Excess air formation as a mechanism for delivering oxygen to groundwater

Authors

  • L. Mächler,

    Corresponding author
    1. Eawag, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
    2. Institute of Biogeochemistry and Pollutant Dynamics, ETH Zürich, Switzerland
    • Corresponding author: L. Mächler, Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-8600 Dübendorf, Switzerland. (lars.qmaechler@eawag.ch)

    Search for more papers by this author
  • S. Peter,

    1. Eawag, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
    2. Institute of Biogeochemistry and Pollutant Dynamics, ETH Zürich, Switzerland
    Search for more papers by this author
  • M. S. Brennwald,

    1. Eawag, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
    Search for more papers by this author
  • R. Kipfer

    1. Eawag, Swiss Federal Institute of Aquatic Science and Technology, Dübendorf, Switzerland
    2. Institute of Biogeochemistry and Pollutant Dynamics, ETH Zürich, Switzerland
    3. Institute of Geochemistry and Petrology, ETH Zürich, Switzerland
    Search for more papers by this author

Abstract

[1] The temporal dynamics and spatial distribution of the concentrations of dissolved gases (He, Ar, Kr, N2, O2, and CO2) in an infiltrating groundwater system fed by the peri-alpine river Thur (Switzerland) were analyzed before, during and after a single, well-defined flood event. The analysis was based on measurements taken in five different groundwater observation wells that were located approximately 10 m apart and tapped the same groundwater body, but were situated in three different riparian zones. The input of O2 into the groundwater as a result of the formation of excess air was found to be of the same order of magnitude as that resulting from the advection of river water, although the amount of excess air formed and the amount of O2 delivered varied significantly among the riparian zones. The results suggest that the input of O2 into groundwater as a result of excess air formation is controlled not only by the hydraulic conditions prevailing in the river and the groundwater, but also by the thickness of the confining bed at the top of the aquifer. The sandy gravel aquifer itself is too coarse to trap a significant amount of air during the water level rise. The clay layer confining the aquifer, however, acts as a barrier hindering the escape of air from the subsoil to the surface, and hence is likely to be a key factor controlling the trapping and dissolution of air in groundwater.

1. Introduction

[2] A significant part of Switzerland's drinking water (over 30% [Hartmann and Michel, 1992; SVGW, 2002]) is produced from riparian aquifers. Rivers which recharge their adjacent aquifers are highly dynamic and are characterized by high variability with regard to discharge rates and the chemical composition of the solutes, e.g., dissolved oxygen [Uehlinger and Naegeli, 1998; Kirchner et al., 2000; Uehlinger, 2006].

[3] In groundwater, dissolved oxygen (O2), carbon dioxide (CO2), and nitrogen (N2) are essential reactants and products of biogeochemical processes, in particular of microbiological activity, which remove nutrients and pollutants and are therefore an important control of water quality [Whitman and Clark, 1982; Korom, 1992; Brunke and Gonser, 1997; Boulton et al., 1998; Malard and Hervant, 1999; Hayashi and Rosenberry, 2002; Rivett et al., 2008]. In particular, CO2 is produced in both aerobic respiration—the consumption of O2—and during denitrification, whereby in case of complete denitrification also N2 is produced. However, CO2 concentrations are not a direct quantitative measure for bacterial activity, as part of the produced CO2 is often transformed into calcium bicarbonate [see e.g., Hynes, 2001].

[4] During floods, biogeochemical changes in the river are transmitted to the groundwater. Furthermore, the varying hydraulic conditions can activate interaction of the groundwater with the soil, e.g., input of organic carbon [Baker et al., 2000; Peter et al., 2012a], and trigger gas input from the soil air into the groundwater [Vogel et al., 1981; Stute et al., 1992; Aeschbach-Hertig et al., 1999; Williams and Oostrom, 2000; Kipfer et al., 2002; Klump et al., 2007].

[5] Therefore, to quantitatively and qualitatively interpret subsurface biogeochemical processes, not only the concentrations of N2, O2, and CO2, need to be measured, but it is also necessary to determine the total amount of each gas injected per water mass (effective gas input). Gas transport by infiltrating river water into the aquifer and gas exchange through the unsaturated zone within the aquifer contribute to the effective gas input. Both contributions are often unknown [Vogel et al., 1981; Stute et al., 1992; Kipfer et al., 2002; Lippmann et al., 2003; Holocher et al., 2003; Klump et al., 2008]. In particular, the initial concentration cannot usually be determined by assuming atmospheric solubility equilibrium, as the equilibrium concentration is controlled mainly by local soil temperature and not necessarily by the temperature prevailing during the last occurrence of gas partitioning [Mazor, 1972; Vogel et al., 1981; Heaton and Vogel, 1981; Beyerle et al., 1999].

[6] Gas partitioning in an aquifer by the entrapment and dissolution of air bubbles, in response to water-table fluctuations or groundwater recharge, leads to the formation of excess air, i.e., a surplus of atmospheric gases relative to atmospheric solubility equilibrium [Heaton and Vogel, 1981; Kipfer et al., 2002; Lippmann et al., 2003; Klump et al., 2007]. Excess air formation therefore contributes substantially to the effective input of O2 and N2 into groundwater [Rose and Long, 1988; Beyerle et al., 1999; Williams and Oostrom, 2000; Massmann and Sueltenfuss, 2008]. In conclusion, initial O2 availability in groundwater may be much larger than expected based on atmospheric equilibrium considerations alone [Beyerle et al., 1999].

[7] Atmospheric noble gases (He, Ne, Ar, Kr, and Xe) are biogeochemically inert and are therefore excellent tracers to analyze gas exchange in porous media, such as excess air formation [e.g., Kipfer et al., 2002]. In particular, measured noble gas concentrations enable the amount of injected excess air to be reconstructed [Beyerle et al., 1999; Aeschbach-Hertig et al., 1999, 2000; Kipfer et al., 2002; Ingram et al., 2007; Klump et al., 2007, 2008], and hence allow the effective O2 and N2 input into groundwater to be estimated.

[8] Both in laboratory experiments [Williams and Oostrom, 2000] and in few field studies, excess air formation was documented to influence O2 availability in groundwater [Beyerle et al., 1999; Massmann and Sueltenfuss, 2008]. However, no field study up to our knowledge investigated in detail O2 injection into riparian groundwater due to the hydraulic fluctuations of the river. Therefore, in this work, we studied excess air formation as a mechanism for the delivery of O2 to groundwater in different riparian zones near an infiltrating peri-alpine-river responding to a flood incident. Furthermore, in order to quantify possible biochemical production of N2 due to denitrification processes, we determined the injection of atmospheric N2 into the groundwater.

[9] Our study was conducted during a small flood event of an alpine river (Thur, Switzerland), and included the quantification of different dissolved gas concentrations (N2, O2, CO2, He, Ar, and Kr) in several groundwater wells. The wells were located, between 1 and 50 m away from the river, covering three different functional process zones, i.e., individual types of hydrogeomorphic patches that differ in their physiochemical characteristics, size, shape, and internal structural complexity, and therefore fulfill different ecological functions [Thorp et al., 2006].

2. Methods

2.1. Study Site

[10] The river Thur is located in the northeast of Switzerland. The catchment area of the river Thur includes an elevation of over 2500 m asl and the rivers estuary leads into the river Rhine at 345 m asl. The high elevation difference and the fact that there is no lake along its course, makes the river Thur a very dynamic river. In particular, precipitation events in the catchment trigger immediate increases in water discharge, which often leads to floods [BAFU, 2010; Vogt et al., 2010]. Therefore, most of the river was channelized for flood protection in the late 19th/early 20th century [Binderheim-Banakay et al., 2000; BAFU, 2010]). In recent years, the overbanks were removed over a length of approximate 2.5 km and the riverbed was widened to allow small meanders to develop in order to revitalize the river [BAFU, 2010; Schneider et al., 2011].

[11] The underlying Thur-Valley aquifer spreads over the entire valley and constantly exchanges water with the river. At our study site (Niederneunforn, Canton Thurgovia) this riparian aquifer is approximately 6 m thick (see Figure 1). It consists of sandy gravel with impervious clay below and with a 0.5–2 m thick alluvial loam layer forming the aquifer-top. Depending on the water table and the thickness of the top layer, the groundwater at our study site exhibits unconfined and confined conditions [Vogt et al., 2010]. The studied observation wells are <50 m away from the river and were set in three different riparian zones (see Figure 2): gravel (well GB1), grass (well G1 and G2), and willow (well W1 and W2). The groundwater residence time in the different wells ranges from <1 day in the gravel zone to more than 4 days in the two wells in the willow zone [Vogt et al., 2010].

Figure 1.

Schematic view of the transversal section of the aquifer. Aspect ratios of the different objects are not to scale.

Figure 2.

Schematic map of the field site (Niederneunforn at river Thur, Switzerland) during low flow. During high flow the gravel banks were flooded. The groundwater wells in the different functional process zones are also shown. Groundwater flow direction is taken from Vogt et al. [2010].

2.2. Measurements

[12] The Gas Equilibrium Membrane Inlet Mass Spectrometer (GE-MIMS) system used for the in situ determination of the gas concentrations in groundwater is described in detail in Mächler et al. [2012]. The system permits to quasi-continuously (every 12 min) determine concentration of dissolved gases (He, Ar, Kr, N2, and O2) in groundwater. For this work, we included dissolved CO2 into the measurement procedure, following the experimental setup and calibration procedure of Mächler et al. [2012]. The overall error of the CO2 concentration measurements is about 3% at atmospheric solubility concentration. The observed CO2 concentration in groundwater was always larger than the detection limit of the system [Mächler et al., 2012]. The overall measurement errors of the other analyzed gases are ±1% for Ar, N2, and O2, ±3% for He, and ±4% for Kr [Mächler et al., 2012].

2.3. Gas Exchange in Porous Media

[13] In open water, gases are exchanged through the interface between the water body and the atmosphere and commonly attain gas partitioning equilibrium which is described by Henry's law [see e.g., Kipfer et al., 2002]:

display math(1)

where math formula is the partial pressure of a gas i in air, math formula is the gas concentration in air-saturated water (ASW), and Hi is the Henry coefficient of gas i at the temperature Tw and salinity Sw.

[14] In contrast, gas exchange in the unsaturated zone of a riparian groundwater not only occurs with the atmosphere, but also with entrapped air bubbles that are captured in the unsaturated zone (i.e., excess air formation), e.g., during water-table fluctuations [Vogel et al., 1981; Stute et al., 1992; Aeschbach-Hertig et al., 1999; Williams and Oostrom, 2000; Kipfer et al., 2002; Klump et al., 2007].

[15] Aeschbach-Hertig et al. [2000] proposed a simple but widely accepted parameter model of excess air formation, which assumes the gases in the gas bubbles and the dissolved gases to be in solubility equilibrium with respect to the elevated pressure in the gas phase. Under such conditions, and further with the given water temperature Tw and by assuming a negligible groundwater salinity (Sw ≈ 0), the dissolved gas concentration math formula of a gas i can be described by the closed-system equilibration model [CE model, see Aeschbach-Hertig et al., 2000]:

display math(2)

where A is the amount of dry air per unit mass of water initially entrapped in the water and zi is the volume fraction of the gas i in dry air. The fractionation F describes the degree of bubble dissolution. In case of F = 0, all the bubbles are completely dissolved and F ≈ 1 means that virtually no gas of the air bubbles dissolved in the surrounding groundwater. For F > 0, the excess air component is elementally fractionated with respect to the complete dissolution of air. During the partial dissolution of air bubbles, the more soluble (noble) gases are enriched in the water phase. Aeschbach-Hertig et al. [2000] showed that the model parameters (Tw, A, F) can be determined from the observed noble gas concentration by least-squares regression using equation (2) [Aeschbach-Hertig et al., 1999, 2000].

[16] In equation (2), the concentrations prevailing before excess air formation math formula are assumed to be equal to the ASW concentration, i.e., math formula. However, this is not necessarily true (see section 1) and hence equation (2) needs to be modified:

display math(3)

[17] Note that the derivation of equation (3) follows exactly the derivation of the CE model given by Aeschbach-Hertig et al. [2000], however with an arbitrary concentration math formula instead of math formula as the initial concentration.

[18] The results attained from equation (3) differs significantly from equation (2), if math formula deviates significantly from math formula. In such a case, only the modified CE model (i.e., equation (3)) is appropriate to qualitatively describe the gas injection. However, if math formula in equation (3) corresponds to an ASW concentration at a similar temperature as Tw, the CE-model regression will deliver reasonable values for A and F. Note that in case of total dissolution (i.e., F = 0), the value A is actually determined correctly. However, in this case the resulting value for math formula needs to be interpreted as the ASW concentration at the water temperature that prevailed during the air/water partitioning before excess air formation, which is in general different from the groundwater temperature during excess air formation. In case of an elemental fractionation (F > 0), the composition of the excess air is in principle temperature dependent. However, in strong contrast to temperature reconstruction from dissolved noble gases, the determination of A and F is mainly controlled by light and hence poorly soluble noble gas species (He). As the Henry coefficient of He hardly depends on the temperature, the determination of A and F are only marginally affected by the water temperature at which the noble gas concentrations prior to the excess air formation were preset. In our case of a losing river the initial concentration for noble gases are given by the ASW concentration at river temperature. Temperature differences of the river (15–19.5°C) and groundwater (e.g., 13–16°C in the willow zone) are reasonably small. Hence, the original CE model (i.e., equation (2)) can be used to determine A and F out of the measured noble gas concentrations, as in the given temperature range changes in the ASW concentration of the less soluble noble gas species are low ( math formula for Ar and math formula for He). Also for calculating N2 injection the original CE model can be used, as the initial N2 concentrations are near the ASW concentrations of the river. However, to estimate O2 input we used equation (3), as initial O2 concentrations in the hypoxic groundwater were very low.

[19] To calculate O2 input by using equation (3) in case of F > 0, we implicitly assume that no O2 is consumed during the actual formation of excess air. In the case of significant O2 consumption during excess air formation, O2 is constantly extracted from the bubbles to attain local equilibrium between the dissolved gases and the entrapped gas phase, until the bubbles are void of O2. In that limiting case, the effective O2 input is equal to math formula. Therefore, the quantification of the effective O2 input using equation (3) represents a conservative estimate.

2.4. Limitations of the CE Model

[20] The CE approach used to parameterize excess air is based on assumptions which are at least in the case of highly dynamic hydraulic conditions not always justified [see Klump et al., 2008]. The underlying assumptions of the CE model are that the exchanging water and the entrapped gas phase are at any time and at any location in solubility equilibrium (equilibrium assumption), and that gas transport due to groundwater advection can be neglected during excess air formation (closed system assumption). Rather than giving a comprehensive discussion on the excess air formation [see Holocher et al., 2003; Klump et al., 2008], we briefly discuss why the simplified CE approximation of gas partitioning in porous media sets a reasonable conceptual frame to quantify aeration of ground water in response to excess air formation [see Klump et al., 2008, and section 3.3].

[21] Due to the equilibrium assumption, the CE approach neglects the kinetics of gas dissolution and the gas transport within the water (and gas) phase. These assumptions are reasonably full-filled if the advective gas transport (with the flowing water) is slow compared to the effective gas transfer from the gas phase into the water, i.e., if the respective Damköhler number, the ratio of the rates of gas exchange and advective gas transport is large [Klump et al., 2008].

[22] Further, it can be shown that the formation of unfractionated excess air (e.g., “pure air injection”) is not necessarily the result of the complete dissolution of entrapped air bubbles [Klump et al., 2008]. Rather, the gas exchange between the advecting water and the dissolving gas reaches a quasi steady state whereby the differences in solubilities of the different gas species are balanced by an opposite elemental fractionation in the remaining gas phase. Such a quasi steady state results in virtually constant gas fluxes out of gas phase into the water [Klump et al., 2008]. The produced noble gas excess has an elemental composition very similar to that of unfractionated air, although a trapped nonatmospheric gas phase is still present and is continuously dissolving [Klump et al., 2008]. Under these conditions, the CE model is shown to (slightly) underestimate the effusively produced excess air amount and therefore the injected O2 due to excess air is slightly underrated [Klump et al., 2008].

3. Results

3.1. Changes in Dissolved Gas Concentrations in Response to a Flood Event

[23] The field campaign started in the middle of May (2011) after a dry period of several weeks. At that time, the discharge of the river Thur was below 10 m3/s (Figure 3). On 15 May, after heavy precipitation in the catchment, a small flood (discharge approximately 100 m3/s) passed our study site at Niederneunforn and forced the water table to rise approximately 1 m. At peak flood, the gravel zone was flooded. The groundwater table in the grass and the willow zones remained below the soil surface. After the peak flow, the discharge rapidly decreased and three days later was again below 20 m3/s.

Figure 3.

Discharge of the river Thur at Niederneunforn during April–May 2011.

[24] The dissolved gas concentrations measured in the three functional process zones are shown in Figure 4. All measured O2 concentrations in all wells were found to be below ASW concentration. Especially low O2 concentrations were found in well W1 throughout the entire measurement period except for 2 days after the flood. Also, particularly, low O2 concentration was found in well W2 during the period after the flood as well as in the grass and gravel zones during the flood. In all zones, the dissolved CO2 concentrations were significantly higher than the ASW concentration and increased during the flood.

Figure 4.

Dissolved gas concentrations in the wells. For each measured gas specimen, two panels are shown: (right) the dissolved gas concentrations determined in the wells positioned in the willow zone and (left) the concentrations in the grass and gravel zones. The gray area corresponds to the flood event as shown in Figure 3.

[25] In the groundwater of the grass and gravel zones, the temporal evolutions of CO2 and O2 were anticorrelated, whereby the observed increase in CO2 concentration during the flood was similar to the O2 concentration decrease. Assuming that the initial concentrations of O2 in the grass and gravel zones remained the same and did not change at all, the CO2 concentration increases can be explained by the consumption of the initially available O2. However, in the willow zone the CO2 concentrations increased much stronger during the flood than in the other zones. Due to the initially prevailing low O2 concentrations, the increase in CO2 concentrations cannot be explained by consumption of the initially available O2.

[26] Noble-gas concentrations also increased in response to the flood event. The synchronous temporal concentration evolution of all dissolved noble-gases in the first day of the flood points to a gas input, triggered by the flood event, i.e., excess air formation. Furthermore, for well W1 and well W2 gas transport from the river (or the hyporheic zone) to the different groundwater wells takes up to 4 days [Vogt et al., 2010] and hence a change in river gas concentrations takes far longer than 1 day to reach well W1 and W2. We therefore conclude that excess air was formed within the aquifer and not in the river during infiltration.

[27] Due to the direction of groundwater flow (see Figure 2), the excess air observed in well GB1 was formed in situ in the gravel zone. Furthermore, the typical Darcy velocity at the field site is approximately 10 m/day [Vogt et al., 2010]. Thus, the water sampled in the willow and gravel zone in the first and probably also in the second day after the peak flood was already in the respective zone during the flood (see Figure 2). In conclusion, the observed excess air was generated in the same zone where the respective observation well is located. However, for longer time scales concentrations might be affected by advective transport from other zones and hence do not only reflect excess air formation or any other process within the zone itself.

[28] To analyze the temporal evolution of excess air formation, the evolution of He and Kr concentrations is illustrated in Figure 5a. Especially during the flood (filled points), the determined He and Kr concentrations are higher than ASW concentrations, as they do not fall on the line the respective ASW concentrations span. The noble gas surpluses during the flood have an elemental composition which follows a pressure trend (dashed arrow 2 in Figure 5) [Beyerle et al., 1999; Kipfer et al., 2002]. That means the noble gas concentrations during the flood event can be interpreted as a combination of an ASW component and an excess noble gas component, generated by the (partial) dissolution of entrapped air bubbles under elevated pressure.

Figure 5.

(a) Two-elemental plot of He and Kr and (b) plot of ΔCO2 against He concentrations. The ΔCO2 is the difference of the measured CO2 concentration and the expected ASW concentration. The arrows in Figure 5a indicate the concentration increase of dissolved noble cases in response to different dissolution processes (see 1 and 2 in the legend, Kipfer et al. [2002]). The arrow origins are set at the atmospheric solubility concentration at the water temperature prevailing during the process of gas dissolution. Arrows of the same process, but with different origins (on different ASW concentrations) are approximately parallel (gray arrows).

[29] Figure 5a also indicates higher Kr and He concentrations in the willow zone than in the grass and gravel zones, which implies an enhanced excess air formation in the willow zone. Furthermore, Kr concentrations in the excess air components of the groundwater in the willow zone are enriched with regard to the He concentrations, i.e., the Kr/He concentration ratio exceeds the ratio of dissolved atmospheric air. The excess air components, hence, are elementally fractionated with regard to complete dissolution of air in groundwater at 15°C. Thus, only partial dissolution of the gas bubbles occurred (i.e., F > 0).

[30] He concentrations and the excess CO2 concentrations (see Figure 5b) show a strong positive correlation, which indicates that the CO2 rise is closely linked to excess air formation. We hypothesize that the CO2 increase during the flood event resulted from O2 input due to excess air formation, whereby the injected O2 was respired to CO2. Under such conditions, the measured CO2 increase therefore must balance the O2 input, unless a significant part of the produced CO2 is mineralized or transferred as a free gas into the remaining gas bubbles.

[31] In summary, the measured dissolved gas concentrations indicate local excess air formation. However, both the amount and the elemental composition of the excess air differ between the different functional process zones, influencing the dissolved O2 and the CO2 concentrations. To confirm and quantitatively discuss these statements, both excess air and effective O2 input needs to be determined.

3.2. Calculations of O2 Input Due to Excess Air Formation

[32] The determined CE-model parameters F and A and the corresponding χ2 values of the respective fitting procedures are given in Table 1. Almost all χ2 values are around 1 or below, hence reasonably agree with the degree of freedom of the fitting parameters, which implies that the measured concentrations can be reproduced by the CE model. The model calculations confirm that only in the willow zone large air entrapment occurred (i.e., math formula) during the flood: the values of A in the other zones are almost two orders of magnitudes smaller than in the willow zone. Furthermore, in the willow zone the composition of the excess air component is elementally fractionated (F > 0), whereby in groundwater of the grass and gravel zone the excess air is of atmospheric composition (F = 0). These findings are in agreement with the observed noble gas excesses in the willow zone during the flood event, which are enriched in the more soluble and heavier noble gas species (see Figure 5).

Table 1. Calculated Model Parameters (With Equation (2)), the Effective O2 Input (With Equation (3)), and the Relative Deviation of Calculated to the Measured N2 Concentration (ΔN2)a
 Well math formulaFb (–)χ2 (–)O2 input (µmol/l)ΔN2c (%)Day in May 2011
  1. a

    Our wells correspond to the wells analyzed by Vogt et al. [2010] (names are given in the brackets after our nomenclature of the wells).

  2. b

    Only if the regression using the unfractionated excess air model (i.e., math formula) did not yield acceptable results, the CE model (i.e., F is unconstrained) was used [see also Klump et al., 2008].

  3. c

    Total error, accounting for analytical and model errors, is around 2%.

Gravel
GB1(r050)0.2 ± 0.200.0042 ± 2211
0.6 ± 0.200.305 ± 2013
1.0 ± 0.200.2110 ± 2−316
2.2 ± 0.201.5820 ± 2017
0.2 ± 0.200.572 ± 2020
Grass
G1(r042)0.7 ± 0.200.017 ± 1211
1.1 ± 0.200.3510 ± 2013
1.8 ± 0.200.3616 ± 2−316
1.3 ± 0.200.1412 ± 2017
0.5 ± 0.200.015 ± 2020
G2(r072)0.2 ± 0.401.72 ± 4211
0.9 ± 0.400.699 ± 4013
1.4 ± 0.400.00413 ± 4016
1.5 ± 0.401.1614 ± 4−317
0.1 ± 0.400.031 ± 4−420
Willow
W1(r041)0 ± 0.100.220 ± 1010
0 ± 0.202.890 ± 2013
0 ± 0.101.360 ±1214
0 ± 0.102.790 ± 1015
89 ± 140.68 ± 0.050.19232 ± 40−116
17 ± 90.55 ± 0.09<0.00169 ± 30017
1.2 ± 0.300.1212 ± 3018
3.5 ± 0.301.1433 ± 3419
0.5 ± 0.100.075 ± 1−120
0.3 ± 0.100.543 ± 1021
0.1 ± 0.100.4031 ± 1022
0.2 ± 0.105.462 ± 1224
W2(r058)0 ±0.109.390 ± 1011
1.6 ± 0.200.0615 ± 2−113
91.8 ± 0.50.69 ± 0.010.41218 ± 9−116
44 ± 300.72 ± 0.05<0.001122 ± 30−117
1.5 ± 0.101.3114 ± 1320
1.5 ± 0.100.8814 ± 1224

[33] The enhanced excess air formation in the willow zone injected significantly more O2 into the groundwater in this zone than in the grass and gravel zones, where hardly any O2 injection occurred (see Table 1). The comparison of calculated effective O2 input with the measured O2 and CO2 concentrations in the groundwater of the willow zone (Figure 6) reveals that the effective O2 input was similar to the CO2 concentration increase during the flood math formula.

Figure 6.

Measured CO2 and O2 concentrations in comparison to the calculated O2 concentration accounting for excess air formation. The error bars indicate 1-σ errors of the calculated values for A and F.

[34] We assume the initial CO2 concentration in well W1 and W2 to be equal to the measured CO2 concentration before the flood, because, before the flood, the sum of the measured O2 and the CO2 concentrations in the grass zone was constant in time, and similar to the measured CO2 concentration in the willow zone. Thus, assuming advection of the gases with water from the grass zone to the willow zone and complete reduction of O2, the initial CO2 concentration of the water in the willow zone must have been very similar to the measured CO2 concentration before the flood. However, the sum of the initial CO2 and the CO2 produced by the local respiration of the injected O2 explain about 90% of the measured CO2 concentration during the flood. The other remaining 10% of CO2 needs to be explained by another CO2 source, e.g., CO2 production due to denitrification. We estimated the N2 injection due to excess air formation in order to determine whether the measured N2 concentrations can be fully explained by the physical exchange during air/water partitioning (e.g., excess air formation), or whether any biogeochemical reactions affected the N2 concentration. The analysis shows that excess air formation completely accounts for the elevated dissolved N2 concentrations in the groundwater body at our field site, i.e., the differences (ΔN2) of measured N2 and the calculated N2 concentrations are within measurement uncertainty in all zones (see Table 1).

[35] Hence, complete denitrification does not seem to significantly contribute to the observed CO2 variations. However, incomplete denitrification which does not produce N2 and therefore cannot be addressed by dissolved gas measurements may contribute to CO2 formation.

3.3. Accuracy and Reliability of the Used CE Model

[36] We will give in this section some arguments why our simplified CE approach, which only approximately describes excess air formation, nevertheless produces valid and acceptable results (see section 3.3).

[37] At our site, the observed groundwater table fluctuation of around 1 m occurred in about 1 day, whereas a local gas equilibrium between the entrapped air bubbles and the surrounding water is established within an hour [see Holocher et al., 2003]. Therefore, the Damköhler number is large, which justifies the equilibrium assumption (see section 3.3 [Klump et al., 2008]).

[38] In our study, in particular in the willow zone where most excess air was generated, the heavy and soluble Kr shows a similar supersaturation as the light and only poorly soluble He (F ≈ 0.7, see Table 1). Even the high groundwater velocities of around 10 m/day [Vogt et al., 2010] are not fast enough to significantly deplete the Kr excess with regard to the He excess. Therefore, the produced excess air is not affected by groundwater velocity and hence the basic assumption underlying the CE approach is justified (see section 3.3).

[39] In summary, in our case the use of the CE model to quantify excess air formation and the modified CE model to estimate the according O2 injection yields reasonable results even when accounting for the sometimes physically unrealistically assumptions of the CE model [see Klump et al., 2008].

4. Discussion

[40] The results of the measurements and the model calculations show that in the case of the riparian groundwater near the river Thur, the O2 increase due to excess air formation in response to flood events can be of the same order of magnitude as the atmospheric solubility concentration of O2. In our study, during the flood the measured O2 concentrations in the groundwater remained low. However, the CO2 concentrations increased considerably. The observed CO2 production was similar to the calculated O2 inputs. Hence, we conclude that the O2 introduced by excess air formation was consumed immediately after injection, whereby CO2 was produced.

[41] Our study confirms that excess air formation is an important mechanism for delivering O2 into groundwater [e.g., Rose and Long, 1988; Beyerle et al., 1999]. Furthermore, the results of our study confirm the laboratory experiments performed by Williams and Oostrom [2000] that a fluctuating water table can oxygenate riparian groundwater, and the field experiments performed by Massmann and Sueltenfuss [2008] near a drinking water production well, where excess air was identified as a major O2 source, whenever water fluctuations were occurring.

[42] However, our study further shows that water-table fluctuations do not always induce excess air formation, as no excess air formation was observed in the grass and gravel zones during the flood (see Table 1). Excess air formation in groundwater is constrained to the quasi-saturated zone at the boundary between the saturated and the unsaturated zone, where air bubbles can be entrapped by water-table fluctuations [Kipfer et al., 2002; Holocher et al., 2003; Klump et al., 2008]. Hydraulic conditions, i.e., the shape and material of the aquifer, of the quasi-saturated zone therefore are assumed to play an important role in the trapping of air [Wilson and McNeil, 1997; Holocher et al., 2003; Klump et al., 2007].

[43] At our field site, spatial variability in the groundwater confinement leads to differences in the hydraulic condition between the different process zones. The thickness and depth of the loam layer covering the Thur aquifer increases with distance from the river (see Figure 1). As a result of the missing loam layer in the first meter, the groundwater body is always unconfined in the gravel zone, whereas in the willow zone a thick loam layer ensures that the groundwater is always confined (see Figure 1). In the grass zone, however, a confined groundwater body is only present during the flood, while during normal flow conditions of the river Thur unconfined conditions prevail. These different hydraulic conditions may set a conceptual frame to explain why excess air formation differs between the different process zones.

[44] The unconfined conditions in the gravel zone do not effectively trap air (due to the high gas permeability of the gravel) and prevent excess air from being generated. Similarly, in the grass zone the soil air can be pushed toward the gravel zone which again prevents effective air trapping. In contrast, in the willow zone, the rising groundwater table in response to enhanced river discharge exerts a higher hydrostatic pressure on the entrapped gas phase, as the confining loam layer prevents the entrapped air from escaping.

[45] Besides the confining loam layer, additional factors that affect the texture of the soil and hence influence the “mobility” of the entrapped air bubbles will impact the formation of excess air. Such differences in soil texture and geomorphology, which may have a certain influence on the ecological function (e.g., due to different vegetative cover or root growth of the willows), may further explain the difference in excess air formation in the grass and the willow zone.

[46] From a methodological point of view, our field study shows that in order to estimate the effective O2 turnover in groundwater, just measuring O2 concentrations is not sufficient [see also Beyerle et al., 1999], as such measurements do not allow the O2 input into groundwater to be quantified. In our study, the measured O2 concentrations in the groundwater remained low even during the flood despite a high O2 input due to excess air formation. Hence, only the quantification of excess air allowed O2 turnover to be quantified. Furthermore, the discrepancy between the estimated O2 injection and the low measured O2 concentrations show that the locally introduced O2 was immediately reduced, which explains a significant part of the observed increase in CO2 concentration during the flood event. Similarly, our quantification of excess air yields an adequate estimate of the dissolved atmospheric N2 concentrations, which need to be known to detect and study complete denitrification processes.

5. Conclusions and Outlook

[47] Our study demonstrates that excess air formation contributes significantly to the O2 input into riparian groundwater and hence crucially controls the redox condition and therefore the quality of shallow groundwater. Hence, for any O2-budget calculation, the O2 being delivered by excess air formation needs to be quantified [see also Massmann and Sueltenfuss, 2008]. We identified that flood induced excess air formation may significantly contribute to groundwater aeration and, thus, improve drinking water quality by enabling faster pollutant degradation. Vice versa, the water quality of a gaining river might be considerably enhanced if the infiltrating groundwater is oxygenated by excess air formation before entering the river [see also Williams and Oostrom, 2000]. Furthermore, our study demonstrates that morphology acts as structural control to excess air formation and substantially constrains O2 input due to excess air formation in groundwater, as e.g., low permeable sediments in the unsaturated zone enhance air trapping and thus foster excess air formation in response to groundwater table rise.

[48] For further studies, the results of our method, i.e., CO2 and O2 concentrations, and O2 and N2 input due to excess air formation, can be embedded in a broader biochemical context. Additional quantification of the major ion chemistry, organic carbon or the analysis of different nitrogen species would enable to determine the influence of excess air on biogeochemical processes like denitrification.

Acknowledgments

[49] We thank A. Bretzler for improving the manuscript. This work was performed within the framework of the RECORD project (Assessment and Modeling of Coupled Ecological and Hydrological Dynamics in the Restored Corridor of a River (Restored Corridor Dynamics)), and was funded by Eawag (Swiss Federal Institute of Aquatic Science and Technology) and CCES (Competence Center Environment and Sustainability of the ETH Domain).

Ancillary