E. Annette Hernandez

A fuzzy simulation–optimization approach for optimal estimation of groundwater availability under decision maker uncertainty

As groundwater is a slowly replenishing resource that can be depleted relatively easily, there is a growing interest in actively managing aquifer resources. Participatory, multi-stakeholder driven approaches are increasingly being adopted to plan groundwater use such that the resource is available for present as well as future needs. The state of Texas requires neighboring GCDs (local regulatory bodies) within a groundwater management area (GMA) to engage in joint planning activities and define desired future conditions (DFCs) for the aquifers they regulate. The DFCs are then used to estimate modeled available groundwater which defines how much water is available within an aquifer in a given region. The groundwater joint planning process was modeled using a combined simulation–optimization modeling scheme in this study. The response surface methodology was used to establish regional-scale aquifer stress-response relationships. In addition to average county-wide drawdown, other aquifer responses including stream-aquifer exchanges, coastal-aquifer exchanges and GMA-wide drawdown were considered to define the DFCs. A constrained linear regression was used in conjunction with a regional groundwater flow model to obtain the necessary response functions which formed the basis for a crisp optimization model whose objective was maximizing groundwater production while ensuring that the prescribed DFCs are not violated (constraints). This model was transformed into a fuzzy linear programming model to account for the fact that groundwater planners find it difficult to specify DFCs with a high degree of precision. Using linear membership functions, the decision makers’ preferences were captured using two values––a minimum preferred cut-off and the maximum allowable value for the metric. For estimating groundwater availability, the fuzzy optimization model reconciles production and maximizes the goal and the constraints representing the DFCs. The developed framework was illustrated by applying it to joint planning in Groundwater Management Area 15 in South Texas. The optimization models were highly sensitive to acceptable average drawdowns, while the coastal-aquifer interactions had secondary impacts. The fuzzy optimization model yielded lower estimates of groundwater availability in comparison to the crisp optimization scheme. The fuzzy optimization model is therefore consistent with the precautionary principle and recommended for use in the early stages of groundwater planning where incomplete understanding of the aquifer dynamics precludes specification of precise limits for the DFCs.

Assessing an intermittently operated household scale slow sand filter paired with household bleach for the removal of endocrine disrupting compounds

Endocrine disrupting compounds (EDCs) are a contaminant of emerging concern throughout the world, including developing countries where centralized water and wastewater treatment plants are not common. In developing countries, household scale water treatment technologies such as the biosand filter (BSF) are used to improve drinking water quality. No studies currently exist on the ability of the BSF to remove EDCs. In this experiment, the BSF was evaluated for the removal of three EDCs, estrone (E1), estriol (E3), and 17α-ethinyl estradiol (EE2). Removal results were compared to the slow sand filter (SSF) from the literature, which is similar to the BSF in principal but comparisons have revealed differences in removal of other water quality parameters between SSF and BSF. In general, the BSF minimally removed the compounds from spiked lake water as removal was less than 15% for all three compounds, though mass removal much higher than other studies in which the SSF was used. Household bleach was added to the rate was BSF effluent as suggested in order to achieve different Cl- concentrations (0.67, 2.0, 5.0, and 10.0 mg/L) and subsequent removal of EDCs by oxidation was examined. Concentrations were reduced > 98% for all compounds when the Cl- concentration was greater than 5 mg/L. Removal efficiency was > 50% at the 0.67 mg/L Cl- concentration, while almost 70% removal was observed for all compounds at the 2.0 mg/L Cl- concentration.

A multi-media planning model for assessing co-located energy and desalination plants

The co-location of desalination plants with existing or proposed power plants can bring forth economic and ecological advantages in terms of reducing the costs of water intake and reducing fish impingement. However, fossil fuel-based power plants are a source of ozone precursors and the added strain of power needed for the energy intensive desalination process increases these pollutants into the atmosphere. Furthermore, withdrawal from brackish water sources puts a stress on slowly replenishing aquifers. Additionally, the resulting concentrate is highly saline and disposal into ecologically sensitive bays and estuaries may be problematic. Balancing these limitations with the need for freshwater is of great importance for sustainability of water scarce arid and semi-arid regions and also requires a holistic multimedia impact evaluation. Therefore, an integrated system of systems approach is adopted in this study and a decision support system that integrates the flow of water, concentrate and resulting pollutants through two engineered (power plant and desalination plant) as well as three natural systems (coastal bay, aquifer and the atmosphere) is developed to study the co-location of a power plant and a desalination plant near the City of Corpus Christi in South Texas. The objective of the model is to minimize the amount of groundwater extraction and minimize the amount of water extracted from the bay to emphasize water and ecosystem conservation, respectively. These objectives, in turn, are subject to various other constraints including (1) conservation of mass; (2) air quality regulations; (3) salinity regulation policies; (4) groundwater management constraints; (5) water demand requirements; and (6) energy demand constraints. The results indicate that when conservation of the aquifer is weighted more, less water is pulled from the aquifer until later time periods. The salinity of the bay increases and creates a need for a greater amount of power necessary to process the saline water which, in turn, enhances the atmospheric loading of ozone precursors. Therefore, the conservation of groundwater scenario is limited by the air quality standards. Alternatively, when the goal is to conserve the ecological integrity of the bay while meeting freshwater demands, the model is bound by the prescribed drawdown constraint that limits the amount of water that can be extracted from the aquifer. The results from the study indicate that blending saline bay water with brackish groundwater and using cleaner burning fossil fuels that have limited air quality impacts will enhance the performance of the co-located power and desalination operations. The results of the study highlight the need for an integrated multimedia evaluation in assessing the feasibility of desalination in areas with marginal air quality.

A successive steady-state model for simulating freshwater discharges and saltwater wedge profiles at Baffin Bay, Texas

Submarine groundwater discharges (SGD) are an important source of freshwater to coastal bays and estuaries in arid and semi-arid regions. Understanding groundwater flows to these ecologically sensitive bodies is important for coastal environmental sustainability. A management-oriented mathematical model capable of simulating the flow of groundwater into a coastal bay (i.e., submarine groundwater discharge) is developed here using the principles of quasi-steady-state flow and the existence of a sharp interface between the freshwater and the saltwater portions of the aquifer. The model is applied to the Baffin Bay in South Texas, a hypersaline coastal body with no major river discharges. Two global sensitivity approaches (the one-at-a time design; OAT) and the grid-based Monte Carlo sensitivity index are used to identify critical model inputs. The sensitivity of the model inputs to the Nash–Sutcliffe Efficiency (NSE) criterion is calculated making use of synoptic observed SGD measurements made over a period of one tidal cycle. The results of the study indicate that global sensitivity analysis methods are particularly sensitive to the number of model realizations. The ability of these techniques to screen out insensitive model inputs increased with increasing number of realizations. The variability in the identified inputs was more prominent with the OAT sensitivity methods than Monte Carlo-based techniques. In general, the aquifer properties (hydraulic conductivity and aquifer thickness) as well as fluid properties (seawater and fresh water densities) along with the antecedent SGD was noted to be the most sensitive parameters. This result indicates that the implementation of sharp-front coastal–aquifer models can be improved through better hydrogeologic characterization and measuring temperature and salinity data to improve density estimation. The global sensitivity methods also help identify reasonable values for model inputs which can serve as a starting point for advanced calibrations. The results, however, indicated that the model is likely over-parameterized with different input sets yielding similar NSE estimates. Based on these initial parameter estimates, the model was able to capture the general trend in the observed SGD but could not capture the dynamic associated with high water levels in the bay. Pre-calibration global sensitivity analysis is recommended in similar applications as it not only provides insights into future data collection efforts but can also help assess the likely success of model calibration. However, given the variability among the techniques, it is suggested that multiple global sensitivity methods be utilized.

Semi-analytical solutions for stream–aquifer interactions under triangular stream-stage variations and its application to study urbanization impacts in an ungaged watershed of south Texas

Simple yet physically based models to evaluate stream–aquifer interactions during a flooding event subject to triangular stream stage variation were developed in this study. The results from the developed models were compared with other analytical and numerical solutions and noted to be very accurate. The study fills an important gap with regard to available analytical and semi-analytical solutions for modeling stream–aquifer interactions, which can be used for evaluating numerical codes. In particular, the developed models are very useful to obtain preliminary insights with regard to bank storage in ungaged watersheds as required for watershed management and planning studies in rapidly urbanizing watersheds. The utility of the model is illustrated by applying it to study the effects of urbanization on stream–aquifer interactions in the Arroyo Colorado River Watershed along the US–Mexico border region. The results indicate that increased urbanization reduces the amount of influx into the banks. The reduction in flood passage time was noted to have a greater impact than the associated rise in stage. The presence of a semi-permeable barrier was seen to mask the effects of urbanization. The model results also implicitly highlight the importance of how water quality variations caused due to urbanization can affect stream–aquifer interactions.

Identifying influencing wells for gradient estimation in the confined portion of the Gulf Coast aquifer near Kingsville, TX

Hydraulic gradient is a fundamental aquifer characteristic required to estimate groundwater flow and quantify advective fluxes of pollutants. Graphical and local estimation schemes using potentiometric head information from three or four wells are used to compute hydraulic gradients but suffer from imprecision and subjectivity. The use of linear regression is recommended when hydraulic head data from a groundwater monitoring network consisting of several wells are available. In such cases, statistical influence analysis can be carried out to evaluate how each well within the network impacts the gradient estimate. A suite of five metrics, namely—(1) the hat-values, (2) studentized residuals, (3) Cook’s distance, (4) DFBETAs and (5) Covariance ratio (COVRATIO) are used in this study to identify influential wells within a regional groundwater monitoring network in Kleberg County, TX. The hat-values indicated that the groundwater network was reasonably well balanced and no well exerted an undue influence on the regression. The studentized residuals and Cook’s distance indicated the wells with the highest influence on the regression predictions were those that were close to high groundwater production centers or affected by coastal-aquifer interactions. However, the wells in the proximity of the production centers had the highest impact on the estimated gradient values as ascertained using DFBETAs. The covariance ratio which indicates the sensitivity of a monitoring well on the estimated standard error of regression was noted to be significant at most wells within the network. Therefore, networks seeking to address changes in groundwater gradients due to climate and anthropogenic influences need to be denser than those used to ascertain synoptic gradient estimates alone. The magnitude of the groundwater velocity was greatly underestimated when the influential wells were excluded from the network. The direction of flow was affected to a lesser extent, but a complete gradient reversal was noted when the network consisted of only four peripheral wells. The influence analysis therefore provides a valuable tool to assess the importance of individual wells within a monitoring network and can therefore be useful when existing networks are to be pruned due to fiscal constraints.

Combined optimization of a wind farm and a well field for wind-enabled groundwater production

Abstract Energy requirements constitute a significant cost in groundwater production and can also add to a large carbon footprint when fossil fuels are used for power. Wind-enabled water production is advantageous as it minimizes air pollution impacts associated with groundwater production and relies on a renewable resource. Also, as groundwater extraction represents a deferrable load (i.e., it can be carried out when energy demands within an area are low), it provides a convenient way to overcome the intermittency issue associated with wind power production. Multiple turbine wind farms are needed to generate large quantities of power needed for large-scale groundwater production. Turbines must be optimally located in these farms to ensure proper propagation of kinetic energy throughout the system. By the same token, well placement must reconcile the competing objectives of minimizing interferences between production wells while ensuring the drawdowns at the property boundary are within acceptable limits. A combined simulation–optimization based model is developed in this study to optimize the combined wind energy and water production systems. The wind farm layout optimization model is solved using a re-sampling strategy, while the well field configuration is obtained using the simulated annealing technique. The utility of the developed model is to study wind-enabled water production in San Patricio County, TX. Sensitivity analysis indicated that identifying optimal placement of turbines is vital to extract maximum wind power. The variability of the wind speeds has a critical impact on the amount of water that can be produced. Innovative technologies such as variable flow pumping devices and aquifer storage recovery must be used to smooth out wind variability. While total groundwater extraction is less sensitive to uncertainty in hydrogeological parameters, improper estimation of aquifer transmissivity and storage characteristics can affect the feasibility of wind-driven groundwater production.

A multidimensional fuzzy least-squares regression approach for estimating hydraulic gradients in unconfined aquifer formations and its application to the Gulf Coast aquifer in Goliad County, Texas

Epistemic uncertainties arise during the estimation of hydraulic gradients in unconfined aquifers due to planar approximation of the water table as well as data gaps arising from factors such as instrument failures and site inaccessibility. A multidimensional fuzzy least-squares regression approach is proposed here to estimate hydraulic gradients in situations where epistemic uncertainty is present in the observed water table measurements. The hydraulic head at a well is treated as a normal (Gaussian) fuzzy variable characterized by a most likely value and a spread. This treatment results in hydraulic gradients being characterized as normal fuzzy numbers as well. The multidimensional fuzzy least-squares regression has an exact analytical form and as such can be implemented easily using matrix algebra methods. However, the method was noted to be sensitive to round-off and truncation errors when the epistemic uncertainties are small. A closeness index based on the cardinality of a fuzzy number is used to evaluate how well the regression model fits the fuzzy hydraulic head observations. A fuzzy Euclidian distance measure is used to compare two fuzzy numbers and to evaluate how fuzziness in the observed hydraulic heads affects the fuzziness in the estimated hydraulic gradients. The Euclidian distance measure is also used to ascertain the influence of each well on the fuzzy hydraulic gradient estimation. The fuzzy regression framework is illustrated by applying it to evaluate hydraulic gradients in the unconfined portion of the Gulf Coast aquifer in Goliad County, TX. The results from the case-study indicate that there is greater uncertainty associated with the estimation of the hydraulic gradients in the vertical (Z-axis) direction. The epistemic uncertainties in the hydraulic head data at the wells have a significant impact on the gradient estimates when they are of the same order of magnitude as the most likely values of the observed heads. The influence analysis indicated that 5 of the 13 wells in the network had a critical influence on at least one of the hydraulic gradients. Three wells along the northeastern section of the study area and bordering the Victoria County were noted to have the least influence on the regression estimates. The fuzzy regression framework along with the associated goodness-of-fit and influence measures provides a useful set of tools to characterize the uncertainties in the hydraulic heads and gradients arising from data gaps and planar water table approximation.

Prioritizing Groundwater Monitoring in Data Sparse Regions using Atanassov Intuitionistic Fuzzy Sets (A-IFS)

Water quality index (WQI) is a single measure that is commonly used to prioritize water wells and manage groundwater resources. WQI is pragmatic as it combines several water quality parameters into a single index. However, the process of aggregation is imprecise and suffers from uncertainties in measurements and subjective specification of weights. The goal of this study is to demonstrate how Atanassov’s Intuitionistic Fuzzy Sets (A-IFS) can be used to aggregate water quality parameters into a composite index to rank and prioritize groundwater wells. The A-IFS weighted geometric mean (A-IFS-WGM) method and the A-IFS based Technique for Order of Preference by Similarity to Ideal Solution (A-IFS-TOPSIS) using Euclidean (A-IFS-TOPSIS-E) and Hamming (A-IFS-TOPSIS-H) are introduced and illustrated to prioritize and rank water supply wells in a fast growing yet poorly studied area in Guntur, Andhra Pradesh, India. The concept of A-IFS entropy is also presented to directly ascertain weights from the data. This objective selection of weights from the data eliminates the subjectivity and difficulties associated with assigning relative importance to different water quality parameters. The results of the study indicate that the weights obtained using the entropy methods are consistent with the geochemical characteristics of the regional aquifer. The A-IFS-WGM method is more sensitive to weights compared to the A-IFS-TOPSIS methods which are influenced to a larger extent by the membership and non-membership values (ratings). Special consideration must be placed on ascribing the hesitation margin of the decision maker and identifying the membership values for non-preference as the methods exhibit greater sensitivity to these factors. The developed methods provide pragmatic data-driven approaches to prioritize and rank groundwater wells within a monitoring network.

Detecting seasonal and cyclical trends in agricultural runoff water quality—hypothesis tests and block bootstrap power analysis

Seasonal and cyclic trends in nutrient concentrations at four agricultural drainage ditches were assessed using a dataset generated from a multivariate, multiscale, multiyear water quality monitoring effort in the agriculturally dominant Lower Rio Grande Valley (LRGV) River Watershed in South Texas. An innovative bootstrap sampling-based power analysis procedure was developed to evaluate the ability of Mann-Whitney and Noether tests to discern trends and to guide future monitoring efforts. The Mann-Whitney U test was able to detect significant changes between summer and winter nutrient concentrations at sites with lower depths and unimpeded flows. Pollutant dilution, non-agricultural loadings, and in-channel flow structures (weirs) masked the effects of seasonality. The detection of cyclical trends using the Noether test was highest in the presence of vegetation mainly for total phosphorus and oxidized nitrogen (nitrite + nitrate) compared to dissolved phosphorus and reduced nitrogen (total Kjeldahl nitrogen—TKN). Prospective power analysis indicated that while increased monitoring can lead to higher statistical power, the effect size (i.e., the total number of trend sequences within a time-series) had a greater influence on the Noether test. Both Mann-Whitney and Noether tests provide complementary information on seasonal and cyclic behavior of pollutant concentrations and are affected by different processes. The results from these statistical tests when evaluated in the context of flow, vegetation, and in-channel hydraulic alterations can help guide future data collection and monitoring efforts. The study highlights the need for long-term monitoring of agricultural drainage ditches to properly discern seasonal and cyclical trends.