HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Heilweil, V. M. Articles by Gardner, P. M. Search for Related Content PubMed Articles by Heilweil, V. M. Articles by Gardner, P. M. GeoRef GeoRef Citation Agricola Articles by Heilweil, V. M. Articles by Gardner, P. M. Related Collections Isotopes Recharge Infiltration

ORIGINAL RESEARCH

Borehole Environmental Tracers for Evaluating Net Infiltration and Recharge through Desert Bedrock

^a U.S. Geological Survey, 2329 Orton Cir., Salt Lake City, UT 84119
^b Univ. of Utah, 135 South 1460 East, Rm. 719 WBB, Salt Lake City, UT 84112
^* Corresponding author (heilweil{at}usgs.gov)

Received 10 January 2005.

	ABSTRACT

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
Permeable bedrock aquifers in arid regions are being increasingly developed as water supplies, yet little is generally known about recharge processes and spatial and temporal variability. Environmental tracers from boreholes were used in this study to investigate net infiltration and recharge to the fractured Navajo Sandstone aquifer. Vadose zone tracer profiles at the Sand Hollow study site in southwestern Utah look similar to those of desert soils at other sites, indicating the predominance of matrix flow. However, recharge rates are generally higher in the Navajo Sandstone than in unconsolidated soils in similar climates because the sandstone matrix allows water movement but not root penetration. Water enters the vadose zone either as direct infiltration of precipitation through exposed sandstone and sandy soils or as focused infiltration of runoff. Net infiltration and recharge exhibit extreme spatial variability. High-recharge borehole sites generally have large amounts of vadose zone tritium, low chloride concentrations, and small vadose zone oxygen-18 evaporative shifts. Annual net-infiltration and recharge rates at different locations range from about 1 to 60 mm as determined using vadose zone tritium, 0 to 15 mm using vadose zone chloride, and 3 to 60 mm using groundwater chloride. Environmental tracers indicate a cyclical net-infiltration and recharge pattern, with higher rates earlier in the Holocene and lower rates during the late Holocene, and a return to higher rates during recent decades associated with anomalously high precipitation during the latter part of the 20th century. The slightly enriched stable isotopic composition of modern groundwater indicates this recent increase in precipitation may be caused by a stronger summer monsoon or winter southern Pacific El Niño storm track.

Abbreviations: GNIP, Global Network of Isotopes In Precipitation • NWQL, National Water Quality Laboratory • TDTP, tritium depth-to-peak method • TMB, tritium mass-balance method • TU, tritium units • TU-m, tritium-unit meters

	INTRODUCTION

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
THE NAVAJO SANDSTONE forms an important part of the Dakota-Glen Canyon aquifer system in the Colorado Plateau region of the United States, which covers an area of about 210 000 km² in Utah, Arizona, Colorado, and New Mexico (Robson and Banta, 1995). It is a very uniform, well-sorted, fine- to medium-grained quartz eolian sand held together by calcite cement with prominent cross-bedding structures and secondary fracturing (Cordova, 1978). The Navajo Sandstone shares characteristics of both unconsolidated sediments and other consolidated rock formations. Because of its high primary permeability, movement of water through the sandstone matrix is more similar to flow through unconsolidated soils than through lower permeability bedrock. However, as in other exposed or shallowly buried bedrock formations, roots are present only in fractures. Conceptually, this limits transpiration losses and enhances net infiltration and recharge compared with unconsolidated basin-fill sediments in similar arid settings. For this paper, the phrases "infiltration" and "net infiltration" are defined as vadose zone water movement below land surface and below the root zone, respectively. The maximum depth of the root zone is considered to be synonymous with the soil–bedrock contact because root penetration into the sparsely fractured sandstone is minimal. The phrase "Groundwater recharge," or "recharge" for short, is defined as water entering the aquifer at the water table. It is assumed that all net infiltration eventually becomes groundwater recharge. Net-infiltration and recharge rates may be equivalent if conditions remain uniform while water is moving through the vadose zone to the water table.

Rapid population growth in the region and further development of groundwater resources requires quantification of groundwater recharge. Direct and focused net infiltration of precipitation on exposed outcrops or shallowly buried bedrock have been identified as the primary sources of recharge to the Navajo (Cordova et al., 1972, Cordova, 1978, 1981; Hood and Danielson, 1981; Eychaner, 1983; Hood and Patterson, 1984) and other fractured bedrock aquifers (Rasmussen and Evans, 1993; Flint et al., 2002). This is supported by vadose zone solute distributions observed along shallow trenches in the Navajo Sandstone of southwestern Utah (Heilweil and Solomon, 2004), indicating a connection between recharge and surficial characteristics such as soil coarseness and outcropping bedrock. Soil water sampling along these trenches showed that net infiltration readily penetrates sandstone beneath eolian sands as much as 3 m thick, whereas infiltration is unable to reach the soil–bedrock interface where finer-grained sandy loam and loamy sand deposits are thicker than 0.7 m.

The quantification of bedrock net infiltration and recharge is difficult, particularly in desert environments where infiltration is spatially and temporally variable. Regional watershed and groundwater-flow modeling often use empirical methods for estimating infiltration and runoff (Maxey and Eakin, 1950) that are not well suited for evaluating heterogeneity in recharge caused by localized climate, surface morphology, and geologic variables. Conversely, soil physics methods such as soil water and Darcy flux measurements (Nimmo et al., 1994) are limited by their time- and site-specific nature and may not accurately represent long-term recharge at the basin scale. Therefore, environmental-tracer methods were chosen for this study because of their ability to both integrate a range of temporal and spatial scales in the evaluation of recharge, as well as to quantify vadose zone travel times. As pointed out by others (Allison, 1988; Allison et al., 1994; Flint et al., 2002; Scanlon et al., 2002), the application of multiple tracer methods addresses this variability by providing flux estimates for these varying spatial and temporal scales. The objective of this study was to apply environmental tracer methods to estimate rates of net infiltration and recharge to the Navajo aquifer in Sand Hollow under current and past climates and to develop a general understanding of processes affecting recharge to exposed bedrock aquifers in arid regions.

	SITE DESCRIPTION

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
The Sand Hollow study area is located in southwestern Utah along the western edge of the Navajo aquifer system of the Colorado Plateau (Fig. 1 ) (northernmost = 37.10°, southernmost = 37.00°, easternmost = 113.22°, westernmost = 113.25°). Altitudes within Sand Hollow range from about 900 to 1300 m. The area is considered arid, with a mean annual precipitation at nearby St. George, UT of 210 ± 70 mm (based on records for 1893 through 2003; Western Regional Climate Center, 2004). The Navajo Sandstone is as much as 350 m thick at Sand Hollow. The basin has fairly gentle topographic relief and the sandstone is either exposed at the surface or covered with a veneer of soil (<1.5 m). About 40% of the basin is covered by sand dunes and fine sands, mostly located in the southern (upland) area (Fig. 2 ). Finer-grained loams and loamy very fine sands cover about 50% of the basin. The Navajo Sandstone crops out in the remaining 10% of the basin. Caliche deposits are often present at the soil–bedrock contact, particularly beneath the loams and loamy sands. A main ephemeral wash drains the higher-altitude southern part of the basin. Even during the largest storms, however, surface water spreads out and infiltrates into the permeable soils at the northern end where the topographic slope decreases, rather than leaving the basin. A reservoir constructed during 2002 in the lower-altitude part of the basin is being conjunctively managed for both surface-water storage and groundwater recharge (Fig. 1 and 2). Monitoring-well construction for observation of vadose zone and groundwater response to this artificial recharge also provided an opportunity for evaluating net infiltration and recharge processes.

View larger version (150K): [in this window] [in a new window]   Fig. 1. Location of Sand Hollow, Washington County, Utah.

View larger version (71K): [in this window] [in a new window]   Fig. 2. Soils map with location of wells, borehole core-collection sites, and meteorology station in Sand Hollow, UT.

The underlying Navajo aquifer is unconfined, with depths to water in the central part of the basin ranging from 15 to 65 m below land surface. Vertical groundwater hydraulic gradients calculated from nested piezometers at Sites 13 through 17 and at Sites 42 and 43 are 0.006 and 0.003, respectively. Based on the distribution of potentiometric contours, these downward gradients at Sand Hollow are consistent with a conceptual model in which net infiltration of precipitation through the vadose zone is the primary source of recharge to the aquifer (as opposed to other potential recharge sources, such as interbasin groundwater flow from higher-elevation areas).

	THEORY

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
Chloride concentrations in atmospheric deposition and pore water (both within the vadose zone and beneath the water table) are often used for estimating rates of net infiltration and recharge. A simpflied form of the Cl^– mass balance (CMB) method (Allison and Hughes, 1978; Allison, 1988; Dettinger, 1989; Wood and Sanford, 1995) can be represented as

[1]

where q_CMB is the net-infiltration or recharge rate (L T^–1), [Cl]_dep is the average Cl^– concentration of atmospheric deposition (M L^–3), [Cl]_pw is the average Cl^– concentration of pore water (M L^–3), and P is the precipitation rate (L T^–1). Atmospheric Cl^– deposition includes Cl^– in both precipitation and dry dust accumulation. The CMB method is based on the assumptions that Cl^– deposition is constant with respect to time, that Cl^– is transported through the vadose zone to the water table in liquid phase and does not partition into solid or gas phases, and that Cl^– is geochemically conservative and has no sources or sinks within the porous media of the vadose zone or underlying aquifer (Scanlon, 2000). The simplified form of the CMB method represented by Eq. [1] additionally assumes that there is no surface water Cl^– run-on or runoff at each borehole site.

Chloride concentrations in atmospheric deposition, average annual precipitation, and the amount of vadose zone Cl^– accumulation are used to estimate the residence time, T_r, of vadose zone Cl^–:

[2]

where M_Cl is the total mass of Cl^– in a column of unit area containing vadose zone material between land surface and the water table (M) and V_P is the average annual volume of precipitation falling on a unit area of land (L³ T^–1). M_Cl is calculated from a vertical profile of vadose zone pore-water concentrations and assumes that all the chloride is dissolved (in liquid state). Similar to the CMB method, the vadose zone Cl^– residence time calculation assumes that atmospheric Cl^– deposition is constant with respect to time and Cl^– behaves conservatively (geochemically) and has no sources or sinks within the vadose zone.

Vadose zone tritium concentrations are also used for evaluating net-infiltration rates (Allison and Hughes, 1978; Allison, 1988; Allison et al., 1994). Two commonly used methods are the tritium depth-to-peak (TDTP) method and the tritium mass balance (TMB) method. The TDTP method is based on depth below land surface of the 1963 ³H precipitation peak. This peak from above-ground nuclear testing was originally more than two orders of magnitude above concentrations from natural ³H production. The net-infiltration rate, q_TDTP (L T^–1), can be estimated by the TDTP method using the equation (Cook et al., 1994)

[3]

where z is the vertical distance (L) between land surface and the 1963 ³H peak, t is the length of time (T) between the 1963 ³H peak and the sample collection time, and _v is the depth-weighted volumetric water content of the vadose zone between land surface and the ³H peak. The TDTP method assumes one-dimensional movement of water through the vadose zone and that volumetric water content throughout the profile does not change with time.

The tritium mass-balance (TMB) method, similar to the CMB method, is an application of the principle of continuity (conservation of mass). The method is based on the fraction of the mass of ³H measured in the vadose zone beneath the root zone compared with the decay-corrected mass of ³H estimated to have fallen as precipitation at the site. The difference between these two ³H masses is assumed to have been recycled back into the atmosphere through evapotranspiration. Because of the very small fractionation between tritiated water (³HHO) and normal water (H₂O), the loss of ³HHO directly mirrors the loss of normal H₂O to evapotranspiration. Multiplying the fraction of vadose zone ³H mass by the average annual precipitation provides the average annual net-infiltration rate. Therefore, higher net-infiltration rates result from larger ratios of the mass of ³H in the vadose zone compared with the mass of ³H in precipitation. The TMB equation (Cook et al., 1994) is defined as

[4]

where q_TMB is the net-infiltration rate (L T^–1), (z) is the volumetric water content at depth z, [³H(z)] is the pore-water ³H concentration at depth z (M L^–3), w_i is a weighting function to correct for variations in annual net infiltration (the weighting function used for Sand Hollow was the ratio of precipitation during each year (i) to mean annual precipitation), and (T^–1) is the decay constant for ³H [ln(2)/(12.32 yr)], and [³H_i]e^–i is the decay-corrected ³H concentration in precipitation i years before present (M L^–3) where is the radioactive decay constant for tritium of 0.05576.

The use of the term mass balance for both the CMB and TMB methods requires further clarification. There is a fundamental difference between the CMB and TMB methods. In the CMB method, the Cl^– is conserved once it enters the subsurface and only water is lost through subsequent evaporation and transpiration. In contrast, ³H is not conserved in the vadose zone, but rather can be evaporated or transpired back into the atmosphere. Therefore, the TMB is a true mass balance, whereas the CMB would be more accurately described as a Cl^– enrichment method because it inherently assumes that all Cl^– is conserved within the subsurface.

The TMB method can also be presented in the simplified form

[5]

where M_VZ is the total mass of tritium present in the vadose zone at the site in tritium-unit meters (M), M_PPT is the total mass of tritium estimated to have fallen as precipitation at the site in tritium-unit meters (M), and P is the average annual precipitation (L T^–1). M_VZ can be calculated by integrating the product of the measured water content (unitless fraction) and tritium concentration in tritium units (TU) (M L^–3) for each vertical sample interval by the interval length (L). M_PPT can be calculated by multiplying the decay-corrected average annual estimated atmospheric tritium concentration for the location by total annual precipitation for that year.

The TDTP method is considered more accurate than the TMB method primarily because it does not require knowledge of the ³H input function (the historical record of ³H in precipitation). The ³H input function is generally not known for each site and must be interpolated from sparsely located Global Network of Isotopes In Precipitation (GNIP) stations (International Atomic Energy Agency Global Network of Isotopes in Precipitation, 2004). For Sand Hollow, the nearest sites are Flagstaff, AZ, Salt Lake City, UT, and Albuquerque, NM, located at distances of 270, 425, and 650 km, respectively, each having different wind and climate patterns. Also, because no early ³H precipitation measurements were made at any of these stations, the Sand Hollow ³H input function before the early 1960s is based only on correlations with the Ottawa, Canada station located more than 3000 km to the northeast at a much more northerly latitude.

A high percentage of ³H within the shallow root zone may eventually be recycled into the atmosphere rather than becoming net infiltration, causing an overestimate of net-infiltration rates when ³H in the root zone is included in the total mass of vadose zone ³H. However, error associated with root zone ³H in the mass-balance calculation is not considered an important factor at the borehole sites in Sand Hollow because of the exposed or shallowly buried sandstone, where minimal root penetration occurs.

At borehole sites without the presence of a clearly discernible 1963 vadose zone ³H peak, the TMB method must be used to estimate net infiltration. The three primary disadvantages of applying the TMB method at Sand Hollow are:

1. It will overpredict net infiltration at sites where surface-water run-on has occurred because this additional surface mass of ³H would increase the denominator of Eq. [4] and [5], thus causing lower actual net infiltration.
   
2. The ³H input function is not well constrained and must be interpolated from sparsely located GNIP stations.
   
3. The seasonal variation in ³H concentrations in precipitation can affect the actual mass of ³H being recharged, particularly at places such as Sand Hollow where net infiltration and recharge have large seasonal variations. In the spring, upward shifts of stability regions of the tropopause allow high ³H concentrations to be mixed into the moist layer and precipitated (Eriksson, 1983), causing a summer maximum and corresponding winter minimum in precipitation ³H concentrations.
   

The equations used for estimating uncertainty of atmospheric Cl^– deposition, net-infiltration, and recharge rates are provided in the Appendix.

	MATERIALS AND METHODS

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
Environmental tracers used during this study include chloride (Cl^–), bromide (Br^–), tritium (³H), deuterium (²H), and oxygen-18 (¹⁸O). These tracers were analyzed in atmospheric deposition (combined dry fall and wet fall), vadose zone pore water, and groundwater. Six-month composite atmospheric-deposition samples were collected with a 75-mm-diameter straight-sided Buchner funnel at a height of about 1 m above ground supported by a stake and connected with copper tubing to a 1-L plastic sample bottle buried about 0.3 m below ground (Friedman and others, 1992). A thin (10 mm) layer of mineral oil in the bottle was used to minimize evaporation of water. The range in Cl^– concentration from 6-mo atmospheric deposition samples was compared with Cl^– concentrations from individual storm events, which were collected with a 150-mm-diameter brass funnel draining into a 250-mL high-density polyethylene bottle (Heilweil, 2003). To sample both wet fall and dust deposition between storms, the funnels for both the individual storms and the multiple-month composite samples were not rinsed between sample collections.

Vadose zone core samples were collected at 18 sites in Sand Hollow to evaluate vadose zone solute accumulations (Fig. 2). Borehole core samples (63.5-mm diameter) from 13 of these sites were analyzed for chloride, stable isotopes, and/or tritium and are reported in this work (Fig. 3 , Table 1). To minimize contamination of these pore waters with drilling fluids, cores were collected with a triple-tube continuous coring system with air. To minimize evaporative loss of water, the core samples were immediately heat-sealed in a layered aluminum/plastic laminate. Vadose zone matric potentials were measured with in situ head-dissipation probe measurements at four depths (4.6, 10.7, 18.3, 30.0 m) at Site 39. Groundwater samples were collected from turbine-shaft production wells, as well as in piezometers and boreholes with an air-driven submersible piston pump after purging a minimum of three casing volumes.

View larger version (75K): [in this window] [in a new window]   Fig. 3. Chloride concentration and volumetric water content of pore-water and groundwater samples collected at selected boreholes in Sand Hollow, UT.

Deuterium (²H) and ¹⁸O isotopic ratios of precipitation, vadose zone pore waters, and groundwater were analyzed with an isotope-ratio mass spectrometer at the University of Utah Stable Isotope Ratio Facility for Environmental Research. Tritium concentrations in precipitation, vadose zone pore water, and groundwater were analyzed at the University of Utah Dissolved Gas Service Center on a mass spectrometer by the helium in-growth method (Clark et al., 1976). Pore waters were extracted from core samples for isotopic analysis by cryodistillation. Pore-water tritium sample volumes were generally around 50 mL, and minimum in-growth holding times were about 12 wk. Uncertainty in the pore-water tritium analysis was generally <0.5 TU, with many samples being between 0.01 and 0.1 TU.

	RESULTS

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
Chloride
The Cl^– concentration of atmospheric deposition is needed to calculate both vadose zone net-infiltration and groundwater recharge rates using the Cl^– mass-balance method. The precipitation-weighted mean concentration of atmospheric Cl^– deposition at the Sand Hollow weather station (elevation 940 m) is 0.8 ± 0.3 mg L^–1, based on eight 6-mo atmospheric-deposition samples collected between June 1999 and September 2004, which ranged from 0.5 to 1.2 mg L^–1. The 0.3 mg L^–1 standard deviation (1 ) results in an uncertainty in Cl^– deposition of about 38%. Three other 6-mo periods were excluded because of high sulfate concentrations, which indicate possible contamination from bird droppings, decaying insect debris, or dust associated with nearby construction (Table 2). The range in 6-mo composite Cl^– deposition values is similar to the range in Cl^– concentrations from individual storm events (Heilweil, 2003). The mean Cl^– concentration is larger than average annual Cl^– values between 1999 and 2003 at higher-elevation National Atmospheric Deposition Program sites in Bryce Canyon National Park (0.07–0.08 mg L^–1; elevation 2480 m) and Grand Canyon National park (0.08–0.15 mg L^–1; elevation 2150 m) (National Atmospheric Deposition Program, 1999–2003). However, the Cl^– concentrations at Sand Hollow are similar to those measured during the same period in Salt Lake City, Utah (0.27–1.29 mg L^–1; elevation 1300 m). Also, the bowl-shaped topography of Sand Hollow may be a natural dust trap, possibly causing an elevated dry-dust component of Cl^– deposition relative to regional values.

With an average estimated precipitation rate of 210 ± 70 mm yr^–1 at Sand Hollow, the precipitation-weighted mean Cl^– concentration of 0.8 ± 0.3 mg L^–1 was used to calculate a Cl^– deposition rate of 17 ± 8.5 µg 100 mm^–2 yr^–1. Leachates prepared from borehole-core samples show that accumulation of Cl^– in the vadose zone at Sand Hollow is quite variable. Total mass of Cl^– accumulation ranged from 4 ± 0.7 mg 100 mm^–2 (Sites 9 and 35) to at least 230 ± 30 mg 100 mm^–2 (Site 4) for a vertical column of 10-mm horizontal length and 10-mm horizontal width by vertical depth (distance from land surface to the water table). As determined by Eq. [2] and [A5], the total masses of vadose zone Cl^– represent about 200 ± 100 to 12 000 ± 6000 yr of accumulation in Sand Hollow (Table 1). A cooler and wetter regional climate during the late Pleistocene (Weng and Jackson, 1999) likely precluded the accumulation of shallow vadose zone Cl^– before about 12 000 yr ago.

Vadose zone water content of the borehole-core samples is generally consistent with trends in vadose zone Cl^– accumulation. Mean vadose zone volumetric water content, calculated from land surface to the capillary fringe, ranged from 1.7 to 7.8% at the 13 borehole sites (Table 1, Fig. 3). Site 4 had the lowest average vadose zone water content (1.7% , consistent with the largest Cl^– accumulation of any of the sites within Sand Hollow. Site 4 is covered with finer-grained soils and is located on East Ridge (Fig. 2, Table 1), where more bare-soil evaporation likely occurs because of its increased exposure to wind. Site 9 had the highest average vadose zone water content (7.8%), consistent with having the smallest Cl^– accumulation. Located along a small side wash, this site regularly receives focused infiltration of surface-water runoff from the exposed sandstone of West Ridge.

The Cl^– profiles (Fig. 3) generally have the characteristic bulge shape observed in many desert soils (Phillips, 1994). The peak concentrations range from 28 mg L^–1 at Site 35 to 29 900 mg L^–1 at Site 4 (Table 1). Most of the profiles (Fig. 3) show shallow Cl^– accumulations within the sandstone beneath the root zone and maximum concentrations occurring between 1.2 and 15 m below land surface (Table 1), with an average depth of about 5 m.

Net infiltration rates and uncertainties were calculated with the CMB method (Eq. [1], [A4]) using both (i) the average vadose zone Cl^– concentration of pore water from borehole samples between the soil–bedrock interface and the bottom of the Cl^– bulge and (ii) the average vadose zone Cl^– concentration below the Cl^– bulge. Because the Cl^– bulge at Sand Hollow is always in the sandstone and beneath the zone of root transpiration, calculations based on Cl^– concentrations within the bulge at Sand Hollow are considered to be net-infiltration rates. These rates likely represent net infiltration during the mid- to late Holocene and depend on the amount of Cl^– accumulation at each site. Based on pore-water Cl^– concentrations between land surface and the bottom of the Cl^– bulge, net-infiltration rates ranged from 0.03 ± 0.01 to 8 ± 4 mm yr^–1; rates ranged from 0.5 ± 0.2 to 13 ± 7 mm yr^–1 using pore-water Cl^– concentrations below the bottom of the Cl^– bulge (Table 3).

It is assumed that all of the pore-water and groundwater Cl^– at Sand Hollow is of atmospheric origin. The Navajo Sandstone is a clean, well-sorted, eolian sandstone containing no known evaporite or other salt deposits. However, because of the upward advective movement into the Navajo Sandstone of Cl^–-rich brines from underlying formations containing evaporite deposits, as documented at other study sites (Kimball, 1992; Naftz et al., 1997; Heilweil et al., 2000), Cl^–/Br^– ratios were examined to evaluate potential Cl^– contributions from geologic sources. Such geologic sources of Cl^– typically have Cl^–/Br^– ratios exceeding 1000, and the ratios increase with increasing Cl^– concentration (Davis et al., 1998). However, no such trend is evident in groundwater from Sand Hollow, and groundwater Cl^–/Br^– ratios are always <500 (Fig. 4 ), with a mean and standard deviation of 230 ± 80 (n = 31). The mean and standard deviation of Cl^–/Br^– ratios for the atmospheric deposition samples is 125 ± 95 (n = 8; Table 2). This is consistent with the range of 100 to 200 reported by Davis et al. (2004) for the southwestern United States. Furthermore, vadose zone pore-water Cl^–/Br^– ratios generally increase from land surface to the water table (Heilweil, 2003), consistent with a doubling in the mean Cl^–/Br^– ratio from 125 for atmospheric-deposition samples to 230 for groundwater samples. This indicates that some unsaturated zone process, such as preferential uptake of Br^– by plant roots, influences groundwater Cl^–/Br^– ratios rather than a geologic source of Cl^–.

View larger version (12K): [in this window] [in a new window]   Fig. 4. Relation of chloride/bromide ratio to chloride concentration in water from selected wells in Sand Hollow, UT.

Tritium
The estimated ³H concentration of precipitation at Sand Hollow from 1950 to 2000 is shown in Fig. 5 , based on a distance-weighted average from monitoring stations at Albuquerque, NM; Flagstaff, AZ; and Salt Lake City, UT. The maximum ³H concentration of precipitation during the aboveground nuclear testing in 1963 was estimated to exceed 2000 TU. Present ³H concentrations in precipitation at Sand Hollow are near or slightly above the pre-bomb background levels. Three samples of recent precipitation (1999–2001) had measured ³H concentrations ranging from 9.1 to 21.0 TU (Table 2). A surface-water sample collected during an ephemeral flow event on 9 Nov. 2002 had a ³H concentration of only 2.3 TU, possibly caused by a dilution with older (lower ³H) shallow vadose zone pore waters flushed into the wash by the infiltrating precipitation.

View larger version (14K): [in this window] [in a new window]   Fig. 5. Estimated tritium concentration of precipitation at Sand Hollow, UT, 1953 to 2000.

View larger version (28K): [in this window] [in a new window]   Fig. 6. Tritium concentration of pore-water and groundwater samples collected from selected boreholes in Sand Hollow, UT.

At the seven sites where both ³H methods were applied, the ratio of the mass-balance to depth-to-peak net-infiltration rates was used as a correction factor for the sites where only the mass-balance method was applied. This correction factor ranged from 1.5 to 2.3; therefore, the reported net-infiltration rates based on borehole ³H in Table 3 for the four sites without TDTP rates (2, 12, 27, 38) are the TMB method rates of Table 4 multiplied by an average correction factor of 2.0. The use of this correction factor assumes that the depth-to-peak method is more accurate than the mass-balance method, which requires an estimate of historical ³H precipitation concentrations. Use of the average annual weighted masses of ³H in precipitation may be the cause of this underestimation of actual net infiltration. As determined from groundwater stable isotope ratios that are more depleted than the average annual concentrations, most net infiltration and recharge at Sand Hollow occurs in the winter. This is also the season of minimum atmospheric ³H concentrations in the northern hemisphere, which would cause the actual mass of ³H in winter recharge to be less than predicted using average annual ³H concentrations. This smaller number in the denominator of Eq. [5] would mean higher actual TMB recharge rates, possibly similar to the TDTP rates.

Groundwater ³H concentrations within Sand Hollow ranged from below detection (0.01) to almost 7 TU (Table 3). Precipitation falling at Sand Hollow more than 50 yr ago would likely have a present-day ³H concentration of <0.5 TU because of radioactive decay. Groundwater ³H concentrations >0.5 TU indicate that precipitation during the past 50 yr has reached the water table. Wells containing water with ³H concentrations >0.5 TU at Sand Hollow are screened within 4 m of the water table (Heilweil, 2003). There is an inverse correlation (r² = 0.77, n = 45) between Cl^– concentrations and ³H concentrations in groundwater at Sand Hollow (Fig. 7 ), indicating that high-recharge areas correspond to sites where infiltrating precipitation undergoes less evaporative solute concentration in the root zone.

View larger version (11K): [in this window] [in a new window]   Fig. 7. Relation of tritium concentration to chloride concentration in water from selected wells in Sand Hollow, UT.

[6]

View larger version (20K): [in this window] [in a new window]   Fig. 8. Relation between stable-isotope ratios of deuterium and oxygen in vadose zone pore water, water samples from selected wells, and precipitation in Sand Hollow, UT.

Stable-isotope ratios in vadose zone pore waters range from –47 to –113 for ²H and from +0.5 to –14.0 for ¹⁸O and show most variability in the shallower part of the vadose zone, converging to the groundwater value at or near the water table (Fig. 9 ). This indicates that little pore-water mixing occurs in the vadose zone compared with that within the aquifer. Pore-water samples from high-recharge sites (Sites 9, 43, 44, 50) have a mean ²H concentration of –88, whereas pore waters from low-recharge sites (Sites 2, 4, 12, 27, 37, 38, 39, 46) have a mean ²H ratio of –75 and follow an evaporative trend below the LMWL (Fig. 8). The slope of this evaporative trend is 3.1, similar to the calculated slope of about 4 for arid settings with low relative humidity (Gat, 1971). The evaporative shift for each pore-water sample was quantified by calculating the expected ¹⁸O ratio with the measured ²H ratio and the LMWL given in Eq. [6]. The evaporative shift calculated for each site is the average value from land surface to the water table (Table 3). The smallest shift of 2.1 at the highest recharge site (Site 9) and the largest shift of 4.7 at the lowest recharge site (Site 4) supports the hypothesis that less evaporative loss occurs where infiltrating water moves quickly beneath the root zone in the more active recharge areas of the basin.

View larger version (26K): [in this window] [in a new window]   Fig. 9. Stable isotope ratio of deuterium in pore-water and groundwater samples collected from selected wells in Sand Hollow, UT.

	DISCUSSION

TOP EXECUTIVE SUMMARY ABSTRACT INTRODUCTION SITE DESCRIPTION THEORY MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS APPENDIX: UNCERTAINTY ANALYSIS REFERENCES

 
Several environmental tracer techniques were used to evaluate vadose zone processes, compare net-infiltration and recharge rates, and examine spatial and temporal variability. This approach provides a higher level of confidence than that obtained from any one method by itself and allows for comparison of net infiltration and recharge at different time scales. At Sand Hollow, vadose zone ³H targets modern (post-1950s) net infiltration whereas vadose zone Cl^– provides a longer-term record of net infiltration (as long as about 12 000 yr at Sand Hollow). Groundwater Cl^– provides more spatially averaged recharge estimates than vadose zone methods because of horizontal hydraulic gradients and mixing in the well screen. Stable isotopes (²H and ¹⁸O), although not providing recharge rates, are useful for evaluating both spatial and temporal differences in net infiltration and recharge.

Smooth vadose zone Cl^– and ³H bulges characteristic of those occurring in unconsolidated desert soils were observed in many of the boreholes, an unusual finding for fractured bedrock and indicative of the predominance of matrix-dominated vadose zone flow in the Navajo Sandstone. This is contrary to an earlier study of shallow vadose zone trenches at Sand Hollow, which found lower solute accumulations in high-angle fractures than in the adjacent matrix at depths between 3 and 6 m below the bedrock surface (Heilweil and Solomon, 2004). This indicates that preferential flow associated with fractures may be most prominent in the shallowest part of the vadose zone, possibly caused by ponding at the bedrock contact during and after large precipitation events and channeling through near-surface caliche-coated fractures. This water may subsequently be imbibed into more-permeable adjacent sandstone at greater depths where fracture coatings were observed to decrease, explaining the apparent matrix-dominated environmental tracer distributions observed in boreholes. This is supported by a recent numerical modeling study of net infiltration in the Navajo Sandstone (Ludwig, 2003) that demonstrated that unlined fractures with 1-mm apertures (typical of those observed in trenches at Sand Hollow) do not readily saturate and therefore do not act as flow conduits. Rather, water is quickly imbibed by the highly porous sandstone.

Spatial Variability
Net-infiltration and recharge rates at the various sites in Sand Hollow vary greatly, with most recharge occurring in areas that receive focused run-on or have exposed bedrock (sites 9, 43, 44; Fig. 2). Interestingly, substantial amounts of direct net infiltration and recharge also occur beneath coarser-grained sandy soils away from washes (sites 35, 50). This is contrary to findings generally reported from alluvial desert basins receiving similar amounts of precipitation, where little or no basin-floor recharge away from washes occurs under present climatic conditions (Phillips, 1994; Prudic, 1994; Tyler et al., 1996, Andraski, 1997; Izbicki et al., 2002). The results of this study indicate that a higher percentage of water is able to pass beyond the maximum rooting depth and become net infiltration where permeable bedrock formations are exposed or shallowly buried by coarse-grained soils. Low net-infiltration and recharge rates are measured at sites with fine-grained soils receiving little surface-water runoff (sites 2, 4, 12, 27, 37, 38, 39, 46). These loamy soils have a larger water storage capacity, making more infiltrating water available for recycling back to the atmosphere via evaporation and transpiration. Large variations in net-infiltration rates at other study sites in the desert southwestern U.S. receiving similar amounts of precipitation have also been attributed to soil coarseness (Scanlon et al., 2003).

The tracers used in this study generally show similar trends in spatial variability, even though net-infiltration and recharge rates at individual sites vary depending on the particular environmental-tracer method. For example, sites having higher Cl^–- and ³H-based net-infiltration and recharge estimates generally display smaller amounts of vadose zone Cl^– accumulation, higher shallow groundwater ³H concentrations, smaller vadose zone ¹⁸O shifts, and less depleted ²H values than sites with lower estimated rates (Table 3). Net-infiltration rates calculated from ³H concentrations are well correlated with rates determined with Cl^– from both within and below the Cl^– bulge (r² = 0.83 and 0.67, respectively). Net-infiltration and recharge rates calculated using the three different Cl^– mass-balance methods at each site (within the bulge, below the bulge, below the water table) are also generally well correlated. However, at Sites 39 and 46, pore-water Cl^– concentrations decrease substantially when crossing from the deep vadose zone into the underlying water table, resulting in much higher groundwater Cl^–-based rates (15 ± 7 mm yr^–1) than those based on vadose zone Cl^– within the bulge (0.1 ± 0.05 and 0.05 ± 0.02 mm yr^–1, respectively) and vadose zone Cl^– beneath the bulge (3 ± 1 and 4 ± 1 mm yr^–1, respectively). This may be caused by higher-elevation areas with higher recharge rates and a large horizontal component of groundwater flow.

Although they cannot provide direct estimates of recharge rates, groundwater stable isotope ratios and groundwater ³H concentrations can be used for evaluating spatial variability of recharge. Using the average ²H values of –75 for the low-recharge site vadose zone pore waters, –88 for the high-recharge site vadose zone pore waters, and –86 for groundwater, a two-end-member mixing model indicates that about 85% of groundwater recharge in Sand Hollow occurs at high-recharge sites. The spatial distribution of groundwater ³H data indicates that this recharge is concentrated primarily in the part of the basin covered by coarser soils or exposed bedrock (Tables 1 and 3, Fig. 2). Only 5 of 45 groundwater sites within Sand Hollow had values of 0.6 TU or more, an indicator of recent recharge arriving at the water table.

Site-Specific Temporal Variability
Temporal variation in the spatial distribution of net infiltration at Sand Hollow occurs and is likely caused by processes such as climate change (variability in precipitation, temperature, vegetation), caliche formation, erosion, the migration of eolian sand dunes, and changing ephemeral wash locations. The latter has likely occurred at Sites 27 and 38. Unlike the other 11 boreholes, these sites are located in or near the main ephemeral wash of Sand Hollow and do not fit the trend of increasing peak concentration of the shallow Cl^– bulge with increasing total vadose zone Cl^– accumulation (Table 1). Considering the total accumulation of Cl^– in the vadose zone at Site 27 (72 ± 8 mg 100 mm^–2), the peak Cl^– concentration is less than expected. Although this site is situated in the main ephemeral wash, solute distributions in nearby trench excavations indicate that the wash had recently migrated from a location farther to the east (Heilweil and Solomon, 2004). The lower-than-expected peak concentration of the shallow Cl^– bulge at this site is consistent with a recent shift to higher net infiltration beneath the active wash channel, which would push downward and dilute a preexisting shallow Cl^– bulge. The high total cumulative vadose zone Cl^– at the site indicates that the previously accumulated Cl^– has not yet been flushed out of the vadose zone.

The opposite trend may have occurred at Site 38. The very shallow 2-m depth of the Cl^– peak and the higher-than-expected 14 700 mg L^–1 peak concentration at this site (Table 1) indicates a recent change to lower net-infiltration rates and increased run-on of Cl^–. This site is located adjacent to a section of the ephemeral wash that was impounded by a small levee built by cattle ranchers about 50 yr ago (L. Jessup, Washington County Water Conservancy District, personal communication, 2001) and is now covered with about 0.5 m of fine-grained silts that have settled out of surface-water runoff. Before the existence of the levee, the site was likely a higher recharge site on the natural wash channel. This would explain the low Cl^– concentrations measured below a 3-m depth, the poor correlation between total Cl^– accumulation and peak Cl^– concentration, the low vadose zone ³H-based net-infiltration rate, and a small average vadose zone ¹⁸O evaporative shift (2.2) similar to the highest recharge sites (Table 3). The very high Cl^– concentration in the shallow subsurface likely represents a combination of salt accumulation from run-on and evaporative enrichment associated with decreased infiltration through the shallow, low-permeability silts at the site.

Basin-Scale Temporal Variability
Net infiltration at Sand Hollow is episodic and dominated by large but infrequent precipitation events. Moving 30-d precipitation totals for St. George show that two-thirds (10 of 15) of the 30-d periods during the 20th century with more than 100 mm of precipitation occurred during 1957 to 1997. The larger amounts of precipitation during the latter part of the 20th century may have flushed down Cl^– that previously accumulated in the shallow soils. This vadose zone Cl^– accumulation in Sand Hollow is driven by evaporatranspiration within the root zone, consisting primarily of soils above the bedrock contact. This would account for the average 5-m depth for vadose zone Cl^– peaks, much greater than the maximum root zone depth of about 1.5 m at the soil–bedrock interface (Table 1). Periodic flushing, or alternating wetter and drier periods, may also explain double peaks in the Cl^– bulges at Sites 2, 4, and 39 in Fig. 3 (and also at Sites 9, 27, 44, and 50, which are not discernible in the figure). Such downward flushing of Cl^– has been both observed (Stonestrom et al., 2003; Scanlon et al., 2005) and modeled (Scanlon et al., 2003) in the Armagosa Desert of southwestern Nevada.

For the group of 11 sites with vadose zone ³H information, the ³H-based net-infiltration rates are much higher (2.4 ± 1.2 mm yr^–1 to more than 57 ± 7 mm yr^–1; geometric mean of 9 mm yr^–1; Table 3) than rates calculated by using shallow vadose zone Cl^– within the bulge (0.08 ± 0.04 to 8 ± 4 mm yr^–1; geometric mean of 0.4 mm yr^–1). At the high-recharge sites, the tritium peak is generally present at greater depths than the bottom of the Cl^– bulge, indicating that the bulge CMB rates should be representative of modern (post-1950s) net infiltration. However, the much-lower CMB rates, compared with ³H-based rates, indicate that differing processes affect the two tracers. Possibilities include (i) downward vapor transport of ³H, (ii) run-on of additional Cl^– at each site, or (iii) relict Cl^– from a previously dry period before the latter half of the 20th century not completely flushed downward or diluted by recently higher net infiltration. Thermally induced vapor transport of tritiated water has been previously invoked to explain observed deeper bomb-pulse ³H (1963) peaks versus ³⁶Cl (1954) peaks in the vadose zone at sites in New Mexico and Texas (Phillips et al., 1988; Scanlon, 1992). However, the ³H-based net-infiltration rates at the high-recharge sites in Sand Hollow are a factor of about seven times higher than the Cl^– bulge CMB rates, which is difficult to explain by vapor transport, especially because the generally high water contents (5.4–7.8% Table 1) indicate the predominance of liquid-phase flow. Alternatively, additional Cl^– input associated with run-on could potentially increase Cl^– concentrations and decrease apparent CMB recharge rates. However, the ratio of ³H versus Cl^–-bulge net-infiltration rates of 4 to 8 at Sites 9 and 44 beneath intermittent washes are similar to ratios ranging from 3 to 13 at Sites 35, 43, and 50 in interdrainage areas away from washes with no expected run-on. Sites having substantial amounts of additional Cl^– input from run-on would presumably have much higher ratios than sites not receiving run-on. Therefore, the explanation of relict Cl^– from previous centuries under drier conditions remains as the most likely hypothesis for the lower CMB rates, although some vapor-phase ³H transport is also possible. This is supported by calculated Cl^– accumulation inventories within the bulges ranging from 50 to 400 yr, representing older net infiltration than the ³H peaks (Table 1).

The ³H/Cl^– bulge net-infiltration rate ratios at the low-recharge sites are much higher (7–80, mean = 40) than the high-recharge site ratios (3 to 13, mean = 7). At these low-recharge sites, the bottom of the Cl^– bulge is much deeper than the bomb-pulse tritium, so this Cl^– clearly represents conditions before the latter half of the 20th century. The Cl^– accumulation within these bulges ranges from about 800 to 7000 yr (Table 1). Even if the CMB rates at the low net-infiltration sites are adjusted upward using an average ³H/Cl^– bulge net-infiltration ratio of 7 for the high-recharge sites, the ³H-based rates are still, on average, about sixfold higher. This indicates that net-infiltration rates during the mid- to late-Holocene were much lower than in recent decades.

Comparison of ³H-based net-infiltration rates to deeper vadose zone CMB net-infiltration rates beneath the Cl^– bulge is not as straightforward. For the group of high-recharge sites (9, 35, 43, 44, 50), vadose zone ³H-based net-infiltration rates (10 ± 2 mm yr^–1 to more than 57 ± 7 mm yr^–1) are higher than vadose zone Cl^–-based rates from beneath the bulge (3 ± 2 to 13 ± 7 mm yr^–1), also indicating higher rates in recent decades than during previous centuries. For the group of low-recharge sites (2, 12, 27, 37, 38, 39), however, vadose zone ³H-based net-infiltration rates (2.4 ± 1.2 to 4.6 ± 2.1 mm yr^–1) are similar to vadose zone Cl^–-based rates beneath the bulge (2 ± 1 to 5 ± 2 mm yr^–1). This deeper Cl^– represents the oldest preserved vadose zone information on net infiltration at Sand Hollow and indicates that rates earlier in the Holocene were similar to the higher rates of recent decades. Unlike the high-recharge sites, the total amount of Cl^– beneath the bulge at these low-recharge sites is a small fraction of the total vadose zone Cl^– accumulation (3–33%), consistent with higher rates earlier in the Holocene.

Four of the 12 borehole sites drilled to the water table have shallow groundwater ³H concentrations of more than 0.5 TU, indicating some recent recharge. The CMB recharge rates based on groundwater chloride at these sites range from 9 ± 4 to 60 ± 29 mm yr^–1, very similar to the vadose zone ³H-based net-infiltration rates of 10 ± 2 to >57 ± 7 mm yr^–1. This close agreement between water-table and vadose zone tracers in areas with short vadose zone residence times indicates an equilibration between net infiltration, vadose zone movement, and groundwater recharge. This contrasts recent studies showing long-term drying conditions and upward water transport in alluvial interbasin areas with deep water tables in response to the Pleistocene–Holocene climate shift (Walvoord et al., 2002a; Scanlon et al., 2003).

Eight of the 12 borehole sites drilled to the water table have shallow groundwater ³H concentrations <0.5 TU, indicating recharge that occurred more than 50 yr ago. Recharge rates based on groundwater Cl^– concentrations at these eight boreholes rang