even the channel capacity of streams may be altered. Tilting can also affect the operation of pumping plants because such plants may be highly sensitive to minute tilting of the land surface. In a critical area such as a coastal bay where bordering lands subside, levees may have to be built. This has been done in the southern San Francisco Bay area in California to prevent flooding of adjacent agricultural, urban, and industrial areas by saline bay waters (Fowler, 1981). In addition, when the consolidating sediments are deep, casings of the water wells are compressed and frequently ruptured, requiring expensive maintenance and replacement. As sediments con- tinue to compact in a groundwater reservoir, reduction in groundwater storage capacity and even in the permeability of the sediments may occur, although the usual case is for the fine grained materials (primarily clays) to consolidate rather than the more important granular materials of the aquifers. The legal aspects of land surface subsidence caused by groundwater withdrawal are additional con- cerns facing water resource managers (Kopper and Finlayson, 1981). 

98 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER Chemical Impacts Hydrogeochemical and biogeochemical reactions may affect physical as- pects of the aquifer and MUS system performance. Clogging via microbial ac- tivity or mineral precipitation, for example, reduces hydraulic conductivity and affects MUS system performance. Mixing of water during MUS activities may lead to dissolution and enhancement of dual porosity. Although these processes are described in this chapter, significantly more detail is provided in Chapters 4 and 6. CONCLUSIONS AND RECOMMENDATIONS Conclusion: To facilitate the siting and implementation of MUS systems, maps of favorable aquifers and hydrogeological characteristics can be prepared using 3D capable geographical information systems (GIS). At a regional or statewide scale, such GIS maps can help visualize and characterize major aqui- fers for future development of MUS systems, map and analyze regional changes in head and flow patterns, and facilitate comprehensive, regional water resources management. At a project scale, they can aid in establishing the design, spacing, orientation, and capacity of wells and recharge basins, evaluating their impact on the environment and existing users, estimating the critical pressure for rock frac- turing, visualizing the movement of stored water throughout the system (espe- cially useful for systems with waters of varying density or quality), and evaluat- ing the extent of potential water quality changes in the aquifer during storage and movement. Recommendation: States, counties, and water authorities considering MUS should consider incorporating 3D capable GIS along with existing hydro- geologic, geochemical, cadastral, and other data in (1) regional mapping efforts to identify areas that are, or are not, likely to be favorable for development of various kinds of MUS systems, and (2) project conception, design, pilot testing, and adaptive management. Conclusion: Long-term local and regional impacts of MUS systems on both native groundwater and surface water have been recognized, including changes in groundwater recharge, flow, and discharge, and effects on aquifer matrix such as compaction of confining layers or clay interlayers during re- charge and recovery cycles. Recommendation: Monitoring and modeling should be performed to pre- dict likely effects—positive or negative—of MUS systems on the physical sys- tem, including inflows, storage, and outflows. Appropriate measures can and should be taken to minimize negative effects during operations. Conclusion: Groundwater numerical modeling at regional and/or high- resolution local scales provides a cost-effective tool for planning, design and 

HYDROGEOLOGICAL CONSIDERATIONS 99 operation of an MUS system. Recommendation: Analyses using groundwater flow and solute transport modeling should become a routine part of planning for, designing, and adap- tively operating MUS systems. Uncertainty analysis should also be incorporated into prediction of a system’s short- and long-term performance, especially re- garding the expected values of recovery efficiency and storage capacity. Conclusion and Recommendation: In addition to the topics above, re- search is particularly needed, and should be conducted, in the following areas: • Hydrologic feasibility. This includes (1) lack of knowledge about stor- age zones and areas favorable for recharge for major aquifers in the United States; (2) limited understanding of how aquifer heterogeneity, scale effects, and other physical, chemical, and biological properties impact recharge rate and recovery efficiency of the MUS system; (3) lack of understanding of matrix behavior, especially fractured aquifers, during recharge versus withdrawal tests (e.g., expansion vs. compac- tion) to prevent or limit artificially induced deformation of the aquifer matrix; (4) need to develop of tools to analyze non-Darcian flow around recharge wells to avoid poor design of recharge wells; and (5) need for overall characterization, system recovery efficiency, optimum placement of monitoring wells, recharge and pumping impacts, and hy- draulic fracturing in an aquifer with dual porosity. • Impacts of MUS systems on surface water. How, in terms of both quan- tity and timing, might a surface spreading or well recharge facility af- fect the flow of neighboring streams? What would be the hydrologic, ecological, and legal consequences of this interaction between the MUS system and surface water? An integrated or system approach should be developed and employed for assessing such impacts. • Technology enhancement and methodology development for determin- ing hydrological properties of the aquifers and their impacts on per- formance of the MUS system. These include (1) surface and borehole geophysical methods to determine hydrological properties and the ex- tent of recharge water volumes during cycle testing; (2) optimization of cycle test design (frequency, duration, and intensity) to improve per- formance of MUS systems for various hydrological settings; (3) better conceptual models for delineation of storage zone and recovery zone; and (4) better understanding of non-Darcian flow near recharge wells through experimental study and field monitoring, and further develop- ment of theories and numerical models to assess the interaction of stored water (especially urban runoff) with native groundwater. 

100 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER REFERENCES Alley, W. M., T. E. Reilly, and O. L. Franke. 1999. Sustainability of Ground Wa- ter. U.S. Geological Survey Circular 1186. Anderson, M. P., and W. W. Woessner. 1992. Applied Groundwater Modeling: Simulation of Flow and Advective Transport. Academic Press. Annan, A. P. 2005. GPR methods for hydrogeological studies. Hydrogeophysics, Water Science and Technology Library 50 : 185-213. Arthur, J. D., H. A. R. Wood, A. E. Baker, J. R. Cichon, and G. L. Raines. 2007. Development and Implementation of a Bayesian-based aquifer vulnerability assessment in Florida. Natural Resources Research 16(2):93-107. Artimo, A., J. Mäkinen, R. C. Berg, C. C. Abert, and V.-P. Salonen. 2003. Three- dimensional geologic modeling and visualization of the Virttaankangas aq- uifer, southwestern Finland. Hydrogeology Journal 11(3):378-386. Artimo, A., S. Saraperä, and I. Ylander. 2005. The role of 3d geologic model- ing and database solutions in the Virttaankangas artificial recharge project, southwestern Finland. In Russell, H., R. C. Berg, and L. H. Thorleifson (convenors), Three-Dimensional Geologic Mapping for Groundwater Ap- plications: Workshop Extended Abstracts. Geological Survey of Canada Open File 5048. Ottawa: Geological Survey of Canada. ASCE (American Society of Civil Engineers). 1987. Groundwater Management. ASCE Manuals and Reports on Engineering Practice 40. Third Edition. Reston, VA: ASCE. ASCE 2001. Standard Guideline for Artificial Recharge of Ground Water. EWRI-ASCE 43-01. Reston, VA: ASCE. Bakalowicz, M. 2005. Karst groundwater: A challenge for new resources. Hydrogeology Journal 13:148-160. Bear, J. 1979. Hydraulics of Groundwater. New York: McGraw-Hill. Bear, J. 1988. Dynamics of Fluids in Porous Media. New York: Dover. Bögli, A., 1980, Karst Hydrology and Physical Speleology, Springer-Verlag, Berlin, 284 p. Boulding, J. R., and J. S. Gin. 2003. Practical Handbook of Soil, Vadose Zone, and Ground-Water Contamination. Boca Raton, FL: Lewis. Bouwer, H.. 2002. Artificial recharge of groundwater: hydrogeology and engi- neering. Hydrogeology Journal 10: 121-142. Bouwer, H., and R. C. Rice. 1976. A slug test method for determining hydraulic conductivity of unconfined aquifers with completely or partially penetrating wells. Water Resources Research 12(3):423-428. Bouwer, H., J. T. Back, and J. M. Oliver. 1999. Predicting infiltration and ground-water mounds for artificial recharge. Journal of Hydrologic Engi- neering 4(4):350-357. Boyle Engineering Incorporation, 1999. Concept Design of ASR Wellfield and Collection Facilities. Prepared for New Mexico- Texas Water Commission. http://www.nm-txwatercomm.org/library/ASR/guide.pdf. 

HYDROGEOLOGICAL CONSIDERATIONS 101 Bristow, C. S., and H. M. Jol. 2002. Ground Penetrating Radar in Sediments, Geological Socoety, London, Special Publications, 211. London, UK: Geo- logical Society London. Bristow, C.S., and H. M. Jol. 2003. An introduction to ground penetrating radar (GPR) in sediments. Pp. 1-7 in C. S. Bristow, and H. M. Jol (eds.) Ground Penetrating Radar in Sediments. Geological Society Special Publication 211. Brown, C. J. 2005. Planning Decision Framework for Brackish Water Aquifer, Storage and Recovery (ASR) Projects. Doctoral Dissertation. University of Florida. Brown, C. J., S. England, G. T. Stevens, H. P. Cheng, and E. Richardson. 2006. ASR regional study—benchscale modeling. U.S. Army Corps of Engineers. Technical Report. FL: USACE. Brown, C. J., S. Itani, and M. Zhang. 2005. Aquifer Storage & Recovery (ASR) Regional Study: A Scientific Evaluation of Potential Pressure Induced Con- straints and Changes in the Floridan Aquifer System and the Hawthorn Group. U.S. Army Corps of Engineers, Final Draft Report. Available online at http://www.evergladesplan.org/pm/projects/project_docs/pdp_asr_com- bined/ 021606_asr_comb_pressure_final.pdf . Accessed May 8, 2007. Charbonneau, R. J. 2000. Groundwater Hydraulics and Pollutant Transport. Up- per Saddle River, NJ: Prentice Hall. CH2M Hill. 1989. Construction and testing of the aquifer storage and recovery (ASR) demonstration project for Lake Okeechobee, Florida. Volume I - Engineering Report and Volume II – Appendixes A-I. Engineering report prepared for South Florida Water Management District. CH2M Hill. 1993. Boynton Beach aquifer storage and recovery system: Engi- neering report prepared for the City of Boynton Beach, Florida. Cserepes, L., and L. Lenkey. 2004. Forms of hydrothermal and hydraulic flow in a homogeneous unconfined aquifer. Geophysical Journal International 158:785-797. Delleur, J.W., 1998. The Handbook of Groundwater Engineering. Boca Raton, FL: CRC Press. de Marsily, G. 1986. Quantitative Hydrogeology. London: Academic. Dillon, P., and R. Molloy. 2006. Developing Aquifer Storage and Recovery (ASR) Opportunities in Melbourne. Technical Guidance for ASR, CSIRO Land and Water Science Report 4/06. Dillon, P., K. Barry, P. Pavelic, T. Sovich, A. Hutchinson, and G. Woodside. 2004. Talbert Gap Saline Intrusion Barrier, Orange County, California. In Final Report to AwwaRF, Project No. 2618, Volume 2, Chapter 3. Denver, CO: AwwaRF. Dobecki, T. L., S. B. Upchurch, and J. L. Hare. 2007 (in review). Geophysical Feasibility Study: Delineation of the Boundaries of an ASR-Injected Water “Bubble” with Controlled Source Audio-Frequency Magnetotelluric Profil- ing: Florida Geological Survey Open File Report No. 90. Doerfliger, N., P.-Y. Jeannin, and F. Zwahlen. 1999. Water vulnerability as- 

102 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER sessment in karst environments: A new method of defining protections areas using a multi-attribute approach and GIS tools (EPIK method). Environ- mental Geology 39(2):165-176. Domenico, P. A., and F. W. Schwartz. 1990. Physical and Chemical Hydrogeol- ogy. New York: Wiley. Driscoll, F. G. 1995. Groundwater and Well, Second ed. St. Paul, MN: USF Johnson Screens. Dudding, M., R. Evans, P. Dillon, and M. Molloy. 2006. Developing Aquifer Storage and recovery (ASR) opportunities in Melbourne: Report on Broad- scale Map of ASR Potential for Melbourne: Technical Report for Victorian Smart Water Fund. Victoria, Australia. Available online at: http://www.smartwater.com.au/projectdocs/project28/Broad%20Scale%20 Mapping%20Report%20of%20ASR%20potential%20in%20Melbourne.pdf. Accessed May 2007. Everett, M. E., and M. A. Meju. 2005. Near-surface controlled-source electro- magnetic induction: Background and recent advances. Hydrogeophysics, Water Science and Technology Library 50:157-183. Faunt, C., D. Sweetkind, and W. Belcher. 2004. Three-dimensional hydro- geologic framework model. Pp. 165-255 In W. Belcher (ed.) Death Valley Region Ground-Water Flow System, Nevada and California; Hydrogeologic Framework and Transient Ground-Water Flow Model. U.S. Geological Survey Scientific Investigations Report, SIR 2004-5205. Reston, Va: U.S. Geological Survey. Fetter, C. W. 2001. Applied Hydrogeology. Upper Saddle River, NJ: Prentice Hall. Fitts, C.R. 2002. Groundwater Science. Amsterdam: Elsevier Science. Fowler, L. C. 1981. Economic consequences of land surface subsidence. Journal of the Irrigation and Drainage Division 107(IR2) Proc. Paper 16335. Gale, I. N., A. T. Williams, I. Gaus, and H. K. Jones. 2002. ASR-UK: Elucidat- ing the hydrogeological issues associated with aquifer storage and recovery in the UK. BGS Report No. CR/02/156/N. London: U.K. Water Industry Research Limited. Galloway, D. L., D. R. Jones, and S. E. Ingebritsen. 1999. Land subsidence in the United States. U.S. Geological Survey Circular 1182. Gonzalez-Herrera, R., I. Sanchez-y-Pinto, and J. Gamboa-Vargas. 2002. Groundwater-flow modeling in the Yucatan karstic aquifer, Mexico. Hydro- geology Journal 10: 539-552. Guo, W., and C.D. Langevin. 2002. User’s Guide to SEAWAT: a Computer Pro- gram for Simulation of Three-Dimensional Variable-Density Ground-Water Flow. Open-File Report 01-434. Tallahassee, FL: U.S. Geological Survey Harvey, R. W. and J. N. Ryan. 2004. Use of PRD1 bacteriophage in groundwater viral transport, inactivation, and attachment studies. FEMS Microbiology Ecology 49:3-16 Helm, D. C. 1994a. Hydraulic forces that play a role in generating fissures at depth. Bulletin of the Association of Engineering Geologists 31(3): 293-304. 

HYDROGEOLOGICAL CONSIDERATIONS 103 Herczeg, A. L., K. J. Rattray, P. J. Dillon, P. Pavelic, and K. E. Barry. 2004. Geochemical processes during five years of aquifer storage recovery. Ground Water 42(3):438-445. Holzer, T. L. 1984. Ground failure induced by groundwater withdrawal from un- consolidated sediments. Pp. 67-105 in T. L. Holzer (ed.) Reviews in Engineer- ing Geology, Volume VI. Geol. Soc. of America. Huaming, G., and Y. Wang. 2004. Specific vulnerability assessment using the MLPL model in Datong city, Shanxi province, China. Environmental Geology 45:401-407. Hubbard, S. S., and Y. Rubin. 2005. Introduction to hydrogeophysics. Hydro- geophysics, Water Science and Technology Library 50:3-21 Ingebritsen, S. E., and W. E. Sanford. 1998. Groundwater in Geologic Proc- esses. Cambridge, UK: University Press. John, D. E., and J. B. Rose. 2005. Review of factors affecting microbial survival in groundwater. Environmental Science and Technology 39:7345. Johnson, C. D., J. W. Lane, Jr., and F. D. Day-Lewis. 2004. Time-series monitor- ing in fractured-rock aquifers. Pp. 295-307in Proceedings Fractured Rock Conference: State of the Science and Measuring Success in Remediation, September 13-15, Portland, ME. : National Ground Water Association. Johnston, R., 1997. Sources of water supplying pumpage from regional aquifer systems of the United States. Hydrogeology Journal 5(2):54-63. Kasteel, R., H. –J. Vogel, and K. Roth. 2000. From local hydraulic properties to effective transport in soil. European Journal of Soil Science 51:81-91. Kimbler, O.K., R. G. Kazmann, and W. R. Whitehead, 1975. Cycle storage of fresh water in saline aquifers, Louisiana Water Resources Research Insti- tute, Bulletin 10. Kipp, Kenneth L. 1997. Guide to the Revised Heat and Solute Transport Simula- tor: HST3D -- Version 2. Water-Resources Investigations Report 97-4157. Denver, CO:USGS. Knappe, A., P. Möller, P. Dulski, and A. Pekdeger. 2005. Positive gadolinium anomaly in surface water and ground water of the urban area Berlin, Ger- many. Chemie der Erde 65:167-189.Mack, T.J. 2003. Preliminary ground- water-flow model of a coastal bedrock-aquifer system, southeastern New Hampshire. In MODFLOW and More 2003: Understanding through model- ing, September 17–19. Denver, CO: International Ground Water Modeling Center. Konikow, L.F., L. L. August, and C. I. Voss. 2001. Effects of clay dispersion on aquifer storage and recovery in coastal aquifers. Tranport in Porous Media 43:45-64. Kopbr, M., S. Mares, and F. Paillet. 2005. Geophysical well logging: borehole geophysics for hydrogeological studies: principles and applications. Hydro- geophysics, Water Science and Technology Library 50(Springer): 291-331. Kopper, W. and D. Finlayson. 1981. Legal Aspects of Subsidence due to well pumping, Journal of the Irrigation and Drainage Division ASCE 107 IR2 Proc. Paper 163.31. 

104 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER Leonard, J. G. 1992. Port Phillip Region Groundwater Resources – Future use and Management. Victoria, Australia: Department of Water Resources. Li, J. 2000. A nonlinear elastic solution for subsidence due to ASR applications to multi-aquifer systems. Pp. 331-342 in L. Carbognin, G. Gambolati, and A. I. Johnson (eds.) Land Subsidence. Li, J., and Z. Sheng. 2002. Sensitivity analysis of nonlinear elastic solutions for land subsidence caused by ASR applications. Pp. 367-373 in P. J. Dilon (ed.) Management of Aquifer Recharge for Sustainability. Proceedings of the 4th International Symposium on Artificial Recharge of Groundwater at South Australia, Australia. Rotterdam, Netherlands: A. A. Balkema. Lin, H.C., D. R. Richards, G. T. Yeh, J. R. Cheng, H. P. Cheng, and N. L. Jones. 1997. FEMWATER: A Three-Dimensional Finite Element Computer Model for Simulating Density-Dependent Flow and Transport in Variably Saturated Media. Technical Report CHL-97-12. Waterways Experiment Station. Vicksburg, MS:U.S. Army Corps of Engineers. Lohman, S.W., et al. 1972 Definitions of Selected Ground-Water Terms— Revisions and Conceptual Refinements. U.S. Geological Survey Water Supply Paper 1988. Maliva, R. G., W. Guo, and T. M. Missimer. 2006. Aquifer storage and recov- ery: Recent hydrogeological advances and system performance. Water Envi- ronment Research 78(13):2428-2435. Mansuy N. 1998. Water Well Rehabilitation – A Practical Guide to Understand- ing Well Problems and Solutions. Mission Woods, KS: CRC Press. de Marsily, G. 1986. Quantitative Hydrogeology; Academic Press, p. 115. Maupin, M.A., and N.L. Barber. 2005. Estimated Withdrawals from Principal Aquifers in the United States, 2000. U.S. Geological Survey Circular 1279. Reston, VA: USGS. McDonald, M. G., and A. W. Harbaugh. 1988. A modular three-dimensional fi- nite difference ground water flow model. Techniques of Water-Resources Investigation Report v. 06-A1. 43. McNeill, J. D. 1980. Electrical Conductivity of Soils and Rocks, Geonics Ltd., Mississauga, Ont., Technical Note TN-5. Mercer, J. W., and C. R. Faust. 1981. Ground-Water Modeling. Columbus, OH: National Water Well Association. Merritt, M.L. 1985. Subsurface Storage of Freshwater in South Florida: A Digi- tal Model Analysis of Recoverability: U.S. Geological Survey Water-Supply Paper 2261. Missimer, T. M., W. Guo, C. W. Walker, and R. G. Maliva. 2002. Hydraulic and density considerations in the design of aquifer storage and recovery sys- tems. Florida Water Resources Journal 30–36. Moore, J. E. 2002. Field Hydrogeology—A Guide for Site Investigations and Report Preparation. Boca Raton, FL: Lewis Publishers. Muniz, A., and W. B. Ziegler. 1994. Aquifer storage and recovery in Southeast Florida. In Artificial Recharge of Groundwater, II: Proceeding of the Sec- 

HYDROGEOLOGICAL CONSIDERATIONS 105 ond International Symposium on Artificial Recharge of Groundwater. New York: ASCE. Murray, E. C., and G. Tredoux. 2002. Karkams borehole injection tests: Results from injection to a low-permeability fractured granitic aquifer. Pp. 301-304 in P. J. Dillon (ed.) Management of Aquifer Recharge for Sustainability. Rotterdam, Netherlands: A.A.Balkema. Nastev, M., M. M. Savard, P. Lapcevic, R. Lefebvre, and R. Martel. 2004. Hy- draulic properties and scale effects investigation in regional rock aquifers, south-western Quebec, Canada. Hydrogeology Journal 12:257-269. NRC (National Research Council). 2001. Aquifer Storage and Recovery in the Comprehensive Everglades Restoration Plan: A critique of the Pilot Projects and Related Plans for ASR in the Lake Okeechobee and Western Hillsboro Areas. Washington, DC:The National Academy Press. OCWD (Orange County Water District). 2006. Groundwater recharge opera- tions (Forebay Conservation). Available online at: http://www.ocwd.com/ html/recharge.htm. Accessed November 14, 2007. Paine, J. G., and B. R. S. Minty. 2005. Airborne hydrogeophysics. Water Sci- ence and Technology Library 50: 333-357. Pavelic, P., P. J. Dillon, and C. T. Simmons. 2002. Lumped parameter estima- tion of initial recovery efficiency during aquifer storage and recovery. In P. J. Dillon (ed.) Management of Aquifer Recharge for Sustainability, Pro- ceedings of the 4th International Symposium on Artificial Recharge (ISAR4), Adelaide, Sept. 22-26, 2002. Lisse: Swets & Zeitlinger. Pavelic, P., P. J. Dillon, and N. Robinson. 2004. Groundwater Modelling to As- sist Well-Field Design and Operation for the ASTR Trial at Salisbury, South Australia. Available online at http://www.clw.csiro.au/publications/technical 2004/tr27-04.pdf. Pavelic, P., P. J. Dillon, and C. T. Simmons. 2006. Multiscale characterization of a heterogeneous aquifer using an ASR operation. Ground Water 44(2):155- 164. DOI:10.1111/j.1745-6584.2005.00135.x Pride, S. 2005. Relationships between seismic and hydrological properties. Water Science and Technology Library 50: 253-290. Prothero, D. R., and F. Schwab. 2004. Sedimentary Geology: An Introduction to Sedimentary Rocks and Stratigraphy. Second Edition, New York: W. H. Freeman. Pyne, R. David. 2005. Aquifer Storage Recovery: A Guide to Groundwater Re- charge Through Wells, ASR Systems. Reese, R. S. 2002. Inventory and Review of Aquifer Storage and Recovery in Southern Florida, prepared as part of the U.S. Geological Survey Place- Based Studies Program. The U.S. Geological Survey Water Resources In- vestigation Report 02-4036. Reston, VA: U.S. Geological Survey. Reese, R. S., and C. A. Alvarez-Zarikian. 2007. Hydrology and Aquifer Storage and Recovery performance in the Upper Floridan Aquifer, Southern Florida: U.S. Geological Survey Scientific Investigations Report 2006-5239. Rollins, P. 2006. Improved Efficiency in Contaminated Groundwater Site Char- 

106 PROSPECTS FOR MANAGED UNDERGROUND STORAGE OF RECOVERABLE WATER acterization: A Minimally Invasive Approach. 13th Annual International Pe- troleum Environmental Conference, San Antonio, TX, October 17-20. Available online at http://ipec.utulsa.edu/Conf2006/Abstracts/Rollins_ 34.pdf. Accessed May 5, 2007. Ross, M., M. Parent, and R. Lefebvre. 2005. 3D geologic framework models for regional hydrogeology and land-use management: A case study from a Quaternary basin of southwestern Quebec, Canada. Hydrogeology Journal 13(5-6):690-707. Rossi, P., M. Aragno, N. Dörfliger, K. Kennedy and I. Müller. 1998. Bacterio- phages as surface and ground water tracers. Hydrology and Earth System Sciences 2:101. Schmidt, D. A., and R. Burgmann. 2003. Time-dependent land uplift and subsi- dence in the Santa Clara valley, California, from a large interferometric syn- thetic aperture radar data set. Jour. of Geophysical Research 108(B9):2416.DOI:10.1029/2002JB002267. Schwartz, F. W., and H. Zhang. 2003. Fundamentals of Groundwater. New York, NY: Wiley. Sheng, Z. 2005. An aquifer storage and recovery systems with reclaimed wastewater to preserve native groundwater resources in El Paso, Texas. Journal of Environmental Management 75:367-377. Sheng, Z., D.C. Helm, and J. Li. 2003. Mechanisms of earth fissuring caused by groundwater withdrawal. Journal of Environmental & Engineering Geo- science 9(4):313-324. Shrier, C. 2002. Survey and analysis of aquifer storage and recovery (ASR) systems and associated regulatory programs in the United States. Denver, CO: American Water Works Association. Soller, D., S. Price, R. Berg, and J. Kempton. 1998. A method for three- dimensional mapping. Pp. 79-84 in D. Soller (ed.) Digital Mapping Tech- niques ‘98 USGS Open-File Report 98-487. Reston, VA: U.S. Geological Survey. Spitz, K. and J. Moreno. 1996. A Practical Guide to Groundwater and Solute Transport Modeling. New York: John Wiley. Thorleifson, L. H., J. A. Berg, R. G. Tipping, Z. Malolepszy, K. L. Harris , B. A. Lusardi, D. R. Setterholm, and F. J. Anderson. 2005. 3D geological model- ing in support of ground water inventory in the Fargo-Moorhead region, Minnesota and North Dakota. Pp. 97-100 in H. Russell, R. Berg, and H. Thorleifson (convenors) Three-Dimensional Geologic Mapping for Groundwater Applications: Reston, VA: U.S. Geological Survey. USACE and SFWMD (U.S. Army Corps of Engineers/South Florida Water Management District), 1999. Central and Southern Florida comprehensive review study – final integrated feasibility report and programmatic envi- ronmental impact statement. U.S. Army Corps of Engineers, Jacksonville District, Jacksonville, Florida. USACE and SFWMD. 2004. Final Pilot Project Design Report/Environmental Impact Statement, Lake Okeechobee ASR Pilot Project, Hillsboro ASR Pi- 

HYDROGEOLOGICAL CONSIDERATIONS 107 lot Project and Caloosahatchee River ASR Pilot Project . Jacksonville, FL: USACE. Vacher, H. L., W. C. Hutchings, and D. A. Budd. 2006. Metaphors and models: The ASR bubble in the Floridan aquifer. Ground Water 44(2):144-154. DOI: 10.1111/j.1745-6584.2005.00114.x Van Overmeeren, R. A. 1998. Radar facies of unconsolidated sediments in the Netherlands: A radar stratigraphy interpretation method for hydrogeology. Journal of Applied Geophysics 40:1-18. Verstraeten, I. M., G. S. Fetterman, M. T. Meyer, T. D. Bullen, and S. K. Se- bree. 2005. Use of tracers and isotopes to evaluate vulnerability of water in domestic wells to septic waste. Ground Water Monitoring and Remediation 25:107-117. Viessman, W., and G. L. Lewis. 2003. Introduction to Hydrology. Upper Saddle River, NJ : Prentice Hall. Voss, C. I. and A. M. Provost. 2002. SUTRA, a Model for Saturated- Unsaturated Variable Density Ground-water Flow with Energy or Solute Transport. U.S. Geological Survey Open-File Report 02-4231. Ward, W. C., K. J. Cunningham, R. A. Renken, M. A. Wacker, and J. I. Carlson. 2003. Sequence-Stratigraphic Analysis of the Regional Observation Moni- toring Program (ROMP) 29A Test Corehole and Its Relation to Carbonate Porosity and Regional Transmissivity in the Floridan Aquifer System, High- lands County, Florida. Prepared as part of the Comprehensive Everglades Restoration Program, U.S. Geological Survey Open-File Report 03-201. Reston, VA: U.S. Geological Survey. WDNR (Wisconsin Department of Natural Resources) Aquifer Storage Recov- ery Technical Advisory Group, 2002. A Review of Aquifer Storage Recov- ery Techniques. Wimpenny J. W. T., N. Cotton, and M. Statham. 1972. Microbes as tracers of water movement. Water Research 6:731-739. Yeh, G.T., H. P. Cheng, J. R. Cheng, and H. C. Lin. 1990. A numerical model to simulate water flow and contaminant and sediment transport in WAterSHed systems of 1-D stream-river network, 2-D overland regime, and 3-D subsur- face media (WASH123D: Version 2.0), Report CHL-98-19 Waterway Ex- perimental Station, U. S. Army Corps of Engineers, Vicksburg, MS. Zheng, C., and P. P. Wang. 1999. MT3DMS: A Modular Three-Dimensional Multi- Species Model for Simulation of Advection, Dispersion and Chemi- cal Reactions of Contaminants in Groundwater Systems: Documentation and User’s guide. SERDP-99-1. U.S. Army Engineer Research and Devel- opment Center, Vicksburg, MS. Zijl, W., and M. Nawalany. 1993. Natural Groundwater Flow. Boca Raton, FL: CRC Press.