Aquifer heterogeneity controls to quality monitoring network performance for the protection of groundwater production wells.

A groundwater monitoring network surrounding a pumping well (such as a public water supply) allows for early contaminant detection and mitigation where possible contaminant source locations are often unknown. This numerical study investigates how the contaminant detection probability of a hypothetical sentinel-well monitoring network consisting of one to four monitoring wells is affected by aquifer spatial heterogeneity and dispersion characteristics, where the contaminant source location is randomized. This is achieved through a stochastic framework using a Monte Carlo approach. A single production well is considered that results in converging non-uniform flow close to the well. Optimal network arrangements are obtained by maximizing a weighted risk function that considers true and false positive detection rates, sampling frequency, early detection, and contaminant travel time uncertainty. Aquifer dispersivity is found to be the dominant parameter for the quantification of network performance. For the range of parameters considered, a single monitoring well screening the full aquifer thickness is expected to correctly and timely identify at least 12% of all incidents resulting in contaminants reaching the production well. This proportion increases to a global maximum of 96% for a network consisting of four wells and very dispersive transport conditions. Irrespective of network size and sampling frequency, more dispersive transport conditions result in higher detection rates. Increasing aquifer heterogeneity and decreasing aquifer spatial continuity also lead to higher detection rates, though these effects are diminished for networks of 3 or more wells. Statistical anisotropy has no effect on the network performance. Earlier detection, which is critical for remedial action and supply safety, comes with a significant cost in terms of detection rate, and should be carefully considered when a monitoring network is being designed.

Parallel Simulation in Subsurface Hydrology: Evaluating the Performance of Modeling Computers.

Monte Carlo uncertainty analysis, model calibration and optimization applications in hydrology, usually involve a very large number of forward transient model solutions, often resulting in computational bottlenecks. Parallel processing can significantly reduce overall simulation time, benefiting from the architecture of modern computers. This work investigates system performance using two realistic flow and transport modeling scenarios, applied to various modeling hardware, to provide information on the expected performance of parallel simulations and inform investment decisions. We investigate how performance, measured in terms of speedup and efficiency, changes with increasing number of parallel processes. We conclude that the maximum performance achieved by parallelization can range from 40% to 100% of the theoretical limit, with the lower increases associated with multi-CPU servers. The number of parallel processes required to maximize performance is application dependent, and in contrast to common practice, often needs to be significantly larger than the total number of system CPU cores. Further testing is required to better understand how the physical problem being simulated affects the optimal number of parallel processes needed. Finally, when laptops are considered for modeling applications, careful consideration should be given not only to the specifications but also to the intended use designated by the manufacturer.

Outcomes of the first combined national survey of pesticides and emerging organic contaminants (EOCs) in groundwater in New Zealand 2018.

The first national survey of Emerging Organic Contaminants (EOCs) involved sampling 121 wells located throughout New Zealand and analysis for a suite of 29 EOCs. This survey was carried out in conjunction with the 2018 national survey of pesticides in groundwater, a survey that is conducted on a four-yearly basis which included the analysis of glyphosate for first time. A total of 227 EOCs were detected in the 85 wells (70%). There were 29 different EOCs in the analytical suite and 25 different EOCs were detected in at least one well. The highest concentration measured was 655 ng/L for sucralose, an artificial sweetener. These results indicate that EOCs, sourced from either animal or human effluents/activities, are making their way into shallow groundwater systems and can be detected at low concentrations. A total of 135 wells were analysed for glyphosate, glufosinate and their principal metabolites. There was only one detection of glyphosate at a concentration of 2.1 µg/L. This well showed evidence of poor wellhead protection and the contamination likely came from containers that were stored near the well. A total of 279 wells were sampled and analysed for pesticides and 68 wells (24.4%) contained detectable residues of pesticides, with 28 of these wells having two or more pesticides detected. The maximum number of pesticides detected in one well was six. None of the sampled wells exceeded the Maximum Acceptable Value (MAV) for drinking water in New Zealand and the concentrations of most of the detected pesticides were equivalent to less than 0.5% of the MAV. Comparisons with earlier National Surveys of pesticides in groundwater in New Zealand indicate the frequencies of pesticide detections have remained similar over the last 16 years, with higher detection frequencies occurring before that time.

Achieving unbiased predictions of national-scale groundwater redox conditions via data oversampling and statistical learning.

An important policy consideration for integrated land and water management is to understand the spatial distribution of nitrate attenuation in the groundwater system, for which redox condition is the key indicator. This paper proposes a methodology to accommodate the computational demands of large datasets, and presents national-scale predictions of groundwater redox class for New Zealand. Our approach applies statistical learning methods to relate the redox class determined on groundwater samples to spatially varying attributes. The trained model uses these spatial variables to predict redox status in areas without sample data. We assembled the groundwater sample data from regional authority databases, and assigned each sample a redox class. A key achievement was to overcome the influence of sample selection bias on model training via oversampling. We removed additional bias imposed by imbalances in the predictor variables by applying a conditional inference random forest classifier. The unbiased trained model uses eight predictors, and achieves a high validation performance (accuracy 0.81, kappa 0.71), providing good confidence in model predictions. National maps are provided for redox class and probability at specified depths. Feature importance rankings indicate that reducing conditions are associated with poorly-drained soils, and to a lesser extent, high hydrological variability, low elevation, and low-permeability lithology. These conditions are common in New Zealand's coastal and lowland plains, where artificial drainage is required to make land suitable for production. The spatial extent of reduced groundwater increases with depth, suggesting a shallow influence of soil infiltration or mobile organic carbon, and a deeper influence of lithological electron donors. Our model provides unbiased predictions at a scale relevant for environmental policy development and legislation. Identifying where the ecosystem service provided by denitrification can be utilised will enable spatially targeted interventions that can achieve the desired environmental outcome in a more cost-effective manner than non-targeted interventions.

The effects of denitrification parameterization and potential benefits of spatially targeted regulation for the reduction of N-discharges from agriculture.

Diffuse nitrate leaching from agricultural areas is a major environmental problem in many parts of the world. Understanding where in a catchment nitrate is removed is key for designing effective land use management strategies that protect water quality, while minimizing the impact on economic development. In this study we assess the effects of spatially targeted nitrate leaching regulation in a basin with limited knowledge of the complexity of chemical heterogeneity. Three alternative nitrate reactivity spatial parameterizations were incorporated in a catchment-scale flow and transport model and used to evaluate the effectiveness of four possible spatially targeted regulation options. Our findings confirm that denitrification parameterization cannot be numerically determined based on model inversion alone. Detailed field based characterization using physical and geochemical methods should be considered and incorporated in the numerical inversion scheme. We also demonstrate that there are potential benefits of implementing spatially targeted regulation compared to spatially uniform regulation. Focusing regulation in areas where nitrate residence time is short, such as riparian zones or areas with low natural N-reduction, results in greater reduction of N-discharges through groundwater. Significantly improved efficiencies can be expected when delineation of management zones considers the chemical heterogeneity and groundwater flow paths. These improved efficiencies are achieved by adopting management rules that regulate land use in discharge sensitive areas, where leaching changes contribute the most to the catchment nitrate discharges. In our case study, regulation in discharge sensitive zones was twice as efficient compared to other management options.

Application of the re-circulating tracer well test method to determine nitrate reaction rates in shallow unconfined aquifers.

Five re-circulating tracer well tests (RCTWTs) have been conducted in a variety of aquifer settings, at four sites across New Zealand. The tests constitute the first practical assessment of the two-well RCTWT methodology described by Burbery and Wang (Journal of Hydrology, 2010; 382:163-173) and were aimed at evaluating nitrate reaction rates in situ. The performance of the RCTWTs differed significantly at the different sites. The RCTWT method performed well when it was applied to determine potential nitrate reaction rates in anoxic, electro-chemically reductive, nitrate-free aquifers of volcanic lithology, on the North Island, New Zealand. Regional groundwater flow was not fast-flowing in this setting. An effective first-order nitrate reaction rate in the region of 0.09 d(-1) to 0.26 d(-1) was determined from two RCTWTs applied at one site where a reaction rate of 0.37 d(-1) had previously been estimated from a push-pull test. The RCTWT method performed poorly, however, in a fast-flowing, nitrate-impacted fluvio-glacial gravel aquifer that was examined on the South Island, New Zealand. This setting was more akin to the hypothetical physiochemical problem described by Burbery and Wang (2010). Although aerobic conditions were identified as the primary reason for failure to measure any nitrate reaction in the gravel aquifer, failure to establish significant interflow in the re-circulation cell due to the heterogeneous nature of the aquifer structure, and natural variability exhibited in nitrate contaminant levels of the ambient groundwater further contributed to the poor performance of the test. Our findings suggest that in practice, environmental conditions are more complex than assumed by the RCTWT methodology, which compromises the practicability of the method as one for determining attenuation rates in groundwater based on tracing ambient contaminant levels. Although limited, there appears to be a scope for RCTWTs to provide useful information on potential attenuation rates when reactants are supplemented to the aquifer system under examination.

Groundwater biofilm dynamics grown in situ along a nutrient gradient.

This paper describes the in situ response of groundwater biofilms in an alluvial gravel aquifer system on the Canterbury Plains, New Zealand. Biofilms were developed on aquifer gravel, encased in fine mesh bags and suspended in protective columns in monitoring wells for at least 20 weeks. Four sites were selected in the same groundwater system where previous analyses indicated a gradient of increasing nitrate down the hydraulic gradient from Sites 1 to 4. Measurements during the current study classified the groundwater as oligotrophic. Biofilm responses to the nutrient gradients were assessed using bioassays, with biomass determined using protein and cellular and nucleic acid staining and biofilm activity using enzyme assays for lipid, carbohydrate, phosphate metabolism, and cell viability. In general, biofilm activity decreased as nitrate levels increased from Sites 1 to 4, with the opposite relationship for carbon and phosphorus concentrations. These results showed that the groundwater system supported biofilm growth and that the upper catchment supported efficient and productive biofilms (high ratio of activity per unit biomass).

Modeling water flow and bacterial transport in undisturbed lysimeters under irrigations of dairy shed effluent and water using HYDRUS-1D.

HYDRUS-1D was used to simulate water flow and leaching of fecal coliforms and bromide (Br) through six undisturbed soil lysimeters (70 cm depth by 50 cm diameter) under field conditions. Dairy shed effluent (DSE) spiked with Br was applied to the lysimeters, which contained fine sandy loam layers. This application was followed by fortnightly spray or flood water irrigation. Soil water contents were measured at four soil depths over 171 days, and leachate was collected from the bottom. The post-DSE period simulations yielded a generally decreased saturated water content compared to the pre-DSE period, and an increased saturated hydraulic conductivity and air-entry index, suggesting that changes in soil hydraulic properties (e.g. via changes in structure) can be induced by irrigation and seasonal effects. The single-porosity flow model was successful in simulating water flow under natural climatic conditions and spray irrigation. However, for lysimeters under flood irrigation, when the effect of preferential flow paths becomes more significant, the good agreement between predicted and observed water contents could only be achieved by using a dual-porosity flow model. Results derived from a mobile-immobile transport model suggest that compared to Br, bacteria were transported through a narrower pore-network with less mass exchange between mobile and immobile water zones. Our study suggests that soils with higher topsoil clay content and soils under flood irrigation are at a high risk of bacteria leaching through preferential flow paths. Irrigation management strategies must minimize the effect of preferential flow to reduce bacterial leaching from land applications of effluent.

Effects of pH, ionic strength, dissolved organic matter, and flow rate on the co-transport of MS2 bacteriophages with kaolinite in gravel aquifer media.

Viruses are often associated with colloids in wastewater and could be transported with colloids into groundwater from land disposal of human and animal effluent and sludge, causing contamination of groundwater. To investigate the role of colloids in the transport of viruses in groundwater, experiments were conducted using a 2m long column packed with heterogeneous gravel aquifer media. Bacteriophage MS2 was used as the model virus and kaolinite as the model colloid. Experimental data were analyzed using Temporal Moment Analysis and Filtration Theory. In the absence of kaolinite colloid, MS2 phage traveled slightly faster than the conservative tracer bromide (Br), with little differences observed between unfiltered and filtered MS2 phage (0.22 microm as the operational cut-off for colloid-free virus). In the presence of kaolinite colloids, MS2 phage breakthrough occurred concurrently with that of the colloidal particles and the time taken to reach the peak virus concentration was reduced, suggesting a colloid-facilitated virus transport in terms of peak-concentration time and velocity. Meanwhile mass recovery and magnitude of concentrations of the phages were significantly reduced, indicating colloid-assisted virus attenuation in terms of concentrations and mass. Decreasing the pH or increasing the ionic strength increased the level of virus attachment to the aquifer media and colloids, and virus transport became more retarded, resulting in lower peak-concentration, lower mass recovery, longer peak-concentration time, and greater apparent collision efficiency. Increasing the concentration of dissolved organic matter (DOM) or flow rate resulted in faster virus transport velocity, higher peak-concentrations and mass recoveries, and lower apparent collision efficiencies. The dual-role of colloids in transport viruses has important implications for risk analysis and remediation of virus-contaminated sites.

Dissipation and sorption of six commonly used pesticides in two contrasting soils of New Zealand.

We investigated dissipation and sorption of atrazine, terbuthylazine, bromacil, diazinon, hexazinone and procymidone in two contrasting New Zealand soils (0-10 cm and 40-50 cm) under controlled laboratory conditions. The six pesticides showed marked differences in their degradation rates in both top- and subsoils, and the estimated DT(50) values for the compounds were: 19-120 (atrazine), 10-36 (terbuthylazine), 12-46 (bromacil), 7-25 (diazinon), 8-92 (hexazinone) and 13-60 days for procymidone. Diazinon had the lowest range for DT(50) values, while bromacil and hexazinone gave the highest DT(50) values under any given condition on any soil type. Batch derived effective distribution coefficient (K(d)(eff)) values for the pesticides varied markedly with bromacil and hexazinone exhibiting low sorption affinity for the soils at either depth, while diazinon gave high sorption values. Comparison of pesticide degradation in sterile and non-sterile soils suggests that microbial degradation was the major dissipation pathway for all six compounds, although little influence of abiotic degradation was noticeable for diazinon and procymidone.

Validation of two innovative methods to measure contaminant mass flux in groundwater.

The ability to quantify the mass flux of a groundwater contaminant that is leaching from a source area is critical to enable us to: (1) evaluate the risk posed by the contamination source and prioritize cleanup, (2) evaluate the effectiveness of source remediation technologies or natural attenuation processes, and (3) quantify a source term for use in models that may be applied to predict maximum contaminant concentrations in downstream wells. Recently, a number of new methods have been developed and subsequently applied to measure contaminant mass flux in groundwater in the field. However, none of these methods has been validated at larger than the laboratory-scale through a comparison of measured mass flux and a known flux that has been introduced into flowing groundwater. A couple of innovative flux measurement methods, the tandem circulation well (TCW) and modified integral pumping test (MIPT) methods, have recently been proposed. The TCW method can measure mass flux integrated over a large subsurface volume without extracting water. The TCW method may be implemented using two different techniques. One technique, the multi-dipole technique, is relatively simple and inexpensive, only requiring measurement of heads, while the second technique requires conducting a tracer test. The MIPT method is an easily implemented method of obtaining volume-integrated flux measurements. In the current study, flux measurements obtained using these two methods are compared with known mass fluxes in a three-dimensional, artificial aquifer. Experiments in the artificial aquifer show that the TCW multi-dipole and tracer test techniques accurately estimated flux, within 2% and 16%, respectively; although the good results obtained using the multi-dipole technique may be fortuitous. The MIPT method was not as accurate as the TCW method, underestimating flux by as much as 70%. MIPT method inaccuracies may be due to the fact that the method assumptions (two-dimensional steady groundwater flow to fully-screened wells) were not well-approximated. While fluxes measured using the MIPT method were consistently underestimated, the method's simplicity and applicability to the field may compensate for the inaccuracies that were observed in this artificial aquifer test.

Modelling the dissipation kinetics of six commonly used pesticides in two contrasting soils of New Zealand.

We used three non-linear bi-phasic models, bi-exponential (BEXP), first-order double exponential decay (FODED), and first-order two-compartment (FOTC), to fit the measured degradation data for six commonly used pesticides (atrazine, terbuthylazine, bromacil, diazinon, hexazinone and procymidone) in two New Zealand soils. Corresponding DT(50) and DT(90) values for each compound were numerically obtained and compared against those estimated by simple first-order kinetic (SFOK) model. All 3 non-linear models gave good fit of the measured data under both soil depths and were well supported by the values obtained for the respective statistical indices (RMSE, CRM and r(2)). The FOTC model gave by far the best fit for most compounds, followed by the FODED and BEXP models. Overall, DT(50) values derived by non-linear models for the six compounds in soils from both sites were lower than the values obtained by the SFOK model. Differences in the SFOK and the three non-linear models derived DT(90)were, however, an order of magnitude higher for some compounds, while for others differences were very small. Although all three non-linear models described most data by giving excellent fits, in a few instances > 5-10% asymptotes hindered the estimation of DT(90) values. This work shows that when degradation deviates from first-order kinetic, application of non-linear decay models to describe the kinetics of degradation becomes important in order to derive the true end-points for pesticides in soil.

Use of tandem circulation wells to measure hydraulic conductivity without groundwater extraction.

Conventional methods to measure the hydraulic conductivity of an aquifer on a relatively large scale (10-100 m) require extraction of significant quantities of groundwater. This can be expensive, and otherwise problematic, when investigating a contaminated aquifer. In this study, innovative approaches that make use of tandem circulation wells to measure hydraulic conductivity are proposed. These approaches measure conductivity on a relatively large scale, but do not require extraction of groundwater. Two basic approaches for using circulation wells to measure hydraulic conductivity are presented; one approach is based upon the dipole-flow test method, while the other approach relies on a tracer test to measure the flow of water between two recirculating wells. The approaches are tested in a relatively homogeneous and isotropic artificial aquifer, where the conductivities measured by both approaches are compared to each other and to the previously measured hydraulic conductivity of the aquifer. It was shown that both approaches have the potential to accurately measure horizontal and vertical hydraulic conductivity for a relatively large subsurface volume without the need to pump groundwater to the surface. Future work is recommended to evaluate the ability of these tandem circulation wells to accurately measure hydraulic conductivity when anisotropy and heterogeneity are greater than in the artificial aquifer used for these studies.

Distance and flow effects on microsphere transport in a large gravel column.

Consumption of microbially contaminated ground water can cause adverse health effects and the processes involved in pathogen transport in aquifers need to be understood. The influences of distance, flow velocity, and colloid size on colloid transport were examined in homogenous pea-gravel media using an 8-m column and three sizes (1, 5, and 10 microm) of microspheres. Experiments were conducted at three flow rates by simultaneously injecting microspheres with a conservative tracer, bromide. Observed concentrations were simulated with CXTFIT and analyzed with filtration theory. The results demonstrate that colloid concentration is strongly log-linearly related to transport distance (as suggested by filtration theory) in coarse gravels, similar to our previous field studies. In contrast, the log-linear relationship is often reported to be invalid in fine porous media. The observed log-linear relationship is possibly because straining is negligible in the coarse gravels investigated. This has implications in predicting setback distances for land disposal of effluent, and suggests that setback distances in gravel aquifers can be estimated using constant spatial removal rates (f). There was an inverse relationship between transport distance and colloidal concentration, but not with temporal attachment rate (katt) and collision coefficient (alpha). Increases in flow velocity result in increasing colloidal recovery, katt and alpha but decreasing f. Increases in sphere size result in decreasing colloidal recovery with increasing katt, f, alpha, and velocity enhancement. Diffusion is the dominant collision mechanism for 1-microm spheres (81-88%), while settling dominates for 5- and 10-microm spheres (> 87%), and interception is very small for all spheres investigated.