Table of Contents

Introduction

Part I

Appendix A

Appendix B

Appendix C

Part II A and B

Part II B and C

Part III

Part IV

References

Index

Summary

Part I. Understanding Ground-water Recharge

Summary of Part I

This part attempts to establish a hydrogeological framework for the understanding of natural ground-water recharge processes in relation to climate, landform, geology, and biotic factors. It begins with the concepts of ground-water flow systems, which form the basis for comprehending recharge processes. This work then concentrates on the sources and mechanisms of ground-water recharge, and stresses the importance of developing correct conceptualizations of recharge. A variety of recharge estimation methodologies are then outlined, with an emphasis on minimizing uncertainty. This contribution then discusses developing predictive relationships for recharge based on the major recharge-influencing factors, and into regionalizing point-recharge data. A discussion of difficulties that face the field of recharge assessment follows with recommendations to minimize these difficulties.

Although various well-established methods for the quantitative estimation of recharge exist, few can be applied successfully in the field. All methods are characterized by large uncertainties. When estimating ground-water recharge, proceeding from a good conceptualization of different recharge mechanisms and their importance in the study area is essential. Besides this conceptualization, the objectives of the study, available data and resources, and possibilities of obtaining supplementary data should guide the choice of recharge-estimation methods. A key to deciding on a recharge-estimation methodology is related to the spatial and temporal scale of interest. If the major concern is obtaining good recharge estimates over a limited area, then the need for detailed information is evident.

However, for regional studies small-scale variability in local recharge ceases to be a major problem. In addition, the inherent temporal variability of recharge has important implications for the measurement techniques adopted. Different measurement techniques provide recharge estimates with different temporal scales. For example, in arid and semiarid areas where deep drainage fluxes are low and water tables are deep, interpreting ground-water hydrographs and water-table rises may be misleading for estimating rates of ground-water recharge; chemical and isotopic methods are likely to be more successful than physical methods in such cases. A recharge-related glossary is presented as Appendix C.

1. Introduction and Terminology

The endless circulation of water as it moves in its various phases through the atmosphere, to the earth, over and through the land, to the ocean, and back to the atmosphere is known as the hydrologic cycle. This cycle is powered by the sun, and, through phase changes of water (i.e., evaporation and condensation) involving storage and release of latent heat, it affects the global circulation of both the atmosphere and oceans, and hence is instrumental in shaping weather and climate. The efficiency of water as a solvent makes geochemistry an intimate part of the hydrologic cycle; all water-soluble elements follow this cycle at least partially. Thus, the hydrologic cycle is the integrating process for the fluxes of water, energy, and the chemical elements. This cycle is the foundation of hydrologic science and occurs over a wide range of space and time scales.

FIGURE I–1—Schematic representation of the hydrologic cycle (from Freeze, 1974).

Figure I–1 illustrates different parts of the land-based portion of the hydrologic cycle that affect an individual watershed or catchment (Freeze and Cherry, 1979). Water enters the hydrologic system as precipitation, in the form of rainfall or snowmelt. Water leaves the system as streamflow or runoff, and as evapotranspiration, a combination of evaporation from open bodies of water, evaporation from soil surfaces, and transpiration from the soil by plants. Precipitation is delivered to streams on the land surface as overland flow to tributary channels and in the subsurface as interflow or lateral subsurface flow and baseflow following infiltration into the soil.

A portion of the infiltrated water enters the ground water or aquifer system by passing through the vadose or unsaturated zone, and it exits to the atmosphere, surface water, or to plants. As figure I-1 shows, the flowlines deliver ground water from the highlands towards the valleys or from the recharge areas to the discharge areas. As figure I-1 also shows, in a recharge area there is a component to the direction of ground-water flow that is downward. Ground-water recharge is the entry to the saturated zone of water made available at the water-table surface. Conversely, in a discharge area there is a component to the direction of ground-water flow that is upward (fig. I-1). Ground-water discharge is the removal of water from the saturated zone across the water-table surface. The patterns of ground-water flow from the recharge to the discharge areas form ground-water flow systems, which constitute the framework for understanding recharge processes. Therefore, ground-water flow systems are examined next.

2. Ground-water Flow Systems

The route ground water takes to a discharge point is known as a flow path. A set of flow paths with common recharge and discharge areas is termed a ground-water flow system. The three-dimensional closed system that contains the entire flow paths followed by all water recharging the ground-water system has been termed a ground-water basin (Freeze and Witherspoon, 1967). Ground water possesses energy mainly by virtue of its elevation (elevation or gravitational head) and of its pressure (pressure head). Ground water also can possess kinetic energy by virtue of its movement, but usually this energy is negligibly small because of ground water’s low velocities. Ground water moves from regions of higher energy to regions of lower energy. A measure of ground water’s energy is the level at which the water stands in a borehole drilled into an aquifer and measured with reference to an (arbitrary) reference level or datum such as sea level. This height that water stands above a reference datum is called hydraulic head or simply head. The hydraulic head, for most practical purposes, is composed of the sum of the pressure head and gravitational or elevation head. Both of these component forms of energy (i.e. elevation energy and pressure energy) are known as potential energy. The change in hydraulic head over a certain (arbitrary) distance along the ground-water flow path is called hydraulic gradient or head gradient and constitutes the driving force for ground-water movement. According to Darcy’s Law, which describes the flow of ground water through an aquifer, the ground-water flow rate is directly proportional to the cross sectional area through which flow is occurring, and directly proportional to the hydraulic gradient. Gravity due to elevation differences is the predominant driving force in ground-water movement. Under natural conditions, ground water moves “downhill” until it reaches the land surface, such as at a spring, or the root zone, where it is evapotranspired to the atmosphere.

Therefore, ground water moves from interstream (higher) areas toward streams or the coast (lower areas). Except for minor surface irregularities, the slope of the land surface also is toward streams or the coast. The depth to the water table is greater along the divide between streams than it is beneath the floodplain. In effect, the water table usually is a subdued replica of the land surface.

A ground-water flow pattern is controlled by the configuration of the water table, and by the distribution of hydraulic conductivity in the rocks. The water table, in turn, is affected by the topography and is controlled by the climate. The flow pattern is therefore a function of topography, geology, and climate. These three parameters have been collectively termed as the hydrogeologic environment (Toth, 1970). In addition, biotic influences affect most aspects of the hydrologic cycle, including ground water. Vegetation, for example, regulates the rate at which a land surface returns water vapor to the atmosphere, and humans alter nearly all aspects of water on land.

Based on their relative position in space, three distinct types of flow systems have been recognized (Toth, 1963, 1999; right-hand side of fig. I–2): (1) a local system, which has its recharge area at a topographic high and its discharge area at the immediately adjacent topographic low; (2) an intermediate system, which is characterized by one or more topographic highs and lows located between its recharge and discharge areas; and (3) a regional system, which has its recharge area at the major topographic high and its discharge area at the bottom of the basin. Regional flow systems are at the top of this hierarchical organization; all other flow systems are nested within them.

FIGURE I–2—Effects and manifestations of gravity-driven flow in a regionally unconfined drainage basin (adapted from Toth, 1999).

Based on a comparative study of variations in selected geometric parameters, such as depth to impermeable basement, slope of the valley flanks, and local relief, the conditions under which local, intermediate, and regional systems may develop were elucidated (Toth, 1963). If local relief is negligible, and there is a general slope of topography, only regional systems will develop. Because no extensive unconfined regional system can exist across valleys of large rivers or highly elevated watersheds, pronounced local relief generally is an indicator of a local system. The greater the relief, the deeper the local systems that develop. Under extended flat areas unmarked by local relief, neither regional nor local systems can develop. Waterlogged areas may develop, and the ground water may be highly mineralized from concentrations of salts.

The recognition that, in topography-controlled flow regimes, ground water moves in systems of predictable patterns, and that various identifiable natural phenomena are regularly associated with different segments of the flow systems, was not made until the 1960’s when the system-nature of ground-water flow was first understood (Toth, 1962, 1963; Freeze and Witherspoon, 1967). This recognition of the system-nature of subsurface water flow has provided a unifying theoretical background for the study and understanding of a wide range of natural processes and phenomena and thus has shown flowing ground water to be a general geologic agent (Toth, 1999).

A schematic overview of ground-water flow distribution and some typical hydrogeologic conditions and natural phenomena associated with it in a gravity-flow environment is presented in figure I-2 (Toth, 1999). On the left side of the figure, a single flow system is shown in a region with insignificant local relief; on the right side, a hierarchical set of local, intermediate, and regional flow systems is depicted in a region of composite topography. Each flow system has an area of recharge, an area of throughflow, and one of discharge. In the recharge areas, the hydraulic heads, representing the water’s potential energy, are relatively high and decrease with increasing depth; water flow is downward and divergent. In discharge areas, the energy and flow conditions are reversed: hydraulic heads are low and increase downward, resulting in ascending and converging water flow. In the areas of throughflow, the water’s potential energy is largely invariant with depth (the isolines of hydraulic heads are subvertical) and, consequently, flow is chiefly lateral. The flow systems operate as conveyor belts with the flow serving as the mechanism for mobilization, transport (distribution), and accumulation of mass and energy thus effectively interacting with their ambient environment (Toth, 1999).

3. Flow System Extensions

Studying flow systems in ground-water basins may help gain an understanding of the interrelations between the processes of infiltration and recharge at topographically high parts of the basin and of ground-water discharge through evapotranspiration and baseflow. For example, at least some of the water derived from precipitation that enters the ground in recharge areas will be transmitted to distant discharge points and thus cause a relative moisture deficiency in soils overlying recharge areas. Water that enters the ground in discharge areas may not overcome the upward potential gradient, and therefore becomes subject to evapotranspiration in the vicinity of its point of entry. Water input to saturated discharge areas generates overland flow, but in unsaturated discharge areas infiltrating water and upflowing ground water are diverted laterally through superficial layers of high hydraulic conductivity. Further, the ramifications of anthropogenic activities in discharge areas are immediately apparent. Some of these include (Domenico, 1972): (1) water-logging problems associated with surface-water irrigation of lowlands; (2) water-logging problems associated with destruction of phreatophytes, or plants discharging shallow ground water; and (3) pollution of shallow ground waters from gravity-operated sewage and waste-disposal systems located in valley bottoms in semiarid basins where surface water is inadequate for dilution.

The spatial distribution of flow systems also will influence the intensity of natural ground-water discharge. From figure I-2, the main stream of a basin may receive ground water from the area immediately within the nearest topographic high and possibly from more distant areas. If baseflow calculations are used as indicators of average recharge, significant error may be introduced in that baseflow may represent only a small part of the total discharge occurring downgradient from the line separating the areas of discharge from the recharge areas.

In ground-water hydrology today, the system concept is fundamental to thinking about a ground-water problem. System thinking is vital to the understanding of practical problems, such as ground-water contamination from point sources, or the impact of a structure such as a dam, waste-disposal facility, or gravel pit. Many such studies suffer irreparably from the failure to place the local site in the context of the larger ground-water system of which the site is only a small part.

4. Sources and Mechanisms of Recharge

The sources of recharge to a ground-water system include both natural and human-induced phenomena. Natural sources include recharge from precipitation, lakes, ponds and rivers (including perennial, seasonal, and ephemeral flows), and from other aquifers. Human-induced sources of recharge include irrigation losses, both from canals and fields, leaking water mains, sewers, septic tanks, and over-irrigation of parks, gardens, and other public amenities. Recharge from these sources has been classified as direct recharge from percolation of precipitation and indirect recharge from runoff ponding. Other classifications include localized or focused recharge, preferential recharge, induced recharge, mountain front recharge, and others (Lerner et al., 1990; Simmers, 1997).

Direct or diffuse recharge is defined as water added to the ground-water reservoir in excess of soil-moisture deficits and evapotranspiration, by direct vertical percolation of precipitation through the unsaturated zone—that is, recharge below the point of impact of the precipitation. This mode of recharge is spatially distributed (diffuse) and results from widespread percolation through the entire vadose zone. It is typical of humid climates because frequent, regular precipitation maintains a high water content in the soil, so that there is little additional storage capacity in the vadose zone; thus, infiltration can be routed quickly through the vadose zone to the saturated zone. This recharge raises the water table, which leads to increased streamflow. Thus, in humid climates, flowing perennial streams are typically ground-water discharge areas sustained by diffuse recharge in the basin.

Indirect recharge results from percolation to the water table following runoff and localization in joints, as ponding in low-lying areas and lakes, or through the beds of surface-water courses. Two distinct categories of indirect recharge are evident: (1) that associated with surface-water courses, and (2) a localized or focused form resulting from horizontal surface concentration of water in the absence of well-defined channels, such as recharge through sloughs, potholes, and playas. Recharge through such topographic depressions, which are common in the Canadian prairies and Great Plains of the United States, also is known as depression-focused recharge, and occurs where surface runoff or lateral flow of subsurface moisture accumulates within or beneath such depressions on the landscape. Thus knowledge of lateral subsurface flow processes becomes important in understanding recharge processes. In arid and semi-arid regions, localized and indirect recharge are often the most important sources of natural recharge.

Percolation to the water table from streambeds takes two forms, depending on whether there is a saturated connection between the stream and the water table. Where no connection exists (fig. I–3a), a situation typical of arid zones where water tables are generally deep, water moves downward from the streambed to the water table, forming a ground-water mound which then dissipates laterally away from the stream. As long as the mound is recharged by unsaturated flow, there is no hydraulic connection between the ground water and the streamflow, in the sense that the recharge rate is almost unaffected by the ground-water levels. Yet, even when the unsaturated condition is present, the stream and aquifer may in fact be hydraulically connected in the sense that further lowering of the regional water table could increase channel losses. At some critical depth to the water table, however, further lowering has no influence on channel losses. At this depth, which depends mostly on soil properties and water stage in the channel, the aquifer becomes hydraulically disconnected from the stream. If the distance from the water table to the stream stage is greater than approximately twice the stream width, the seepage begins to rapidly approach the maximum seepage for an infinitely deep water table. The parameters determining the recharge process are the width, depth, and duration of streamflow, and the hydraulic characteristics of the local material in and below the streambed.

Figure I-3—Recharge from streambeds (a) with hydraulic connection, and (b) with no hydraulic connection.

In less arid areas, water-table levels tend to rise closer to the streambed. In these situations, a hydraulic connection will usually exist between the stream and the ground water (fig. I–3b), and the recharge rate will decrease as the water table rises. The recharge process will be dominated by horizontal rather than vertical flow, and will have a much shorter turnover or transit time than when there is no hydraulic connection. In these less arid environments, recharge from general catchment percolation also is likely, and the mix between the two mechanisms may be hard to predict.

Mountain-front recharge typically involves complex processes of unsaturated and saturated flow in fractured rocks, as well as infiltration along channels flowing across alluvial fans. On a large scale, mountain-front recharge through fractured bedrock is primarily a diffuse recharge process, whereas infiltration from mountain streams is considered a localized recharge process. Vertical leakage across low-permeability strata and underflow from adjacent aquifers (interaquifer flows) can be important sources of recharge but typically they do not involve the vadose zone.

In areas where the potential recharge rate exceeds the rate at which water can flow laterally through the aquifer, the aquifer becomes overfull and available recharge is rejected, a condition known as rejected recharge. In this situation, ground-water pumping in recharge areas can increase the rate of underground flow from the area and more water could be drawn into the aquifer as induced recharge.

Two different flow mechanisms, called capillary and viscous flow, drive potential ground-water recharge through the vadose zone (Hendrickx and Walker, 1997). Capillary flow takes place in pores with a diameter less than approximately 3 mm in which capillary forces, together with gravity, determine the flow process. A porous medium in which capillary forces are dominant behaves like a sponge; i.e. no free drainage occurs even at high water content, and capillary rise causes water to move upwards against the pull of gravity. The capillary flow process normally leads to stable wetting fronts, but sometimes unstable wetting fronts form that are characterized by fingered flow. (Fingered flow is unstable flow whereby the percolating water may concentrate at certain points to break into the sublayer in the form of finger-like or tongue-like protrusions.) Theoretical and experimental research results demonstrate that dry sandy soils are prime sites for the occurrence of unstable wetting fronts (i.e., boundaries between the wetted and dry regions of soil during infiltration), which may be expected in dune fields that often provide a large portion of the recharge under semi-arid conditions. This type of flow also occurs in the transition of percolating water from a fine-textured top layer to a coarser-textured sublayer. Unfortunately, it is not yet possible to quantify the effects of fingered flow on recharge rates (Hendrickx and Walker, 1997).

Macropore flow occurs in pores with a diameter or width larger than 3 mm, such as cracks in clay soils, rock fractures, fissures in sediments, solution channels, worm holes, and old root channels. In macropore flow, the effects of capillarity are no longer felt, and the flow process is dominated by viscous forces and gravity. Flow through macropores also is known as preferential or bypass flow, and the resulting recharge is called preferential recharge, which preferentially takes place through such macropores, as opposed to diffuse recharge, which takes place through the entire vadose porous medium. The velocity with which water moves from the soil surface to the water table often is several orders of magnitude higher through macropores than through the soil matrix. Saturated flow through macropores can be quantified using Poiseuille’s equation as opposed to Darcy’s equation for diffuse flow. However, capillary and macropore flow frequently occur simultaneously within the same soil mass without the presence of clearly defined macropores. The depth to which preferential flow is effective depends on the nature and connectivity of the macropores or preferred pathways, but rarely are they effective beyond the root-zone depth of approximately 2 m (Hendrickx and Walker, 1997).

The process of macropore flow, shown in fig. I–4, is somewhat similar to localized recharge, albeit on a much smaller scale, because horizontal water movement is required. When the overall water input from precipitation or irrigation, q*(t), exceeds the infiltration capacity of the soil, i(t), a horizontal overland flow, o(t), is generated that causes a water flux into the macropores, q_s(0,t). This flux causes water content inside the macropore, w(z,t), to increase. A fraction of the water, r, that occupies a macropore at a given depth will be absorbed by the soil matrix through the macropore walls whereas the remainder will percolate downwards into the macropore, q(z,t). When the infiltration rate, i(t), decreases with time and with increasing antecedent soil-water content, the opportunity for overland flow, o(t), and macropore flow, q_s(0,t), increases.

FIGURE I–4—Schematic representation of the fluxes involved during infiltration into a macroporous soil. See text for explanation of symbols (from Germann and Beven, 1985).

5. Conceptual Models of Recharge (Hatton, 1998)

The key to successful hydrological measurement and modeling is the appropriate conceptualization of the system of interest. The conceptual model includes the recognition of important hydrological processes, path-ways, boundary conditions, spatial and temporal limits, inputs and outputs, and constraints. If the conceptual model is wrong to start with, then recharge estimates based on this model will be unreliable (Hatton, 1998). For example, at the plot scale, important elements of a conceptual water-balance model aimed at recharge prediction might be (1) the pattern and amount of evaporation with respect to land cover; (2) the importance of overland flow; (3) the existence of any lateral throughflow; (4) the datum in the profile beyond which drainage will become ground-water recharge; (5) the transience and frequency of recharge events; and (6) the hydraulic pathway(s) that water may take through the profile.

At the catchment scale, the potential complexity of the correct conceptual model increases dramatically, for it includes not only all the considerations of the plot-scale recharge phenomena, but also the distribution of these phenomena in space, as well as the interaction of the water-balance components of adjacent plots (Hatton, 1998). For example, overland flow or shallow through-flow can become ground-water recharge downslope. The complexity of lateral flow systems, and their definition, becomes paramount at the catchment scale. Important considerations include (1) the presence of any confining bed(s), their depth, hydraulic conductivity and distribution across the catchment; (2) the hydraulic-head surface of ground-water system(s) and the degree of confinement of aquifers across the catchment; and (3) the geomorphic and geologic features associated with the discharge of ground water, which define, locate, and control saturated areas within the catchment.

Successful estimation of ground-water recharge depends on first identifying the probable flow mechanisms and important features influencing recharge for a given locality, because it cannot be assumed that a procedure successfully developed for one area will prove equally reliable for another. Thus, in each case involving recharge-estimation modeling, conceptual models must be based on local data and experience.

In summary, the vital aspects of a conceptual model of catchment-recharge processes must consider (1) what parts of the landscape contribute to ground-water recharge; (2) how these areas change with time; (3) if the topographic catchment is the same as the ground-water catchment; (4) what controls recharge rates from place to place; and (5) the importance of lateral redistribution of runoff and shallow throughflow to recharge downslope (Hatton, 1998).

If appropriate, one can use a mean recharge rate over the entire catchment, or at least over that portion of the catchment subject to recharge. Expected rates or changes in the rate of recharge, however estimated or modeled, can be applied uniformly over this area. In other words, the recharge across the landscape can be treated one-dimensionally. The assumption here is that the lateral redistribution of water in the catchment takes place only after the recharge reaches the ground-water table, and that the subsequent discharge of this ground water does not in turn change the area subject to recharge (that is, the discharge area does not grow significantly in size). The conditions where such an approach might be appropriate are areas with deep, uniformly permeable soils, deep ground water, and a very low topographic (hydraulic) gradient (Hatton, 1998).

However, most catchments are heterogeneous in their topography, soil, geology, and land cover. To model catchment recharge in these systems, the spatial pattern of these influences on recharge must be taken into account. There are two basic ways to approach this problem, depending on the nature of the recharge modeling to be undertaken. In the first general approach, the catchment is broken up into land units in which recharge can be expected to respond similarly to climate inferences on recharge and its relation to land use; these units are then distributed spatially on this basis. In the second approach, the individual controls on recharge are distributed independently and serve as input into a spatially explicit water-balance model yielding recharge (Hatton, 1998).

In either approach, an appreciation and understanding of scaling hydrologic parameters is essential (Hatton, 1998). As one looks at larger areas of the landscape and thus incorporates natural heterogeneity into the modeling, the parameter values used to represent hydrologic processes often change. For example, the saturated hydraulic conductivity of a soil profile will be different from the inferred conductivity of a hillslope, which in turn will be different from the inferred conductivity of an entire catchment, even if climate, geology, vegetation, and other variables are held constant. Thus, it is not reasonable to assume scale invariance in model parameters as one moves from point measurements to entire catchments. This even holds true for the mean of many point estimates of model parameters. The search for scale-invariant model representations of hydrologic phenomena for catchments has yet to yield a solution. Indeed, it is unlikely that any general scaling theory can be developed because of the dependence of hydrological systems on historic and geological perturbations (Beven, 1995). In most watershed models, equations representing hydrologic processes across scales usually involve “effective” parameters—that is, the parameter values change with scale.

Finally, in characterizing ground-water recharge, a distinction between potential and actual recharge needs to be made. Potential recharge is soil water that percolates below the root zone, whereas actual recharge is soil water that reaches the aquifer. Most potential recharge water will be stored in the vadose zone at negative pressures (suctions) and is not available for exploitation. In addition, it still may be lost at a later time by an increase in vegetation rooting depth, capillary rise, or upward vapor transport. Conversely, actual recharge is the amount of water that indeed reaches the water table and can be pumped.

6. Methodologies for Recharge Estimation

A number of methodologies are used to estimate recharge. These can be classified as (1) direct or indirect; (2) physical, chemical, or isotopic; (3) methods based on the analysis of inflow, outflow, or aquifer response; (4) methods based on the unsaturated or saturated zones; and (5) methods based on numerical modeling of ground-water flow, soil-water flow, both soil- and ground-water flows, or modeling of the hydrologic balance at plot, field, or watershed scales. Additional classifications also exist. Within each methodology, a number of estimation techniques are available (see also Scanlon, Healy, and Cook, 2002, for a recent, comprehensive review of recharge methodologies, as well as other related articles in the special theme issue of the Hydrogeology Journal—v. 10, no 1, February 2002—on ground-water recharge). Here we lump these methodologies into physical methods and tracer methods, and describe them in Appendices A and B, respectively.

7. Accuracy of Recharge Estimates

Because recharge is not easy to measure directly, estimates of it are prone to large errors. Four common types of error are discussed below (Lerner et al., 1990). The most serious and most common type of error is an error in the conceptual model. It arises when the recharge process is not fully understood, or when too many simplifying assumptions are made. For example, in a given study it may be assumed that excess irrigation water applied in parks becomes recharge, whereas in reality, a low-conductivity layer causes perching and horizontal flow to a surface drain. Or a monthly time step might be used for a soil-moisture budgeting model in a semi-arid area, resulting in zero recharge being estimated, whereas occasional short wet spells overcome soil-moisture deficits to cause some recharge.

Another common error relates to temporal and spatial variability. Most recharge processes are nonlinear in relation to time. For example, a low-intensity rainfall might cause no recharge because of a high rate of evapotranspiration, whereas the same amount over a shorter time period might be sufficient to saturate the soil and cause recharge. Thus, errors will arise if temporal variations are ignored—for example by using monthly, annual, or long-term average data. Recharge also is nonlinear with respect to spatial variations of inputs and physical properties of soils and aquifers.

Measurement error is another type of error and has to do with the equipment used to make measurements. This kind of error generally is not overlooked. The final type of error, calculation errors, can be avoided by care and checking, especially of units. A particularly difficult type of error can occur with numerical computer models unless they are rigorously tested under a wide range of conditions.

Error analysis or sensitivity analysis can show which variables in an equation lead to the highest errors, and special effort can be concentrated on obtaining the most accurate estimates for these. However, this approach will not help if the conceptual model is wrong. More than one method of estimation using other data should be used to provide an independent check. Table I-1 summarizes six different methods for estimating natural ground-water recharge from precipitation. These methods were tested and compared in a sandy till area in southeastern Sweden (Johannson, 1988). As mentioned earlier, the desired resolution in time is an important criterion in method selection. The interest may vary from estimation of instantaneous recharge to long-time averages. Table I-2 classifies the methods indicated in table I-1 according to the time and areal resolution. Clearly, comparative studies, in which several methods are applied to minimize the uncertainty in estimations of ground-water recharge, are needed.

TABLE I–1—Comparison of six different methods for estimation for ground-water recharge that were tested in southeastern Sweden^a.

TABLE I–2—Classificaton of the applied methods for estimation of ground-water recharge (shown in table I–1) according to the resolution in time of their results. A dashed line indicates point values of ground-water recharge and a solid line indicates an areally integrated value^a.

8. Factors Influencing Recharge, Predictive Relationships, and Recharge Regionalization

8.1 Factors Influencing Recharge

The key environmental factors controlling recharge are climate, soils and geology, vegetation and land use, topography, and depth to water table. The water-balance equation is commonly used to quantify the components of the hydrologic cycle:

P + I = RO + D +ET + S . . . . . . . . . . .(I–1)

where P is precipitation, I is irrigation, RO is surface runoff, D is deep drainage and recharge, ET is evapotranspiration, and S is water stored in the soil. Under nonirrigated conditions, where I = 0, the left-hand side of equation (I–1) is fixed in the sense that it is outside human control. Hence, a decrease in any of the variables on the right-hand side forces an equal increase in the other terms to maintain the equality (i.e., the water balance). For example, a decrease in surface runoff, RO (e.g., as a result of increased infiltration through better tillage practices) may increase the amount stored in the soil profile, S; an increase in S would tend to increase deep drainage (and recharge), D, and evapotranspiration, ET. Clearly, to understand and estimate aquifer recharge, a basic understanding of this water cycling is needed.

Most direct measurements of hydrologic variables related to recharge assessments provide only point measurements or estimates and do not integrate such variables (shown in equation (I–1) in relation to space and time. Recharge varies across the landscape because the aforementioned controlling factors vary, but finding ways to estimate and predict this spatial and temporal variability and to regionalize point measurements remains a major problem in recharge assessments (Sophocleous, 1992).

A daily water-balance modeling analysis (based on equation (I–1) for the Rattlesnake Creek basin in south-central Kansas, an approximately 1,300-mi² semiarid to subhumid agricultural basin, demonstrated that soil factors, plant cover, and land-use practice are important controls on ground-water recharge (Sophocleous and McAllister, 1987, 1990). The importance of each of these factors is detailed below. Although such results are highlighted for the case study of the aforementioned Kansas basin, they are general enough to be valid for any agricultural plain region of similar climate in the world.

Soil factors, such as the available-water capacity (AWC) of soil profiles, exert a dominating influence. AWC is the volume of water available to plants if the soil were at field capacity, i.e., the moisture content held by soil against the pull of gravity after the excess water has drained out of a saturated or nearly saturated soil. The AWC of each soil determines the maximum limit of actual evapotranspiration (ET) that can be extracted without additional infiltration and the maximum soil-moisture deficit possible. (Soil-moisture deficit is an estimate of the degree to which soil moisture content has dropped below field capacity.) Thus, given the same hydroclimatic conditions and crop cover, a soil with a relatively low AWC will exhibit a relatively small soil-water deficit, and smaller amounts of water will be lost through ET compared to losses from a soil with higher awc (figs. I–5 and I–6). The AWC also determines the amount of water that can infiltrate into the soil before deep drainage occurs. The AWC acts as a buffer for infiltrating water. Given the same initial-moisture conditions, a soil with higher AWC can absorb more infiltrating water than low-AWC soils. Thus, deep drainage decreases with increasing AWC (fig. I–7).

FIGURE I–5—Soil-moisture deficit versus available-water capacity for grassland, dryland, and irrigated cropland for the upper two-thirds of the Rattlesnake Creek basin in south-central Kansas (from Sophocleous and McAllister, 1987).

The water balance also is greatly influenced by the plant cover and land-use practice. The largest element of the water balance in equation (I–1) in the south-central Kansas basin under study is the ET component, as can be seen for native grassland in fig. I–8. The impact of vegetation on the hydrologic balance is complex and depends on factors such as crop coefficients (i.e., empirically determined coefficients relating potential ET to crop ET), growth stages, rooting depths, soil, water, and climatic conditions as used in soil-water balance simulation model.

The crop coefficients vary with the stage of crop growth. Mature plants have greater ability to extract soil moisture from all soil horizons and, thus, have larger crop coefficient than young plants. The crop with the largest crop coefficients employed in the soil-water balance model in south-central Kansas is alfalfa. In addition, alfalfa is continuously grown from one year to the next with multiple harvests without replanting or land fallowing. Prairie grasses have the next highest overall crop coefficients in the study region with a long growing season. All other crops have lower crop coefficients and are grown only part of the year.

FIGURE I–6—Actual evapotranspiration versus available-water capacity for grassland, dryland, and irrigated cropland for the upper two-thirds of the Rattlesnake Creek basin in south-central Kansas (from Sophocleous and McAllister, 1987).

FIGURE I–7—Deep drainage versus available-water capacity for grassland, dryland, and irrigated cropland for the lower one-third of the Rattlesnake Creek basin in south-central Kansas (from Sophocleous and McAllister, 1987).

FIGURE I–8—Grassland water-balance components for (b) the lower one-third and (a) the rest of the Rattlesnake Creek basin in south-central Kansas (from Sophocleous and McAllister, 1987).

The greatest deep drainage occurred in irrigated wheat fields, mainly because of the shallow rooting depth of wheat, while the lowest values occurred in alfalfa and grassland acreages (fig. I–9). Decreased amounts of deep drainage in the northeastern portion of the Kansas study basin (fig. I–8), where precipitation is lower, are from grasslands, indicating the dominant effect precipitation and vegetation exert on deep drainage.

In general, the effect of irrigation is to significantly increase evapotranspiration and also increase deep drainage, as can be seen in figs. I–6 and I–7, respectively. For summer crops, such as sorghum and soybeans, as well as alfalfa, most of the irrigation amounts are spent in evapotranspiration activities, with negligible amounts for deep drainage. The effects of grasslands are reduced deep drainage and runoff, and increased soil-moisture deficits compared to cropland acreages (figs. I–5, I–6, and I–7). From the areal distribution of the various components of the water balance, it was concluded that single average values of hydrologic variables used in management practices are not realistic, and that a spatial-discrimination attempt in managing water resources is needed (Sophocleous and McAllister, 1987, 1990).

A computerized water-balance procedure such as the one used in the Rattlesnake Creek basin study can be used to demonstrate and predict human and natural impacts on the hydrologic cycle (Sophocleous and McAllister, 1987, 1990). The hydrologic effects of vegetation changes, weather modification, extreme weather conditions, and so on can be readily estimated during the planning process using the methodology of combining classification techniques (to identify hydrologically “homogeneous” unit areas within the heterogeneous basin) and water-balance modeling employed in the Rattlesnake Creek basin study in Kansas. Thus, had the Rattlesnake Creek basin been entirely covered by prairie grasses, as it probably was during predevelopment time, and had the 1982–83 precipitation pattern and amount prevailed (which is about 10% below average), the overall basin deep drainage is model-predicted to have been 1.13 inches/yr, compared to less than 0.15 inch/yr if alfalfa were planted exclusively in the basin. If the entire basin were planted with dryland wheat under 1982–83 precipitation conditions, the overall basin deep drainage is model-predicted to have been 5.1 inches/yr. Such figures can be arrived at by multiplying the deep-drainage amounts for the corresponding crop and soil complex by the planted area, summing up these figures, and dividing by the area of interest. Similarly, the hydrologic effects of manipulating the proportion of various crops and the amounts of irrigation within any soil-association area can thus be assessed (Sophocleous and McAllister, 1987, 1990).

Provided that future precipitation patterns can be established, then, under known vegetation and land-use practices, various components of the water balance, such as deep drainage and surface runoff within the basin, can be predicted using the presented methodology. An example of the relative effects of an approximately 19% precipitation difference on the components of the water balance, keeping the precipitation time-pattern constant, is shown in fig. I–8. This figure represents actual grassland data from the northeastern portion of the Kansas basin study area, which received 18.9 inches of annual precipitation (fig. I–8b) and the rest of the basin, which received an annual precipitation average of 23.3 inches (fig. I–8a). Note the large increase in deep drainage in the higher precipitation region, especially in low-AWC soils, compared to the deep drainage in the lower precipitation region.

FIGURE I–9—Effects of vegetation on deep drainage in the lower one-third of the Rattlesnake Creek basin in south-central Kansas (from Sophocleous and McAllister, 1987).

8.2 Predictive Relations and Recharge Regionalization

Although numerous studies to estimate recharge in specific areas have been carried out, no systematic attempt has been made to develop generic, predictive relationships for quantifying recharge based on the aforementioned controlling environmental factors. This is important for ground-water management and protection. Recharge studies in the agricultural plains of central Kansas resulted in the development of such an approach based on classification and statistical analyses and taking advantage of Geographic Information Systems (GIS) capabilities for “mapping” recharge and its controlling factors (Sophocleous, 1992, 2000c). Because of the generality and applicability of the methodology to most semiarid to humid plain regions of the world with relatively shallow water table, the Kansas case study will be outlined below.

Although geostatistical methodologies and multivariate statistical techniques such as cluster analyses are useful tools for regionalizing point measurements, the usually small number of experimental sites for recharge estimation precludes usage of such techniques. To develop practical relationships between annual recharge and easily measured, independent recharge-controlling factors for the south-central Kansas plains (Great Bend Prairie region), advantage was taken of recently completed multi-year (1985–1992) field-based recharge-assessment studies at 10 sites in that region. As a result, a number of multiple-regression analysis models were developed depending on the number of controlling factors considered. Most of the 10 recharge-assessment field sites were located in grassland and adjacent to irrigated cropland fields. This analysis (Sophocleous, 1992, 2000c) showed that, given the vegetation cover considered, the most influential variables in recharge estimation were, in order of decreasing importance, annual precipitation (PCP; major climatic variable), average maximum soil-profile water storage (AWC; major soil variable), average shallowest depth to water table (DTW; major ground-water condition variable), and average springtime precipitation rate (RATE; secondary climatic variable).

Each of these factors then was used to zone the region for recharge estimation and was mapped as a separate GIS layer or coverage. Thus, four GIS (ARC-INFO) data layers were constructed for the region based on the results of the multiple-regression analysis as shown in fig. I–10. Each data layer was classified into the same number of data classes (six in all) and assigned a class rank. Then the overlays were combined to produce a master map of “homogeneous” zones. GIS technology is ideally suited for such overlay analysis.

FIGURE I–10—Four recharge-related GIS coverages for the Great Bend Prairie region of south-central Kansas. Solid circles indicate recharge-assessment sites. Crosses indicate climatic stations (adapted from Sophocleous, 1992).

An ARC-INFO overlay analysis procedure was conducted to identify areas of differing recharge in the south-central Kansas study region. The regression coefficients of the developed multiple regression models, normalized to 1, were used to weigh the class rankings of each recharge-affecting variable. Based on this classification scheme, an area-wide recharge map (fig. I–11) indicating five differing recharge regions was derived. The recharge zonation agreed well with the field-estimated recharge values at the sites (Sophocleous, 1992, 2000c).

In addition, the fact that the GIS-based regression estimates for recharge were of the same magnitude as other independent estimates, even during an extreme flooding period during the summer of 1993 (Sophocleous et al., 1996), attest to the robustness of the methodology, although additional tests are desirable. It was concluded that the combination of multiple regression and GIS overlay analyses is a powerful, robust (even under extreme conditions), and practical approach to regionalizing small samples of recharge estimates.

FIGURE I–11—GIS coverage showing recharge zonations for the Great Bend Prairie region of south-central Kansas. Solid circles indicate recharge-assessment sites (adapted from Sophocleous, 1992).

9. Difficulties and Challenges in Recharge Estimation

Quantification of the rate of natural and human-induced ground-water recharge is a basic prerequisite for efficient ground-water resource management, and is particularly vital in arid and semi-arid regions where such resources are often the key to economic development. However, the rate of aquifer replenishment is one of the most difficult factors in the evaluation of ground-water resources to measure.

Although there are various well-established methods for the quantitative estimation of recharge, few can be applied successfully in the field. A 1988 international recharge-estimation workshop (Simmers, 1988) concluded that “no single comprehensive estimation technique can yet be identified from the spectrum of methods available; all are reported to give suspect results.”

Difficulties in reliably quantifying ground-water recharge stem from a variety of factors. These include the limited capability to identify and quantify the probable recharge mechanisms and important features influencing recharge for a given locality, the nonlinear recharge response with time, the highly variable areal distribution of ground-water recharge, the scarcity of hydrogeologic data, and the complexities of the hydrologic balance in general.

Because of these uncertainties, project designs and management strategies need to be flexible enough not to require radical change if initial predictions prove wrong, due to incorrect assumptions about recharge rates and other hydrogeological factors (Foster, 1988). Ground-water recharge estimation must be treated as an iterative process that allows progressive collection of aquifer-response data and resource evaluation. In addition, more than one technique needs to be used to verify results.

When estimating ground-water recharge, one must start with a good conceptualization of different recharge mechanisms and their importance in the study area. To identify the probable flow mechanisms, the field evidence must be examined carefully. The recharge mechanism at a particular location can depend on a variety of factors, which may be different from the influencing factors at another location. Therefore, just because a method has successfully estimated the recharge in one locality, one should not assume the same method can be used elsewhere, even if the situation appears to be similar (Rushton, 1988). Once the recharge mechanisms have been defined, calculations can be carried out to estimate the recharge. In addition to being based on a good conceptualization, the choice of methods should be guided by the objectives of the study, available data and resources, and possibilities of obtaining supplementary data.

Consideration of the following questions may facilitate recharge estimation (Lerner et al., 1990). (1) How much recharge can the aquifer accept? A full aquifer will reject further water, which must then find another destination. (2) How much water can the unsaturated zone transmit? High potential recharge rates (for example, from rivers or irrigation canals) may not be able to pass through low conductivity layers. (3) What other destinations are there for potential recharge, and how large are they? (4) How much potential recharge is there? (5) What is the actual recharge? This step considers the balance and destinations of all water from the source, based upon the first four questions. (6) How do other estimates compare? As previously mentioned, whenever possible, more than one method should be used.

The key to deciding on a recharge-estimation methodology is the spatial and temporal scale of interest. If the major concern is obtaining good recharge estimates over a limited area (e.g., for waste disposal or local water-supply purposes), then the need for detailed information is evident. In this situation multiple site investigations are needed, which also require identification of preferential flow contributions. Conversely, for projects on a regional scale, or those requiring only preliminary recharge estimates, ground-water-based methods (such as those involving interpretation of fluctuations in ground-water levels) are relevant and small-scale variability in local recharge ceases to be a problem (Simmers, 1997).

The inherent temporal variability of recharge has important implications for the measurement techniques adopted (Cook, 1993; Scanlon, Healy, and Cook, 2002). Different measurement techniques provide recharge estimates with different temporal scales. For example, applied tracers and lysimeters are only able to provide information on recharge over the period of measurement, usually no more than a few years. Meteorological water-balance techniques, and those involving interpretation of fluctuations in ground-water levels, likewise can provide information only on recharge over the period of record. Chloride-displacement techniques provide a mean recharge rate since the change in land use, while bomb tracers give a mean recharge rate since peak fallout (a period of about 40 years). Chloride mass-balance methods have a much longer temporal scale, typically in the order of hundreds to thousands of years (dependent on the recharge rate and the thickness of the unsaturated zone). The techniques adopted will depend upon the purpose of the study. Where interest is in estimating long-term recharge rates, a long temporal scale of measurement is desirable. On the other hand, if interest is on the effect of land management on recharge, those techniques with smaller temporal scales are required.

In areas where the annual variability of recharge is very high, measurement techniques with long time scales will be required to estimate the long-term mean annual recharge rates with any accuracy. Where the annual variability of recharge is lower, measurement techniques with shorter time scales will be suitable.

The temporal variability of the soil-water flux should decrease with depth (Cook, 1993). If the recharge rate is sufficiently low, and the water table sufficiently deep, then below some depth, the temporal variability of drainage will approach zero. At these depths, even the measurement of the soil-water flux over a short time scale should be sufficient to infer the long-term drainage rate. This decrease in temporal variability of soil-water flux with depth may cause problems in estimating recharge from hydrograph records. If much of the temporal variability is lost during passage through the unsaturated zone, then these methods may underestimate the recharge rate. For example, the coefficient of variation (CV) of annual drainage below the root zone of Banksia woodland (33 ft deep) was found to be 64%, whereas the CV for drainage at 66 ft (water-table recharge level) was 11%. Thus, in areas where deep drainage fluxes are low and water tables are deep, ground-water-based methods may be inappropriate for estimating rates of ground-water recharge. In particular, methods which involve interpretation of hydrograph records will underestimate recharge. Individual recharge events will not usually be seen as rises in water tables if the time lag is greater than a few days. Seasonal variations in drainage will likewise not be reflected in water-table variations if the time lag is much greater than a few months.

Tracer methods seem the most reliable for point- recharge measurements in arid areas. The best method will depend on the magnitude of recharge flux. Chloride appears most reliable over drainage rates from less than 0.04 to 4 inches/yr. At deep drainage rates of more than 4 inches/yr, measurement errors and anion exclusion may become important. Bomb tracers ³H and ³⁶Cl are suitable for recharge rates greater than 0.8 inch/yr. The accuracy of drainage estimates obtained with natural tracers should generally not be assumed to be better than ±50%.

Temporal variability of recharge is related to temporal variability of precipitation. The variability of annual recharge increases rapidly as the mean annual recharge decreases. For mean annual recharge of 1.2 inches/yr, measurements over at least 15 to 20 years have been suggested (Cook, 1993).

When recharge rates are only a few millimeters per year or less, chemical and isotopic methods are likely to be more successful than physical methods, such as water- balance methods, which rely on measured or estimated values of water flux. Water-flux estimates are often in error by as much as one order of magnitude or more, especially when measuring physical parameters in the drier ranges. An advantage of tracers is that they integrate all the processes that combine to affect water flow in the unsaturated zone. Tracer behavior is generally a much more robust indicator of water movement in a porous medium than is the solution of the equations of water flow, especially when soils are relatively dry (e.g., for arid sites).

In estimating recharge, a method such as ³H-profiling relies on estimating the amount of the tracer beneath the soil surface; thus, the precision of the estimate of recharge will increase with recharge rate. In contrast, for a tracer such as Cl, the concentration of which is inversely proportional to recharge rate, the precision of the estimate will increase with decreasing recharge rate. Combining Cl (tracer) and suction-profile (physical) information, at sites where changing land use has significantly altered recharge rates is useful in quantifying both past and present recharge rates at such locations (Allison et al., 1994).

Indirect, physical approaches, such as water balance and Darcy flux measurements, were found the least successful, whereas methods using tracers (e.g., Cl, ³H, and ³⁶Cl) have been found to be the most successful in estimating ground-water recharge in arid regions. Of the tracer techniques available, Cl-balance techniques seem to be the simplest, least expensive, and most universal for recharge estimation.

Nevertheless, advances in recent years show that the value of water-balance and Darcian methods should not be underestimated. Reliability of water-balance methods for recharge estimation depends on the precision with which the water-balance components have been determined. In arid and semiarid regions, application of this method is more difficult than in humid regions because precipitation is frequently only slightly different from actual evapotranspiration; small errors in these two components thus cause large errors in recharge estimates. To minimize such errors, one can use a combination methodology such as the hybrid water-fluctuation methodology, which uses a storm-by-storm water-balance analysis in combination with analyses of vadose-zone moisture and water-table fluctuations (see section A1.1). However, the hybrid water-fluctuation methodology is applicable predominantly to relatively flat, semiarid to humid regions with relatively shallow water table.

Despite their importance for ground-water management and protection, no generic, predictive relationships for quantifying recharge based on the major controlling factors of climate, soils, vegetation, and land use have been usefully developed. In addition, the problem of regionalizing point measurements, given the spatial and temporal variability of recharge and aquifer heterogeneity, remains a serious one. Although a methodology to address such problems has been developed (see Section 8.2), additional studies and approaches are needed to tackle these challenges.