Geology topics

Andrew J. Long

Publications and source records attributed to Andrew J. Long.

At least 37 records · Page 2Linked to original sources

USGS2013

Comparison of a karst groundwater model with and without discrete conduit flow

Karst aquifers exhibit a dual flow system characterized by interacting conduit and matrix domains. This study evaluated the coupled continuum pipe-flow framework for modeling karst groundwater flow in the Madison aquifer of western South Dakota (USA). Coupled conduit and matrix flow was simulated within a regional finite-difference model over a 10-year transient period. An existing equivalent porous medium (EPM) model was modified to include major conduit networks whose locations were constrained by dye-tracing data and environmental tracer analysis. Model calibration data included measured hydraulic heads at observation wells and estimates of discharge at four karst springs. Relative to the EPM model, the match to observation well hydraulic heads was substantially improved with the addition of conduits. The inclusion of conduit flow allowed for a simpler hydraulic conductivity distribution in the matrix continuum. Two of the high-conductivity zones in the EPM model, which were required to indirectly simulate the effects of conduits, were eliminated from the new model. This work demonstrates the utility of the coupled continuum pipe-flow method and illustrates how karst aquifer model parameterization is dependent on the physical processes that are simulated.

South Dakota↗

USGS2013

Prediction, time variance, and classification of hydraulic response to recharge in two karst aquifers

Many karst aquifers are rapidly filled and depleted and therefore are likely to be susceptible to changes in short-term climate variability. Here we explore methods that could be applied to model site-specific hydraulic responses, with the intent of simulating these responses to different climate scenarios from high-resolution climate models. We compare hydraulic responses (spring flow, groundwater level, stream base flow, and cave drip) at several sites in two karst aquifers: the Edwards aquifer (Texas, USA) and the Madison aquifer (South Dakota, USA). A lumped-parameter model simulates nonlinear soil moisture changes for estimation of recharge, and a time-variant convolution model simulates the aquifer response to this recharge. Model fit to data is 2.4% better for calibration periods than for validation periods according to the Nash–Sutcliffe coefficient of efficiency, which ranges from 0.53 to 0.94 for validation periods. We use metrics that describe the shapes of the impulse-response functions (IRFs) obtained from convolution modeling to make comparisons in the distribution of response times among sites and between aquifers. Time-variant IRFs were applied to 62% of the sites. Principal component analysis (PCA) of metrics describing the shapes of the IRFs indicates three principal components that together account for 84% of the variability in IRF shape: the first is related to IRF skewness and temporal spread and accounts for 51% of the variability; the second and third largely are related to time-variant properties and together account for 33% of the variability. Sites with IRFs that dominantly comprise exponential curves are separated geographically from those dominantly comprising lognormal curves in both aquifers as a result of spatial heterogeneity. The use of multiple IRF metrics in PCA is a novel method to characterize, compare, and classify the way in which different sites and aquifers respond to recharge. As convolution models are developed for additional aquifers, they could contribute to an IRF database and a general classification system for karst aquifers.

South Dakota, Texas↗

USGS2012

Multivariate statistical approach to estimate mixing proportions for unknown end members

A multivariate statistical method is presented, which includes principal components analysis (PCA) and an end-member mixing model to estimate unknown end-member hydrochemical compositions and the relative mixing proportions of those end members in mixed waters. PCA, together with the Hotelling T 2 statistic and a conceptual model of groundwater flow and mixing, was used in selecting samples that best approximate end members, which then were used as initial values in optimization of the end-member mixing model. This method was tested on controlled datasets (i.e., true values of estimates were known a priori) and found effective in estimating these end members and mixing proportions. The controlled datasets included synthetically generated hydrochemical data, synthetically generated mixing proportions, and laboratory analyses of sample mixtures, which were used in an evaluation of the effectiveness of this method for potential use in actual hydrological settings. For three different scenarios tested, correlation coefficients ( R 2 ) for linear regression between the estimated and known values ranged from 0.968 to 0.993 for mixing proportions and from 0.839 to 0.998 for end-member compositions. The method also was applied to field data from a study of end-member mixing in groundwater as a field example and partial method validation.

Journal of Hydrology↗

USGS2012

Groundwater flow, quality (2007-10), and mixing in the Wind Cave National Park area, South Dakota

A study of groundwater flow, quality, and mixing in relation to Wind Cave National Park in western South Dakota was conducted during 2007-11 by the U.S. Geological Survey in cooperation with the National Park Service because of water-quality concerns and to determine possible sources of groundwater contamination in the Wind Cave National Park area. A large area surrounding Wind Cave National Park was included in this study because to understand groundwater in the park, a general understanding of groundwater in the surrounding southern Black Hills is necessary. Three aquifers are of particular importance for this purpose: the Minnelusa, Madison, and Precambrian aquifers. Multivariate methods applied to hydrochemical data, consisting of principal component analysis (PCA), cluster analysis, and an end-member mixing model, were applied to characterize groundwater flow and mixing. This provided a way to assess characteristics important for groundwater quality, including the differentiation of hydrogeologic domains within the study area, sources of groundwater to these domains, and groundwater mixing within these domains. Groundwater and surface-water samples collected for this study were analyzed for common ions (calcium, magnesium, sodium, bicarbonate, chloride, silica, and sulfate), arsenic, stable isotopes of oxygen and hydrogen, specific conductance, and pH. These 12 variables were used in all multivariate methods. A total of 100 samples were collected from 60 sites from 2007 to 2010 and included stream sinks, cave drip, cave water bodies, springs, and wells. In previous approaches that combined PCA with end-member mixing, extreme-value samples identified by PCA typically were assumed to represent end members. In this study, end members were not assumed to have been sampled but rather were estimated and constrained by prior hydrologic knowledge. Also, the end-member mixing model was quantified in relation to hydrogeologic domains, which focuses model results on major hydrologic processes. Finally, conservative tracers were weighted preferentially in model calibration, which distributed model errors of optimized values, or residuals, more appropriately than would otherwise be the case The latter item also provides an estimate of the relative effect of geochemical evolution along flow paths in comparison to mixing. The end-member mixing model estimated that Wind Cave sites received 38 percent of their groundwater inflow from local surface recharge, 34 percent from the upgradient Precambrian aquifer, 26 percent from surface recharge to the west, and 2 percent from regional flow. Artesian springs primarily received water from end members assumed to represent regional groundwater flow. Groundwater samples were collected and analyzed for chlorofluorocarbons, dissolved gasses (argon, carbon dioxide, methane, nitrogen, and oxygen), and tritium at selected sites and used to estimate groundwater age. Apparent ages, or model ages, for the Madison aquifer in the study area indicate that groundwater closest to surface recharge areas is youngest, with increasing age in a downgradient direction toward deeper parts of the aquifer. Arsenic concentrations in samples collected for this study ranged from 0.28 to 37.1 micrograms per liter (μg/L) with a median value of 6.4 μg/L, and 32 percent of these exceeded 10 μg/L. The highest arsenic concentrations in and near the study area are approximately coincident with the outcrop of the Minnelusa Formation and likely originated from arsenic in shale layers in this formation. Sample concentrations of nitrate plus nitrite were less than 2 milligrams per liter for 92 percent of samples collected, which is not a concern for drinking-water quality. Water samples were collected in the park and analyzed for five trace metals (chromium, copper, lithium, vanadium, and zinc), the concentrations of which did not correlate with arsenic. Dye tracing indicated hydraulic connection between three water bodies in Wind Cave.

South Dakota↗

USGS2012

Microgravity methods for characterization of groundwater-storage changes and aquifer properties in the karstic Madison aquifer in the Black Hills of South Dakota, 2009-12

A study of groundwater storage in the karstic Madison aquifer in the Black Hills of South Dakota using microgravity methods was conducted by the U.S. Geological Survey in cooperation with West Dakota Water Development District, South Dakota Department of Environment and Natural Resources, and Lawrence County. Microgravity measurements from 2009 to 2012 were used to investigate groundwater-storage changes and effective porosity in unconfined areas of the Madison aquifer. Time-lapse microgravity surveys that use portable high-sensitivity absolute and relative gravimeters indicated temporal-gravity changes as a result of changing groundwater mass. These extremely precise measurements of gravity required characterization and removal of internal instrumental and external environmental effects on gravity from the raw data. The corrected data allowed groundwater-storage volume to be quantified with an accuracy of about plus or minus 0.5 foot of water per unit area of aquifer. Quantification of groundwater-storage change, coupled with water-level data from observation wells located near the focus areas, also was used to calculate the effective porosity at specific altitudes directly beneath gravity stations. Gravity stations were established on bedrock outcrops in three separate focus areas for this study. The first area, the Spring Canyon focus area, is located to the south of Rapid City with one gravity station on the rim of Spring Canyon near the area where Spring Creek sinks into the Madison aquifer. The second area, the Doty focus area, is located on outcrops of the Madison Limestone and Minnelusa Formation to the northwest of Rapid City, and consists of nine gravity stations. The third area, the Limestone Plateau focus area, consists of a single gravity station in the northwestern Black Hills located on an outcrop of the Madison Limestone. An absolute-gravity station, used to tie relative-gravity survey data together, was established on a relatively impermeable bedrock outcrop to minimize groundwater-storage change at the reference location. Data from the three focus areas allow for interpretation of groundwater-storage characteristics using microgravity measurements. Gravity measurements, together with water-level data from an observation well located 2 miles from the Spring Canyon focus area and measured streamflow in Spring Creek, provided evidence that rapid groundwater-storage change, responding to changes in sinking streamflow over the recharge area of the aquifer, occurred in the Madison aquifer directly beneath the gravity station at Spring Canyon. This phenomenon likely was a result of groundwater movement through caverns, conduits, and fractures, which are common in karst aquifers. Spatially and temporally separated microgravity data for the Doty focus area indicated horizontal and vertical heterogeneity of effective porosity for the Madison aquifer. One such example of this was indicated by water-level measurements at an observation well and gravity measurements at four gravity stations in the southeastern part of the Doty area, which were used to estimate effective porosity values ranging from greater than 0 to 0.18. A decrease in groundwater storage determined by microgravity measurements during the spring recharge period for five upgradient stations in the Doty focus area indicated the possibility of rapid release and downgradient cascading of perched groundwater. Evidence for similar phenomena was documented for Wind Cave and Brooks Cave in the Black Hills. Absolute-gravity measurements at the Limestone Plateau focus area confirmed the relation between water levels in an observation well and changes in groundwater storage. Comparison of these gravity measurements with water levels in a nearby observation well resulted in an effective porosity estimate of 0.02 for the Madison aquifer beneath the gravity station.

South Dakota↗

USGS2011

Multivariate analyses with end-member mixing to characterize groundwater flow: Wind Cave and associated aquifers

Principal component analysis (PCA) applied to hydrochemical data has been used with end-member mixing to characterize groundwater flow to a limited extent, but aspects of this approach are unresolved. Previous similar approaches typically have assumed that the extreme-value samples identified by PCA represent end members. The method presented herein is different from previous work in that (1) end members were not assumed to have been sampled but rather were estimated and constrained by prior knowledge; (2) end-member mixing was quantified in relation to hydrogeologic domains, which focuses model results on major hydrologic processes; (3) a method to select an appropriate number of end members using a series of cluster analyses is presented; and (4) conservative tracers were weighted preferentially in model calibration, which distributed model errors of optimized values, or residuals, more appropriately than would otherwise be the case. The latter item also provides an estimate of the relative influence of geochemical evolution along flow paths in comparison to mixing. This method was applied to groundwater in Wind Cave and the associated karst aquifer in the Black Hills of South Dakota, USA. The end-member mixing model was used to test a hypothesis that five different end-member waters are mixed in the groundwater system comprising five hydrogeologic domains. The model estimated that Wind Cave received most of its groundwater inflow from local surface recharge with an additional 33% from an upgradient aquifer. Artesian springs in the vicinity of Wind Cave primarily received water from regional groundwater flow.

South Dakota↗

USGS2010

Simulated groundwater flow in the Ogallala and Arikaree aquifers, Rosebud Indian Reservation area, South Dakota – Revisions with data through water year 2008 and simulations of potential future scenarios

The Ogallala and Arikaree aquifers are important water resources in the Rosebud Indian Reservation area and are used extensively for irrigation, municipal, and domestic water supplies. Drought or increased withdrawals from the Ogallala and Arikaree aquifers in the Rosebud Indian Reservation area have the potential to affect water levels in these aquifers. This report documents revisions and recalibration of a previously published three-dimensional, numerical groundwater-flow model for this area. Data for a 30-year period (water years 1979 through 2008) were used in steady-state and transient numerical simulations of groundwater flow. In the revised model, revisions include (1) extension of the transient calibration period by 10 years, (2) the use of inverse modeling for steady-state calibration, (3) model calibration to base flow for an additional four surface-water drainage basins, (4) improved estimation of transient aquifer recharge, (5) improved delineation of vegetation types, and (6) reduced cell size near large capacity water-supply wells. In addition, potential future scenarios were simulated to assess the potential effects of drought and increased groundwater withdrawals. The model comprised two layers: the upper layer represented the Ogallala aquifer and the lower layer represented the Arikaree aquifer. The model’s grid had 168 rows and 202 columns, most of which were 1,640 feet (500 meters) wide, with narrower rows and columns near large water-supply wells. Recharge to the Ogallala and Arikaree aquifers occurs from precipitation on the outcrop areas. The average recharge rates used for the steady-state simulation were 2.91 and 1.45 inches per year for the Ogallala aquifer and Arikaree aquifer, respectively, for a total rate of 255.4 cubic feet per second (ft 3 /s). Discharge from the aquifers occurs through evapotranspiration, discharge to streams as base flow and spring flow, and well withdrawals. Discharge rates for the steady-state simulation were 171.3 ft 3 /s for evapotranspiration, 74.4 ft 3 /s for net outflow to streams and springs, and 11.6 ft 3 /s for well withdrawals. Estimated horizontal hydraulic conductivity used for the numerical model ranged from 0.2 to 84.4 feet per day (ft/d) in the Ogallala aquifer and from 0.1 to 4.3 ft/d in the Arikaree aquifer. A uniform vertical hydraulic conductivity value of 4.2x10 -4 ft/d was estimated for the Ogallala aquifer. Vertical hydraulic conductivity was estimated for five zones in the Arikaree aquifer and ranged from 8.8x10 -5 to 3.7 ft/d. Average rates of recharge, maximum evapotranspiration, and well withdrawals were included in the steady-state simulation, whereas the time-varying rates were included in the transient simulation. Inverse modeling techniques were used for steady-state model calibration. These methods were designed to estimate parameter values that are, statistically, the most likely set of values to result in the smallest differences between simulated and observed hydraulic heads and base-flow discharges. For the steady-state simulation, the root mean square error for simulated hydraulic heads for all 383 wells was 27.3 feet. Simulated hydraulic heads were within ±50 feet of observed values for 93 percent of the wells. The potentiometric surfaces of the two aquifers calculated by the steady-state simulation established initial conditions for the transient simulation. For the transient simulation, the difference between the simulated and observed means for hydrographs was within ±40 feet for 98 percent of 44 observation wells. A sensitivity analysis was used to examine the response of the calibrated steady-state model to changes in model parameters including horizontal and vertical hydraulic conductivity, evapotranspiration, recharge, and riverbed conductance. The model was most sensitive to recharge and maximum evapotranspiration and least sensitive to riverbed and spring conductances. To simulate a potential future drought scenario, a synthetic recharge record was created, the mean of which was equal to 64 percent of the average estimated recharge rate for the 30-year calibration period. This synthetic recharge record was used to simulate the last 20 years of the calibration period under drought conditions. Compared with results of the calibrated model, decreases in hydraulic-head values for the drought scenario at the end of the simulation period were as much as 39 feet for the Ogallala aquifer. To simulate the effects of potential increases in pumping, well withdrawal rates were increased by 50 percent from those estimated for the 30-year calibration period for the last 20 years of the calibration period. Compared with results of the calibrated model, decreases in hydraulic-head values for the scenario of increased pumping at the end of the simulation period were as much as 13 feet for the Ogallala aquifer. This numerical model is suitable as a tool to help understand the flow system, to help confirm that previous estimates of aquifer properties were reasonable, and to estimate aquifer properties in areas without data. The model also is useful to help assess the effects of drought and increases in pumping by simulations of these scenarios, the results of which are not precise but may be considered when making water management decisions.

South Dakota↗

USGS2009

Hydrograph separation for karst watersheds using a two-domain rainfall-discharge model

Highly parameterized, physically based models may be no more effective at simulating the relations between rainfall and outflow from karst watersheds than are simpler models. Here an antecedent rainfall and convolution model was used to separate a karst watershed hydrograph into two outflow components: one originating from focused recharge in conduits and one originating from slow flow in a porous annex system. In convolution, parameters of a complex system are lumped together in the impulse-response function (IRF), which describes the response of the system to an impulse of effective precipitation. Two parametric functions in superposition approximate the two-domain IRF. The outflow hydrograph can be separated into flow components by forward modeling with isolated IRF components, which provides an objective criterion for separation. As an example, the model was applied to a karst watershed in the Madison aquifer, South Dakota, USA. Simulation results indicate that this watershed is characterized by a flashy response to storms, with a peak response time of 1 day, but that 89% of the flow results from the slow-flow domain, with a peak response time of more than 1 year. This long response time may be the result of perched areas that store water above the main water table. Simulation results indicated that some aspects of the system are stationary but that nonlinearities also exist.

Journal of Hydrology↗

USGS2009

Age-distribution estimation for karst groundwater: Issues of parameterization and complexity in inverse modeling by convolution

Convolution modeling is useful for investigating the temporal distribution of groundwater age based on environmental tracers. The framework of a quasi-transient convolution model that is applicable to two-domain flow in karst aquifers is presented. The model was designed to provide an acceptable level of statistical confidence in parameter estimates when only chlorofluorocarbon (CFC) and tritium ( 3 H) data are available. We show how inverse modeling and uncertainty assessment can be used to constrain model parameterization to a level warranted by available data while allowing major aspects of the flow system to be examined. As an example, the model was applied to water from a pumped well open to the Madison aquifer in central USA with input functions of CFC-11, CFC-12, CFC-113, and 3 H, and was calibrated to several samples collected during a 16-year period. A bimodal age distribution was modeled to represent quick and slow flow less than 50 years old. The effects of pumping and hydraulic head on the relative volumetric fractions of these domains were found to be influential factors for transient flow. Quick flow and slow flow were estimated to be distributed mainly within the age ranges of 0-2 and 26-41 years, respectively. The fraction of long-term flow (>50 years) was estimated but was not dateable. The different tracers had different degrees of influence on parameter estimation and uncertainty assessments, where 3 H was the most critical, and CFC-113 was least influential.

Journal of Hydrology↗

USGS2009

A one-dimensional heat-transport model for conduit flow in karst aquifers

A one-dimensional heat-transport model for conduit flow in karst aquifers is presented as an alternative to two or three-dimensional distributed-parameter models, which are data intensive and require knowledge of conduit locations. This model can be applied for cases where water temperature in a well or spring receives all or part of its water from a phreatic conduit. Heat transport in the conduit is simulated by using a physically-based heat-transport equation that accounts for inflow of diffuse flow from smaller openings and fissures in the surrounding aquifer during periods of low recharge. Additional diffuse flow that is within the zone of influence of the well or spring but has not interacted with the conduit is accounted for with a binary mixing equation to proportion these different water sources. The estimation of this proportion through inverse modeling is useful for the assessment of contaminant vulnerability and well-head or spring protection. The model was applied to 7 months of continuous temperature data for a sinking stream that recharges a conduit and a pumped well open to the Madison aquifer in western South Dakota. The simulated conduit-flow fraction to the well ranged from 2% to 31% of total flow, and simulated conduit velocity ranged from 44 to 353 m/d.

Journal of Hydrology↗

USGS2009

Numerical groundwater-flow model of the Minnelusa and Madison hydrogeologic units in the Rapid City area, South Dakota

The city of Rapid City and other water users in the Rapid City area obtain water supplies from the Minnelusa and Madison aquifers, which are contained in the Minnelusa and Madison hydrogeologic units. A numerical groundwater-flow model of the Minnelusa and Madison hydrogeologic units in the Rapid City area was developed to synthesize estimates of water-budget components and hydraulic properties, and to provide a tool to analyze the effect of additional stress on water-level altitudes within the aquifers and on discharge to springs. This report, prepared in cooperation with the city of Rapid City, documents a numerical groundwater-flow model of the Minnelusa and Madison hydrogeologic units for the 1,000-square-mile study area that includes Rapid City and the surrounding area. Water-table conditions generally exist in outcrop areas of the Minnelusa and Madison hydrogeologic units, which form generally concentric rings that surround the Precambrian core of the uplifted Black Hills. Confined conditions exist east of the water-table areas in the study area. The Minnelusa hydrogeologic unit is 375 to 800 feet (ft) thick in the study area with the more permeable upper part containing predominantly sandstone and the less permeable lower part containing more shale and limestone than the upper part. Shale units in the lower part generally impede flow between the Minnelusa hydrogeologic unit and the underlying Madison hydrogeologic unit; however, fracturing and weathering may result in hydraulic connections in some areas. The Madison hydrogeologic unit is composed of limestone and dolomite that is about 250 to 610 ft thick in the study area, and the upper part contains substantial secondary permeability from solution openings and fractures. Recharge to the Minnelusa and Madison hydrogeologic units is from streamflow loss where streams cross the outcrop and from infiltration of precipitation on the outcrops (areal recharge). MODFLOW-2000, a finite-difference groundwater-flow model, was used to simulate flow in the Minnelusa and Madison hydrogeologic units with five layers. Layer 1 represented the fractured sandstone layers in the upper 250 ft of the Minnelusa hydrogeologic unit, and layer 2 represented the lower part of the Minnelusa hydrogeologic unit. Layer 3 represented the upper 150 ft of the Madison hydrogeologic unit, and layer 4 represented the less permeable lower part. Layer 5 represented an approximation of the underlying Deadwood aquifer to simulate upward flow to the Madison hydrogeologic unit. The finite-difference grid, oriented 23 degrees counterclockwise, included 221 rows and 169 columns with a square cell size of 492.1 ft in the detailed study area that surrounded Rapid City. The northern and southern boundaries for layers 1-4 were represented as no-flow boundaries, and the boundary on the east was represented with head-dependent flow cells. Streamflow recharge was represented with specified-flow cells, and areal recharge to layers 1-4 was represented with a specified-flux boundary. Calibration of the model was accomplished by two simulations: (1) steady-state simulation of average conditions for water years 1988-97 and (2) transient simulations of water years 1988-97 divided into twenty 6-month stress periods. Flow-system components represented in the model include recharge, discharge, and hydraulic properties. The steady-state streamflow recharge rate was 42.2 cubic feet per second (ft3/s), and transient streamflow recharge rates ranged from 14.1 to 102.2 ft3/s. The steady-state areal recharge rate was 20.9 ft3/s, and transient areal recharge rates ranged from 1.1 to 98.4 ft3/s. The upward flow rate from the Deadwood aquifer to the Madison hydrogeologic unit was 6.3 ft3/s. Discharge included springflow, water use, flow to overlying units, and regional outflow. The estimated steady-state springflow of 32.8 ft3/s from seven springs was similar to the simulated springflow of 31.6 ft3/s, which included 20.5 ft3

South Dakota↗

USGS2007

Environmental tracers as indicators of karst conduits in groundwater in South Dakota, USA

Environmental tracers sampled from the carbonate Madison aquifer on the eastern flank of the Black Hills, South Dakota, USA indicated the approximate locations of four major karst conduits. Contamination issues are a major concern because these conduits are characterized by direct connections to sinking streams, high groundwater velocities, and proximity to public water supplies. Objectives of the study were to estimate approximate conduit locations and assess possible anthropogenic influences associated with conduits. Anomalies of young groundwater based on chlorofluorocarbons (CFCs), tritium, and electrical conductivity (EC) indicated fast moving, focused flow and thus the likely presence of conduits. ??18O was useful for determining sources of recharge for each conduit, and nitrate was a useful tracer for assessing flow paths for anthropogenic influences. Two of the four conduits terminate at or near a large spring complex. CFC apparent ages ranged from 15 years near conduits to >50 years in other areas. Nitrate-N concentrations >0.4 mg/L in groundwater were associated with each of the four conduits compared with concentrations ranging from <0.1 to 0.4 mg/L in other areas. These higher nitrate-N concentrations probably do not result from sinking streams but rather from other areas of infiltration. ?? Springer-Verlag 2007.

Hydrogeology Journal↗

USGS2007

Characterization of ground-water flow and water quality for the Madison and Minnelusa aquifers in northern Lawrence County, South Dakota

The Madison and Minnelusa aquifers are used extensively for water supplies for the city of Spearfish and other users in northern Lawrence County, South Dakota. Ground water in the Madison and Minnelusa aquifers in the study area generally flows north from outcrop areas where recharge from sinking streams and infiltration of precipitation occurs. Ground water that moves northward and eastward around the Black Hills enters the study area from the west and results in hydraulic heads that are several hundred feet higher on the western side of the study area than on the eastern side. The estimated average recharge rate of 38 cubic feet per second (ft3/s) on outcrops of the Madison Limestone and Minnelusa Formation is less than the total estimated average spring discharge rate of 51 ft3/s in the northwestern part of the study area. Sixteen pounds of fluorescein dye were injected into Spearfish Creek on March 25, 2003, when streamflow was 6.6 ft3/s. The dye was detected in water samples from four wells completed in the Madison aquifer ranging from 2.6 to 4.5 miles north of the injection site. First arrival times ranged from 5 to 169 days, and ground-water velocities ranged from about 0.1 to 0.5 mile per day. Sixty-four pounds of Rhodamine WT was injected into Spearfish Creek at the same location on May 9, 2003, when streamflow was 5.6 ft3/s. Rhodamine WT dye concentrations measured in samples from the same four wells were about an order of magnitude less than measured fluorescein concentrations. Oxygen- and deuterium-isotope values for samples from Cox Lake and McNenny Pond springs indicated a probable component of spring discharge that originates from outcrops of the Madison Limestone and Minnelusa Formation on the Limestone Plateau south of the study area. Oxygen- and deuterium-isotope values for samples from Mirror Lake spring indicated possible contributions from overlying aquifers and local recharge. Oxygen- and deuterium-isotope values for the combined springflow contributing to Crow Creek in the northwestern part of the study area indicated that the primary source of water is the Madison and Minnelusa aquifers. Oxygen- and deuterium-isotope values for Old Hatchery and Higgins Gulch springs, located north of Spearfish, indicated a source water originating from the outcrops of the Madison Limestone and Minnelusa Formation within the study area. Concentrations of three chlorofluorocarbons (CFC-11, CFC-12, and CFC-113) were used to characterize ground-water residence times in the study area. For the four wells where dye was detected, CFC-11 apparent ages ranged from 12 to 26 years, indicating that the wells contained months-old water mixed with years- to decades-old water. Logarithmic regression analysis of the CFC-11 apparent ages for water from 10 wells and distance to a possible conduit trending north through the area where dye was detected, yielded an r2 value of 0.71. Straight-line regression analysis of the CFC-11 apparent ages for the six wells closest to the possible conduit had an r2 value of 0.96. Two wells located relatively close to the outcrop areas had no or very low tritium values indicating relatively long residence times and diffuse ground-water flow. The tritium value of 7.2 TU in water from well COL where dye was detected, indicated that the water probably is a bimodal mixture, with a substantial portion that is older than 50 years. Water from well ELL, where dye was detected, had a tritium value of 19.7 TU and a CFC apparent age of 15 years, indicating that the sample from this well probably is a unimodal mixture with very little water older than 50 years. Comparison of the CFC apparent age for three spring sites (Cox Lake, 26 years; McNenny Pond, 26 years; Mirror Lake, 13 years) also indicated that Mirror Lake spring probably has a component of local recharge from formations that overlie the Minnelusa Formation. In the Madison aquifer, specific conductance ranges from 18 to 945 microsiemens per cen

South Dakota↗

USGS2007

Analysis of ground-water flow in the Madison aquifer using fluorescent dyes injected in Spring Creek and Rapid Creek near Rapid City, South Dakota, 2003-04

The Madison aquifer, which contains fractures and solution openings in the Madison Limestone, is used extensively for water supplies for the city of Rapid City and other suburban communities in the Rapid City, S. Dak., area. The 48 square-mile study area includes the west-central and southwest parts of Rapid City and the outcrops of the Madison Limestone extending from south of Spring Creek to north of Rapid Creek. Recharge to the Madison Limestone occurs when streams lose flow as they cross the outcrop. The maximum net loss rate for Spring and Rapid Creek loss zones are 21 and 10 cubic feet per second (ft3/s), respectively. During 2003 and 2004, fluorescent dyes were injected in the Spring and Rapid Creek loss zones to estimate approximate locations of preferential flow paths in the Madison aquifer and to measure the response and transit times at wells and springs. Four injections of about 2 kilograms of fluorescein dye were made in the Spring Creek loss zone during 2003 (sites S1, S2, and S3) and 2004 (site S4). Injection at site S1 was made in streamflow just upstream from the loss zone over a 12-hour period when streamflow was about equal to the maximum loss rate. Injections at sites S2, S3, and S4 were made in specific swallow holes located in the Spring Creek loss zone. Injection at site R1 in 2004 of 3.5 kilograms of Rhodamine WT dye was made in streamflow just upstream from the Rapid Creek loss zone over about a 28-hour period. Selected combinations of 27 wells, 6 springs, and 3 stream sites were monitored with discrete samples following the injections. For injections at sites S1-S3, when Spring Creek streamflow was greater than or equal to 20 ft3/s, fluorescein was detected in samples from five wells that were located as much as about 2 miles from the loss zone. Time to first arrival (injection at site S1) ranged from less than 1 to less than 10 days. The maximum fluorescein concentration (injection at site S1) of 120 micrograms per liter (ug/L) at well CO, which is located adjacent to the loss zone, was similar to the concentration in the stream. Fluorescein arrived at well NON (injection at site S1), which is located about 2 miles northeast of the loss zone, within about 1.6 days, and the maximum concentration was 44 ug/L. For injection at site S4, when streamflow was about 12 ft3/s, fluorescein was detected in samples from six wells and time to first arrival ranged from 0.2 to 16 days. Following injection at site S4 in 2004, the length of time that dye remained in the capture zone of well NON, which is located approximately 2 miles from the loss zone, was almost an order of magnitude greater than in 2003. For injection at site R1, Rhodamine WT was detected at well DRU and spring TI-SP with time to first arrival of about 0.5 and 1.1 days and maximum concentrations of 6.2 and 0.91 ug/L, respectively. Well DRU and spring TI-SP are located near the center of the Rapid Creek loss zone where the creek has a large meander. Measurable concentrations were observed for spring TI-SP as many as 109 days after the dye injection. The direction of a conduit flow path in the Spring Creek area was to the northeast with ground-water velocities that ranged from 770 to 6,500 feet per day. In the Rapid Creek loss zone, a conduit flow path east of the loss zone was not evident from the dye injection.

South Dakota↗

USGS2006

Translating CFC-based piston ages into probability density functions of ground-water age in karst

Temporal age distributions are equivalent to probability density functions (PDFs) of transit time. The type and shape of a PDF provides important information related to ground-water mixing at the well or spring and the complex nature of flow networks in karst aquifers. Chlorofluorocarbon (CFC) concentrations measured for samples from 12 locations in the karstic Madison aquifer were used to evaluate the suitability of various PDF types for this aquifer. Parameters of PDFs could not be estimated within acceptable confidence intervals for any of the individual sites. Therefore, metrics derived from CFC-based apparent ages were used to evaluate results of PDF modeling in a more general approach. The ranges of these metrics were established as criteria against which families of PDFs could be evaluated for their applicability to different parts of the aquifer. Seven PDF types, including five unimodal and two bimodal models, were evaluated. Model results indicate that unimodal models may be applicable to areas close to conduits that have younger piston (i.e., apparent) ages and that bimodal models probably are applicable to areas farther from conduits that have older piston ages. The two components of a bimodal PDF are interpreted as representing conduit and diffuse flow, and transit times of as much as two decades may separate these PDF components. Areas near conduits may be dominated by conduit flow, whereas areas farther from conduits having bimodal distributions probably have good hydraulic connection to both diffuse and conduit flow. ?? 2006 Elsevier B.V. All rights reserved.

Journal of Hydrology↗

USGS2006

Statistical analyses of hydrologic system components and simulation of Edwards aquifer water-level response to rainfall using transfer-function models, San Antonio region, Texas

In 2003 the U.S. Geological Survey, in cooperation with the San Antonio Water System, did a study using historical data to statistically analyze hydrologic system components in the San Antonio region of Texas and to develop transfer-function models to simulate water levels at selected sites (wells) in the Edwards aquifer on the basis of rainfall. Water levels for two wells in the confined zone in Medina County and one well in the confined zone in Bexar County were highly correlated and showed little or no lag time between water-level responses. Water levels in these wells also were highly correlated with springflow at Comal Springs. Water-level hydrographs for 35 storms showed that an individual well can respond differently to similar amounts of rainfall. Fourteen water-level-recession hydrographs for a Medina County well showed that recession rates were variable. Transfer-function models were developed to simulate water levels at one confined-zone well and two recharge-zone wells in response to rainfall. For the confined-zone well, 50 percent of the simulated water levels are within 10 feet of the measured water levels, and 80 percent of the simulated water levels are within 15 feet of the measured water levels. For one recharge-zone well, 50 percent of the simulated water levels are within 5 feet of the measured water levels, and 90 percent of the simulated water levels are within 14 feet of the measured water levels. For the other recharge-zone well, 50 percent of the simulated water levels are within 14 feet of the measured water levels, and 90 percent of the simulated water levels are within 27 feet of the measured water levels. The transfer-function models showed that (1) the Edwards aquifer in the San Antonio region responds differently to recharge (effective rainfall) at different wells; and (2) multiple flow components are present in the aquifer. If simulated long-term system response results from a change in the hydrologic budget, then water levels would be difficult to simulate accurately.

Texas↗

USGS2004

Linear model describing three components of flow in karst aquifers using 18O data

The stable isotope of oxygen, 18O, is used as a naturally occurring ground-water tracer. Time-series data for ??18O are analyzed to model the distinct responses and relative proportions of the conduit, intermediate, and diffuse flow components in karst aquifers. This analysis also describes mathematically the dynamics of the transient fluid interchange between conduits and diffusive networks. Conduit and intermediate flow are described by linear-systems methods, whereas diffuse flow is described by mass-balance methods. An automated optimization process estimates parameters of lognormal, Pearson type III, and gamma distributions, which are used as transfer functions in linear-systems analysis. Diffuse flow and mixing parameters also are estimated by these optimization methods. Results indicate the relative proximity of a well to a main conduit flowpath and can help to predict the movement and residence times of potential contaminants. The three-component linear model is applied to five wells, which respond to changes in the isotopic composition of point recharge water from a sinking stream in the Madison aquifer in the Black Hills of South Dakota. Flow velocities as much as 540 m/d and system memories of as much as 71 years are estimated by this method. Also, the mean, median, and standard deviation of traveltimes; time to peak response; and the relative fraction of flow for each of the three components are determined for these wells. This analysis infers that flow may branch apart and rejoin as a result of an anastomotic (or channeled) karst network.

Journal of Hydrology↗

USGS2003

Simulated ground-water flow in the Ogallala and Arikaree aquifers, Rosebud Indian Reservation area, South Dakota

South Dakota↗