Identifying the influential aquifer heterogeneity factor on nitrate reduction processes by numerical simulation
Introduction
At the field scale, oxidation–reduction reactions, commonly known as redox reactions, are strongly influenced by various types of aquifer heterogeneity since hydraulic conductivity, soil and chemical reactive substances are not uniformly distributed. A heterogeneous hydraulic conductivity field, for example, can cause a wide range of groundwater travel times and flow patterns which influence transport and distribution of mobile species. Heterogeneous aquifer mineralogy affects local geochemical conditions and, therefore, the spatial distribution of reactive substances that contribute to the redox reactions. These effects may account for the fact that most reaction rates derived from homogeneous closed systems (e.g., a laboratory experiment and its corresponding numerical simulation) are significantly different from field measurements.
These potential relationships are especially important for nitrate contamination problems in shallow groundwater systems since transport and redox transformation processes of nitrate are closely linked to complex and spatially-distributed physical and geochemical reactions (Hansen et al., 2014, Wriedt and Rode, 2006). Nitrate (NO3−) is one of the most widely spread contaminants in groundwater primarily as a result of intensive use of nitrogen-containing fertilizers and/or contamination with human or animal organic waste (e.g., leaching from septic tanks and sewers). High concentrations of nitrate have been suggested as a major cause of accelerated eutrophication in lakes, reservoirs and rivers (Rivett et al., 2008; Zan et al., 2010; Rinke et al., 2013). It is also dangerous for human health, particularly the increase in risk of methemoglobinemia, which prevents oxygen transport in the bloodstream of infants (Fan and Steinberg, 1996; USEPA 2009). Therefore, the European Union (EU) Drinking Water Directive (98/83/EC) set a maximum allowable concentration for nitrate at 50 mg/L nitrate (11.3 mg/L nitrate-N) (European Union 1998). Regardless, nitrate concentration in groundwater has gradually increased in many countries (Oyarzun et al., 2007; Kludt et al., 2016). Moreover, it has been reported that a new interest in biofuel crops to meet energy needs is likely to elevate nitrate concentration in shallow groundwater resources (Twomey et al., 2010).
In principle, denitrification processes in the soil, unsaturated zone and in the aquifer may contribute substantially to the reduction of nitrate load in the groundwater. This reduction of nitrate and nitrite-bound nitrogen to gaseous products like N2O and N2 occurs under anaerobic conditions. It is controlled by two main processes which require the availability of adequate reactive electron donors (Postma et al., 1991; Zhang et al., 2009; Smith and Duff, 1988; Hiscock et al., 1991), (i) hetero-chemoorganotrophic (heterotrophic) denitrification with organic carbon as electron donor (Starr and Gillham, 1993; Bragan et al., 1997) and (ii) autolithotrophic denitrification with iron(II) sulfide as the electron donor, often described as more relevant (Tesoriero et al., 2000; Kölle et al., 1987). Pyrite (FeS2) is the most abundant sulfide mineral in the natural system and acts as a potential electron donor. The controlling factor for the nitrate reduction process by pyrite oxidation is the pH of the groundwater (Torrentó et al., 2010) and spatial distribution of pyrite in the aquifer (Zhang et al., 2013; Miotliński, 2008).
The extent of nitrate removal can be verified through comparison with in situ field measurements (extensive field sampling data) (Postma et al., 1991; Zhang et al., 2009; Kinzelbach et al., 1991; Lee et al., 2006) and laboratory experiments (Bauer et al., 2009; Rodríguez-Escales et al., 2016; Jokela et al., 2002). However, such experiments are necessarily limited to relatively short timescales, and complex coupled physical-geochemical systems are difficult to elucidate. Therefore, numerical simulation can potentially be used to develop a better understanding of these complexities (Steefel et al., 2014). Since the 1980s, a variety of numerical tools has been developed and applied to simulate nitrate fate, such as BIOMOC (Essaid and Bekins, 1997), CFPv2 with UMT3D (Xu et al., 2015), GeoSysBRNS (Centler et al., 2010), MODFLOW-PHT3D (Zhang et al., 2013), MODFLOW with RT3D (Bailey et al., 2013; Rolle et al., 2008), OpenGeoSys-GEM (Kosakowski and Watanabe, 2014), ParCrunchFlow (Beisman et al., 2015), RISK-N (Oyarzun et al., 2007), SF-Monod (Cui et al., 2014), Streamline approach (Eckert et al., 2012) and TOUGHREACT-N (Maggi et al., 2008). Although a significant body of knowledge of the fate and transport of nitrogen species has been acquired by the aforementioned existing methods, one of the main challenges is a proper characterization of the subsurface since geological and sedimentological structure are inherently heterogeneous (Bayer et al., 2015). Most previous models typically neglect the effect of the spatial variability of the subsurface due to limited data availability. Moreover, these high complexities require expensive computational demands for simulation. Stochastic approaches (Rubin, 2003; Dagan, 2002) can then be used to obtain random fields using Spatial Random Function (SRF) models which are employed to model spatial variability using statistical tools such as the mean, variance, or a probability distribution function. Each of the random fields is then used to analyze flow and transport processes. This approach has been used by a number of authors, e.g., Bellin et al. (1993), Bosma and van derZee (1993), Espinoza and Valocchi (1997), de Dreuzy et al. (2007), Mohamed et al. (2010), Heße et al. (2015) and Fiori et al. (2011).
The objective of this work is to clarify the influence of the aquifer heterogeneity on nitrate transport and redox transformation processes by using reactive transport modeling. Among the various types of aquifer heterogeneity, physical and chemical aquifer heterogeneities are considered. Physical aquifer heterogeneity is represented by spatial variability in hydraulic conductivity and chemical aquifer heterogeneity in the subsurface is represented as spatial variability in electron donor availability; distribution of the electron donor concentration that is critical for redox reaction. Previous modeling studies of aquifer heterogeneities have illustrated the contaminant fate and transport under physical (Cui et al., 2014; Fiori et al., 2011; Robinson et al., 2009) and chemical heterogeneities (Li et al., 2007; Fakhreddine et al., 2016). However, none of these studies have explored and described the most influential aquifer heterogeneity factors under coupled physical and chemical aquifer heterogeneity.
For this, OpenGeoSys (OGS), a finite element groundwater flow model including a multi species transport code (Kolditz et al., 2012), was coupled to the IPhreeqc module of the geochemical solver PHREEQC (Charlton and Parkhurst, 2011). This new coupling scheme (hereinafter referred to as OGS#IPhreeqc (He et al., 2015)) is capable of simulating chemical reactions and processes such as water flow and solute transport. Also, a parallelization scheme using MPI (Message Passing Interface) is implemented to improve computational efficiency. The heterogeneous aquifers are generated by inducing stochastically-generated fields using the random field generator of the gstat project (Pebesma, 2004).
The hypothetical aquifer systems generated in this study are based on data from Hessian Ried, an important groundwater reservoir in upper Rhine Graben in Germany. However, assumed conditions such as mean uniform flow, contaminants input patterns, distribution of reactive substances and its spatial characteristics constitute limitations which do not make yet possible direct verification processes. Still, we believe this is a necessary step and the results of this study can be interpreted from two different perspectives. First, one might be interested in modeling redox reactions (e.g., nitrate reduction reactions). This study can show how nitrate reactive transport processes with coupled physical and chemical aquifer heterogeneity has been achieved. The second perspective involves designing remediation strategy and risk assessment. Due to inherent complexities, quantifying the uncertainty of the prediction has been considered as a crucial point when choosing a model (Bolster et al., 2009). This study can show how aquifer heterogeneity influences the overall reaction efficiency, how to estimate beforehand its impact and suggests that the most influential aquifer heterogeneity factor must be considered.
This paper is organized as follows: first, we describe the methods used in this study (Section 2). Specifically, the coupling schemes for reactive transport simulation and stochastic method are described. In Section 3, the application of the model for the simulation is defined and a series of simulation scenarios is described. Finally, simulation results are presented and discussed in Sections 4 and 5.
Section snippets
Methods
In this following section, we provide a short overview of the methods and numerical tools that we used in our study.
Model application
The present study is a based on results from field study in the Hessian Ried that is located between the Odenwald in the east and river Rhine in the west in the Upper Rhine Graben (Kludt et al., 2016). This major rift system is filled with thick unconsolidated sediments and groundwater resources are extensively used for public water supply in the Rhine-Main region as well as for agricultural irrigation. Due to intensive agriculture in the Ried, nitrogen-based fertilizers have been applied for
Results and discussion
In this section, simulation results are described and compared by a series of scenarios having different aquifer heterogeneity characteristics. First, the reference model is presented. The results of the reference model serve as a basis for comparison with all other scenarios. Then, the impacts of the physical (Scenario 2) and chemical (Scenario 3) aquifer heterogeneity factors are discussed with a focus on the nitrate reduction capacity. The final step includes the discussion of the coupled
Discussions
The results of the numerical simulation showed that physical and chemical aquifer heterogeneity significantly influences on the distribution of the dissolved species and therefore the nitrate removal efficiency by changing heterogeneity variance. Besides, the correlation length of the random fields and transverse dispersion are also important factors to delineate the transport and fate of the dissolved species.
The statistical factors determine the strength of heterogeneity in the random fields.
Summary and conclusion
The present study deals with two aquifer characteristics, spatial heterogeneity of hydrological parameters versus spatial heterogeneity of geochemical properties, with a focus on nitrate reduction processes. Since aquifer data are limited, a geostatistical approach is used in this study to generate stochastic realizations of aquifer heterogeneity parameters. This coupled reactive transport code (OGS#IPhreeqc) is applied to pyrite-driven denitrification of nitrate-contaminated groundwater
Acknowledgments
This work has been enabled on the German side by a grant from the Helmholtz Centre for Environmental Research (UFZ) as well as in the Helmholtz research program “Terrestrial Environment (POF-3)” and on the U.S. side, by a grant from the National Science Foundation under grant EAR-1011336, “The Method of Anchored Distributions (MAD): Principles and Implementation as a Community Resources.”. Any opinions, findings and conclusions of recommendations expressed in this material are those of the
References (85)
- et al.
Enhanced biodegradation by hydraulic heterogeneities in petroleum hydrocarbon plumes
J. Contam. Hydrol.
(2009) - et al.
GeoSysBRNS—a flexible multidimensional reactive transport model for simulating biogeochemical subsurface processes
Comput. Geosci.
(2010) - et al.
Modules based on the geochemical model PHREEQC for use in scripting and programming languages
Comput. Geosci.
(2011) An assessment of first-order stochastic dispersion theories in porous media
J. Hydrol.
(1997)- et al.
Modeling nitrogen transport and transformation in aquifers using a particle-tracking approach
Comput. Geosci.
(2014) - et al.
Contaminant degradation in physically and chemically heterogeneous aquifers
J. Contam. Hydrol.
(2007) - et al.
Numerical simulation of isotope fractionation in steady-state bioreactive transport controlled by transverse mixing
J. Contam. Hydrol.
(2012) - et al.
Imaging geochemical heterogeneities using inverse reactive transport modeling: an example relevant for characterizing arsenic mobilization and distribution
Adv. Water Resour.
(2016) - et al.
Health implications of nitrate and nitrite in drinking water: an update on methemoglobinemia occurrence and reproductive and developmental toxicity
Regul. Toxicol. Pharm.
(1996) - et al.
The impact of local diffusion upon mass arrival of a passive solute in transport through three-dimensional highly heterogeneous aquifers
Adv. Water Resour.
(2011)