Multi-Objective Optimization of Managed Aquifer Recharge

This study demonstrates the utilization of a multi-objective hybrid global/local optimization algorithm for solving managed aquifer recharge (MAR) design problems, in which the decision variables included spatial arrangement of water injection and abstraction wells and time-variant rates of pumping and injection. The objective of the optimization was to maximize the efficiency of the MAR scheme, which includes both quantitative and qualitative aspects. The case study used to demonstrate the capabilities of the proposed approach is based on a published report on designing a real MAR site with defined aquifer properties, chemical groundwater characteristics as well as quality and volumes of injected water. The demonstration problems include steady state and transient scenarios. The steady state scenario demonstrates optimization of spatial arrangement of multiple injection and recovery wells, whereas the transient scenario was developed with the purpose of finding optimal regimes of water injection and recovery at a single location. Both problems were defined as multi-objective problems. The scenarios were simulated by applying coupled numerical groundwater flow and solute transport models: MODFLOW-2005 and MT3D-USGS. The applied optimization method was a combination of global (the non-dominated sorting genetic algorithm [NSGA-2]) and local (the Nelder-Mead downhill simplex search algorithms). The analysis of the resulting Pareto optimal solutions led to the discovery of valuable patterns and dependencies between the decision variables, model properties, and problem objectives. Additionally, the performance of the traditional global and the hybrid optimization schemes were compared.     

Mechanism of Suspended Kaolinite Particle Clogging in Porous Media During Managed Aquifer Recharge

Managed aquifer recharge is an effective strategy for urban stormwater management. Chemical ions are normally retained in stormwater and groundwater and may accelerate clogging during the recharge process. However, the effect of water chemistry on physical clogging has not previously been investigated. In this study, we investigated the hydrogeochemical mechanism of saturated porous media clogging in a series of column experiments. The column was packed with river sand and added suspensions of kaolinite particles. Calcium chloride and sodium chloride are used as representative ions to study chemical effects. We found that an increase in ionic strength resulted in retention of kaolinite solids in the column, with a breakthrough peak of C/C₀ value of 1 to 0.2. The corresponding hydraulic conductivity decreased with increased solids clogging. Divalent cations were also found to have a greater influence on kaolinite particle clogging than monovalent cations. The enhanced hydrochemical-related clogging was caused by kaolinite solids flocculating and increasing the deposition rate coefficient by 1 to 2 times in high ionic strength conditions. Three clogging mechanisms of kaolinite solids are proposed: surface filtration, inner blocking, and attachment. This study further deepens the understanding of the mechanisms of solids clogging during aquifer recharge and demonstrates the significance of ionic strength on recharge clogging risk assessments.     

Facilitating Open Pit Mine Closure with Managed Aquifer Recharge

Dewatering of open pit mines can lower the regional water table for distances of several kilometers from the pit. When the mine is closed, dewatering operations usually cease, and the water table near the pit begins to rise. If the pit is backfilled, the water table will eventually recover, but this recovery may take several hundred years. However, if the extracted water is re-injected into the subsurface, then this may accelerate recovery of the water table. We show that there is an optimal distance for re-injection, which is sufficiently far from the mine to minimize the amount of groundwater that flows back to the pit during mine operations (and hence necessitate additional pumping) but is still close enough to speed up the water table recovery post-mine closure. The optimal injection distance increases with the aquifer hydraulic diffusivity and the mine life (duration of dewatering and injection), and typically ranges between about two and nine times the radius of the mine pit. Where the mine pit is not backfilled, the relative reduction in drawdown due to injecting all the pumped water at the optimal distance is between approximately 10% and 50% after a recovery time equal to the mining period, increasing to 30% to 90% after a recovery time five times the mining period. The relative drawdown reduction due to managed aquifer recharge will be even greater for a pit which is backfilled when mining ceases.     

Estimating Groundwater Mounding in Sloping Aquifers for Managed Aquifer Recharge

Design of managed aquifer recharge (MAR) for augmentation of groundwater resources often lacks detailed data, and simple diagnostic tools for evaluation of the water table in a broad range of parameters are needed. In many large‐scale MAR projects, the effect of a regional aquifer base dip cannot be ignored due to the scale of recharge sources (e.g., wadis, streams, reservoirs). However, Hantush's (1967) solution for a horizontal aquifer base is commonly used. To address sloping aquifers, a new closed‐form analytical solution for water table mound accounts for the geometry and orientation of recharge sources at the land surface with respect to the aquifer base dip. The solution, based on the Dupiuit‐Forchheimer approximation, Green's function method, and coordinate transformations is convenient for computing. This solution reveals important MAR traits in variance with Hantush's solution: mounding is limited in time and space; elevation of the mound is strongly affected by the dip angle; and the peak of the mound moves over time. These findings have important practical implications for assessment of various MAR scenarios, including waterlogging potential and determining proper rates of recharge. Computations are illustrated for several characteristic MAR settings.     

Modeling Stream-Aquifer Interactions Under Seasonal Groundwater Pumping and Managed Aquifer Recharge

In South Korea, a significant amount of groundwater is used for the heating of water-curtain insulated greenhouses during the winter dry season, which had led to problems of groundwater depletion. A managed aquifer recharge (MAR) project is currently underway with the goal of preventing such groundwater depletion in a typical cultivation area, located on an alluvial aquifer near the Nam River. In the present study, FEFLOW, a three-dimensional finite element model, was used to evaluate different strategies for MAR of the cultivation areas. A conceptual model was developed to simulate the stream-aquifer dynamics under the influence of seasonal groundwater pumping and MAR. The optimal rates and duration of MAR were assessed by analyzing the recovery of the groundwater levels and the change in the groundwater temperature. The simulation results indicate that a MAR rate of 8000 m³/d effectively restores the groundwater level when the injection wells are located inside the groundwater depletion area. It is also demonstrated that starting the MAR before the beginning of the seasonal pumping is more effective. Riverbank filtration is preferable for securing the injection water owing to plentiful source of induced recharge from the river. Locating the pumping wells adjacent to the river where there are thick permeable layers could be a good strategy for minimizing decreases in the groundwater level and temperature.     

Assessment of Managed Aquifer Recharge Site Suitability Using a GIS and Modeling

We completed a two‐step regional analysis of a coastal groundwater basin to (1) assess regional suitability for managed aquifer recharge (MAR), and (2) quantify the relative impact of MAR activities on groundwater levels and sea water intrusion. The first step comprised an analysis of surface and subsurface hydrologic properties and conditions, using a geographic information system (GIS). Surface and subsurface data coverages were compiled, georeferenced, reclassified, and integrated (including novel approaches for combining related datasets) to derive a spatial distribution of MAR suitability values. In the second step, results from the GIS analysis were used with a regional groundwater model to assess the hydrologic impact of potential MAR placement and operating scenarios. For the region evaluated in this study, the Pajaro Valley Groundwater Basin, California, GIS results suggest that about 7% (15 km²) of the basin may be highly suitable for MAR. Modeling suggests that simulated MAR projects placed near the coast help to reduce sea water intrusion more rapidly, but these projects also result in increased groundwater flows to the ocean. In contrast, projects placed farther inland result in more long‐term reduction in sea water intrusion and less groundwater flowing to the ocean. This work shows how combined GIS analysis and modeling can assist with regional water supply planning, including evaluation of options for enhancing groundwater resources.     

Sulfur Hexafluoride and Potassium Bromide as Groundwater Tracers for Managed Aquifer Recharge

Sulfur hexafluoride (SF₆) is an established tracer for use in managed aquifer recharge projects. SF₆ exsolves from groundwater when it encounters trapped air according to Henry's law. This results in its retardation relative to groundwater flow, which can help determine porous media saturation and flow dynamics. SF₆ and the conservative, nonpartitioning tracer, bromide (Br⁻ added as KBr), were introduced to recharge water infiltrated into stacked glacial aquifers in Thurston County, Washington, providing the opportunity to observe SF₆ partitioning. Br⁻, which is assumed to travel at the same velocity as the groundwater, precedes SF₆ at most monitoring wells (MWs). Average groundwater velocity in the unconfined aquifer in the study area ranges from 3.9 to 40 m/d, except in the southwestern corner where it is slower. SF₆ in the shallow aquifer exhibits an average retardation factor of 2.5 ± 3.8, suggesting an air-to-water ratio on the order of 10⁻³ to 10⁻² in the pore space. Notable differences in tracer arrival times at adjacent wells indicate very heterogeneous conductivity. One MW exhibits double peaks in concentrations of both tracers with different degrees of retardation for the first and second peaks. This suggests multiple flowpaths to the well with variable saturation. The confining layer between the upper two aquifers appears to allow intermittent connection between aquifers but serves as an aquitard in most areas. This study demonstrates the utility of SF₆ partitioning for evaluating hydrologic conditions at prospective recharge sites.     

Quantifying Modern Recharge and Depletion Rates of the Nubian Aquifer in Egypt

Egypt is currently seeking additional freshwater resources to support national reclamation projects based mainly on the Nubian aquifer groundwater resources. In this study, temporal (April 2002 to June 2016) Gravity Recovery and Climate Experiment (GRACE)-derived terrestrial water storage (TWS_GRACE) along with other relevant datasets was used to monitor and quantify modern recharge and depletion rates of the Nubian aquifer in Egypt (NAE) and investigate the interaction of the NAE with artificial lakes. Results indicate: (1) the NAE is receiving a total recharge of 20.27 ± 1.95 km³ during 4/2002?2/2006 and 4/2008–6/2016 periods, (2) recharge events occur only under excessive precipitation conditions over the Nubian recharge domains and/or under a significant rise in Lake Nasser levels, (3) the NAE is witnessing a groundwater depletion of ? 13.45 ± 0.82 km³/year during 3/2006–3/2008 period, (4) the observed groundwater depletion is largely related to exceptional drought conditions and/or normal baseflow recession, and (5) a conjunctive surface water and groundwater management plan needs to be adapted to develop sustainable water resources management in the NAE. Findings demonstrate the use of global monthly TWS_GRACE solutions as a practical, informative, and cost-effective approach for monitoring aquifer systems across the globe.     

Distributed Temperature Sensing to Measure Infiltration Rates Across a Groundwater Recharge Basin

Managed aquifer recharge is used to augment groundwater resources and provide resiliency to water supplies threatened by prolonged droughts. It is important that recharge facilities operate at their maximum efficiency to increase the volume of water stored for future use. In this study, we evaluate the use of distributed temperature sensing (DTS) technology as a tool to measure high-resolution infiltration rates at a large-scale recharge facility. Fiber optic cable was laid out inside a spreading basin in a spiral pattern, at two different depths. The cables measured the propagation of diurnal surface water temperature oscillations into the basin depth. The rate of heat propagation is proportional to the velocity of the water, making it possible to estimate the infiltration rate from the temperature measurements. Our results showed that the infiltration rate calculated from DTS, averaged over the entire basin, was within 5% of the infiltration rate calculated using a conventional metering method. The high-resolution data obtained from DTS, both spatially and temporally, revealed heterogeneous infiltration rates throughout the basin; furthermore, tracking the evolution of infiltration rates over time revealed regions with consistently high infiltration rates, regions with consistently low infiltration rates, and regions that evolved from high to low rates, which suggested clogging within that region. Water utilities can take advantage of the high-resolution information obtained from DTS to better manage recharge basins and make decisions about cleaning schedule, frequency, and extent, leading to improved basin management strategies, reduced O&M costs, and increased groundwater recharge.     

Comparison of Recharge Estimation Methods During a Wet Period in a Karst Aquifer

Management of water resources, implying their appropriate protection, calls for a sound evaluation of recharge. Such assessment is very complex in karst aquifers. Most methods are developed for application to detrital aquifers, without taking into account the extraordinary heterogeneity of porosity and permeability of karst systems. It is commonly recommended to estimate recharge using multiple methods; however, differences inherent to the diverse methods make it difficult to clarify the accuracy of each result. In this study, recharge was estimated in a karst aquifer working in a natural regime, in a Mediterranean‐type climate, in the western part of the Sierra de las Nieves (southern Spain). Mediterranean climate regions are characterized by high inter‐annual rainfall variability featuring long dry periods and short intense wet periods, the latter constituting the most important contribution to aquifer water input. This paper aims to identify the methods that provide the most plausible range of recharge rate during wet periods. Six methods were tested: the classical method of Thornthwaite‐Mather, the Visual Balan code, the chloride balance method, and spatially distributed methods such as APLIS, a novel spatiotemporal estimation of recharge, and ZOODRM. The results help determine valid methods for application in the rest of the unit of study and in similar karst aquifers.     

Evaluation of Noble Gas Recharge Temperatures in a Shallow Unconfined Aquifer

Water table temperatures inferred from dissolved noble gas concentrations (noble gas temperatures, NGT) are useful as a quantitative proxy for air temperature change since the last glacial maximum. Despite their importance in paleoclimate research, few studies have investigated the relationship between NGT and actual recharge temperatures in field settings. This study presents dissolved noble gas data from a shallow unconfined aquifer heavily impacted by agriculture. Considering samples unaffected by degassing, NGT calculated from common physically based interpretive gas dissolution models that correct measured noble gas concentrations for "excess air" agreed with measured water table temperatures (WTT). The ability to fit data to multiple interpretive models indicates that model goodness-of-fit does not necessarily mean that the model reflects actual gas dissolution processes. Although NGT are useful in that they reflect WTT, caution is recommended when using these interpretive models. There was no measurable difference in excess air characteristics (amount and degree of fractionation) between two recharge regimes studied (higher flux recharge primarily during spring and summer vs. continuous, low flux recharge). Approximately 20% of samples had dissolved gas concentrations below equilibrium concentration with respect to atmospheric pressure, indicating degassing. Geochemical and dissolved gas data indicate that saturated zone denitrification caused degassing by gas stripping. Modeling indicates that minor degassing (<10% ΔNe) may cause underestimation of ground water recharge temperature by up to 2°C. Such errors are problematic because degassing may not be apparent and degassed samples may be fit by a model with a high degree of certainty.