Published January, 1994

Vadose-Zone Techniques for Estimating Groundwater Recharge

in Arid and Semiarid Regions
G. B. Allison, G. W. Gee,* and S. W. Tyler

ABSTRACT METHODS
Both physical and chemical methods have been used to estimate Recent reviews of a range of techniques for, and problems
recharge in arid and semiarid areas. Our review indicates that indi- associated with, estimation of groundwater recharge have been
rect, physical approaches, such as water balance and Darcy flux mea- given by Allison (1981, 1988) and Gee and Hillel (1988). In
surements, are the least successful, while methods using tracers (e.g., addition, Keith et al. (1982) have prepared a bibliography of
Cl, 3H, and 36CI) have been the most successful in estimating ground- recharge studies in semiarid areas. In their review on the use
water recharge in dry regions. Lysimeters, which can directly measure of environmental isotopes for modeling in hydrology, Dincer
root-zone drainage, have been useful in quantifying recharge, partic- and Davis (1984) outlined some of the techniques available for
ularly for coarse soils, but are costly to construct and operate. Of the recharge estimation using these isotopes.
tracer techniques available, Cl balance techniques appear to be the To estimate local recharge, some techniques rely on mea-
simplest, least expensive, and most universal for recharge estimation. surements made in the unsaturated zone and, for others, in the
In Australian studies, under native vegetation in semiarid areas, Cl saturated zone. One of the major problems in studying the
profiles were found to be remarkably uniform, indicating very low unsaturated zone in semiarid areas is that measurements need
and relatively uniform rates of groundwater recharge. Following changes to be made beneath the zone where uptake of water by roots
is significant. For annual crops and pastures, this is not a great
in land use, recharge appeared to become much more variable, in- problem as rooting depths usually are < 2 m and loss of water
creasing more than two orders of magnitude. Methods for scaling from depths greater than this are negligible. However, native
point estimates of recharge to large areas using indirect techniques vegetation and many trees and shrubs may have living roots
(such as nondestructive electromagnetic induction) have also been de- to great depth. The significance of these deep roots in modi-
veloped. In deep unsaturated zones, the pressure response in the soil fying annual water balance and recharge is uncertain.
water may be recorded in the profile, and simple field measurements
may be used to obtain semi-independent verification of recharge rates
determined by using Cl balance techniques. Indirect Physical Methods
In principle, one of the simplest methods used for estimating
local recharge, R, is empirical expressions of the type

I N ARID AND SEMIARID AREAS (where precipitation is

less than -700 mm/yr), the native vegetation has
often developed such extensive root systems that al-
R = k, (P - k2) [1]
most all of the precipitation is consumed by evapo-
where P is precipitation and k^ and k2 are constants for a par-
transpiration. As a result, the local recharge flux under ticular area. Such expressions have been used with varying
native vegetation is very low, but generally constant degrees of success and are probably most useful for making
in time. Following clearing of native vegetation and "first-guess" estimates of recharge where annual recharge is
establishment of comparatively shallow-rooted pas- fairly high, >50 mm/yr, and thus should seldom be used in
tures and crops, local recharge increases. Variability arid or semiarid regions.
of rainfall will result in recharge being highly variable Methods relying on estimates of soil physical parameters
with time. fall into the following three classes: (i) soil water balance, (ii)
In arid and semiarid regions, physical parameter zero-flux plane method, and (iii) estimation of water fluxes
methods, which rely on measurements of hydrological beneath the root zone using unsaturated hydraulic conductivity
parameters, are problematic for several reasons. First, and the gradient in soil water potential.
These techniques will be discussed only briefly here because
the fluxes are still low and changes in these parameters extensive literature is available on them.
will be small and difficult to detect. Second, the tem-
poral variability is such that measurements must be
made for several years to obtain an estimate of mean Soil Water Balance
values. Third, spatial variation brought about by local The soil water balance can be represented, in the absence
topography and soil texture changes requires a large of significant surface runoff, by
number of sampling locations to assess recharge var-
iability.
This review emphasizes chemical and isotopic
R = P - E, + S [2]
methods for estimating local recharge in semiarid areas
because they show more promise than do physical where Ea is actual evapotranspiration and S is the change in
methods in such dry climates. Physical parameter soil water storage, which usually may be ignored if calcula-
tions are made on an annual basis. A number of techniques
methods assume more importance in wetter areas and for estimating £a, based on Penman-type equations (Howard
for estimating artificial recharge where the fluxes are and Lloyd, 1979; Rushton and Ward, 1979) and other methods
higher and the inputs more constant with time. (e.g. Alley, 1984), have been used. The data requirement of
these methods is large and Howard and Lloyd (1979) suggested
G.B. Allison, CSIRO, Private Bag no. 2, Glen Osmond, SA that large errors can occur unless the accounting period is < 10
5064, Australia; G.W. Gee, Pacific Northwest Laboratory, P.O. d. The soil water balance technique has been used extensively
Box 999, Richland, WA 99352; and S.W. Tyler, Desert Research in temperate areas. However, it is unlikely to be successful in
Institute, P.O. Box 60220, Reno, NV 89506. Received 1 Oct. semiarid areas because of long periods of less than potential
1992. 'Corresponding author.
evaporation, when errors in £a are greatest and P and £a are
Published in Soil Sci. Soc. Am. J. 58:6-14 (1994). nearly equal (Gee and Hillel, 1988).
 14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ALLISON ET AL.: REVIEW OF TECHNIQUES FOR ESTIMATING GROUNDWATER RECHARGE

Zero-Flux Plane Method coarse-textured soils, free of vegetation, were observed to drain
(recharge) as much as one-half of the annual precipitation un-
The zero-flux plane (ZFP) method relies on locating a plane der strikingly different climatic conditions (Gee et al., 1993).
of zero hydraulic gradient in the soil profile. Recharge during Lysimeters are expensive and permanent instruments, and
a time interval is obtained by summation of the changes in typically are filled with disturbed soils, which generally have
water content below the plane. Unfortunately, the method breaks water content profiles that differ in some degree from those
down in periods of high infiltration when the hydraulic gradient found in surrounding soils. Drainage can occur only when a
becomes positive downward throughout the profile. This is water table develops at the base of the lysimeter, unless so-
when recharge fluxes are likely to be highest. lution samplers (e.g., ceramic cup extractors) and a vacuum
Use of this technique can give good estimates of recharge system are installed at the base of the lysimeter. This last
for periods during the year when the ZFP exists, as has been factor, however, is unlikely to be a problem if the lysimeter
shown by studies in the United Kingdom (e.g., Wellings, 1984). is relatively deep and the vegetation is shallow rooted (Kitch-
However, it suffers from the same problems as the other two ing and Shearer, 1982). While lysimeters have been useful in
classes if annual values of recharge are required, because one quantifying drainage at waste sites under arid conditions, they
of the other two methods must be used when the ZFP disap- are limited in their ability to document the spatial variability
pears. produced by natural and human-induced changes in surface
and subsurface flow pathways. Construction cost and logistics
Estimation of Water Fluxes limit size and depth to generally no more than a few square
Several studies have reported attempts to use unsaturated meters of areal extent and seldom more than a 3-m depth,
zone hydraulic conductivity (K(0) or K(ili), and hydraulic po- although lysimeters as deep as 18 m have been constructed
tential data, t/^ff), to solve either Darcy's Law or Richards' (Alien et al., 1991; Gee et al., 1992, 1993).
equation in the unsaturated zone and to estimate soil water flux All of the physical techniques (both direct and indirect meth-
for periods of months to years (Sophocleous and Perry, 1985; ods) discussed above have limitations in arid and semiarid
Stephens and Knowlton, 1986). If the water flux is calculated areas, principally because recharge fluxes are likely to be low,
at such a depth in the profile that no further extraction by roots on the order of a few millimeters or less, necessitating long
occurs, then the flux will be equal to groundwater recharge: periods of measurement. Lysimetry, however, has the potential
to overcome problems of low flux, if lysimeters are large enough
and monitoring is long enough. Remaining problems for phys-
R = K(Q) A//t [3] ical techniques include measurement uncertainties, expense,
where A//T is the total head gradient. For most soil systems, and inadvertent modification of the soil moisture regime.
HT = Hg + Hm, where He = gravity head and Hm = matric
suction head. Natural Tracers for Estimation of Local Recharge
Both K(ff) and K(i//) relationships are difficult and time con-
suming to determine, both in the field and in the laboratory, Natural tracers have both advantages and disadvantages over
with difficulty and uncertainty increasing with soil dryness. physical methods. On the positive side, they represent a spa-
Slight differences in measured water content translate into large tially uniform (at least to a first approximation) input to the
differences in unsaturated hydraulic conductivity. The work of soil water-groundwater system. In many cases a history of
Stephens and Knowlton (1986) highlighted the problem of using tracer fallout in precipitation is known, and, because the aqui-
these techniques for estimating groundwater recharge, espe- fer and the overlying unsaturated zone usually store sufficient
cially where the fluxes are low. They found that the annual water to represent many years of recharge, a historical record
recharge flux varied by more than a factor of five (7-37 of recharge may be derived. The principal disadvantage of
mm/yr), depending on how they computed mean hydraulic isotopes is that they may offer only an indirect measure of
conductivity. recharge and mechanisms of infiltration will affect the inter-
Sophocleous (1991, 1992) estimated recharge in the Great pretation of results. However, the use of multiple tracers can
Bend Prairie of Kansas using a combination of physically based often provide corroborative information needed for correct
methods (measured precipitation, soil water tension, and water interpretation.
content profiles coupled with water table fluctuations). The The natural tracers most commonly used in recharge studies
most influential variables on recharge were total annual pre- are 3H, 14C, 36C1, 15N, 18O, 2H, 13C, and Cl. Of these, the
cipitation, average maximum soil-profile water storage in spring first three are radioactive, with half lives of 12.3, 5700, and
months, average shallowest depth to water table, and spring 301,000 yr, respectively. Their concentrations in the hydro-
rainfall rate. Using geographic information systems (GIS) tech- logic
3
cycle have been modified greatly by nuclear testing. Both
nology, Sophocleous (1992) developed an overlay mapping H and 36C1 from atmospheric testing have been used for soil
routine using available precipitation, soils, and water table data. water tracing and recharge studies (Zimmerman et al., 1967;
From this, he was able to prepare area-wide maps of differing Gvirtzman and Margaritz, 1986; Phillips et al., 1988). Chlor-
recharge regions that were in reasonable agreement with field ine-36 has been used increasingly as more analytical facilities
measurements. have become available. Input concentrations of the other iso-
topes mentioned above have also changed in time, but across
Direct Physical Methods a much longer time scale, due to changes in temperature and
rainfall patterns. As far as we are aware, little is known of the
Lysimetry has long been used successfully to make direct temporal changes in the fallout of Cl.
estimates of drainage in temperate areas. Lysimeters are also Of the tracers mentioned above, 3H, 2H, and 18O most ac-
widely used as a reference instrument from which evapotran- curately simulate the movement of water36because they form
spiration is measured for irrigated agriculture in arid and sem- part of the water molecule. In most soils, C1 and NO3 move
iarid regions (Alien et al., 1992). Where appropriate, recharge as the water does, but in some heavier textured soils anion
can be estimated from lysimeter water balance. Gee et al. exclusion may be a problem (Gvirtzman and Margaritz, 1986),
(1992, 1993) have successfully used lysimeters to estimate and the tracer may move more rapidly than the water being
recharge at arid sites in the western USA. Drainage collected traced.
from lysimeters clearly shows that recharge can occur in desert ^Although piston flow is often able to explain the behavior
climates, even where potential evapotranspiration exceeds pre- of tracers in the field, there is mounting evidence (particularly
cipitation by as much as a factor of 10 or more. At sites near at more humid sites) that water movement along preferred
Las Cruces, NM, and Richland, WA, lysimeters filled with pathways is the rule rather than the exception (Gish and Shir-
 14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SOIL SCI. SOC. AM. J., VOL. 58, JANUARY-FEBRUARY 1994

mohammadi, 1991). Thus, non-piston-type flow must be dealt for water vapor of high 3H concentration to diffuse into soil
with in any comprehensive analysis of recharge. Using isotopic water 3or groundwater systems. Estimates of recharge made
analysis, Allison and Hughes (1983) found 3H much deeper using H in such areas therefore would be anomalously high.
than the recharge rate would imply in native forest, suggesting However, this is significant only at R < 1 mm/yr. Careful
preferred flow of water along root channels. consideration should be given to this possibility when inter-
Three techniques have been used for estimating recharge preting data such as those given by Dincer et al. (1974).
rates from tracer profiles in the unsaturated zone.
1. From the position of the tracer peak. The water in the
profile above the peak in tracer concentration represents Chlorine-36
recharge since the time that peak occurred. Any bypass In the last decade, a number of studies (Morris et al., 1987;
(preferential) flow will result in recharge being under- Phillips et al., 1984, 1988; Scanlon, 1992; Cecil et al., 1992;
estimated. It is recommended that the amount of water Walker36 et al., 1992) have presented soil water profiles of bomb-
in the profile be estimated at the time of year that the pulse C1. In studies under arid or semiarid conditions (Phil-
soil moisture deficit is at its maximum. lips et al., 1988; Scanlon, 1992), the 36C1 profile appeared to
2. From the shape of the tracer profile in the soil. This is match that of the input signal; however, the pulse was still
generally more reliable than Method 1 since information very near the soil surface. For example, Phillips et al. (1988)
about flow mechanisms can be obtained. In order to ob- found that the 36C1 fallout 36 peak at one site was very closely
tain estimates of mean annual recharge, R, a weighting reflected in the soil water C1 and was still within the root
function taking into account year-to-year variations of zone at about 1-m depth. A Cl mass balance at their site gave
recharge, is needed. A comparison of several possible a recharge flux of =0.03 mm/yr36at a depth of 5 m beneath the
weighting schemes was given by Allison and Hughes surface. Using the depth to the C1 peak resulted in an esti-
(1978). mated recharge of 2.6 mm/yr. A cumulative Cl depth profile
3. From the total amount of tracer, Tp stored in the profile. suggests that the bottom of the effective root zone at their site
This is given by was36 = 1.5 m and that it would take > 125 yr for the bomb 36 peak
in C1 to reach this depth. Scanlon (1992) reported a C1
profile from an unvegetated arroyo. The 36C1 was found at 0.5-
T(z) 0(z) dz [4] m depth, suggesting a flux to that depth of 1.4 mm/yr. At the
same site, the Cl mass balance, however, suggested a net re-
where T(z) is the tracer concentration of water in the unsatu- charge of 0.04 mm/yr at a depth of 2 m. Thus the bomb peak
rated zone at a distance z beneath the surface and 0(z) is the of 36C1, like 3H, is not an ideal tracer for areas of low recharge
volumetric water content. For evaporative tracers, such as 3H, or in areas of changing land use because it can still retain the
mean annual recharge can be estimated by imprint of the root zone velocity. It will probably be best suited
for regions where the local recharge is expected to be higher
than =30 to 50 mm/yr, but this will depend on the water-
R =T holding capacity of the soil and the rooting depth.
[5] Tyler and Walker (1993) presented a simplified model of
water transport and root water uptake in the36root zone. Their
where T^ is the tracer concentration of recharge water i years model illustrates that fallout tracers (such as C1) found in the
before the present; w, is the annual recharge weighting factors, root zone probably overpredict recharge rates unless the re-
and A is the tracer decay constant; In this analysis nonpiston charge rate is a significant fraction of the precipitation. This
flow can be handled since R is independent of the distribution suggests that, for arid sites, where recharge is often <0.1 times
of the tracer in the profile. the precipitation, 36C1 would not be the tracer of choice since
it will probably overpredict recharge. Other drawbacks to fall-
Tritium out 36C1 are the difficulty in analyzing the near-background
concentrations and the cost of analysis (only a few machines
Many studies on the estimation of recharge using natural 3H are available for the specialized analysis required and the cost-
in the unsaturated zone have been given in the literature (Schmalz per-sample ranges from $500 to $1000 or more).
and Polzer, 1969; Smith et al., 1970; Allison and Hughes,
1974; Foster and Smith-Carington, 1980; Tyler et al., 1992). Chloride
Almost all studies reported have made use of the fact that the
peak in 3H in precipitation due to nuclear fallout has been Input of Cl occurs at the soil surface both in rainfall and as
preserved in the unsaturated zone. dry fallout. Samples collected from most stations include both
To estimate recharge from 3H profiles, the 3H concentration wet and dry fallout, although the estimate of dry deposition
of effective input to the soil water system must be known. may be modified by the collection vessel. The Cl may be of
Work by Gvirtzman and Margaritz (1986) and3 Gvirtzman et either atmospheric or terrestrial origin but, for the hydrologic
al. (1986) has demonstrated the usefulness of H profiles for studies described here, the net accession is required. Thus in
estimating recharge beneath an irrigated area. Their 3H profiles areas that are laterally uniform it is possible that only Cl of
show remarkable preservation with depth and enable annual oceanic origin need be considered. Hutton (1976) showed that
recharge cycles to be seen. The profile results from differences Cl of atmospheric origin dominated up to 300 km from the
in 3H of irrigation water (summer) and rainfall (winter). Such coast in southern Australia. Blackburn and McLeod (1983)
discrimination indicates a very low effective diffusion coeffi- suggested that =50% of the Cl deposition at distances on the
cient for 3H in this soil. The data also suggested that a very order of 700 km from the coast was of oceanic origin.
high percentage (=50%) of immobile water existed in the Most plant species do not take up significant quantities of
draining soil that was not readily exchanged with the mobile Cl from soil water, thus Cl is concentrated by evapotranspir-
water. This led to 3H profiles quite different from those pre- ation in the root zone. In the simplest case of piston flow of
dicted by piston flow. water in the unsaturated zone, the Cl concentration in soil
The techniques for measuring 3H are not valid in areas where water should increase through the root zone to a constant value
the root zone is deep, and interpretation of data may become (Gardner, 1967). Provided the water table is deep or has the
difficult if flow regimes are not found to be reasonably uni- same concentration as the soil water, a depth-concentration
form. Also, it has been suggested that, in arid areas where profile such as shown in Fig. la should result. Under steady-
soils are sandy and have a high air-filled porosity, it is possible state conditions, following Eriksson and Khunakasem (1969),
 14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ALLISON ET AL.: REVIEW OF TECHNIQUES FOR ESTIMATING GROUNDWATER RECHARGE

(a) (b) (c)

Q.
<O
Q

W.T. W.T.

j_

Cl Concentration
Fig. 1. Schematic depth profiles of the Cl concentration of soil water: (a) piston flow with extraction of water by roots; (b)
extraction of water by roots, but with either preferred flow of water through and below the root zone, or diffusive loss of Cl
to the water table (W.T.); and (c) a profile reflecting past (paleo-) recharge conditions (after Allison, 1988).

the flux of Cl at the surface is equal to the flux of Cl beneath could give rise to lower Cl concentrations deeper in the profile.
the root zone, thus Stone (1992) suggested that paleoclimate records support the
PCp = RZs [6] hypothesis that ancient recharge differed significantly from
present in the Murray Basin, South Australia, and that ob-
served Cl profiles with distinct bulges corroborate the paleo-
where P and_R are mean annual precipitation and recharge, record. In contrast, Cook et al. (1989a) attributed the shape of
and Cp and Cs are the mean Cl concentrations in precipitation Cl profiles (similar to Fig. Ic) simply to slow diffusion of Cl
and soil water, respectively. from below a deep (> 20 m) root zone to the underlying water
A more comprehensive model that describes water move- table.
ment in terms of piston flow below a surface mixing layer has
been used by Cook et al. (1992a) to estimate paleorecharge
and paleoclimate from vadose-zone profile data. They reported Oxygen-18 and Deuterium
that Cl profiles from Cyprus and northern Senegal appeared to
show changes in recharge rates for periods up to 400 yr with These stable isotopes have proved particularly valuable for
variations in Cl well correlated with known variations in rain- determining the origin of groundwater; consequently, hundreds
fall and fluctuations in lake levels. of studies have been reported on this topic. The variation of
The Cl mass balance technique in the unsaturated zone has the isotopic composition of rainfall with elevation has been
been used successfully to evaluate recharge in a range of en- used (Arnason, 1977). Variations of isotopic composition with
vironments (Edmunds and Walton, 1980; Allison and Hughes, rainfall intensity and the changes that occur following evapo-
1978; Sharrna and Hughes, 1985). However, many of the depth ration have led to a determination of the possible sources of
profiles of Cl concentration in soil water show a more complex groundwater in arid areas (Gat, 1984; Gat and Naor, 1979;
shape than that shown in Fig. la. Some idealized examples of Issar et al., 1984).
these more complex shapes are given for comparison in Fig. Relatively few studies using stable isotopes have been car-
Ib and Ic. Examples of a range of Cl profiles are given by ried out in the unsaturated zone. Thoma et al. (1979) found
Dimmock et al. (1974), Peck et al. (1981), and Phillips (1993). that the seasonal variations of the 2H composition of rainfall
Several explanations have been offered for the more complex were preserved in a sand dune in France. A knowledge of water
Cl profiles. Bulges in deep Cl profiles have been attributed to content would then enable estimates of recharge to be made.
preferential flow (Peck et al., 1981), to diffusion of Cl to the Bath et al. (1982) and Saxena and Dressie (1984) found that
water table (Cook et al., 1989a), and to paleoclimate-induced some of their profiles of 18O and 2H in soil water showed
changes in recharge (Stone, 1992). cyclical variations in depth that corresponded to seasonal var-
As mentioned above, perennial native vegetation usually de- iations in rainfall. Knowledge of water content and an as-
velops much deeper root systems than the annual vegetation sumption of piston flow enabled estimates of local recharge to
that often replaces it following agricultural development. Al- be made.
lison and Hughes (1983) found Cl profiles of the type shown The three studies referred to above were all carried out in
in Fig. Ic beneath native eucalyptus vegetation where maxi- temperate areas where recharge is =200 mm/yr or more. In
mum Cl concentrations in the soil solution were = 14 000 drier climates, strongly positive values of the displacement of
g/m3. Very similar profiles were found at several locations either 2H or 18O concentration from the concentration found in
beneath the native vegetation. Chloride profiles taken beneath meteoric (rain) water (expressed as 52 or 518) may occur near
land cleared for agriculture (=80 yr previously) also showed the surface due to evaporation through the soil surface (Mun-
very similar profiles, but with the very high concentrations nich et al., 1980), leading to the possibility of identifying an
displaced deeper in the profile, suggesting a piston flow process annual2 marker in the soil water. However, depth profiles of
had occurred at this site. Allison and Hughes (1983) estimated 5lg or H in soil water, shown in Fig. 2, where the mean annual
that recharge at increased from <0.1 mm/yr (under native veg- recharge, R, is estimated from Cl profiles to be =50 and =3
etation) to =3 mm/yr following land clearing. A possible ex- mm/yr, do not show cyclic variations in either of these iso-
planation of the shape of the chloride profile shown as Fig. Ic topes. The most likely reason for the absence of such peaks in
is that it represents a record of the change in recharge with the soil water profile is that, if piston flow occurs, the low
time (Allison et al., 1985). Much higher recharge in the past recharge will ensure that peaks and troughs in isotopic values
 14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 SOIL SCI. SOC. AM. J., VOL. 58, JANUARY-FEBRUARY 1994

Sharma et al. (1985) applied Br at the soil surface, and their

data show strong evidence of nonpiston flow for several meters
1 depth, through the root zone of native vegetation in a sandy
soil. McCord and Stephens (1987) applied Br at shallow depths
2 on the crown (top), side slope, and toe (bottom) of a sand dune
in New Mexico and observed significant differences in down-
ward displacement (recharge) of the Br tracer. Little lateral
1 3 spreading occurred at the crown or toe location, while signif-
icant lateral displacement was observed on the side slope. Dif-
I* ferences in shape and extent of the Br plume were attributed
to topographic (slope) features and moisture-dependent ani-
sotropy of the nearly uniform, but slightly layered sand.

R-50mm/yr Observed Solute and Pressure Fronts in Fields

i i i i i i
with Altered Land Use
-8 -4 Clearing of native eucalyptus vegetation in the Murray River
Basin in South Australia has resulted in large increases in soil-
water flux. This is due to the lower water requirements and
Fig. 2. Depth profiles of 18O S-values (Slg) in soil water at two shallower depths of rooting of the crop and pasture species
sites in southern Australia where the mean annual recharge grown following clearing. It is estimated that fluxes beneath
is 50 and 3 mm/yr, respectively (after Aliison, 1988). the root zones (i.e., groundwater recharge) have increased from
=0.05 to 0.20 mm/yr (Aliison and Hughes, 1983; Leaney and
Aliison, 1986) to =3 to 40 mm/yr (Aliison and Hughes, 1983;
will be separated by only short distances in the soil profile, Aliison et al., 1985).
and diffusional redistribution in both liquid and vapor phases Jolly et al. (1989) and Walker et al. (1991) used a combi-
will occur. Aliison et al. (1984) developed a technique based nation of tracer and-physical methods to estimate recharge.
on the displacement of the SJ8 and S2 values of soil water from Matric suction profiles (obtained using the filter-paper method
the local meteoric line, which may enable estimates of local of Greacen et al., 1989) were combined with Cl profile data
recharge to be made under conditions of low recharge. They and used to interpret the changes in recharge rates due to land
showed that the displacement of S2 from the meteoric line clearing. Cores were taken from two holes, one drilled in a
should be linearly related to R-1'2 and found this relationship site with native vegetation and the other from a cleared field
to hold for the four sites they studied. The nature of this sug- nearby. Figures 3a and 3b show the profiles of water content,
gests more precise estimates as recharge becomes lower, but Cl, and matric suction. The Fig. 3a profile is typical of those
the technique requires further testing. Barnes and Aliison (1988) found under native eucalyptus vegetation in this region. The
developed detailed models for the movement of stable isotopes mean annual recharge at this site, calculated using the Cl tech-
in both the liquid and vapor phases in soils. nique of Aliison and Hughes (1983), is 0.08 mm/yr. In con-
trast, Fig. 3b shows water content, Cl, and matric suction
Nitrate profiles for an adjacent field, which was cleared for 9 yr. A
pressure front has developed to a depth of =7.5 m and the
Another involuntary tracer that may be used to give infor- solute front is lower in the profile (now between 4 and 5 m)
mation on the rate of water movement in the unsaturated zone than that under native vegetation. Using the technique of Al-
is NO3, which has come into increasing agricultural use in iison and Hughes (1983), the postclearing recharge rate is es-
many areas since about 1950. Because some of this added NO3 timated to be approximately 45 mm/yr. Assuming that the
is leached below the root zone, the change from higher to lower pressure and solute fronts were at the surface and 2 m, re-
concentrations of NO3 in soil water at depth is an indication spectively, prior to clearing, then the ratio of the rate of move-
of the position in the profile of recharge originating at the time ment of the solute front to that of the pressure front is between
of increased use of NO3. A knowledge of the amount of water 0.27 and 0.40.
stored in the profile should then enable an estimate of R (see Evidence of preferred pathway flow, such as reported by
also Barnes et al., 1993). In some situations the reverse occurs Johnston et al. (1983), was not found in any of the cores taken
where, following clearing, NO3 associated with the native veg- at this site and a ratio of Q.33 for solute to pressure is in
etation may possibly be a marker associated with the time of reasonable agreement with the theory developed by Warrick
clearing. et al. (1971) for the transfer of solute in unsaturated soils that
do not exhibit preferential flow. The solute front represents
Applied Tracers the boundary between water stored in the soil before clearing
and water that has entered since clearing. The water that has
In contrast with the tracers discussed above, which are input entered the profile postclearing has moved downward and dis-
each year to the unsaturated zone in precipitation, use of an placed the preclearing water ahead of it. This has resulted in
applied tracer necessitates a one-time application, followed by the downward propagation of a pressure front ahead of the
sampling of the tracer pulse with time. Ideally, the tracer should solute front. Below the pressure front, the soil water is still
be applied beneath the root zone and a sufficient time allowed moving at the preclearing rate, while the soil water above the
to elapse between injection and sampling to allow the depth pressure front is moving at the new rate. When the pressure
interval traversed by the tracer peak to be measured accurately. front reaches the water table, water of high salt concentration
These conditions make the technique valuable in temperate will be added to the aquifer at the postclearing recharge rate.
regions where the rooting depth is shallow and recharge rates Such observations provide clear evidence for the deleterious
are high. In semiarid regions where roots are usually deeper effects of land clearing on the salinization of groundwater in
and recharge fluxes lower, this technique will be less useful the Murray River Basin (Aliison et al., 1990).
and will not be discussed in detail here. Artificially added A comprehensive analysis has been developed by Walker et
tracers that have been used with success are 18O, 2H, 3H, and al. (1991) to estimate increases in recharge following land
Br (Blume et al., 1967; Zimmerman et al., 1967; Saxena and clearing. The method, which accounts for transient movement
Dressie, 1984; Athavale et al., 1980; Sharma et al., 1985). of Cl, does not rely on piston-flow assumptions. It uses mea-
 14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ALLISON ET AL.: REVIEW OF TECHNIQUES FOR ESTIMATING GROUNDWATER RECHARGE 11

-2

-4

-6

-8

-10
-12

-14
-16

-18

-20 I I I I I I I
0

-2

-4

-6
-8
-10

-12

-14
-16
-18

-20 I I I
0.08 0.16 0.24 4 8 12 16 20 0
Gravimetric Water (thousands) (thousands)
Content (g/g) Chloride (mg/L) Matnc Suction (kPa)
Fig. 3. Gravimetric water content, Cl in soil water, and matric suction profiles of a core hole drilled (a) beneath eucalyptus
vegetation and (b) in a field cleared of native eucalyptus vegetation for 9 yr (after Jolly et al., 1989).

sured Cl and matric suction profiles to quantify drainage below While nuclear subsidence craters may be a somewhat anom-
the root zone. Test cases from a semiarid region of South alous alteration of the soil surface, analogies may be drawn to
Australia, where land clearing had influenced recharge, dem- other more natural arid land forms. Any features that tend to
onstrated the utility of the method. concentrate ephemeral runoff into areas of permeable soils are
In an extreme case of altered land use, Tyler et al. (1992) likely sites for recharge. It is therefore likely that landforms
observed significant soil water flux beneath a nuclear subsid- such as arroyos, washes, and some closed basin features (de-
ence crater on the Nevada Test Site. Subsidence craters, some siccated playas) may act as local sources of recharge in spite
in excess of 200 m in diameter, have formed as a result of of the regional aridity.
belowground nuclear detonations. While the region is gener-
ally arid (annual precipitation is approximately 150 mm), pe- Variability of Local Recharge
riodic ponding and the presence of riparian species at the bottoms
of the craters suggest that the craters may be localized sources It is now realized that there can be considerable variation in
of recharge. Borehole data from beneath a crater formed in rates of local recharge over the scale of a few meters in many
1971 revealed post-1952 soil water 3H to a depth of 36 m and soil types (Johnston, 1987; Sharma and Hughes, 1985). Figure
wetting front in excess of 50 m. No 3H was observed below a 4 shows four profiles of Cl concentration in soil water from
depth of 3.4 m in the undisturbed soils adjacent to the crater, what appears to be a uniform field, where the surface soils
suggesting that the soil water flux under natural conditions was were sands and sandy loams. All cores were taken within a
small. By applying several transport models, Tyler et al. (1992) circle of 100-m radius, and it is believed that the large varia-
calculated the local recharge flux below the crater to be ap- tions in Cl concentration reflect correspondingly large changes
proximately 0.6 m/yr, or four times in excess of the local in recharge rate (Cook et al., 1989a). The native vegetation
precipitation. It was postulated that the altered surface topog- was removed from the field about 50 yr ago and information
raphy, coupled with the coarse-textured alluvial soils beneath from this field suggest that Cl profiles, and hence recharge,
the crater, allowed deep infiltration of ephemeral runoff con- are quite spatially variable. Other sites are equally nonuniform
centrated at the bottom of the crater. (Allison and Hughes, 1983).
 14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 SOIL SCI. SOC. AM. J., VOL. 58, JANUARY-FEBRUARY 1994

1986). However, because these techniques respond primarily

to soil electrical conductivity, the best correlation between soil
conductivity, as measured by the instrument, and some field
characteristic, is expected to be a function of both water con-
tent and salt concentration. Unfortunately, the magnitude of
the recharge flux is inversely proportional to the Cl (or salt)
concentration in soil water and not directly related to soil water
content. Cook et al. (1989b, 1992b) using a portable EM unit,
developed a calibrated model of soil conductivity using the
Q EM unit in a field with widely varying recharge and found a
Q. good fit between modeled and measured values. Their results
0)
.c suggest that EM methods are promising for obtaining spatial
variations in recharge when coupled with a Cl mass balance
0)
method.

1<1>
.>
DISCUSSION AND CONCLUSIONS
This review has concentrated on methods to estimate
5 recharge in areas of reasonably low rainfall (< 700 mm/
3
yr). In general, the greater the aridity, the smaller and
O potentially more variable the recharge flux. When re-
charge 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
5,000 10,000 15,000 much as an order of magnitude or more. An advantage
of tracers is that they integrate all of the processes that
Chloride Concentration3 in combine to affect water flow in the unsaturated zone.
the Soil Solution (g nr ) Tracer behavior is generally a much more robust indi-
Fig. 4. The relationship between the Cl concentration of soil cator of water movement in a porous medium than is the
water and cumulative water stored in the profile in a field, solution of the equations of water flow, especially when
cleared about 50 yr ago, under a pasture—crop rotation. soils are relatively dry (e.g., for arid sites).
Rainfall is =300 mm/yr. All sites are within a circle of 100- In estimating recharge, a method such as 3H profiling
m radius. Curve numbers reflect different cores (after Allison, relies on estimating the amount of the tracer beneath the
1988).
soil surface, thus the precision of the estimate of re-
charge will increase with recharge rate. In contrast, for
The reason for this variability is unclear. It may be due to a tracer such as Cl, the concentration of which is in-
changes in the water-holding capacity of the soil, preferential versely proportional to recharge rate, the precision of the
flow (from deep cracks and root channels), or, alternatively, estimate will increase with decreasing recharge rate.
the infiltration characteristics of the surface may be such that Combining Cl (tracer) and suction profile (physical) in-
runoff followed by localized recharge may occur. Whatever formation, at sites where changing land use has signifi-
the reason, identification, quantification, and development of cantly altered recharge rates has been useful in quantifying
statistical and probably soil and geomorphological techniques both past and present recharge rates at such locations.
for estimating recharge across a sizable area remain seminal
problems in recharge evaluation. Probably the most difficult and important problem to
Small-scale variability of local recharge is not always a be overcome in estimating recharge is the prediction and
problem. Athavale et al. (1980) were able to make what ap- assessment of its variability. Across some quite large
pears to be 2a reliable estimate of local recharge for an area of areas it appears to show little lateral variability, while in
= 1600 km in India from estimates at 28 sites. Allison and other, apparently similar, areas it can range across at
Hughes (1978) found that natural 3H profiles and, hence, es- least an order of magnitude. Use of rapid and nondes-
timates of local recharge, varied from one soil type to another, tructive techniques such as EM methods may offer some
but 3did not vary greatly within a soil type. They were able to hope here.
use H profiles together with a soil map to estimate total local
recharge to an area of similar size as the3 Indian work. Foster ACKNOWLEDGMENTS
and Smith-Carington (1980) found that H profiles up to 10
km apart in Norfolk, England, showed "remarkable lateral This work was supported by the Commonwealth Scientific
uniformity." Industrial Research Organization of Australia, the U.S. Nu-
At present it appears impossible to predict areas where spa- clear Regulatory Commission, and the U.S. Department of
tial variability will be high. Therefore, there seems to be no Energy under Contract DE-AC06-76RLO 1830.
alternative but to make estimates of local recharge at many Part of this review was originally developed by the senior
sites within an area of interest and to then assess the degree author at the North Atlantic Treaty Organization (NATO) ad-
of spatial variability. vanced research workshop on estimation of natural recharge
of groundwater held in Antalya, Turkey, 8-15 Mar. 1987, and
subsequently published in the 1987 workshop proceedings.
Assessments of Spatial Variations in Recharge
Frequency-domain electromagnetic (EM) or transient elec-
tromagnetic (TEM) methods have been used with success to
assess soil salinity (Williams and Baker, 1982; Buselli et al.,
14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
13
ALLISON ET AL.: REVIEW OF TECHNIQUES FOR ESTIMATING GROUNDWATER RECHARGE
14350661, 1994, 1, Downloaded from https://acsess.onlinelibrary.wiley.com/doi/10.2136/sssaj1994.03615995005800010002x by Lund University, Wiley Online Library on [04/04/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14SOIL SCI. SOC. AM. J., VOL. 58, JANUARY-FEBRUARY 1994