Explicit Estimation of Aquifer Diffusivity from Linear Stream Stage

Different approaches are available for estimation of aquifer hydraulic diffusivity from a linear stream-stage variation and corresponding groundwater heads. These approaches require interpolation from tabulated values or computation of hydraulic gradient at the stream aquifer interface. Certain methods use approximation or interpolation of tabulated values for an infinite series. These methods are prone to errors in the estimated aquifer hydraulic diffusivity. An alternative approach is to use a closed-form solution of the problem and develop an explicit expression for the aquifer hydraulic diffusivity, which is free from errors of using infinite series, interpolation, and computation of hydraulic gradient. Such an alternative method is developed for a linearly varying stream stage. The new method can yield the estimate of hydraulic diffusivity even from a single observation. The proposed method would be applicable to practical hydraulic engineering problems keeping in view that most of a rising part of a stream stage hydrograph can be approximated by a linear rise and certain rivers may show a linearly varying stream stage. Use of the new method is demonstrated on published and field data, which shows that the estimates obtained using the new method are comparable to that obtained using an optimization approach.     

Aquifer Response to Sinusoidal or Arbitrary Stage of Semipervious Stream

Analytical expressions for the aquifer responses, viz., groundwater head, rate of flow and cumulative volume of flow, to a generalized sinusoidal stage of semipervious streams considering the stream boundary resistance, are derived. The analytical aquifer responses to a linear stream stage and to a typical analytical flood wave that was used by Cooper and Rorabaugh, are also derived. For a zero-stream resistance, the aquifer responses converge to those for a fully penetrating stream. Also, two analytical methods, a “ramp kernel method” and a “Fourier series method,” for obtaining the aquifer responses to an arbitrary temporal stage of sempervious stream, are developed. The analytical expressions of the ramp kernels for different aquifer responses are developed. The ramp kernel method is found superior to the conventional convolution that uses numerical integration or pulse kernels for obtaining the convolution integral. In the Fourier series method, the aquifer responses to sinusoidal stage are used along with Fourier series. The results obtained using both methods are in close agreement. The new methods are also applicable to fully penetrating streams by assigning a zero value to the stream resistance.     

Ramp Kernels for Aquifer Responses to Arbitrary Stream Stage

Analytical expressions for ramp kernels (new kernels) for an improved convolution for obtaining aquifer responses, viz, groundwater head, rate, and cumulative volume of groundwater flow, to an arbitrary stage, are obtained. The use of the ramp kernels gives accurate aquifer responses and is superior to the conventional convolution in which numerical integration or pulse kernels are used. The extent of improvement in the results with the use of the ramp kernels is discussed and quantified for three examples, where the results are compared to analytical solutions. For the comparisons, the analytical solutions for linear and sinusoidal stream stages are derived. The use of the ramp kernels reproduces accurately the analytical solutions. The concept of ramp kernels can also be used for obtaining an accurate solution of convolution integrals observed in other fields.     

Stream Multiaquifer Well Interactions

In this paper, unsteady flow into a multiaquifer well due to stream stage changes and varying pumping rate is analyzed. The well is located at such a distance that the radius of influence touches the stream boundary; hence, pumping induces seepage from the stream to the aquifer. The discrete kernel approach, which is based on Duhamel’s principle, has been applied to find the interaction among stream, aquifers, and pumping well for constant as well as varying stream stage. The analytical expression for a damped sinusoidal flood wave passing in a fully penetrating stream has been used for obtaining the aquifer response. By applying image-well theory, the finite aquifer and well system has been transformed into an infinite aquifer and well system. The principle of superposition, which is applicable to a linear system, has been used to analyze the interaction processes among the three components of the system. The interaction of the stream, aquifers, and well with each other are analyzed during pumping, after stoppage of pumping, as well as during passage of a flood wave in the stream.     

Rate and Volume of Stream Flow Depletion due to Unsteady Pumping

Analytical but approximate methods are developed for obtaining pumping induced rate and volume of stream flow depletion, which can account for unsteady (any variation) pumping discharge and are also applicable for intermittent pumping and recovery. Exact analytical solutions for a sinusoidal variation in the pumping discharge are proposed; the proposed methods are verified using these solutions. The proposed methods use ramp kernels that give results superior to the conventional convolution. These ramp kernels assume the linear variation in pumping discharge between the two consecutive discretized points as opposed to the uniform variation assumed in the conventional convolution. The proposed solutions are applicable for homogeneous and isotropic aquifers hydraulically connected to streams.     

Drawdown due to Temporally Varying Pumping Discharge: Inversely Estimating Aquifer Parameters

A ramp kernel method is proposed for accurately calculating the drawdown due to any temporal variation in pumping discharge. The use of the ramp kernels assumes the linear variation between the two consecutive measured pumping discharges. The prior studies assume a rectangular variation between the two consecutive measured pumping discharges. In the rectangular variation, a uniform pumping rate is assumed during a time span. An analytical equation for calculating the ramp kernel is derived. An optimization method is used with the proposed ramp kernels for inversely estimating the aquifer parameters from drawdown due to an arbitrary unsteady pumping discharge. Unlike the prior methods, the proposed method accurately identifies the parameters even when the sampling interval for the drawdown and pumping discharge is longer than that needed for assuming a linear variation. The proposed method outperforms the prior method. Application of the proposed method is illustrated using examples.     

Aquifer Boundaries and Parameter Identification Simplified

A simple method is proposed for simultaneous and explicit identification of confined aquifer parameters and boundaries from drawdowns measured at an observation well during a constant rate pumping test. The method requires the determination of peaks of unimodal curves. Only a little subjectivity is involved in the method as the peak is a well defined point. The method is applicable even when the first and second straight lines on the semilogarithmic characteristic drawdown curve are not fully developed. The calculations involved can be performed on a calculator. Results have also been presented for small arguments of the well functions defining the development of the straight lines. The times for initiation and development of straight lines are quantified. The minimum duration of a pumping test for a reliable identification of the aquifer parameters and boundaries using the proposed method is also quantified. Use of the new method suggests a much shorter duration pumping test for the accurate identification of aquifer parameters and boundaries. This would save considerable time and money. At least a 100-fold savings in time and money involved in a pumping test to locate a boundary is observed when compared to the use of the law of times in the Cooper-Jacob method. Application of the method to published data sets shows that reliable estimates of the aquifer parameters and distance to a boundary are obtained.     

Diagnostic Curve for Confined Aquifer Parameters from Early Drawdowns

A diagnostic curve of unimodal shape is developed for identifying the confined aquifer parameters from early drawdowns. A scaled well function is proposed for the diagnostic curve and computationally simple functions are developed for its accurate approximation. The diagnostic curve may be viewed as an alteration of the Theis’ curve or as the generalization of a previous approach proposed by the writer. Plotting the pumping test data in a convenient form and matching it to the diagnostic curve with a parallel shift of axes identify the aquifer parameters. The unimodal shape of the diagnostic curve facilitates matching and reduces the personal errors. The proposed method is simple, easy to apply, and yields accurate estimates of aquifer parameters from only early drawdowns, which would save considerable time and money involved in conducting a long-duration pumping test. The estimates obtained using the new method are as good as those obtained from much more complex methods. The new method does not require either the initial guess for the parameter values or repetitive evaluation of the well function.     

New Methods for Aquifer Parameters from Slug Test Data

An objective method and a diagnostic curve method are developed for estimating the aquifer parameters (storage coefficient and transmissivity) from slug test data on a fully penetrating well. In the objective method, explicit equations are developed for estimating the aquifer parameters. In the diagnostic curve method, a set of unimodal diagnostic curves is developed along with a guiding straight line. The rise or fall in water level of the well is plotted diagnostically on a double logarithmic graph and matched to one of the diagnostic curves plotted on the same scale with a parallel shift of axes, to estimate the aquifer parameters from the dual coordinates of a selected point on the matched portion of the graphs. The unimodal shape of the diagnostic curves and the guiding straight line facilitate the matching and limit the subjectivity. The proposed methods can easily identify a nonideal condition. The estimates of the aquifer parameters obtained using the proposed methods are more accurate than those obtained using the prior curve matching methods. The proposed methods are also able to identify nonideal conditions. It is hoped that the new methods will be of help to field and practicing engineers.     

Potential for 17β-Estradiol and Estrone Degradation in a Recharge Aquifer System

The potential for degradation of two important endocrine disrupting chemicals, 17β-estradiol (E2) and estrone (E1) in a wastewater effluent recharged aquifer system was assessed using lab scale microcosms. Biotransformation by microorganisms present in the aquifer material was found to be the major mode of removal for both organic contaminants in the microcosms. The addition of ultrafiltered secondary effluent, simulating recharge of aquifer, did not significantly alter the behavior of both E2 and E1 in the aquifer microcosms. A significant difference in E2 removal rates in microcosms that received E2 as a concentrated stock in methanol as opposed to direct dissolution of E2 into groundwater shows that the presence of methanol could alter the biodegradation kinetics of the organic pollutant of interest. The rapid utilization of methanol by indigenous methanol degraders, which was present as an alternative carbon substrate, caused rapid depletion of oxygen and, hence, development of anoxic conditions in the microcosms. Increased oxygen supply led to corresponding improved E2 biotransformation rates, indicating that E2 degradation under anoxic conditions was much slower than under aerobic conditions. On the other hand, in the presence of a high initial concentration of methanol, E1 was more persistent in the medium under aerobic conditions than under anoxic conditions. Apparently, the initial high concentration of methanol had an inhibitory effect on aerobic E1 degraders in the microcosms.     

Simple Equations for Aquifer Parameters from Drawdowns in Large-Diameter Wells

Simple equations are proposed for estimating storage coefficient and transmissivity of an aquifer from drawdowns in large- diameter wells. The proposed method requires determination of the peak and time to peak of a unimodal curve. Using these values and utilizing the provided set of equations, the aquifer parameters are estimated through an iterative procedure. The proposed method is void of subjectivity involved in the previously proposed curve matching methods. Also, the new method can be used when the conventional curve matching methods cannot be applied to estimate the aquifer parameters. The new method can be used to estimate the aquifer parameters from the drawdown data observed only up to a time so that the peak could be determined.     

Water Table Profiles and Discharges for an Inclined Ditch-Drained Aquifer under Temporally Variable Recharge

This paper presents the solution of the linearized Boussinesq equation for an inclined, ditch-drained aquifer, with a temporally varying recharge rate. Water-table profiles and flow rates into the ditches are calculated. As an initial condition the steady-state profile for a constant recharge rate is used, and the linearized Boussinesq equation is solved for a different recharge rate. Then, at a specified time, the transient water table profile is used as initial condition for the Boussinesq equation with a new recharge rate. The transient solution at a new specified time is then used as the initial condition for the Boussinesq equation with a different recharge rate, and so on. Using the Darcy equation, analytical expressions for the flow rates into the ditches can be obtained. The solution allows the calculation of the transient behavior of the groundwater table and its flow rates due to temporally variable recharge rates.     

Estimation of Aquifer Parameters from Pumping Test Data by Genetic Algorithm Optimization Technique

Adequate and reliable estimates of aquifer parameters are of utmost importance for proper management of vital groundwater resources. The pumping (aquifer) test is the standard technique for estimating various hydraulic properties of aquifer systems, viz., transmissivity (T), hydraulic conductivity (K), storage coefficient (S), and leakance (L), for which the graphical method is widely used. In the present study, the efficacy of the genetic algorithm (GA) optimization technique is assessed in estimating aquifer parameters from the time-drawdown pumping test data. Computer codes were developed to optimize various aquifer parameters under different hydrogeologic conditions by using the GA technique. Applicability, adequacy, and robustness of the developed codes were tested using 12 sets of the published and unpublished aquifer test data. The aquifer parameters were also estimated by the graphical method using AquiferTest software, and were compared with those obtained by the GA technique. The GA technique yielded significantly low values of the sum of square errors (SSE) for almost all the datasets under study. The results revealed that the GA technique is an efficient and reliable method for estimating various aquifer parameters, especially in the situation when the graphical matching is poor. Also, it was found that because of its inherent characteristics, GA avoids the subjectivity, long computation time and ill-posedness often associated with conventional optimization techniques. Furthermore, the performance evaluation of the developed GA-based computer codes showed that the fitness value (SSE) of the best point in a population reduces with increasing generation number and population size. The analysis of the sensitivity of the parameters during the performance of GA indicated that a unique set of aquifer parameters was obtained for all three aquifer systems. The GA-based computer programs with interactive windows developed in this study are user-friendly and can serve as a teaching and research tool, which could also be useful for practicing hydrologists and hydrogeologists.     

Secondary Benefits of Aquifer Storage and Recovery: Disinfection By-Product Control

The potential of biological processes during aquifer storage to reduce disinfection by-products (DBP), and DBP precursors were examined under controlled conditions. Finished water treated by conventional water treatment practice was pumped into a sand media column for up to 34 days of residence time. Two experiments were conducted where the finished water was chlorinated or ozonated prior to injection. Chlorination of water withdrawn from simulated aquifer storage conditions resulted in reduced formation of trihalomethane (THM) concentrations for all three treated water types. Ozonation of finished water resulted in a 70% decrease in TTHM formation. Aquifer storage of finished water resulted in a 26–28% reduction in TTHM formation and the removal of preformed THM species was as high as 40%. Overall, aquifer storage of chlorinated finished water resulted in a 44% reduction in TTHM formation when additionally chlorinated after withdrawal. Bromate formed during ozonation was reduced by approximately 54%. This study indicates that the sequencing of chlorination or ozonation with respect to aquifer storage and recovery operations can impact DBP formation.     

Flow Depletion Induced by Pumping Well from Stream Perpendicularly Intersecting Impermeable/Recharge Boundary

Analytical solutions for rate and volume of flow depletion induced by pumping a well from a stream that intersects an impermeable or a recharge boundary at right angles are derived using the basic flow depletion factor defined earlier by the author. A new concept of directly obtaining stream flow depletion using the method of images is proposed. The solutions are derived for five different management cases of a stream and boundary intersecting at right-angles, assuming the aquifer to be confined with semi-infinite areal extent. A computationally simple function is proposed for accurately approximating the error function. The existing analytical solution in the case of a right-angle bend of stream given by Hantush was obtained for unconfined aquifers using a linearization of the governing partial differential equation. The solution for this case obtained using the proposed method for confined aquifer is the same as obtained by Hantush for unconfined aquifers, which shows that the linearization adopted by Hantush does not actually solve this problem for unconfined aquifers.     

Water Table Rise in Sloping Aquifer due to Canal Seepage and Constant Recharge

Analytical solutions of the linearized Boussinesq equation and a fully implicit finite-difference numerical solution of the nonlinear Boussinesq equation were obtained to study transient and steady-state water table rise in a homogeneous, isotropic, and incompressible unconfined sloping aquifer. The rise was due to seepage from two canals located at different elevations above the sloping impermeable barrier and constant recharge from the land surface. Proposed analytical solutions were verified with existing analytical solutions for a horizontal aquifer and were found in close agreement. The effect of recharge and slope of the impermeable barrier on water table rise predicted by both analytical and numerical solutions was studied by considering a numerical example. The effect of the linearization of the Boussinesq equation on water table rise was also studied by comparing the water table heights predicted by the numerical solution with those computed from the analytical solution. The analytical solution overestimates water table elevations compared to those obtained from the numerical solution, and the difference in water table in the middle region decreases with increase in time.     

EDTA, NTA, Alkylphenol Ethoxylate and DOC Attenuation during Soil Aquifer Treatment

Removals of dissolved organic carbon (DOC), ethylenediaminetetraacetic acid (EDTA), nitrilotriacetic acid (NTA) and carboxylated alkylphenol polyethoxylate metabolites (APECs) were studied at the Sweetwater (Tucson, Ariz.) soil aquifer treatment site that treated chlorinated secondary effluent. The site was operated in a first phase by flooding irregularly for weeks interrupted by days of drying and in a second phase by a regular schedule of flooding for 3 days and drying for 4 days. The average hydraulic loading rates were 0.13 and 0.17?m/day in the first and second phases, respectively. During drying, oxygen intruded at least 3?m deep into the unsaturated subsurface causing nitrification of the ammonium that was retained in the top layer during flooding. Nitrification increased nitrate concentrations to >200?mg/L but most was removed to <10?mg/L during transport to 38?m depth. At 38?m depth, removals of DOC, EDTA, NTA, and APECs during the first phase were 85, 80, 90, and 98%, slightly higher (<7%) than during the second phase. Most of the DOC removal occurred during transport to 3?m and most of the trace organics removal occurred during transport from 3 to 38?m depth.     

Comparative Study of SVMs and ANNs in Aquifer Water Level Prediction

In this research, a data-driven modeling approach, support vector machines (SVMs), is compared to artificial neural networks (ANNs) for predicting transient groundwater levels in a complex groundwater system under variable pumping and weather conditions. Various prediction horizons were used, including daily, weekly, biweekly, monthly, and bimonthly prediction horizons. It was found that even though modeling performance (in terms of prediction accuracy and generalization) for both approaches was generally comparable, SVM outperformed ANN particularly for longer prediction horizons when fewer data events were available for model development. In other words, SVM has the potential to be a useful and practical tool for cases where less measured data are available for future prediction. The study also showed high consistency between the training and testing phases of modeling when using SVM compared to ANN. While for the proposed SVM model the relative error of mean square error increased by an average of 42% from the training phase to testing the phase, the corresponding testing error of the ANN model raised by approximately seven times the training error.