Changes in hydraulic head at Devils Hole due to regional tectonic deformation were estimated using aquifer properties and the volumetric strain field present through the Great Basin in the western United States. Devils Hole is a large fault cavern located along a 15 km spring discharge line in a carbonate-rock aquifer, 240 km west of Las Vegas, Nevada. Geodetic measurements indicate that extensional strain in the Devils Hole area is oriented N 65 W and has a rate of 8 nanostrain/yr. Changes in hydraulic head due to strain were calculated and then used as initial conditions in a calibrated numerical ground-water flow model. Results of this analysis show that tectonic deformation could produce up to 0.02 cm/yr of water-level decline in Devils Hole. However, this rate is relatively small in comparison to rates caused from other factors such as natural recharge and ground-water pumping.
 Devils Hole provides a fragile environment for an endangered species of fish, Cyprinodon diabolis, commonly known as the Devils Hole pupfish. The pupfish are dependent on a critical water level in Devils Hole for survival as fluctuations in the water level can have effects on reproduction, nutrients and temperatures [Riggs and Deacon, 2002]. Water levels are measured every 15 minutes to the nearest 0.3 cm and indicate that the monthly mean water levels have been declining at a rate of about 0.7 cm/yr since 1988.
 It is speculated that hydraulic head in the Devils Hole area of Nevada may be under the influence of tectonic deformation caused by extensional crustal strain throughout the Basin and Range province of southern Nevada. Hydrodynamic responses to poroelastic strain have been modeled as pore pressure changes due to coseismic crustal contraction and extension [e.g., Cutillo and Ge, 2006]. In this study, the effects of regional tectonic deformation were investigated on ground-water levels by applying strain to the porous media that houses Devils Hole to predict a hydraulic-head change in Devils Hole.
 Previous studies on water levels in Devils Hole have focused on earthquake induced water-level fluctuations and ground-water pumping. Earthquakes were shown to have instantaneous effects on hydraulic head with recovery periods from days to months [Cutillo and Ge, 2006]. Ground-water pumping near Devils Hole during the late 1960's to early 1970's drew the water level down about 1 m and was subsequently reduced [Dudley and Larson, 1976]. Additional studies have shown that distant ground-water pumping can also cause decline in hydraulic head at Devils Hole [Bedinger and Harrill, 2006].
2. Study Area
 Devils Hole is a fluid-filled fault cavern formed in a carbonate-rock aquifer near the California-Nevada border (Figure 1). The carbonate-rock aquifer housing Devils Hole ranges from about 1,500 m to 9,150 m in thickness [Harrill and Prudic, 1998] and is areally extensive comprising a large portion of the Great Basin. Adjacent to Devils Hole the carbonate rock that houses the aquifer outcrops to form a ridge oriented generally northwest-southeast. The highly fractured and faulted outcrop is indicative of the aquifer below ground surface where the primary conduits of flow are the fractures and joints [Winograd and Thordarson, 1975; Dudley and Larson, 1976; Plume, 1996].
 The ground water that fills Devils Hole originates from the northeast as recharge in areas of high elevation as infiltrating precipitation or yearly melt [Laczniak et al., 1996; Harrill and Prudic, 1998; Fenelon and Moreo, 2002]. Ground water then disperses through the carbonate-rock aquifer, which is confined, in a general southwestern direction towards Ash Meadows. Devils Hole and other natural springs reside where the carbonate-rock aquifer is offset by the Ash Meadows fault system, which includes the Gravity fault (Figure 1). Juxtaposition on the fault plane places less permeable Quaternary and Tertiary sediments adjacent to the highly transmissible carbonate-rock aquifer forcing ground water to be ejected to the surface through fractures and faults (Figure 2).
 The carbonate-rock aquifer is overlain by interbedded alluvial and volcanic units that act as both aquifers and aquitards [Denny and Drewes, 1965; Hunt and Mabey, 1966; Winograd and Thordarson, 1975; Carr, 1988]. These units are semi-consolidated to unconsolidated deposits that have varying hydraulic conductivity in the vertical and lateral directions due to hydrogeologic changes [Harrill and Prudic, 1998]. The hydraulic conductivities of these units are minimal when compared to the fractured network of the carbonate-rock aquifer.
 Crustal strain rates vary throughout the Basin and Range Province of the western United States. Strain monitored at the Nevada Test Site (Figure 1) by geodetic measurements indicates a strain rate of 8 to 50 nanostrain/yr with extension occurring in the northwest orientation [Wernicke et al., 1998; Savage, 1998]. This measurement was taken roughly 40 km north of Devils Hole (25 km from study area) and is used in the following numerical calculations. Devils Hole is also located on the eastern side of the Walker Lane tectonic belt which propagates northwest from Arizona to California. Extensional rates present in the Walker Lane Belt nearest the study area are in agreement with the measurements taken at the Nevada Test Site [Bellier and Zoback, 1995].
3. Conceptual and Analytical Model
 To investigate the implications of tectonic deformation on the hydraulic head in Devils Hole, the current strain rate, fluid density, matrix compressibility and Skempton's coefficient are considered. The strain rate is used to determine volumetric strain. The remaining parameters are used to evaluate water level fluctuations in conjunction with the volumetric strain. The primary input parameter is the horizontal strain rate, which can be converted into a volumetric strain, ɛv, if effects in the vertical orientation are assumed to be negligible and the strain rate is assumed constant for a length of time yielding units of strain. Initial total volume, of a cube is determined using cube dimensions dx, dy and dz. If axes of a frame of reference are aligned with cube orientations, cube dimension dx becomes dx + ɛsinθdx due to strain and cube dimension dy becomes dy − ɛcosθdy. A new total volume, is computed per cube using,
where dx, dy and dz represent cube dimensions [L], θ is the orientation of strain, and ɛ is strain [strain]. The difference between the initial total volume and the new total volume is the volume change due to strain. Therefore, volumetric strain, ɛv, is calculated using the following equation:
 Volumetric strain is attributed to water-level fluctuations through changes in pore pressure. Formulations by Biot  followed byRice and Cleary  demonstrate this in a simple equation:
where ΔP is fluid pressure change [M L−1 T−2] and C is a function of the properties of the porous media defined by Rice and Cleary  as:
where B is Skempton's coefficient and K is the bulk modulus [M L−1 T−2]. For undrained conditions, Skempton's coefficient is defined as the ratio of induced pore pressure to the change in applied stress, and has values between zero and one. Typical values of Skempton's coefficient for saturated rock are between 0.5 and 1.0 [Wang, 2000]. The bulk modulus is the inverse of matrix compressibility [Turcotte and Schubert, 2002]. The negative sign indicates that an increase in compressive stress with a negative value would lead to a positive value for an increase in pore pressure. Fluid pressure change, derived in equation 3, is used to compute a hydraulic-head change using the following equation [Domenico and Schwartz, 1990]:
where Δh is hydraulic-head change [L], ρw is the density of water [M L−3] and g is the gravitational constant [L T−2].
4. Numerical Model
 A finite-difference numerical model was developed to investigate the effects of volumetric strain on the rate of water-level decline in Devils Hole due to the complex ground-water flow system present in the study area (Figure 3). The three dimensional numerical model was constructed using Visual MODFLOW [Schlumberger, 2005]. The model grid consists of 120 columns in the east-west and 100 rows in the north-south orientations. Each grid cell is 279.9 m east-west and 271.9 m north-south. The model has six layers vertically. From top down they are: basin-fill aquifer, alluvial confining unit, upper volcanic and sedimentary rock, lower volcanic and sedimentary rock, carbonate-rock aquifer, and lower confining unit (aquitard). These represent the hydrogeologic units described by Sweetkind et al. . The north and northeast boundaries of the model are assigned a constant hydraulic head condition determined from water table maps. The remaining eastern boundary and southern boundary are considered as no flow representing a ground-water divide and the Montgomery fault, respectively. The western boundary of the model is set an arbitrary distance away from the Ash Meadows fault system as a no flow boundary. In setting the boundary at a distance, the impacts of the Ash Meadows fault system and Ash Meadows spring line on hydraulic head could be modeled without compromising boundary condition effects. The Ash Meadows fault system is represented by cells of low hydraulic conductivity.
 Hydraulic conductivities, evapotranspiration rates and storage values are based on values from the larger regional flow model by San Juan et al. . These values were subsequently adjusted to calibrate the model; however, they remain within the bounds specified by San Juan et al. . Typical values for hydraulic conductivity ranged from 0.004 to 0.05 m/d for alluvial units and 0.1 to 100 m/d for the carbonate-rock aquifer. Evapotranspiration was set to 1.0 × 10−4 m/d and storage values resided between 1.5 × 10−5 and 4.5 × 10−5 m−1. Recharge rates used in the transient model were derived from the precipitation data in the Spring Mountains [Fenelon and Moreo, 2002]. Infiltration of 5 to 10 percent of precipitation was used.
 A steady state model was calibrated using nine observation wells in the model domain (Figure 3) and the mass balance of flow for the study area. A high correlation was achieved upon calibration between observation wells and observed well values noted by a R2 value of 0.98. Volumetric recharge determined by steady state calibration for the purpose of establishing initial conditions was 9.2 × 104 m3/d. This value is similar to an estimate made by Winograd and Thordarson  of 9.3 × 104 m3/d. The total estimated ground-water discharge from the study area can be broken down into components of spring discharge, underflow and evapotranspiration, where values of 5.4 × 104 m3/d, 1.0 × 104 m3/d and 5.5 × 104 ± 2.5 × 104 m3/d are reported, respectively [Winograd and Thordarson, 1975]. Discharge occurred through drains, underflow in the southwest, and evapotranspiration in the model. Calibrated results for these values were 4.6 × 104 m3/d, 9.6 × 103 m3/d, and 3.5 × 104 m3/d, respectively, which are in general agreement with observed values. Ground-water pumping was not considered.
 Multiple one-year simulations were coupled to form the transient model. The reported strain rate of 8 nanostrain/yr trending N 65 W (equivalent to approximately 1.7 mm/yr of crustal elongation over 34 km) [Wernicke et al., 1998] at the Nevada Test Site was used to derive a volumetric strain value (equations 1 and 2) of −3.87 × 10−9 for the duration simulated. The rate of hydraulic-head change, derived from volumetric strain, applied in the numerical model was −0.001 m/yr. Equations 3 through 5 were used to produce this rate with values of B = 1 and matrix compressibility = 4 × 10−10 m2/N [Bredehoeft, 1992]. This strain value represents a maximum case scenario of hydraulic-head change due to extension, which was used to explore the extent of tectonic deformation on hydraulic head in Devils Hole. Assuming B = 1 maximized pore pressure change. The value for matrix compressibility was at the lower range of values determined near the study area to have a maximum effect.
 Results of transient numerical simulations indicated that the rate of water-level decline that may be attributed to tectonic deformation in the Great Basin area is small relative to the observed water-level decline in Devils Hole. The average rate of water-level decline from imposed tectonic extension was estimated to be 0.02 cm/yr (Figure 4). This is shown by a water level difference of only 0.2 cm at the conclusion of 10-year simulations – one with the imposed strain field and one without. Recharge had more influence than tectonic deformation on water levels indicated by pronounced fluctuations in both curves that overpower the difference between the curves (Figure 4). It is assumed that these fluctuations would have diminished if strain had a greater influence. Additional ground-water-level decline rates were simulated for a range of strain orientations and magnitudes consistent with the observed rates and orientations near the study area. The results of these simulations yielded little change from the case shown in Figure 4, and are discussed further below. The overall simulated rise in water level apparent on Figure 4 is attributed to an un-modeled sink in the model that would cause a declining trend. A likely sink is distant ground-water pumping [Bedinger and Harrill, 2006].
 The study area is located in an extensional environment where effects in the horizontal are much more significant than in the vertical direction. Horizontal opening and closing of fractures and faults in the carbonate-rock aquifer is well documented at the surface and sub-surface by Carr ; therefore, volumetric strain was calculated only in the horizontal plane while the vertical component was ignored and assumed to have a minimal effect. This was further supported by sensitivity studies of altering strain magnitudes and orientations where horizontal strain orientations were varied and given greater strain rates [Robertson, 2006]. Results showed that for multiple orientations and strain rate simulations, Devils Hole water levels remained relatively constant.
 The transient numerical model was driven by recharge representing infiltrating average annual precipitation in the principal source areas for the Ash Meadows drainage basin nearest the study area. It is realized that there is a lag in ground-water travel time from monitored precipitation locations to the boundaries of the study area [Fenelon and Moreo, 2002]. During this lag time, the monitored fluctuations in recharge (from varying precipitation events) will likely dissipate to create relatively constant recharge values at the study area boundaries. Therefore, a low percentage of the observed yearly precipitation rates at source areas were used as recharge in the numerical model to maintain relatively constant recharge. The numerical simulations show that although recharge was at a minimum, it still had a dominating effect over volumetric strain on the water levels in Devils Hole (Figure 4).
 Sensitivity analyses conducted by Robertson indicated that Devils Hole can be treated as a monitoring well in numerical simulations at the scale of this study . This was simulated by altering the vertical hydraulic conductivities of the grid cells containing Devils Hole. Hydraulic conductivities were raised 10 orders of magnitude allowing water to flow freely in the vertical direction. Results of these numerical simulations had little to no change from prior simulations where vertical hydraulic conductivities were similar to the surrounding cells.
 The Ash Meadows fault system, in the immediate vicinity of Devils Hole, has large effects on the Ash Meadows spring line. Faulting juxtaposes the carbonate-rock aquifer against less permeable alluvial sediments (Figure 2). Secondary fault and fracture networks allow ground water to exit to the surface at spring locations. The transmissivity of the Ash Meadows fault system is currently unknown. It was assumed by Winograd and Thordarson  that there was an amount of ground water dissipating through the fault zone (underflow), at depth, rather than finding an egress through the spring line (Figure 2). During numerical simulations, the volume of ground water loss through the Ash Meadows fault system was calibrated to 1.0 × 104 m3/d in the south-west section of the study area [Winograd and Thordarson, 1975]. Information regarding the lateral transmissivity of the fault zone could further refine estimates of ground-water seepage and better constrain properties of the spring line.
 It was estimated that the contribution of tectonic deformation to the rate of water-level decline in Devils Hole could be as high as 0.02 cm/yr. This was determined by using the currently observed strain rates near the study area with a numerical model that incorporated geologic properties of rocks, such as hydraulic conductivity, storativity and compressibility; geological features, such as faults and ground-water divides; evapotranspiration; and recharge. Direct measurements of strain rates, precipitation rates, and evaporation rates in Devils Hole were not available at the time of study. In the future, these measurements along with fault properties could provide further information to better constrain numerical models and understand ground-water-level fluctuations.
 Support for this research was made possible by the National Park Service and the University of Colorado at Boulder. Comments from two anonymous reviewers have greatly improved the quality of the paper.