Understanding process dynamics at aquifer-surface water interfaces: An introduction to the special section on new modeling approaches and novel experimental technologies

Authors


Abstract

[1] This paper introduces the special section on “new modeling approaches and novel experimental technologies for improved understanding of process dynamics at aquifer-surface water interfaces.” It is contextualizing the framework for the 27 research papers of the special section by firth identifying research gaps and imminent challenges for ecohydrological research at aquifer-surface water interfaces and then discussing the specific paper contributions on (i) new developments in temperature/heat tracing at GW-SW interfaces, (ii) new methods to capture the temporal and spatial variability of groundwater—surface water exchange, (iii) new approaches in modeling aquifer-river exchange flow, and (iv) new concepts and advanced theory of groundwater—surface water exchange.

1. Imminent Challenges for Ecohydrological Research at Aquifer-Surface Water Interfaces

[2] Investigations of groundwater-surface water (GW-SW) interactions have become increasingly interdisciplinary as evidenced by a growing number of interdisciplinary case studies as well as review papers [Winter, 1995; Boulton et al., 1998; Woessner, 2000; Sophocleous, 2002; Boulton, 2007; Krause et al., 2011a; Robertson and Wood, 2010] and special issues [Krause et al., 2009, 2011b; Fleckenstein et al., 2010; Robertson and Wood, 2010]. Although aquifer-river interfaces are probably the most intensively investigated interfaces, there is also a growing body of research focusing on the analysis of lacustrine groundwater discharge (LGD) and its role for catchment-wide biogeochemical budgets [Cherkauer et al., 1992; La Baugh et al., 1995; Buso et al., 2009; Neumann et al., 2013].

[3] While interdisciplinary research of recent years has significantly advanced the mechanistic process understanding of physical controls on GW-SW exchange flow [Thibodeaux and Boyle, 1987; Cuthbert et al., 2010; Krause et al., 2012a] and their impacts on biochemical cycling [Lewandowski and Nützmann, 2010; Zarnetske et al., 2011a, 2011b; Krause et al., 2009, 2013; Bardini et al., 2012] and on the ecological functioning of GW-SW water interfaces [Brunke and Gonser, 1997; Boulton et al., 1998; Acuna et al., 2008; Malcolm et al., 2005], the spatial patterns and temporal dynamics of these interactions are still often unknown. The introduction to this special section hence summarizes research questions arising from this lack of knowledge and current studies addressing them.

[4] A fundamental open question is how exchange between GW and SW is organized over both space and time. A large number of studies highlighted that the significant heterogeneity of exchange fluxes is characterized by complex spatial patterns [Conant, 2004; Fleckenstein et al., 2006; Kennedy et al., 2009a, 2009b; Lewandowski et al., 2013; Angermann et al., 2012a, 2012b; Krause et al., 2012a; Stonedahl et al., 2012] as well as intense temporal dynamics [Rosenberry and Morin, 2004; Malcolm et al., 2006; Kennedy et al., 2009b]. However, the spatiotemporal interactions as well as scales of different drivers and controls of exchange fluxes (e.g., surface (bedform) driven advective forcing versus subsurface controlled hydrostatic forcing of upwelling GW) are still debated controversially [Endreny and Lautz, 2012; Krause et al., 2012b; Bardini et al., 2013].

[5] Several studies identified flow path-dependent residence time distributions that impact on hyporheic, benthic, or riparian chemical processing rates [Lohse et al., 2009; Pinay et al., 2009; Boano et al., 2010; Zarnetske et al., 2011a, 2012; Bardini et al., 2012]. In addition, the chemical milieu of the environment, its temperature and access to reaction partners have been shown to directly control biochemical turnover [Birgand et al., 2007; Acuna et al., 2008; Zarnetske et al., 2011b; Krause et al., 2013]. Although the understanding of such drivers and controls substantially improved within the last decade, there is still a somewhat limited comprehension of their interactions as well as organizational principles in space and time [Fleckenstein et al., 2010; Krause et al., 2011a,b]. Consequently, despite an increasing recognition of distinctive threshold behavior and hotspot functioning at GW-SW interfaces, the scale dependence and large-scale relevance of enhanced biogeochemical activity in small-scale hotspots [McClain et al., 2003; Lautz and Fanelli, 2008; Bardini et al., 2013; Krause et al., 2013] is still poorly understood. With new emerging pollutants such as pharmaceuticals in rivers and groundwater, there is a growing need to better understand the processes that control their fate [Kidmose et al., 2010; Lewandowski et al., 2011]. The effective assessment of the functioning of GW-SW interfaces under the influence of environmental change requires further improvement of current monitoring and modeling practices, including the adequate representation of dynamic system behavior across multiple scales [Fleckenstein et al., 2010]. The papers in this special section address these challenges and reflect the breadth and vibrancy of current research at GW-SW interfaces.

2. Novel Concepts Based on Innovative Experimental and Modeling Approaches

[6] The advancement of interdisciplinary process understanding of GW-SW interactions across multiple spatiotemporal scales requires (i) novel sets of tools and technological approaches but also (ii) improved conceptual models. This special section provides an overview of innovations and progress made in conceptualizing physical controls on GW-SW exchange, heat transport, biochemical cycling, and ecological functioning. The 27 papers combined in this section reshape the thought of multiscale interactions between GW and SW by presenting novel developments of active and passive tracing techniques, distributed sensor networks and modeling approaches.

2.1. New Developments in Temperature/Heat Tracing at GW-SW Interfaces

[7] Detailed identification of exchange flow patterns across GW-SW interfaces relies on observation techniques with high spatial and temporal resolution. The application of distributed sensor networks as for instance fiber-optic distributed temperature sensing (FO-DTS) has seen rapidly growing number of applications [see Krause and Blume, 2013; Sebok et al., 2013; Blume et al., 2013]. Besides FO-DTS applications with a focus on aquifer-river interfaces [Krause and Blume, 2013], the technology finds increasing applications at aquifer-lake interfaces, highlighting small-scale heterogeneity and temporal variability in GW-SW exchange [Sebok et al., 2013] as well as promising up-scaling potential [Blume et al., 2013]. In addition, there is an increased focus on the capacities and actual limitations of this technology [Rose et al., 2013] and new algorithms for data analysis and calibration are suggested [Hausner et al., 2011; Van de Giessen et al., 2012; Krause and Blume, 2013].

[8] Sebok et al. [2013] apply FO-DTS to investigate seasonal variability in LGD patterns to a Danish lake. Based on the analysis of daily minimum temperature, diel amplitude, and daily standard deviation of temperature, they are able to identify a discrete discharge zone in the nearshore area which was validated by vertical temperature profiling at the top 50 cm of the lakebed and seepage meters. The variability in extent and location of the FO-DTS monitored temperature signal indicates that the groundwater discharge zone varies in time and confirmes that its movement can be adequately traced by FO-DTS.

[9] In a further application of FO-DTS in a groundwater-fed lake system, Blume et al. [2013] present an approach for up-scaling LGD derived based on single transects of either sediment temperature profiles or vertical hydraulic gradients and transfer functions. The transfer functions are tested for their ability to adequately identify 2-D LGD patterns, concluding that both FO-DTS based up-scaling approaches are able to reproduce distinct small-scale heterogeneities in LGD, providing proof of concept for the potential of this methodology and encouraging its further application at larger scales.

[10] In order to assess the limitations and capabilities of FO-DTS surveys, Rose et al. [2013] provide a quantitative analysis of the accuracy in detecting actual signal location and size (spatial extent) and signal intensity (temperature anomaly) of known temperature signals. In controlled lab experiments, they prove that accuracy in detection of signal strength critically declines with decreasing signal size, in particular for signal sizes close to the spatial sampling interval. Their investigations highlight that signal size dependent decreases in detection accuracy may result in potential ambiguity of interpretations of signal size and intensity.

[11] In a study analyzing uncertainties and limitations of different FO-DTS monitoring modes, Krause and Blume [2013] investigate the impacts of (i) seasonal variability in signal strength (temperature difference) and (ii) monitoring modes on the accuracy of FO-DTS surveys. They show that monitoring mode has significant impact on survey accuracy, with the newly proposed two-way single-ended averaging mode proving to be more accurate than single-ended or double-ended surveys. The new monitoring mode will improve in particular the detection of complex and small-scale temperature patterns.

[12] In addition to FO-DTS technologies, other novel heat tracing techniques have been developed and are applied for quantification of exchange fluxes across GW-SW interfaces.

[13] In order to quantify the impact of small-scale variability in streambed hydraulic conductivity on hyporheic exchange at a lowland river, Angermann et al. [2013] combine surveys with a novel heat pulse sensor with the mapping of vertical hydraulic gradients. They show that bedform-induced forcing is super-imposed by spatially variable groundwater upwelling, highlighting (i) the importance of small-scale streambed variability and (ii) the impact of hydrostatic forcing in groundwater upwelling zones that can exceed bedform-induced hydrodynamic forcing.

[14] A somewhat different picture is presented by Gariglio et al. [2013] in their field study of a pool-riffle-pool sequence in a fourth-order headwater stream. They employ temperature series analysis to determine vertical exchange fluxes of water through the streambed surface over 9 months. By using different steady and unsteady approaches to analyze the collected temperature data, they propose a conceptual model of the seasonal dynamics of water exchange in the system, which is characterized by the presence of shallow topography-induced hyporheic flow cells that are altered by the time-varying rate of water upwelling or downwelling depending on seasonal snowmelt and precipitations.

[15] A comparison among the suitability of different tracers to quantify water fluxes is discussed by Engelhardt et al. [2013]. They consider artificial sweeteners released by wastewater treatment plants for their role as emerging microcontaminants but also as potential tracers, and they focus in particular on acesulfame which is expected to exhibit a conservative behavior due to its low potential for sorption and decay. By comparing field observations of hydraulic heads, water temperature, and acesulfame concentrations in subsurface water they conclude that the use of acesulfame as a tracer is hindered by its high background concentrations in groundwater, and that additional (e.g., thermal) observations are needed to correctly investigate the transport of such microcontaminants.

[16] A growing body of literature demonstrates how analytical solutions to 1-D flow and heat transport equations [Hatch et al., 2006; Keery et al., 2007; Schornberg et al., 2010] can give robust estimates of vertical streambed flux, assuming flows are sufficiently uniform [Cuthbert and Mackay, 2013]. Luce et al. [2013] extend existing theory in this area and formulate new explicit analytical solutions for estimating both streambed flux and thermal diffusivity as well as new ways of characterizing the uncertainty in such estimates. Their paper demonstrates how the amount of information derived from temperature time series can be increased compared with existing methods.

2.2. New Methods to Capture the Temporal and Spatial Variability of Groundwater—Surface Water Exchange

[17] The assessment of the ecohydrological functioning and biogeochemical turnover at GW-SW interfaces requires detailed understanding of the temporal dynamics and spatial patterns of GW-SW exchange fluxes as well as residence time distributions [Conant, 2004; Zarnetske et al., 2011a; Stonedahl et al., 2012; Krause et al., 2013]. The analytical challenges associated with the quantification of exchange fluxes and residence times across a wide range of spatiotemporal scales [White, 1993; Hayashi and Rosenberry, 2002; Fleckenstein et al., 2010; Krause et al., 2011a, 2011b] triggered the development of new methodological approaches and innovative combinations of new tracer technologies.

[18] Based on concurrent conservative and reactive stream tracer tests Lemke et al. [2013] propose a methodology that allows differentiating between in-stream and hyporheic transient storage. High-frequency concentrations of conservative (fluorescein) and reactive tracers (resazurin, resorufin) are obtained by online field fluorometry. An advective-dispersive solute transport model is fitted to the tracer breakthrough curves (BTC). The joint analysis of a conservative and a reactive tracer reveals that hyporheic exchange strength is in fact higher and in-stream mixing and dispersion smaller than what would be inferred from using the BTC of the conservative tracer alone. The authors conclude that the joint use of both tracers improves the physical basis of the model.

[19] The impact of groundwater upwelling on potential suppression of hyporheic exchange fluxes and maximum streambed penetration depth of downwelling surface water is the subject of a number of experimental investigations.

[20] Binley et al. [2013] quantify vertical and horizontal water fluxes of a 200 m reach by coupling Darcian flow estimates with in-stream piezometer tracer dilution tests. Their results indicate varying contributions of flow components along the reach with active suppression of hyporheic exchange flow at locations with localized connectivity to the regional groundwater and rather horizontal hyporheic flow where there is reduced groundwater upwelling. Although some variability in streambed hydraulic conductivity is observed at their study site, they conclude that the localized variation in hydraulic conductivity alone cannot account for the spatial variability in flow pathways.

[21] Liao et al. [2013] present and discuss an advanced methodology for the identification of residence time distributions from observed stream concentrations of reactive tracers (resazurin and resorufin). The inversion method does not require assumptions on the shape of the residence time distribution, and it also considers rates of tracer decay reactions and the possibility of equilibrium and kinetic sorption to sediments. An application to a stream tracer experiment shows that the method is able to identify the presence of distinct characteristic temporal scales in the residence time distribution, which can be related to morphological features of different size.

[22] Langston et al. [2013] quantify GW-SW interactions in a proglacial moraine lake in a partially glaciated watershed in the Canadian Rockies based on the joint application of heat and solute tracers. They analyze lake water table fluctuation for interpreting groundwater recharge dynamics and provide a detailed water balance of the groundwater inflow and outflow dominated lake. The mass and energy balance of the lake is used to estimate the large-scale hydraulic conductivity of the proglacial moraine in which the lake is embedded and thus advance alpine groundwater hydrology.

[23] Rosenberry et al. [2013] identify substantial temporal variability of seepage at the sediment-water interface of several lakes, a large river, and an estuary based on high-frequency (resolution on the scale of minutes) direct measurement with an electromagnetic seepage meter. Rain events and water withdrawal caused by evapotranspiration results in seepage changes in the order of 10%. Variability of seepage increases orders of magnitude in response to tides, seiches, and waves. These examples reveal that temporal variability in seepage in response to common hydrological events is much larger than previously realized. Likely implications include effects on water residence time, geochemical transformations, and ecological conditions at and near the sediment-water interface.

2.3. New Approaches in Modeling Aquifer-River Exchange Flow

[24] Recent advances in numerical modeling do not only greatly improve the ability to quantify exchange fluxes and reactive transport of solutes across GW-SW interfaces but also provide important insights into drivers and controls of spatial patterns and temporal dynamics in exchange fluxes, residence time distributions, and chemical turnover.

[25] As Naranjo et al. [2013] identify, based on the application of a 2-D longitudinal flow and solute transport model, heterogeneity and anisotropy in streambed physical conditions can have a strong influence on the spatial distribution of mean residence times in the hyporheic zone. This result contradicts recent findings of other authors [e.g., Bardini et al., 2013] and is expected to have significant implications for biogeochemical cycling in the hyporheic zone as turnover rates are highly dependent on the residence time of solutes in the hyporheic zone.

[26] In their modeling study, Stonedahl et al. [2013] investigate the contribution of different geomorphological features to hyporheic exchange pattern by considering a synthetic streambed topography composed of meanders, bars, and dunes that is based on established geomorphological spectra of meander and bedform sizes. Their results show that the overall exchange flux is similar, but not exactly equal, to the sum of the exchange contributions that would be separately driven by each geomorphological feature, suggesting that the evaluation of up-scaled values of exchange flux at the reach scale is not critically affected by the nonlinear processes that govern subsurface water flow.

[27] The temporal variability of hyporheic exchange due to streamflow dynamics is the subject of a study by Boano et al. [2013]. They develop a mathematical model for hyporheic exchange induced by stream dunes that includes changes in bedform size and bedform migration as well as the feedback of streambed morphology on surface flow depth. The modeling results show that (i) flood events enhance the penetration of surface water deeper and for longer times in the sediments, and that (ii) time-averaged values of exchange flux and penetration depth of surface water in streambed sediments are very close to those calculated using steady state models that only consider mean discharge, while residence times are slightly overestimated if the temporal variability of stream discharge is neglected.

[28] Marzadri et al. [2013] analyze how heat transport along hyporheic flow paths control the thermal regime of streambed sediments determined by daily fluctuations of stream water temperature. The work builds on a previous model of water flow induced by pool-riffle morphologies, and develops semianalytical solutions for water temperature in sediments as a function of water residence time along flow paths. The study results evidence that streambed temperatures are highly influenced by stream size and morphology, with larger lowland streams being characterized by more stable sediment temperatures compared to smaller headwater streams.

[29] Trauth et al. [2013] investigate the effects of different stream flow and ambient groundwater flow conditions on 3-D hyporheic exchange patterns and residence times in streams with pool-riffle morphology. Their simulations highlight the importance of ambient groundwater flow for 3-D hyporheic exchange in complex stream channels. Developing undular hydraulic jumps are shown to create pressure anomalies at the streambed surface affecting hyporheic flow patterns for specific stream flow and bedform configurations. Hyporheic residence times follow a log-normal distribution with median residence times in the order of several hours. Simulation results suggest a threshold beyond which hyporheic exchange becomes insensitive to further increases in bedform height.

[30] Based on hydraulic head and temperature data as well as results from a 2-D coupled flow and heat transport model Constantz et al. [2013] evaluate the characteristic spatial and temporal patterns of stream-aquifer exchange along tributary streams. They identify distinct exchange patterns for upland (close to headwaters), central (alluvial midsections), and distal (in proximity to the main stem) reaches of a tributary stream in Wyoming. Their work provides a general framework to characterize stream-aquifer exchange patterns along tributary streams.

2.4. New Concepts and Advanced Theory of Groundwater—Surface Water Exchange

[31] Technological development and application of innovative methodological approaches have not only lead to more accurate descriptions of process dynamics at GW-SW interfaces at multiple scales but continue to improve interdisciplinary process understanding, challenging previous concepts, and causing paradigm shifts.

[32] Since parameters of commonly used solute transport models cannot be obtained uniquely from physical attributes both scaling and predicting solute transport processes are highly uncertain. By analyzing 98 tracer experiments reported in the literature González-Pinzón et al. [2013] identify inconsistencies of the classic solute transport theory. They use temporal moment analysis, a physically based method of model reduction, and find the coefficient of skewness of breakthrough curves is scale variant and equal to approximately 1.18 while previous models assumed it to decrease over time.

[33] In a combined application of tracer injections and solute transport modeling, Cozzetto et al. [2013] investigate the influence of preferential flow paths and stream thermal regimes on hyporheic exchange fluxes in an Antarctic glacial meltwater stream. Based on results of tracer experiments the authors propose a conceptual model of the investigated reach with discrete preferential flow paths that intersect hydraulically isolated areas and which are affected by streambed heterogeneity and temperature related viscosity.

[34] Transient storage measurements made using injected tracers are most often reported for low-flow periods. In contrast, Ward et al. [2013a] carry out multiple injections through a storm event, revealing the complexity of transient storage interpretations due to dynamic in-stream and near-stream processes. The results highlight the importance of stream morphology in controlling the transport of solute tracers and provide new insights into time variant 3-D GW-SW exchanges at a range of temporal and spatial scales.

[35] Harvey et al. [2012] demonstrate the significance of the hyporheic component of denitrification to whole-stream N budgets based on tracer tests with 15NO3. They find that the active subsurface volume for denitrification differs from the full hyporheic zone, which is one reason why previous investigations could not find significant relationship between stream denitrification and metrics such as hyporheic-zone size or residence time.

[36] Wetlands systems located at GW-SW interfaces can be important environments for biogeochemical cycling. Durst et al. [2013] describe a series of lab-scale experiments to further understand the processes controlling transport and fate of pesticides in such environments. As well as revealing the fundamental transport properties of some common pesticides, the results show the relative merits of various artificial tracers as proxies and the importance of preferential flow of pesticides through root holes in vegetated wetlands.

[37] McCallum et al. [] combine temperature derived riverbed Darcy fluxes and reach losses from differential river gauging to investigate the river-aquifer interactions in a semi-arid environment under highly transient river flow conditions. Significantly, it is shown that high-flow events account for 64% of the reach loss (or 43% if overbank events are excluded) despite occurring only 11% of the time. The results indicate that multiple factors control the exchange flows and highlight the difficulties in up-scaling estimates from point to reach scale.

[38] Ward et al. [2013b] compare a transient storage model (TSM) to the channel mass balance based on field investigations along a series of stream reaches. Their results support the proposed decrease of gross water loss from the channel with increasing discharge. A clear relationship between discharge and transient storage, however, could not be identified, likely due to uncertainty and equifinality in the TSM parameters. The authors suggest that the physical interpretation of either method (water balance and TSM) remains difficult because of shifts in advective transport times, suggesting the conjunctive use of both methods for description of water exchange and solute transport along streams.

[39] Based on a combination of flume experiments and computational fluid dynamic (CFD) simulations, Zhou and Endreny [2013] find that in-stream restoration structures alter both surface water hydraulics and subsurface hyporheic flow. They indicate that restoration structures create backwater effects which increase upwelling and downwelling forces, relatively high vertical fluxes across the sediment-water interface and decrease both the depth of hyporheic flow and reverse subsurface flow.

3. Outlook

[40] The combination of new experimental technologies (including novel distributed sensor networks and advanced tracing techniques) and innovative modeling approaches (with advanced representation of governing parameter distributions and temporal dynamics) continue to improve mechanistic understanding of process dynamics at GW-SW interfaces, creating new concepts and advanced theory that challenge current paradigms.

[41] Although these technological advancements improve the potential to account for scale dependencies in process dynamics, the scale transfer of process understanding, including up-scaling and down-scaling remains a largely unresolved challenge. In this respect, significant potential for enhancing process representations across multiple scales may exist in the coupling of technological and methodological advancements that support the integrated assessment of solute transport and transformation along a stream with the currently available improved local methods. The progress presented in this special section encourages for further development of combinations of data-driven top-down and process-based bottom-up approaches.

Ancillary