Stratification of reactivity determines nitrate removal in groundwater.

Biogeochemical reactions occur unevenly in space and time, but this heterogeneity is often simplified as a linear average due to sparse data, especially in subsurface environments where access is limited. For example, little is known about the spatial variability of groundwater denitrification, an important process in removing nitrate originating from agriculture and land use conversion. Information about the rate, arrangement, and extent of denitrification is needed to determine sustainable limits of human activity and to predict recovery time frames. Here, we developed and validated a method for inferring the spatial organization of sequential biogeochemical reactions in an aquifer in France. We applied it to five other aquifers in different geological settings located in the United States and compared results among 44 locations across the six aquifers to assess the generality of reactivity trends. Of the sampling locations, 79% showed pronounced increases of reactivity with depth. This suggests that previous estimates of denitrification have underestimated the capacity of deep aquifers to remove nitrate, while overestimating nitrate removal in shallow flow paths. Oxygen and nitrate reduction likely increases with depth because there is relatively little organic carbon in agricultural soils and because excess nitrate input has depleted solid phase electron donors near the surface. Our findings explain the long-standing conundrum of why apparent reaction rates of oxygen in aquifers are typically smaller than those of nitrate, which is energetically less favorable. This stratified reactivity framework is promising for mapping vertical reactivity trends in aquifers, generating new understanding of subsurface ecosystems and their capacity to remove contaminants.

Deep denitrification: Stream and groundwater biogeochemistry reveal contrasted but connected worlds above and below.

Excess nutrients from agricultural and urban development have created a cascade of ecological crises around the globe. Nutrient pollution has triggered eutrophication in most freshwater and coastal ecosystems, contributing to a loss in biodiversity, harm to human health, and trillions in economic damage every year. Much of the research conducted on nutrient transport and retention has focused on surface environments, which are both easy to access and biologically active. However, surface characteristics of watersheds, such as land use and network configuration, often do not explain the variation in nutrient retention observed in rivers, lakes, and estuaries. Recent research suggests subsurface processes and characteristics may be more important than previously thought in determining watershed-level nutrient fluxes and removal. In a small watershed in western France, we used a multi-tracer approach to compare surface and subsurface nitrate dynamics at commensurate spatiotemporal scales. We combined 3-D hydrological modeling with a rich biogeochemical dataset from 20 wells and 15 stream locations. Water chemistry in the surface and subsurface showed high temporal variability, but groundwater was substantially more spatially variable, attributable to long transport times (10-60 years) and patchy distribution of the iron and sulfur electron donors fueling autotrophic denitrification. Isotopes of nitrate and sulfate revealed fundamentally different processes dominating the surface (heterotrophic denitrification and sulfate reduction) and subsurface (autotrophic denitrification and sulfate production). Agricultural land use was associated with elevated nitrate in surface water, but subsurface nitrate concentration was decoupled from land use. Dissolved silica and sulfate are affordable tracers of residence time and nitrogen removal that are relatively stable in surface and subsurface environments. Together, these findings reveal distinct but adjacent and connected biogeochemical worlds in the surface and subsurface. Characterizing how these worlds are linked and decoupled is critical to meeting water quality targets and addressing water issues in the Anthropocene.

A hillslope-scale aquifer-model to determine past agricultural legacy and future nitrate concentrations in rivers.

The long-term fate of agricultural nitrate depends on rapid subsurface transfer, denitrification and storage in aquifers. Quantifying these processes remains an issue due to time varying subsurface contribution, unknown aquifer storage and heterogeneous denitrification potential. Here, we develop a parsimonious modelling approach that uses long-term discharge and river nitrate concentration time-series combined with groundwater age data determined from chlorofluorocarbons in springs and boreholes. To leverage their informational content, we use a Boussinesq-type equivalent hillslope model to capture the dynamics of aquifer flows and evolving surface and subsurface contribution to rivers. Nitrate transport was modelled with a depth-resolved high-order finite-difference method and denitrification by a first-order law. We applied the method to three heavily nitrate loaded catchments of a crystalline temperate region of France (Brittany). We found that mean water transit time ranged 10-32 years and Damköhler ratio (transit time/denitrification time) ranged 0.12-0.55, leading to limited denitrification in the aquifer (10-20%). The long-term trajectory of nitrate concentration in rivers appears determined by flows stratification in the aquifer. The results suggest that autotrophic denitrification is controlled by the accessibility of reduced minerals which occurs at the base of the aquifer where flows decrease. One interpretation is that denitrification might be an interfacial process in zones that are weathered enough to transmit flows and not too weathered to have remaining accessible reduced minerals. Consequently, denitrification would not be controlled by the total aquifer volume and related mean transit time but by the proximity of the active weathered interface with the water table. This should be confirmed by complementary studies to which the developed methodology might be further deployed.

What do we need to predict groundwater nitrate recovery trajectories?

Nitrate contamination affects many of the Earth's aquifers and surface waters. Large-scale predictions of groundwater nitrate trends normally require the characterization of multiple anthropic and natural factors. To assess different approaches for upscaling estimates of nitrate recovery, we tested the influence of hydrological, historical, and biological factors on predictions of future nitrate concentration in aquifers. We tested the factors with a rich hydrogeological dataset from a fractured bedrock catchment in western France (Brittany). A sensitivity analysis performed on a calibrated model of groundwater flow, denitrification, and nitrogen inputs revealed that trends in nitrate concentration can effectively be approximated with a limited number of key parameters. The total mass of nitrate that entered the aquifer since the beginning of the industrial period needs to be characterized, but the shape of the historical nitrogen input time series can be largely simplified without substantially altering the predictions. Aquifer flow and transport processes can be represented by the mean and standard deviation of the residence time distribution, offering a tractable tool to make reasonable predictions at watershed to regional scales. Apparent sensitivity to denitrification rate was primarily attributable to time lags in oxygen depletion, meaning that denitrification can be simplified to an on/off process, defined only by the time needed for nitrate to reach the hypoxic reactive layer. Obtaining these key parameters at large scales is still challenging with currently available information, but the results are promising regarding our future ability to predict nitrate concentration with integrated monitoring and modeling approaches.

Dating groundwater with dissolved silica and CFC concentrations in crystalline aquifers.

Estimating intermediate water residence times (a few years to a century) in shallow aquifers is critical to quantifying groundwater vulnerability to nutrient loading and estimating realistic recovery timelines. While intermediate groundwater residence times are currently determined with atmospheric tracers such as chlorofluorocarbons (CFCs), these analyses are costly and would benefit from other tracer approaches to compensate for the decreasing resolution of CFC methods in the 5-20 years range. In this context, we developed a framework to assess the capacity of dissolved silica (DSi) to inform residence times in shallow aquifers. We calibrated silicate weathering rates with CFCs from multiple wells in five crystalline aquifers in Brittany and in the Vosges Mountains (France). DSi and CFCs were complementary in determining apparent weathering reactions and residence time distributions (RTDs) in shallow aquifers. Silicate weathering rates were surprisingly similar among Brittany aquifers, varying from 0.20 to 0.23 mg L-1 yr-1 with a coefficient of variation of 7%, except for the aquifer where significant groundwater abstraction occurred, where we observed a weathering rate of 0.31 mg L-1 yr-1. The silicate weathering rate was lower for the aquifer in the Vosges Mountains (0.12 mg L-1 yr-1), potentially due to differences in climate and anthropogenic solute loading. Overall, these optimized silicate weathering rates are consistent with previously published studies with similar apparent ages range. The consistency in silicate weathering rates suggests that DSi could be a robust and cheap proxy of mean residence times for recent groundwater (5-100 years) at the regional scale. This methodology could allow quantification of seasonal groundwater contributions to streams, estimation of residence times in the unsaturated zone and improve assessment of aquifer vulnerability to anthropogenic pollution.

Autotrophic denitrification supported by biotite dissolution in crystalline aquifers: (2) transient mixing and denitrification dynamic during long-term pumping.

We investigated the mixing and dynamic of denitrification processes induced by long-term pumping in the crystalline aquifer of Ploemeur (Brittany, France). Hydrological and geochemical parameters have been continuously recorded over 15 boreholes in 5km2 on a 25-year period. This extensive spatial and temporal monitoring of conservative as well as reactive compounds is a key opportunity to identify aquifer-scale transport and reactive processes in crystalline aquifers. Time series analysis of the conservative elements recorded at the pumped well were used to determine mixing fractions from different compartments of the aquifer on the basis of a Principal Component Analysis approach coupled with an end-member mixing analysis. We could reveal that pumping thus induces a thorough reorganization of fluxes known as capture, favoring infiltration and vertical fluxes in the recharge zone, and upwelling of deep and distant water at long-term time scales. These mixing fractions were then used to quantify the extent of denitrification linked to pumping. Based on the results from batch experiments described in a companion paper, our computations revealed that i) autotrophic denitrification processes are dominant in this context where carbon sources are limited, that ii) nitrate reduction does not only come from the oxidation of pyrite as classically described in previous studies analyzing denitrification processes in similar contexts, and that iii) biotite plays a critical role in sustaining the nitrate reduction process. Both nitrate reduction, sulfate production as well as fluor release ratios support the hypothesis that biotite plays a key role of electron donor in this context. The batch-to-site similarities support biotite availability and the role by bacterial communities as key controls of nitrate removal in such crystalline aquifers. However, the long term data monitoring also indicates that mixing and reactive processes evolve extremely slowly at the scale of the decade.

Simulation and analysis of solute transport in 2D fracture/pipe networks: the SOLFRAC program.

The Time Domain Random Walk (TDRW) method has been recently developed by Delay and Bodin [Delay, F. and Bodin, J., 2001. Time domain random walk method to simulate transport by advection-dispersion and matrix diffusion in fracture networks. Geophys. Res. Lett., 28(21): 4051-4054.] and Bodin et al. [Bodin, J., Porel, G. and Delay, F., 2003c. Simulation of solute transport in discrete fracture networks using the time domain random walk method. Earth Planet. Sci. Lett., 6566: 1-8.] for simulating solute transport in discrete fracture networks. It is assumed that the fracture network can reasonably be represented by a network of interconnected one-dimensional pipes (i.e. flow channels). Processes accounted for are: (1) advection and hydrodynamic dispersion in the channels, (2) matrix diffusion, (3) diffusion into stagnant zones within the fracture planes, (4) sorption reactions onto the fracture walls and in the matrix, (5) linear decay, and (6) mass sharing at fracture intersections. The TDRW method is handy and very efficient in terms of computation costs since it allows for the one-step calculation of the particle residence time in each bond of the network. This method has been programmed in C++, and efforts have been made to develop an efficient and user-friendly software, called SOLFRAC. This program is freely downloadable at the URL (labo.univ-poitiers.fr/hydrasa/intranet/telechargement.htm). It calculates solute transport into 2D pipe networks, while considering different types of injections and different concepts of local dispersion within each flow channel. Post-simulation analyses are also available, such as the mean velocity or the macroscopic dispersion at the scale of the entire network. The program may be used to evaluate how a given transport mechanism influences the macroscopic transport behaviour of fracture networks. It may also be used, as is the case, e.g., with analytical solutions, to interpret laboratory or field tracer test experiments performed in single fractures.

General database for ground water site information.

In most cases, analysis and modeling of flow and transport dynamics in ground water systems require long-term, high-quality, and multisource data sets. This paper discusses the structure of a multisite database (the H+ database) developed within the scope of the ERO program (French Environmental Research Observatory, http://www.ore.fr). The database provides an interface between field experimentalists and modelers, which can be used on a daily basis. The database structure enables the storage of a large number of data and data types collected from a given site or multiple-site network. The database is well suited to the integration, backup, and retrieval of data for flow and transport modeling in heterogeneous aquifers. It relies on the definition of standards and uses a templated structure, such that any type of geolocalized data obtained from wells, hydrological stations, and meteorological stations can be handled. New types of platforms other than wells, hydrological stations, and meteorological stations, and new types of experiments and/or parameters could easily be added without modifying the database structure. Thus, we propose that the database structure could be used as a template for designing databases for complex sites. An example application is the H+ database, which gathers data collected from a network of hydrogeological sites associated with the French Environmental Research Observatory.

Anomalous diffusion exponents in continuous two-dimensional multifractal media.

We study diffusion in heterogeneous multifractal continuous media that are characterized by the second-order dimension of the multifractal spectrum D2, while the fractal dimension of order 0, D0, is equal to the embedding Euclidean dimension 2. We find that the mean anomalous and fracton dimensions, d(w) and d(s), are equal to those of homogeneous media showing that, on average, the key parameter is the fractal dimension of order 0 D0, equal to the Euclidean dimension and not to the correlation dimension D2. Beyond their average, the anomalous diffusion and fracton exponents, d(w) and d(s), are highly variable and consistently range in the interval [1,4]. d(w) can be consistently either larger or lower than 2, indicating possible subdiffusive and superdiffusive regimes. On a realization basis, we show that the exponent variability is related to the local conductivity at the medium inlet through the conductivity scaling.

Persistence of incomplete mixing: a key to anomalous transport.

Anomalous dispersion in heterogeneous environments describes the anomalous growth of the macroscopic characteristic sizes of scalar fields. Here we show that this phenomenon is closely related to the persistence of local scale incomplete mixing. We introduce the mixing scale ε as the length for which the scalar distribution is locally uniform. We quantify its temporal evolution due to the competition of shear action and diffusion and compare it to the evolution of the global dispersion scale σ. In highly heterogeneous flow fields, for which the temporal evolution of σ is superdiffusive, we find that ε evolves subdiffusively. The anomalous evolutions of the dispersion and mixing scales are complementary, εσ ∝ t. This result relates anomalous global dispersion to the dynamics of local mixing.