Impact of microforms on nitrate transport at the groundwater–surface water interface in gaining streams

Advances in Water Resources 73 (2014) 185–197

Contents lists available at ScienceDirect

Advances in Water Resources journal homepage: www.elsevier.com/locate/advwatres

Impact of microforms on nitrate transport at the groundwater–surface water interface in gaining streams Haizhu Hu a,⇑, Andrew Binley b, Catherine M. Heppell c, Katrina Lansdown c,d, Xiaomin Mao a a

College of Water Resources & Civil Engineering, China Agricultural University, No. 17 Qinghua East Road, Beijing 100083, China Lancaster Environment Centre, Lancaster University, Lancaster LA1 4YQ, United Kingdom c School of Geography, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom d School of Biological and Chemical Sciences, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom b

a r t i c l e

i n f o

Article history: Received 21 October 2013 Received in revised form 23 July 2014 Accepted 28 July 2014 Available online 8 August 2014 Keywords: Hyporheic flow Nitrate transfer Riffle-pool sequence Microform Nitrification Denitrification

a b s t r a c t Small streambed structures (or microforms, 0.01–1 m in length) exist ubiquitously in riverbed systems. Small-scale topography is potentially important in controlling hyporheic exchange flow and transport of conservative and reactive solutes at the groundwater–surface water interface. The role of microforms on NO 3 transfer in a riffle-scale (macroforms of >1 m length) hyporheic zone within a gaining river setting is investigated using a 2-D flow and transport model which accounts for both nitrification and denitrification. Results show that the short pathlines caused by microforms lead to more NO 3 discharge to the river compared with a macroform-only condition due to shortened residence times of both surface water and groundwater in mixing zones. Short hyporheic exchange flow pathways caused by microforms could remain oxic along their entire length or switch from nitrate producing to nitrate consuming as oxygen concentrations decline. Microforms affect net NO 3 flux by the combined effect of introducing more stream mass flux and reducing their residence time in mixing zones under different hydrological and biogeochemical conditions. Our findings underscore that ignoring microforms in river beds may underestimate NO 3 load to the river and have practical implications for pore water sampling strategies in groundwater–surface water studies. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction The hydrologic exchange of stream water and groundwater (GW) underlying a stream channel plays an important role in biogeochemical cycles in streambed sediments, where reactive solutes undergo physio-chemical transformations and thus influence nutrient supply and benthic habitat quality in riverine ecosystem [1–5]. Hyporheic exchange flow (HEF) is the process by which stream water invades the subsurface and rejoins the downstream channel [6,7]. HEF is driven essentially by variations of the hydraulic gradient along the stream-subsurface interface, which can occur due to topographic features ranging from individual particles, ripples, dunes, bars up to riffle-pool sequences and meanders [8]. The streambed morphology can be a key factor in controlling upwelling and downwelling fluxes, increasing hyporheic residence time (RT), expanding the extent of the hyporheic zone (HZ), and

⇑ Corresponding author. E-mail addresses: [email protected] (H. Hu), [email protected] (A. Binley), [email protected] (C.M. Heppell), [email protected] (K. Lansdown), [email protected] (X. Mao). http://dx.doi.org/10.1016/j.advwatres.2014.07.013 0309-1708/Ó 2014 Elsevier Ltd. All rights reserved.

has important biogeochemical implications on stream water chemistry [3,4,9]. Here we apply the same classification of HEF processes as Käser et al. [10], i.e. microform HEF (0.01–1 m), which tends to be induced by hydrodynamic pressure variations, and macroform HEF (>1 m), which is more likely generated by hydrostatic pressure. As a means of evaluating hyporheic exchange, a number of GW models have been used to simulate the spatial variability of movement of surface water (SW) into the subsurface and examine the effect of morphologic features on HEF from a process-based perspective, rather than a lumped model based on the stream behaviour [11–13]. Flow-simulation results from MODFLOW [14] and MODPATH [15], for example, suggest that channel unit, size, and sequence are all important in determining hyporheic exchange patterns [16]. By using the same models combined with MT3DMS [17], Saenger et al. [18] showed that hyporheic exchange in a rifflepool sequence increased with increasing SW flow, and that mass transfer is more influenced by the hydraulic conductivity of riverbed sediments than SW flow. Lautz and Siegel [19] simulated the hyporheic exchange around debris dams and meanders along a semi-arid stream using MODFLOW and MT3D [20]; their results indicate a predominant role of advective processes.

186

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

Conservative tracer experiments conducted in a river with poolriffle morphology indicate that a major mechanism for hyporheic exchange is bed form-induced advection [3]. Tonina and Buffington [13] arrived at a similar conclusion based on observations in their laboratory experiment simulating a gravel pool-riffle channel. Advection was also found to be a major control in micro-scale induced HEF by Jin et al. [21] and Vollmer et al. [22]. Mass transfer and spatial evolution of reactive solutes in the HZ have received less attention, although several studies deserve note and are discussed below. For reactive species, the fate, transport and concentration distribution in the HZ is not only regulated by various bed topographies of a river but also biogeochemical reactions [5,23–25]. The transport of nitrogen, being a nutrient essential to sustain life, has important implications for quality of both SW and GW. In most freshwaters, nitrate (NO 3 ), is the dominant form of nitrogen present and, as such, the removal of NO 3 by denitrification, the microbially mediated reduction of NO 3 to nitrogen gas (N2), is of interest [26]. HZs have been recognised as hotspots of denitrification in the landscape because of the potential for anoxic conditions in the HZ and the availability of labile carbon [27,28]. For example, Hill et al. [3] reported that the hyporheic zone of a small N-rich stream in Ontario served as a NO 3 sink. Pinay et al. [29] noted that the streambed can serve as NO 3 sink, and residence time plays an important role in allowing denitrification to decrease NO 3 concentration. Zarnetske et al. [30] revealed changes in redox conditions from oxic to anoxic with nitrate produced at the start of the flow path and consumed at the end of the flow path across a gravel bar in western Oregon. Given the constraints of field measurements, concentration distributions in HZ have been well investigated via the simulation of coupling between hyporheic flow and biogeochemical activity. For example, significant spatial variations in concentration of reactive solutes have been observed below a riffled sediment bed by Shum [25]; Bardini et al. [31] reported on chemical zonation of nutrients in a duned streambed. Recently, hyporheic nitrogen cycling in gravel bed rivers with riffle-pool morphologies has been investigated by process-based models in Lagrangian coordinates [4,32], and a one-dimensional (1D) model with coupled nitrification–denitrification dynamics has been applied to simulate the fate of NO 3 in HZ by Sheibley et al. [33]. In addition to these approaches, Monte Carlo sensitivity analyses with a non-dimensional form of a 1D reactive transport model has been used to identify whether the HZ is a net source or sink of NO 3 across different temporal and spatial scales [34]. Among the various types of topography triggering HEF, rifflescale HEF is well documented because of its common occurrence in natural streams. Recently, HEF induced by the roughness of the stream bed or in-stream obstacle-induced oscillation has received attention. Although the role of small-scale HEF, in comparison to large-scale HEF, has been less reported in field studies, the influence of microforms on interfacial exchange system is recognised [10,22]. Microforms embedded in a riffle-pool sequence can potentially affect HZ development and biogeochemical process at larger scales and have important implications for the hyporheos [6,35]. However, the evidence of scale of microforms and macroforms and their coupling impact on NO 3 transfer between surface and subsurface water has received little attention to date. Fig. 1 illustrates physical evidence of microforms along a short (40 m) sub-reach of the River Leith in Cumbria, UK. A 250 m reach (also shown in Fig. 1) has been the focus of a number of recent studies on spatial patterns of groundwater–surface water exchange and nutrient transport (e.g. [36,37]). Distinct macroforms are evident in the stream bed topography along the reach, however, from high resolution topographic surveying (i.e. greater sampling density) over a 40 m sub-reach, considerable microform structure can also be seen. Our study here is not specifically

Fig. 1. Evidence of microforms in a sub-reach of the River Leith. The upper image shows bed topography along the 250 m study reach; the lower image shows bed topography along a sub-reach (location is marked by the dashed lines in the upper image) derived from higher resolution spatial sampling. Flow is from left to right. Microforms, embedded in macroforms, are clearly seen in the downstream section of the sub-reach. maOD stands for metres above ordnance datum (UK sea level measurement).

targeted at the River Leith site; we use this example as an illustration of the presence of microforms and, through a generic model, explore the implications of neglecting such features in the study of reactive transport between groundwater and surface water. In their modelling study, Käser et al. [10] verified that microforms can generate a higher total water flux through the subsurface, and reduce the mean residence time of HEF. Furthermore, distinct flow patterns are induced by the interaction between microforms and macroforms. Käser et al. [10] focused their work on conservative transport; here we build on their work and examine: (1) whether microforms can significantly affect the exchange and spatial distribution of NO 3 across the streambed within a rifflepool sequence; (2) which hydrological and biogeochemical factors influence NO 3 delivery to a gaining stream in a streambed topography with microforms embedded in a macroform. In order to address these questions, a vertical 2D flow and transport model coupled with Monod kinetics is developed. Both advective and dispersive transport, along with nitrification and denitrification, are considered in the model. A base (reference) case is first established to evaluate the net mass exchange and spatial distribution of solute in the streambed. The impacts of microforms are investigated for this base case by comparing responses with and without microforms. Four vertical sampling profiles of streambed chemistry  (Cl and NO 3 ) within 1 m depth are taken to illustrate NO3 transformation in this base case. Sensitivity analyses are then performed to evaluate the effect of hydrological and chemical properties on the net nitrate transfer with and without the presence of microforms. The properties of the base case and the configurations of the models used for the subsequent sensitive analysis are shown in Table 1.

187

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

4 2 4 1 Denitrification CH2 O þ NO3 ! N2 þ HCO3 þ CO2 þ H2 O 5 5 5 5

2. Model development 2.1. Conceptual model

 Nitrification of NHþ 4 to NO3 by nitrifying bacteria occurs under aerobic condition and thus is a source of NO 3 to aquatic ecosystem [34]. DOC is assumed to be labile and available to heterotrophic biomass [39,40]. The heterotrophic microorganisms preferentially utilise DO as the electron acceptor, and utilise NO 3 to fuel their metabolism in the absence of oxygen [41]. According to Marzadri et al. [32] and Böhlke and Denver [42], the threshold concentration of DO for denitrification in river sediment is 3 mg L1. Complete reduction of nitrate to nitrogen gas is assumed to occur [43]. Noncompetitive inhibition is incorporated into the utilisation rate expression to suppress denitrification while DO is present [41]. Biomass in sediment is assumed constant [34,44]. The reaction rates described by multiple-Monod kinetics are expressed by

To investigate the effect of microforms on mass flux in SW–GW exchange, a 2D (vertical section) finite element flow and multi-component reactive transport model was established using FEFLOW™. The numerical model comprises of a steady-state flow module and a transient transport module. Four solutes are  included, i.e. NHþ 4 , NO3 , dissolved organic carbon (DOC), and dissolved oxygen (DO), combined with heterotrophic denitrifiers and microbial functional groups for aerobic respiration and nitrification. A conservative solute, chloride, was also included in the model to compare with the reactive behaviour of NO 3 . The vertical-longitudinal view model domain is a rectangle 75 m long and 10 m deep as shown in Fig. 2. A 1 m-thick reactive layer was set immediately beneath the interface of SW and sediment to make sure that reactions occur within hyporheic zones even in low GW discharge scenarios. The model does not account for turbulent stream flow over river beds.

RNO ¼ lDEN

2.2. Mathematical model

RNH ¼ lNI

2.2.1. Model equations For a homogeneous riverbed, steady state flow in a 2-D vertical section is governed by the equation

ar2 h ¼ 0

C NO CC K IO CO C NH X 2 þ lNI X1 K NO þ C NO K C þ C C K IO þ C O K O þ C O K NH þ C NH ð3Þ

CO C NH X1 K O þ C O K NH þ C NH

RC ¼ y0 lOXI

ð1Þ



ð4Þ

CC CO C NO CC X 3  lDEN K C þ CC K O þ CO K NO þ C NO K C þ C C

K IO X2 K IO þ C O

ð5Þ

1

where a (m s ) is the hydraulic diffusivity, given by the ratio of hydraulic conductivity K (m s1) and specific storage Ss (–). Reactive transport in saturated porous media can be expressed as:

@ C j  rðDrC j Þ þ rðqC j Þ ¼ Rj @t

RO ¼ y0 lOXI

CC CO CO C NH X 3  ð1  y0 ÞlNI X1 K C þ CC K O þ CO K O þ C O K NH þ C NH ð6Þ

where lDEN, lNI, lOXI (d1) are the maximum substrate utilisation rates for denitrification, nitrification and aerobic respiration, respectively; CNO, CC, and CO (mg L1) are the concentration of nitrate, DOC, and oxygen, respectively; X1, X2 and X3 (mg L1) are the biomass of nitrification functional group, denitrifiers and aerobic respiration functional group; KNO, KNH, KC, KO (mg L1) are the þ half-saturation constants for NO 3 , NH4 , DOC and O2, respectively; yO (–) is the partition coefficient for aerobic respiration, and 1  yO (–) is for nitrification; kIO (mg L1) is the oxygen inhibition constant. The values of biochemical parameters are within the range of reported values as listed in Table 2.

ð2Þ

where j (–) is the species indicators, j = 1, 2, 3, . . . , N; Cj (mg L1) is the concentration of species j; D (m2 s1) is the tensor of hydrodynamic dispersion; q (m s1) is the Darcy flux and Rj (mg(Ls)1) is the bulk rate of chemical reaction of species j. The reactive processes considered in the model are shown in Fig. 3. DOC-bearing organic matter is represented with the simple chemical formula, ‘‘CH2O’’ and the reaction equations [31,34] are expressed as:

2.2.2. Boundary and initial conditions Both the upstream and downstream boundaries are considered to be no flow boundaries. A spatially varied head distribution is imposed on the top of the model domain which represents flat

Aerobic respiration CH2 O þ O2 ! CO2 þ H2 O Nitrification NHþ4 þ 2O2 ! NO3 þ 2Hþ þ H2 O

Table 1 Base case setting and scenarios used for the sensitivity analysis. Property

Base case

Scenarios

Concentration (mg L1)

Base case

Scenarios

GW discharge q (m d1)

0.14

Low to high 0.08–0.2

SW DOC

4

Ambient 4

Hydraulic conductivity K ( 104 m s1)

1

Coarse to fine 0.8–1.5

SW NHþ 4

0.05

Ambient 0.05

Inhomogeneity K ( 104 m s1)

n

Inhomogeneous Riffle 1.2/Pool 0.8

SW NO 3

8.8

Stream velocity (m s1)

0.8

Low 0.3

SW DO

Stream depth (m) Microform height (m)

0.3 0.16

n n

SW Cl Microform wavelength (m)

High 0.8

Concentration (mg L1)

Base case

Scenarios

Enriched 16

GW DOC

2

n

Enriched 3.85

GW NHþ 4

0.05

n

n

GW NO 3

14

Ambient 0

10

n

GW DO

2

n

8.8 0.6

n n

GW Cl

14

Enriched 14

188

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

Fig. 2. (a) Schematic of 2D vertical simulation domain, and (b) the head distribution in macroform (MA) and macroform and microform (MA + MI) conditions.

8  3 > > H=d 8 U < 0:34 hm ¼ 0:28 >  H=d 32 2g > : 2

0:34

H d

6 0:34

H d

P 0:34

ð8Þ

1

Fig. 3. Schematic diagram of main reaction processes in the simulation.

sediment bed. In the macroform (MA) condition, the hydraulic head induced by a riffle is represented by a segment slope with a 0.1 m head drop over a 3.3 m riffle length, which is located in the middle of the domain. The stream stage is considered to be flat in the rest of the model domain. In the macroform combined with microforms (MA + MI) condition, the head induced by microforms is superimposed on that of MA condition. The dynamic head fluctuation caused by microforms is presented by a fluctuation of the dynamic head h [10,46], which follows sine function described by:

h ¼ hm sin kx

ð7Þ

where hm (m) is the half-amplitude of the head variation; k (–) is the wavenumber of the head disturbance equal to 2p/k, where k (m) is the bedform wavelength, and x (m) is the downstream coordinate parallel to the bed surface. The half-amplitude is taken from an empirical correlation based on the data of Fehlman [47]:

where U (m s ) is the stream velocity; H (m) is the bedform height (trough to crest); d (m) is the stream depth and g (m s2) is the acceleration due to gravity. The parameters used here are based on the base case presented by Käser et al. [10] as shown in Table 1. The sinusoidal pressure distribution together with riffle-pool sequence was superposed along a flat riverbed. A constant flux boundary was specified along the bottom of the domain as the GW discharge (0.14 m d1). The hydraulic conductivity and specific yield were set to 1  104 m s1 and 0.2, respectively, for the total domain, and a porosity of 0.3 was assigned. Values for the longitudinal dispersivity aL and transverse dispersivity aT were set as 0.1 m and 0.01 m, respectively. The molecular diffusion is set as 1  109 m2 s1. A constant concentration boundary was assigned along the bottom boundary and the inlet zone of upper boundary as sources of mass release from GW and SW. In the outlet zones of the upper boundary, a zero concentration gradient was set. The concentration of each solute specified along the top and bottom boundary were shown in Table 1 and Fig. 2. For better understanding of the behav iour of NO 3 , a conservative solute Cl was specified using the same   boundary condition as NO3 . The Cl level is probably lower than that in real systems, but we use it here as a reference for reactive transport. The initial concentrations of all transport variables,   NHþ 4 , NO3 , DO, DOC and Cl , were assumed to be uniformly zero over the entire domain. 2.3. Calculation of net mass flux

Table 2 Parameters used for Monod kinetics.

The net mass flux at the interface of subsurface and surface water in a riffle-pool sequence is presented here by

Parameter

Value used in model

Literature range

References

lDEN (h1) lNI (h1) lOXI (h1)

1.8 1.08 1.8 2.5 0.5 5.2 10 0.1 0.64

0.42, 1.67, 1.6, 3.98 1.08 1.25, 1.9, 1.97 0.5, 1.64, 2, 2.6 0.43, 1 0.2, 0.77, 5.28 1 6, 8.68, 10, 40 0.01, 0.24, 1 0.64

[34,38,44,45] [34] [34,38,44] [34,38,39,44] [34,38] [34,38,39,44] [34,38,39,44,45] [34,38,45] [34]

1

KNO (mg L ) KNH (mg L1) KO (mg L1) KC (mg L1) KIO (mg L1) yO (–)

Q n ¼ Q inSW  Q outSW 1

ð9Þ

where Qn (g d ) is the net mass flux, positive value denotes a mass gain, while negative mass lost; QinSW (g d1) is the mass flux of downwelling SW that has entered the system; QoutSW (g d1) is the mass flux of upwelling subsurface water that has left the system. According to the mass balance, all the mass flux components in the system satisfy the following relationship:

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

Q inSW  Q outSW þ Q inGW  Q reac ¼ DQ

ð10Þ

1

where QinGW (g d ) is the mass flux entering the system from GW; Qreac (g d1) is the reaction-induced mass removal in the system; DQ (g d1) is the variation of mass storage in the bedform system. 3. Results 3.1. Advective hyporheic exchange in the base case The macroform-induced surface water pathlines define the interfacial exchange zone which is a bowl-shaped area (Fig. 4). Microforms produce small scallop-shaped flow cells within shallow sediments, promote shallow water exchange, and reduce path length. Thus, the microforms embedded in the macroform shorten the overall mean residence time of SW in the streambed sediments, as shown in Fig. 4(a) and (b) (red pathlines) and Table 3. As shown in Fig. 4(b), small circulations induced by a periodical pressure distribution smooth the flow path around break-in-slope points of the macroform which are characterised by a steeper hydraulic gradient. The microform-caused flow cells are compressed by upwelling GW downstream. Beneath the interfacial exchange zone in which pathlines originate from and return to the river bed interface, a deeper ambient flow zone dominated by GW upwelling flow is also visible. Solutes in GW are introduced to the SW along upwelling flow paths. GW flow pathlines are sparse close to the area with a locally steep hydraulic gradient, while dense downstream of the riffle, indicating that the upwelling GW is regulated by the interfacial exchange. 3.2. Mixing zones in the base case A mixing zone is developed by tracer mass undergoing dispersion, and creating unique biogeochemical conditions that can attenuate contaminants from either upstream surface water or groundwater under gaining condition [48]. A mixing zone can be defined as the presence of surface water in the subsurface, or the mixing of SW and GW [48], and even the combination of the two [19]. We adopted the combined definition of a mixing zone here, which is visualised by the area with a final concentration equal to or higher than 10% SW concentration and the area with the concentration between 10–90% of GW concentration (if applicable), as well as the area in between. A larger mixing zone is generated by a microforms-embedded macroform than the macroform-only condition, because microforms produce small mixing zones around the macroform (Fig. 4(c)–(n)). The mixing zones of reactive solutes are larger than conservative solute in both cases. For the area where flow paths leave and return to SW, microforms could introduce more DO- and DOC-enriched SW into the mixing zone than a macroform-only condition but reduce the overall residence time in the SW induced mixing zone (Table 3). The SW–GW mixing takes place at the boundary between the regions of advected surface water and upwelling groundwater where flow paths from these two source waters occur in parallel (Fig. 4(a) and (b)). Microforms also shorten the upwelling groundwater residence time in the SW–GW mixing zone, as the maximum residence time from GW shown in Table 3. 3.3. Concentration profiles in the base case The concentration front reflects the water exchange area, and is different in size and shape for MA and MA + MI conditions. For the concentration distribution in the MA + MI system (Fig. 4(d)–(n)), periodical concentration oscillations are present within the uppermost layer, which are associated with the advective

189

transport driven by the sinusoidal pressure distribution. The oscillations extend down to deeper sediments within the riffle-induced large HEF system and fade with depth. At the deeper locations outside the interfacial exchange zones, the concentrations in both cases are affected by upwelling solutes (Figs. 4(c)–(n) and 5). The reactive solute concentration downstream of the macroform is higher than that upstream because regulated GW flow into the river. The Cl distribution in both of the HEF systems reveals the dominance of advection in the hyporheic zone (Fig. 4(c)–(f)). Fig. 4(c) and (d) shows the hyporheic mixing zones by just adding the conservative tracer to SW. Solutes, originating from the surface water, penetrate to a maximum depth of approximately 40 cm. The blurred interface between the hyporheic zone and deeper ambient flow zone demonstrates a dispersion effect. In the upstream and downstream area of the riffle, both the advection due to the upward groundwater flow and the diffusion depending on the concentration difference between SW and GW control the Cl concentration distribution. Diffusive transport is directed upwards in Fig. 4(e) and (f), as the Cl concentration in pore water is higher than in SW. The spatial distribution of DOC concentration corresponds to the pathline patterns (Fig. 4(i) and (j)). The concentration decreases with increasing residence time, so the lowest concentration is found in the outmost area of the exchange zone in both MA and MA + MI cases. DOC is delivered from the river to sediment by advection and dispersion, and is consumed by aerobic respiration and then denitrification. The DOC concentration in the MA + MI case just beneath the bed river surface is higher than that in the MA only condition. The lowest concentration in MA case is shown at a shallower subsurface depth than that in MA + MI case. This is because the short residence time caused by the microforms leads to less decline in DOC concentration. However, microforms bring more mass from SW into the sediment, so the total loss of DOC flux is higher in the MA + MI case (Table 4). DO displays similar behaviour (Fig. 4(k) and (l)) to DOC in both MA and MA + MI cases. It is consumed by both aerobic respiration and nitrification simultaneously. The DO concentrations decrease to approximately 3 mg L1 at a depth of 45 cm in both cases. NHþ 4 concentration exhibits remarkable variations in the both cases (Fig. 4(m) and (n)). The concentration drops rapidly with depth until all DO has been consumed, suggesting spatial distribution of NHþ 4 in the subsurface is controlled by DO zonation and nitrification, supporting the findings of Bardini et al. [31]. The minimum concentration is 0.0064 mg L1 in the MA-only case due to  more oxidisation of NHþ 4 to NO3 following longer residence times in comparison with the MA + MI case. The spatial distribution of NO 3 concentration (from 8.9 to 14 mg L1) is wider than that for Cl due to the combined effect of nitrification and denitrification. The profiles of the ratio of  NO concentration at four points (locations marked in 3 to Cl Fig. 4(a)) show the production and consumption of NO 3 in the hyporheic sediments (Fig. 5). At profile A, NO 3 production is low with a concentration ratio slightly higher than 1 at 20 cm depth in both MA + MI and MA cases. As the NHþ 4 concentration is quite small rel ative to NO 3 , the production of NO3 is limited. The microforms show little impact on nitrification in this base case setting. The shift from nitrification to denitrification occurs at a shallower depth in the MA case, in comparison to the MA + MI case, possibly because the longer residence time leads to DO falling below the threshold concentration for denitrification in the MA case. Denitrification prevails between 30 cm and 60 cm depth in both cases, so NO 3 concentration from GW decreases along the pathlines and eventually exits through the sediment-SW interface at a lower concentration (Fig. 4). The profile A is similar to the high-resolution observed profile B3 in the study of Briggs et al. [49] that a net  NO 3 production in the depth of 0–20 cm and a net NO3 uptake

190

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

Fig. 4. Pathlines and solute spatial distribution for the GW fed river in both macroform only bedform and macroform with microform bedform, including (a,b) pathlines; (c,d) Cl concentration for one end member case; (e,f) Cl concentration distribution; (g,h) NO 3 concentration distribution; (i,j) DOC concentration distribution; (k,l) DO concentration distribution; (m,n) NHþ 4 concentration distribution. Red pathlines originate from SW while blue ones start from the GW. Location of sampling profiles is represented by a black dot. Arrows denotes the flow direction of SW and GW. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

at a greater depth (to at least 55 cm) occur at Dam 2 on day 194 of sampling event. At profile B, where the residence time is longer than profile A, there is a zone of net NO 3 consumption extending

from the streambed interface to a depth of at least 60 cm. Microforms reduce NO 3 consumption at the same depth. Similar profiles can also be found at profiles B1 and B2 at Dam 1 on day 194 and 203

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

191

Fig. 4 (continued)

Table 3 Comparison of residence time (RT, d) between MA and MA + MI systems in base cases. Source

Bedform

SW

MA MA + MI

GW

MA MA + MI

Median RT 7.15 1.43 23.3 17.3

Maximum RT 14.3 2.85 46.5 34.5

investigated by Briggs et al. [49]. In the downstream area (profiles C and D), where upwelling GW dominates, two sources of NO 3 mix and undergo denitrification at shallower depth in MA + MI case, while larger proportion of NO 3 decline occurs in MA case. 3.4. Net mass flux in the base case Both MA and MA + MI systems serve as a net source of NO 3 for a stream, and the net mass flux of NO discharging into SW is higher 3 in the latter case (Table 4), though the effect is somewhat small. This result seems to be fortunate because of the small influence of microforms on NO 3 transfer at the sediment–water interface.

For the net mass flux of DOC and DO (Table 4), the MA + MI serves as a sink for the stream while the MA serves as a source, because the consumption of these two solutes (Qreac) is significantly higher in the MA + MI system than the MA system. Less net NHþ 4 flux discharges from the MA + MI system into SW compared with the MA case, due to the more consumption of NHþ 4 by nitrification. According to the mass balance of MA and MA + MI systems, Eq. (10), Qreac is the major factor contributing to the different net mass flux between MA and MA + MI systems, as QinGW is the same in both cases and DQ is small (Table 4). 3.5. Controls on microforms performance in NO 3 removal We conducted a sensitivity analysis for both MA-only and MA + MI systems to quantify how microforms influence NO 3 load to a gaining stream under different hydrogeochemical conditions. The Qn-based analysis addresses hydraulic conductivity (including homogeneous and heterogeneous distributions), GW discharge,  stream velocity, SW DOC and NHþ 4 concentration and GW NO3 concentration. All the model parameters were chosen to be within the

192

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

12

Cl- concentration (mgL-1) 6 9 12 15

15

0.2

0.2

0.2

0.2

0.4 0.6

MA MA+MI

0.4 0.6

Depth (m)

0.0

Depth (m)

0.0

Depth (m)

0.4 0.6

0.4 0.6

0.8

0.8

0.8

1.0

1.0

1.0

NO3- concentration (mgL-1)

NO3- concentration (mgL-1) 6 9 12 15

6

9

12

NO3- concentration (mgL-1)

15

6

9

12

NO3- concentration (mgL-1)

15

6 0.0

0.2

0.2

0.2

0.2

0.4 0.6

0.6 0.8

0.8

NO3

0.6

1.0 0.8

1.2

1.2

1.0

0.8

1.0

NO3-/Cl1.2

0.8 0.0

0.2

0.2

0.2

0.2

0.6

0.6

Depth (m)

0.0

Depth (m)

0.0

0.4

0.4 0.6

1.2

0.6

0.8

0.8

0.8

1.0

1.0

1.0

1.0

B

1.0

0.4

0.8

A

15

0.6

0.0

0.4

12

0.4

NO3-/Cl-

1.0

9

0.8

NO3-/Cl-

-/Cl-

1.0

0.4

0.8

1.0

1.0

0.8

0.4

Depth (m)

0.0

Depth (m)

0.0

Depth (m)

0.0

Depth (m)

Depth (m)

9

0.0

1.0

Depth (m)

6

0.0

0.8

Depth (m)

Cl- concentration (mgL-1)

Cl- concentration (mgL-1) 6 9 12 15

Cl- concentration (mgL-1) 6 9 12 15

C

D

Fig. 5. Comparison of concentrations profiles of Cl and NO 3 in MA and MA + MI bedforms. Profiles A, B, C and D are collected in the range of the macroform. Point A and D correspond to the starting and ending points (i.e. break-in-slope point) of the riffle shown in Fig. 4.

Table 4 Water flux (m3 d1) and mass flux (g d1) for MA and MA + MI systems in base case. A positive net mass flux Qn denotes net solute entry, while negative value indicates exit. Water and solutes

Qin

QinSW

QinGW

QoutSW

Qn = QinSW  QoutSW

Qreac

DQ

MA + MI case Water Cl NO 3 DOC DO NHþ 4

43.89 440.94 440.91 210.19 478.48 2.25

33.39 293.94 293.91 189.19 457.48 1.73

10.5 147 147 21 21 0.525

43.89 442.43 433.18 174.93 435.00 1.85

10.5 148.49 139.27 14.26 22.48 0.13

n 0 8.96 34.53 40.88 0.39

0 1.49 1.23 0.73 2.6 0.01

MA case Water Cl NO 3 DOC DO NHþ 4

10.60 147.89 147.89 21.52 22.13 0.53

0.1 0.89 0.89 0.52 1.13 0.005

10.5 147 147 21 21 0.525

10.60 147.89 139.10 12.28 8.04 0.31

10.5 147.00 138.21 11.76 6.91 0.305

n 0 8.79 9.24 14.09 0.22

0 0 0 0 0 0

193

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

range of values reported in the literature. The impact of microform geometry, such as obstacle height, obstacle length and riffle length on HEF exchange was studied by Käser et al. [10] for conservative transport, and so these factors are not analysed here. 3.5.1. Impact of GW discharge q For the MA case, the net mass discharge of NO 3 and the other solutes to SW increase significantly with increasing q (Fig. 6), because upwelling GW dominates the total water exchange (Table 4). The net NO 3 flux in the MA + MI case shows the same trend with that in the MA case, but microforms tend to load more NO 3 to SW under each GW discharge condition. The MA + MI system serves as a DOC and DO sink for SW with q in low levels, and switches to a source role when q is equal to 0.2 m d1. The result indicates that the shallower interfacial exchange zones are compressed by high q and reactions are thus limited. 3.5.2. Impact of hydraulic conductivity K The hydraulic conductivity, K, of sediments can affect mass transfer by regulating solute input from SW and residence time in a hyporheic zone [50]. In our simulation, both SW and GW residence time is longer with increasing K (Table 5). This result does not at first appear intuitive as one would expect residence times to be short in high permeability sediments. However, GW discharge regulates the head gradient and hyporheic pathlines in both MA and MA + MI cases, so a combination of K, head gradient, and flow path length determines the hyporheic residence time [11]. Net NO 3 flux almost remains the same with increasing K in the MA case (Fig. 7(a)), as upwelling NO 3 from GW dominates total  þ NO 3 exchange. Unlike NO3 , less DOC, DO and NH4 are loaded to SW due to the combined effect of prolonged residence time and increased mass input. In the MA + MI system, the net flux of NO 3 , DOC and DO shows a non-monotonous change with increasing K. In this system, a higher K significantly enhances SW mass

NO3-

Cl-

MA

0.8 1 1.5

GW

SW

GW

4.03 7.15 14.5

17.3 23.3 25

1.2 1.43 2.76

15.4 17.3 22.1

inflow into the sediment due to a large proportion of SW exchange. However, the increased supply of nutrients and prolonged residence time may not necessarily improve nutrients removal in the MA + MI system. For example, the net NO 3 flux to SW increases to 144 g d1 with K equal to 2  104 m s1. This is because the proportion of nutrients supply also affects NO 3 removal in sediments. Our results indicate, therefore, that sediment permeability can have a greater impact on NO 3 cycling when microforms are present, and the role of microforms depends on biogeochemical conditions and residence time in hyporheic sediment. 3.5.3. Impact of K inhomogeneity Typically the sediments at the location of a riffle are more permeable than in a pool [51]. In order to test the effect of inhomogeneous sediment on mass transfer in MA and MA + MI systems, we simply divided the model domain into two zones of different hydraulic conductivity: a relatively high K (1.2  104 m s1) in the upstream area of the riffle and a low K (0.8  104 m s1) in the downstream area. The vertical dividing line is located at the riffle tail (Point D, see Fig. 4). The results show that inhomogeneity has little impact on solutes transfer in the MA condition (Fig. 8(a)), due to a small proportion of SW exchange. By contrast, þ inhomogeneity increases delivery of NO 3 and NH4 to SW and weakens the role of sediment as a DOC and DO sink in the MA + MI

DO

q=0.8 md q=1.4 md-1 q=2.0 md-1

-120 0.0 -0.2 -0.4 -0.6

-160 -200

NH4+

Cl-

DOC

DO

0

-1

-80

MA + MI

SW

NO3-

-40

Qn (gd-1)

DOC

K (104 m s1)

K=0.8×10-4 ms-1 K=1.0×10-4 ms-1 K=1.5×10-4 ms-1

-40

Qn (gd-1)

0

Table 5 The median residence time (d) of SW and GW in MA and MA + MI conditions with different K.

-80 -0.29 -120

NH4+

-0.30 -0.31

-240

-160

(a) NO3-

Cl-

(a) DOC

DO

40

40

NO3-

Cl-

DOC

DO

0 0

-80 -120 0.2 0.0 -0.2 -0.4

-160 -200

NH4+

-240

Qn (gd-1)

Qn (gd-1)

-40

-40 0.0

-80

NH4+

-0.1 -120 -0.2 -160

(b) Fig. 6. Net mass flux (Qn = QinSW  QoutSW) of each solute in (a) MA and (b) MA + MI conditions under different GW discharges q.

(b) Fig. 7. Net mass flux (Qn = QinSW  QoutSW) of each solute in (a) MA and (b) MA + MI conditions under different hydraulic conductivities K.

194

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

system (Fig. 8(b)). This is because low permeability in the downstream area decreases the inflow of SW mass into the MA + MI system and thus limits reactions. These results indicate that neglecting microforms in inhomogeneous sediments might overestimate the NO 3 removal.

3.5.5. Impact of SW quality We simulated DOC and NHþ 4 enriched conditions respectively, allowing comparisons with the base case. No DOC limiting in the river was modelled by setting the SW DOC concentration to 16 mg L1. Similarly, a large release from the pore water pool of NHþ 4 in the upper sediments to overlying water can be much higher than 0.05 mg L1 (as in the base case) due to mineralization of dissolved organic matter (ammonification) [52]. The NHþ 4 concentration in the SW is specified as 3.85 mg L1 in this NHþ 4 -enriched scenario, while the other conditions remain the same as the base case. SW quality has little impact on net mass flux in the MA-only case, again because SW accounts for a small proportion of total exchange (Fig. 10(a) and (c)). In the MA + MI condition, microforms increase nitrification significantly with abundant NHþ 4 in the stream, and thus more NO is loaded to the SW (Fig. 10(b)). The 3 NO3-

Cl-

DOC

DO

0

Homogeneous Inhomogeneous

Qn (gd-1)

-40 -80

-0.30

NH4+

NO3-

Cl-

DOC

DO

0

Qn (gd-1)

3.5.4. Impact of stream velocity U Here we compare the net mass flux in two stream velocity conditions, i.e. 0.3 m s1 and 0.8 m s1 (base case) in the MA + MI system. Under a higher velocity condition, the MA + MI loads less þ NO 3 and NH4 to SW, and enhance its role as a DOC and DO sink (Fig. 9). The results agree with the study by Bardini et al. [31] that both nitrification and denitrification rates increase with stream velocity. The results also demonstrate that the increase of inward solute fluxes mainly contributes to the increased reaction rates while the decrease of residence times is less important in the system.

40

U=0.3 ms-1 U=0.8 ms-1

-40 -80

0.0

-120

-0.1

-160

NH4+

-0.2

Fig. 9. Net mass flux (Qn = QinSW  QoutSW) of each solute in MA + MI case under low and high stream velocity U conditions.

sink role of DOC is weakened, as higher proportion of DO is consumed by nitrification. A DOC-enriched river enhances aerobic respiration in the MA + MI case (Fig. 10(d)), so less DO is consumed by nitrification and more NHþ 4 exports to SW. Denitrification is slightly limited in this system, because more SW DOC tends to be consumed by aerobic respiration, and then by denitrification. The results imply that higher DOC availability in the stream does not necessarily enhance NO 3 reduction by denitrification when microforms are present. 3.5.6. Impact of GW quality Here we focus on the impact of microforms in an ambient (in contrast to enriched) groundwater quality condition by assigning the GW NO 3 concentration as zero, while the other conditions remain the same as the base case. Both bedforms shift from NO 3  source to NO 3 sink when there is no upwelling NO3 from GW (Fig. 11), though the net NO 3 flux is almost zero in the MA case. The results indicate that microforms may have a positive impact  on NO 3 removal under ambient groundwater NO3 condition. For both cases, more DOC exports from sediments to SW because denitrification is significantly decreased in SW–GW mixing zones. Nitrification decreases in both MA and MA + MI systems, as more DO is involved in aerobic respiration. 4. Discussion 4.1. Base case: nitrate vertical profile

-120 -0.31 -160

(a) NO3-

Cl-

DOC

DO

40

Qn (gd-1)

0 -40 -80

-0.12

NH4+

-120 -0.13 -160

(b) Fig. 8. Net mass flux (Qn = QinSW  QoutSW) of each solute in (a) MA and (b) MA + MI cases under inhomogeneous (K = 1.2  104 m s1 upstream and K = 0.8  104 m s1 downstream of riffle) and homogeneous (K = 1  104 m s1) conditions.

Within our modelling study, microforms affect the pattern of solute distribution within the subsurface. The scallop shape of concentration profiles associated with microforms on the flat streambed surrounding the riffle-pool shows that even when the macroform is absent the (upstream and downstream of the macroform) vertical profiles of nitrate (for example) will vary depending on the exact location from which the samples are taken. Thus the depths of hyporheic exchange interpreted from mixing models will reflect the position of sampling relative to the fine scale structure of the river bed. Consequently, there are important implications for positioning of pore water samplers during field studies if the depth of hyporheic mixing is of interest, or when investigating pore water profiles for the purposes of identifying biogeochemical hotspots. High-resolution solution sampling of chemical species is needed near a riffle embedded with microforms to fully characterise hyporheic water quality. 4.2. Base case: residence time effects In our model, microforms reduce overall residence time in HEF system and lead to more NO 3 load to the SW compared with

195

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

Cl-

DOC

DO

NH4+

0

-40

-40

-80

Ambient SW NH4+ enriched SW

Qn (gd-1)

Qn (gd-1)

NO30

-120

NO3-

Cl-

DO

NO3- enriched GW Ambient GW

-80 -0.28 -120

NH4+

-0.30 -0.32

-160

-160

(a)

NO3-

Cl-

(a)

DOC

DO

NH4+

40

40

0

NO3-

Cl-

DOC

DO

0

-40

Qn (gd-1)

Qn (gd-1)

DOC

-80

-40 -80

-120

-120 -160

-0.12

NH4+

-0.13 -0.14

-160

(b)

(b) 0

Cl-

NO3-

DOC

Ambient SW DOC enriched SW

-40

Qn (gd-1)

DO

-80 -0.30

NH4+

-120 -0.31 -160

(c)

NO3-

Cl-

DOC

DO

40

Qn (gd-1)

0 -40

Fig. 11. Net mass flux (Qn = QinSW  QoutSW) of each solute in (a) MA and (b) MA + MI cases in ambient groundwater quality conditions and NO 3 enriched groundwater conditions.

electron acceptor in subsequent denitrification [31,53]. More proportion of NO 3 is removed in the MA-only case, because the longer residence time allows more denitrification. From our field measurements of the River Leith, for example, we have evidence of nitrate production through mixing model analysis along with direct measurement of nitrification processes in the river bed [54]. Nitrification–denitrification coupling has been also observed in a third-order stream in western Oregon [30]; in the study of Zarnetske et al., short residence times were dominated by aerobic metabolic processes such as the rapid utilisation of DO and DOC and the production of nitrate, whereas the anaerobic process of denitrification dominated the system when residence time were beyond a threshold of 15 h, resulting in a net removal of nitrate from the stream. 4.3. Impact of microforms on mixing zone

-0.10

-80

NH4+

-0.12

-120

-0.14 -160

(d) Fig. 10. Net mass flux (Qn = QinSW  QoutSW) of each solute in (a) MA and (b) MA + MI cases under ambient stream quality and NHþ 4 -enriched stream conditions; (c) MA and (d) MA + MI cases under ambient stream quality and DOC-enriched stream conditions.

macroform-only case. Short HEF pathways caused by microforms could remain oxic along their entire length or switch from nitrate producing (nitrification) to nitrate consuming as oxygen concentrations decline. Thus, microforms can enhance nitrate production, though the effect might be minimal. The nitrification process generates NO 3 under oxic conditions which can serve as an anaerobic

In this gaining stream, the removal of upwelling NO 3 is mixingdependent. The upwelling NO 3 will mix with DOC-riched SW along mixing fringes, and subsequently be consumed by denitrification. These processes are controlled by mass flux and residence time within the mixing zone [48]. The maximum residence time in Table 3 suggests that both SW and upwelling GW spend less time in the mixing zone of the MA + MI case than MA case. Thus, microforms can reduce both GW NO 3 removal efficiency in the SW–GW mixing zone and SW NO removal efficiency in the mixing zone 3 where flow paths leave and return to SW. 4.4. Impact of GW quality on nutrient flux Although it may be expected that increased concentrations of nitrate in GW results in enhanced nitrate removal by denitrification in the streambed and hence a reduction in nitrate flux to

196

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197

SW, our model results indicates that the NO 3 removal can actually decrease as GW nitrate concentrations increases. This is because other important reactants that drive denitrification (such as DOC) are absent. In reality DOC will be present within GW, but may be of poorer quality than the DOC in the SW and thus may not be available for denitrifiers [55] – so effectively labile DOC may be low in the GW, as modelled here. As noted by Heppell et al. [37] in their study of the River Leith, a localised zone of preferential groundwater discharge (20% of their study reach length) exists, where GW discharge is relatively high and nitrate concentrations are correspondingly high (possibly from regional GW). The modelling results here suggests that under such conditions accounting for microforms within macroforms will make little difference to the relative mass flux of nitrate from GW to SW, however, at lower GW nitrate concentrations and GW discharge, the effect of microforms on denitrification should be taken into account. Our modelling considers the hyporheic exchange induced by a macroform and a microforms-embedded macroform by a simple superposition of head variation on a flat bed. The drop of water head in the model may be larger than that generated by incorporating the same-size bed elevation change in the model domain’s geometry. Besides, inclusion of microforms in a macroform may result in more complex head variation than given by the simple superposition of a linear head drop and a sinusoidal head variation.

5. Conclusions We conducted numerical modelling of nitrate transport in the hyporheic zone induced by a macroform only and microformsembedded macroform in a gaining stream. Both nitrification and denitrification were included in the model. We have demonstrated that microforms can regulate the vertical concentration profiles of solutes by its role in enhancing the water exchange at the groundwater–surface water interface and reducing overall residence time of HEF in a riffle-pool sequence. Even when the macroform is absent, the concentration profiles of solutes vary depending on the exact location of sampling. Thus, the depths of hyporheic exchange interpreted from mixing models will reflect the position of sampling relative to the fine scale structure of the river bed. High-resolution solution sampling of chemical species is needed in a microforms-embedded riffle to fully characterise hyporheic water quality. Short HEF pathways caused by microforms could remain oxic along their entire length or switch from nitrate producing (nitrification) to nitrate consuming as oxygen concentrations decline. Thus, microforms can enhance nitrate production. Microforms can reduce both the proportion of GW NO 3 removal in the SW– GW mixing zone and the proportion of SW NO 3 removal in the mixing zone where flow paths leave and return to SW. Ignoring microforms in the river bed, therefore, may lead to over-predictions of the effects of HEF on nitrate removal. Nutrient cycling in the presence of microforms appears more sensitive to sediment permeability than in a macroform-only system, and the overall effect is an interaction of hydrodynamic and biochemical factors. Neglecting microforms in a heterogeneous sediment may overestimate NO 3 removal. Under various SW and GW quality conditions, microforms affect net mass flux mainly by regulating reactions with different inward solute fluxes. Microforms tend to enhance both nitrification and denitrification under a high stream velocity condition. Our results suggest that microforms affect net NO 3 flux by the combined effect of introducing more stream mass flux and reducing their residence time in mixing zones.

Acknowledgements The study was supported by China Scholarship Council (CSC) funding for the first author to study in Lancaster University from October 2011 to October 2012. Data from the River Leith were made available through the UK Natural Environment Research Council (NERC) grant NE/F006063/1. The microtopography survey in the River Leith was completed by Patrick Byrne, working on the above grant. The authors would like to thank three anonymous reviewers, who provided insightful comments and greatly improved the quality of the manuscript. References [1] Boulton AJ, Findlay S, Marmonier P, Stanley EH, Valett HM. The functional significance of the hyporheic zone in streams and rivers. Annu Rev Ecol Syst 1998;29:59–81. http://dx.doi.org/10.1146/annurev.ecolsys.29.1.59. [2] Gu C, Hornberger GM, Herman JS, Mills AL. Influence of stream–groundwater interactions in the streambed sediments on NO 3 flux to a low-relief coastal stream. Water Resour Res 2008;44:W11432. http://dx.doi.org/10.1029/ 2007WR006739. [3] Hill AR, Labadia CF, Sanmugadas K. Hyporheic zone hydrology and nitrogen dynamics in relation to the streambed topography of a N-rich stream. Biogeochemistry 1998;42:285–310. http://dx.doi.org/10.1023/ A:1005932528748. [4] Marzadri A, Tonin D, Bellin A. Morphodynamic controls on redox conditions and on nitrogen dynamics within the hyporheic zone: application to gravel bed rivers with alternate-bar morphology. J Geophys Res 2012. http://dx.doi.org/ 10.1029/2012JG001966. [5] Packman AI, Brook NH, Morgan J. A physicochemical model for colloid exchange between a stream and a sand streambed with bed forms. Water Resour Res 2000;36(8):2351–61. http://dx.doi.org/10.1029/2000WR900059. [6] Harvey JW, Bencala K. The effect of streambed topography on surfacesubsurface water exchange in mountain catchments. Water Resour Res 1993;29(1):89–98. http://dx.doi.org/10.1029/92WR01960. [7] Harvey JW, Wagner BJ. Quantifying hydrologic interactions between streams and their subsurface hyporheic zones. In: Jones JB, Mulholland PJ, editors. Streams and ground waters. San Diego, CA: Academic Press; 2000. p. 3–44. [8] Stonedahl SH, Harvey JW, Wörman A, Salehin M, Packman AI. A multiscale model for integrating hyporheic exchange from ripples to meanders. Water Resour Res 2010;46(12):W12539. http://dx.doi.org/10.1029/2009WR008865. [9] Duff JH, Triska FJ. Nitrogen biogeochemistry and surface-subsurface exchange in streams. In: Jones JB, Mulholland PJ, editors. Streams and ground waters. San Diego, CA: Academic Press; 2000. p. 197–220. [10] Käser DH, Binley A, Heathwaite AL. On the importance of considering channel microforms in groundwater models of hyporheic exchange. River Res Appl 2013;29(4):528–35. http://dx.doi.org/10.1002/rra.1618. [11] Kasahara T, Wondzell SM. Geomorphic controls on hyporheic exchange flow in mountain streams. Water Resour Res 2003;39(1):1–14. http://dx.doi.org/ 10.1029/2002WR001386. [12] Packman AI, Bencala KE. Modeling methods in the study of surface-subsurface hydrologic interactions. In: Jones JB, Mulholland PJ, editors. Streams and ground waters. San Diego, CA: Academic Press; 2000. p. 45–80. [13] Tonina D, Buffington JM. Hyporheic exchange in gravel bed rivers with poolriffle morphology: laboratory experiments and three-dimensional modeling. Water Resour Res 2007;43(1):W01421. http://dx.doi.org/10.1029/ 2005WR004328. [14] Harbaugh AW, McDonald MG. User’s documentation for MODFLOW-96, an update to the US Geological Survey modular finite difference ground-water flow model. US Geological Survey Open-File Report 96–485, US Geological Survey, Denver, CO; 1996. [15] Pollock DW. User’s guide for MODPATH/MODPATH-PLOT, version 3: a particle tracking post-processing package for MODFLOW, the US Geological Survey finite-difference ground-water flow model. US Geological Survey Open-File Report, US Geological Survey, Denver, CO; 1994. [16] Gooseff MN, Anderson JK, Wondzell SM, LaNier J, Haggerty R. A modelling study of hyporheic exchange pattern and sequence, size and spacing of stream bedforms in mountain stream networks, Oregon, USA. Hydrol Process 2006;20:2443–57. http://dx.doi.org/10.1002/hyp.6349. [17] Zheng C, Wang PP. MT3DMS: a modular three-dimensional multispecies transport model for simulation in groundwater systems, documentation and user’s guide. US Army Corps of Eng., Washington, DC; 1999. [18] Saenger N, Kitanidis PK, Street RL. A numerical study of surface–subsurface exchange processes at a riffle-pool pair in the Lahn River, Germany. Water Resour Res 2005;41:W12424. http://dx.doi.org/10.1029/2004WR003875. [19] Lautz LK, Siegel DI. Modeling surface and ground water mixing in the hyporheic zone using MODFLOW and MT3D. Adv Water Resour 2006;29(11):1618–33. http://dx.doi.org/10.1016/j.advwatres.2005.12.003. [20] Zheng C. MT3D: a modular three-dimensional transport model for simulation of advection, dispersion and chemical reactions of contaminants in groundwater systems. Report to the US Environmental Protection Agency; 1990.

H. Hu et al. / Advances in Water Resources 73 (2014) 185–197 [21] Jin G, Tang H, Gibbes B, Li L, Barry DA. Transport of nonsorbing solutes in a streambed with periodic bedforms. Adv Water Resour 2010;33:1402–16. http://dx.doi.org/10.1016/j.advwatres.2010.09.003. [22] Vollmer S, de los Santos Ramos F, Daebel H, Kühn G. Micro scale exchange processes between surface and subsurface water. J Hydrol 2002;269(1–2): 3–10. http://dx.doi.org/10.1016/S0022-1694(02)00190-7. [23] Boano F, Demaria A, Revelli R, Ridolfi L. Biogeochemical zonation due to intrameander hyporheic flow. Water Resour Res 2010;46:W02511. http:// dx.doi.org/10.1029/2008WR007583. [24] Curie F, Ducharne A, Sebilo M, Bendjoudi H. Denitrification in a hyporheic riparian zone controlled by river regulation in the Seine river basin (France). Hydrol Process 2009;23:655–64. http://dx.doi.org/10.1002/hyp.7161. [25] Shum KT. The effects of wave-induced pore water circulation on the transport of reactive solutes below a rippled sediment bed. J Geophys Res 1993;98(C6):10289–301. http://dx.doi.org/10.1029/93JC00787. [26] Seitzinger S, Harrison JA, Böhlke JK, Bouwman AF, Lowrance R, Peterson B, Tobias C, Van Drecht G. Denitrification across landscapes and waterscapes: a synthesis. Ecol Appl 2006;16(6):2064–90. http://dx.doi.org/10.1890/10510761(2006)016[2064:dalawa]2.0.co;2. [27] Triska FJ, Duff JH, Avanzino RJ. The role of water exchange between a stream channel and its hyporheic zone in nitrogen cycling at the terrestrial aquatic interface. Hydrobiologia 1993;251(1–3):67–184. http://dx.doi.org/10.1007/ BF00007177. [28] Zarnetske JP, Haggerty R, Wondzell SM, Baker MA. Labile dissolved organic carbon supply limits hyporheic denitrification. J Geophys Res 2011;116(G4):G04036. http://dx.doi.org/10.1029/2011JG001730. [29] Pinay G, O’Keefe TC, Edwards RT, Naiman R. Nitrate removal in the hyporheic zone of salmon river in Alaska. River Res Appl 2009;25:367–75. http:// dx.doi.org/10.1002/rra.1164. [30] Zarnetske JP, Haggerty R, Wondzell SM, Baker MA. Dynamics of nitrate production and removal as a function of residence time in the hyporheic zone. J Geophys Res 2011;116:G01025. http://dx.doi.org/10.1029/2010JG001356. [31] Bardini L, Boano F, Cardenas MB, Revelli R, Ridolfi L. Nutrient cycling in bedform induced hyporheic zones. Geochim Cosmochim Ac 2012;84:47–61. http://dx.doi.org/10.1016/j.gca.2012.01.025. [32] Marzadri A, Tonina D, Bellin A. A semianalytical three-dimensional processbased model for hyporheic nitrogen dynamics in gravel bed rivers. Water Resour Res 2011;47(11):W11518. http://dx.doi.org/10.1029/2011WR010583. [33] Sheibley RW, Jackman AP, Duff JH, Triska FJ. Numerical modeling of coupled nitrification-denitrification in sediment perfusion cores from the hyporheic zone of the Shigobee River, MN. Adv Water Resour 2003;26(9):977–87. http:// dx.doi.org/10.1016/S0309-1708(03)00088-5. [34] Zarnetske JP, Haggerty R, Wondzell SM, Bokil VA, González-Pinzón R. Coupled transport and reaction kinetics control the nitrate source-sink function of hyporheic zones. Water Resour Res 2012;46:W11508. http://dx.doi.org/ 10.1029/2012WR011894. [35] Maier HS, Howard KW. Influence of oscillating flow on hyporheic zone development. Ground Water 2011;49(6):830–44. http://dx.doi.org/10.1111/ j.1745-6584.2010.00794.x. [36] Binley A, Ullah S, Heathwaite AL, Heppell CM, Byrne P, Lansdown K, Trimmer M, Zhang H. Revealing the spatial variability of water fluxes at the groundwater–surface water interface. Water Resour Res 2013;49:1–15. http://dx.doi.org/10.1002/wrcr.20214. [37] Heppell C, Heathwaite AL, Binley A, Byrne P, Ullah S, Lansdown K, Keenan P, Trimmer M, Zhang H. Interpreting spatial patterns in redox and coupled water-nitrogen fluxes in the streambed of a gaining river reach. Biogeochemistry 2013. http://dx.doi.org/10.1007/s10533-013-9895-4. [38] Lee M-S, Lee K-K, Hyun Y, Clement TP, Hamilton D. Nitrogen transformation and transport modeling in groundwater aquifers. Ecol Model 2006;192(1–2): 143–59. http://dx.doi.org/10.1016/j.ecolmodel.2005.07.013.

197

[39] Kinzelbach W, Schäfer W, Herzer J. Numerical modeling of natural and enhanced denitrification processes in aquifers. Water Resour Res 1991;27(6):1123–35. http://dx.doi.org/10.1029/91WR00474. [40] Baker MA, Vervier P. Hydrological variability, organic matter supply and denitrification in the Garonne River Ecosystems. Freshwater Biol 2000;49:181–90. http://dx.doi.org/10.1046/j.1365-2426.2003.01175.x. [41] Widdowson MA, Molz FJ, Benefield LD. A numerical transport model for oxygen- and nitrate-based respiration linked to substrate and nutrient availability in porous media. Water Resour Res 1988;24(9):1553–65. http:// dx.doi.org/10.1029/WR024i009p01553. [42] Böhlke JK, Denver JM. Combined use of groundwater dating, chemical, and isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural watersheds, Atlantic coastal plain, Maryland. Water Resour Res 1995;31(9):2319–39. http://dx.doi.org/10.1029/95WR01584. [43] Beaulieu JJ, Tank JL, Hamilton SK, Wollheim WM, Hall RO, Mulholland PJ, et al. Nitrous oxide emission from denitrification in stream and river networks. Proc Natl Acad Sci USA 2010;108:214–9. http://dx.doi.org/10.1073/pnas.1011464108. [44] Gu C, Hornberger GM, Mills AL, Herman JS, Flewelling SA. Nitrate reduction in streambed sediments: effects of flow and biogeochemical kinetics. Water Resour Res 2007;43:W12413. http://dx.doi.org/10.1029/2007WR006027. [45] MacQuarrie KT, Sudicky EA, Robertson WD. Numerical simulation of a finegrained denitrification layer for removing septic system nitrate from shallow groundwater. J Contam Hydrol 2001;52:29–55. http://dx.doi.org/10.1016/ S0169-7722(01)00152-8. [46] Elliott AH, Brooks NH. Transfer of nonsorbing solutes to a streambed with bed forms: theory. Water Resour Res 1997;33(1):123–36. http://dx.doi.org/ 10.1029/96WR02784. [47] Fehlman HM. Resistance components and velocity distributions of open channel flows over bedforms [MSc thesis]. Colorado State University: Fort Collins; 1985. [48] Hester ET, Young KI, Widdowson MA. Mixing of surface and groundwater induced by riverbed dunes: Implications for hyporheic zone definitions and pollutant reactions. Water Resour Res 2013;49(9):5221–37. http://dx.doi.org/ 10.1002/wrcr.20399. [49] Briggs MA, Lautz LK, Hare DK. Residence time control on hot moments of net nitrate production and uptake in the hyporheic zone. Hydrol Process 2014;28(11):3741–51. http://dx.doi.org/10.1002/hyp.9921. [50] Grimaldi C, Chaplot V. Nitrate depletion during within-stream transport: effects of exchange processes between streamwater, the hyporheic and riparian zones. Water Air Soil Poll 2000;124:95–112. http://dx.doi.org/ 10.1023/A:1005222513626. [51] Käser DH, Binley A, Heathwaite AL, Krause S. Spatio-temporal variations of hyporheic flow in a riffle-step-pool sequence. Hydrol Process 2009;23:2138–49. http://dx.doi.org/10.1002/hyp.7317. [52] Shibata H, Sugawara O, Toyoshima H, Wondzell SM, Nakamura F, Kasahara T, et al. Nitrogen dynamics in the hyporheic zone of a forested stream during a small storm, Hokkaido, Japan. Biogeochemistry 2004;69:83–104. http:// dx.doi.org/10.1023/B:BIOG.0000031042.90585.0a. [53] Cardenas MB, Cook PLM, Jiang H, Traykovski P. Constraining denitrification in permeable wave-influenced marine sediment using linked hydrodynamic and biogeochemical modelling. Earth Planet Sci Lett 2008;275:127–37. http:// dx.doi.org/10.1016/j.epsl.2008.08.016. [54] Lansdown K, Heppell CM, Dossena M, Ullah S, Heathwaite AL, Binley A, Zhang H, et al. Fine-scale in situ measurement of riverbed nitrate production and consumption in an armored permeable riverbed. Environ Sci Technol 2014;48(8):4425–34. http://dx.doi.org/10.1021/es4056005. [55] Lansdown K, Trimmer M, Heppell CM, Sgouridis F, Ullah S, Heathwaite AL. Characterization of the key pathways of dissimilatory nitrate reduction and their response to complex organic substrates in hyporheic sediments. Limnol Oceanogr 2012;57(2):387–400. http://dx.doi.org/10.4319/lo.2012.57.2.0387.