Heterogeneity in Permeability and Particulate Organic Carbon Content Controls the Redox Condition of Riverbed Sediments at Different Timescales

The hydrological and biogeochemical properties of the hyporheic zone in stream and riverine ecosystems have been extensively studied over the past two decades. Although it is widely acknowledged that sediment heterogeneity can influence biogeochemical reactions, little effort has been made to understand the role of heterogeneity on the spatiotemporal variability of riverbed redox conditions under changing flow dynamics at different timescales. Here we integrate a mechanistic model and field data to demonstrate that heterogeneity in permeability plays a vital role in modulating sediment redox conditions at both seasonal (annual) and event (daily‐to‐weekly) timescales, whereas heterogeneity in particulate organic carbon (POC) content only has a comparable influence on redox conditions at the seasonal timescale. These findings underscore the importance of accurately characterizing sediment heterogeneity, in terms of permeability and POC content, in quantifying biogeochemical dynamics in the riverbed and hyporheic zones of riverine ecosystems.


Introduction
Hyporheic zone (HZ) science has progressed rapidly over the past two decades with the advancements in both modeling tools and field observation techniques (Boano et al., 2014;Cardenas, 2015;Hammett et al., 2022;Krause et al., 2022).Such advances allow for an increasingly mechanistic understanding of the HZ from hydrologic and biogeochemical perspectives.Early studies conceptualized HZ as a homogeneous reactor that receives nutrients from flowing surface water, which led to the conclusion that residence time can be used as an indicator of biogeochemical reaction such as denitrification (Cardenas et al., 2008).Later on, the potential influence of sediment heterogeneity on the formation of biogeochemical hot spots was identified by analyzing the distribution of residence time in HZ (Sawyer & Cardenas, 2009).More recently, the crucial role of sediment particulate organic carbon (POC) as a source of electron donors to sustain biogeochemical reactions has been recognized (Corson-Rikert et al., 2016;Trimmer et al., 2012).Currently, there is a consensus that, for better understanding HZ, the coupled hydrologic and biogeochemical processes need to be integrated with a realistic conceptualization of physiochemical properties of HZ across scales (Krause et al., 2022).
While the influence of heterogeneity in sediment permeability on HZ flow and residence time has been extensively studied, the combined influence of heterogeneity in both permeability and POC on biogeochemical cycling under dynamic flow conditions is less understood.The consideration of heterogeneity in sediment POC may lead • A reactive transport model was developed to quantify the impact of heterogeneity in permeability and particulate organic carbon (POC) concentration on sediment redox conditions • Heterogeneity in permeability controls sediment redox conditions at both seasonal (annual) and event (daily-toweekly) timescales • The effects of heterogeneity in POC occur over the monthly timescale, reflecting a balance between POC metabolism and the influx of oxygen

Supporting Information:
Supporting Information may be found in the online version of this article.
to conclusions that differ from earlier studies.For example, Sawyer (2015) incorporated variations in POC content while evaluating the denitrification capability of heterogeneous sediments by assigning high and low organic carbon contents to silt and sand hydrofacies, respectively.This more realistic conceptualization for HZ sediments led to the conclusion that sediment texture-induced heterogeneity in POC enhanced denitrification by creating small, dissolved organic carbon (DOC)-rich anoxic zones that would not otherwise exist in a well-mixed system.In contrast, models that consider heterogeneity in permeability but not POC have concluded that heterogeneity had little influence on the nutrient cycling (Bardini et al., 2013).In addition to sediment textureinduced heterogeneity in POC, the abundance of POC also often exhibits declining trend with depth in relatively homogeneous sediments, especially at shallow depths in the HZ (Brunke, 1999;Krause et al., 2009;Roden et al., 2023;Trimmer et al., 2012).This is because suspended small-size POC above the sediment-water interface (SWI) can naturally infiltrate into the sediments with advective transport (Brunke, 1999;Roden et al., 2023).Incorporating both textured-induced and infiltration-induced heterogeneity in POC may provide new insights into the complex biogeochemical behavior occurring in heterogeneous riverbed sediments under dynamic flow conditions.
The spatiotemporal pattern of redox conditions in riverbed sediments is an extremely important but understudied problem.Lack of techniques for continuous monitoring of redox-sensitive species in sediments at different depths and time scales poses great challenges for understanding biogeochemical cycling under dynamic flow conditions.Various studies have contradictory conceptualization of the redox boundary conditions of the HZ.For example, both oxic (Newcomer et al., 2018) and anoxic (Knights et al., 2017;Liu et al., 2017;Ping et al., 2023;Winnick, 2021) conditions have been assumed for the sediment lower boundary.Field studies indicate that both situations may be reasonable, depending on the observation depth (Krause et al., 2009).However, for modeling studies, the choice of oxic or anoxic bottom boundary condition plays a vital role as it directly influences whether, and to what degree, redox-sensitive reactions (e.g., denitrification) may occur.So far, the choice seems to be ad hoc, depending on the study purpose.The influence of bottom boundary dissolved oxygen (DO) concentration on HZ sediment redox condition needs to be further evaluated in a quantitative manner.
In this study, we employed a reactive transport model to address two key questions related to the above discussion: (a) How does combined heterogeneity in both sediment permeability and POC content affect sediment redox conditions under dynamic flow condition at different timescales?(b) To what extent does the assumption of oxic or anoxic condition for the bottom boundary affect simulated redox condition in sediments.We are particularly interested in the seasonal (annual) timescale which reflects the combined influence of climate change and hydrologic cycle on river discharge, and the event (daily-to-weekly) timescale which is driven by variations in precipitation as well as anthropogenic manipulation of river stage in the case of regulated rivers.A comprehensive biogeochemical model was developed to explicitly simulate redox processes occurring in the heterogeneous riverbed sediments, which yields new insights into how the physical and geochemical heterogeneity may affect the redox condition compared with the surrogate modeling approaches (e.g., travel-time).The magnitude and distribution of redox reactions that may occur in sediments (e.g., aerobic respiration, NO 3 /Fe(III)/SO 4 2 reduction, methanogenesis) are constrained based on field aqueous and geochemical data.We demonstrate that heterogeneity in permeability plays a pivotal role in controlling sediment redox conditions at both the seasonal and event timescales, whereas the heterogeneity in POC only has a comparable influence on redox condition at the seasonal timescale.In addition, redox condition at shallow depths where POC is abundant is controlled mainly by heterogeneity in permeability and POC, and is not strongly affected by the concentration of DO in the upwelling groundwater at the seasonal timescale.

Sediment Biogeochemical Model
Riverbed sediments exhibit complex physical and biogeochemical dynamics due to the frequent mixing of surface water and groundwater.Hence, the sediment model in fluvial environments like rivers and streams is distinct from the widely used marine sediment model, or early diagenesis model (Wang & Van Cappellen, 1996), in a few key aspects.First, the model needs to consider dynamic groundwater flow, where variably saturated flow is optional, depending on whether riverbeds are exposed to the air during the period of low river stage.Second, coupled simulation of groundwater flow and heat transport is necessary because diel and seasonal variation of sediment temperature affect both rates of microbial metabolism as well as water viscosity, and water viscosity could influence permeability-controlled fluid flux in turn.Third, because the geochemical properties of groundwater in riparian aquifer often varies significantly (e.g., depending on the vegetation, lithology, and hydrogeological conditions), care must be taken in specifying the chemical composition of upwelling groundwater.
A two-dimensional sediment model that couples hydrology, heat and reactive transport was developed to investigate redox dynamics in riverbed sediments under dynamic flow condition using the field data collected at the 300 Area in Hanford reach of Columbia River.The model domain is 2 × 0.2 m in vertical and horizontal directions, respectively, with the top representing the sediment-water interface.The grid dimension in both directions is 0.01 m to capture centimeter-scale heterogeneity.Saturated groundwater flow is driven by measured vertical hydraulic gradients from piezometers installed at different depths in sediments.Heat transport is constrained by the measured river temperature and groundwater temperature, respectively.The biogeochemical reaction network is adapted from a previous study where a microbial population-based reaction framework was employed to simulate microbial redox metabolism (Chen, Chen, et al., 2023;Chen, Yang, et al., 2023).There are ten microbial functional groups in the model, including aerobic respiration, denitrification, iron reduction, sulfate reduction, methanogenesis and methane oxidation.For each individual process, there may exist multiple pathways with respect to different electron donors.For example, the three aerobic populations (or facultative denitrifiers) can consume glucose, acetate and hydrogen using O 2 and NO 3 as the electron acceptor under aerobic condition and anaerobic condition, respectively.For iron reduction, sulfate reduction and methanogenesis, acetate and hydrogen produced by fermentation are utilized as electron donors.Details of the reaction network and kinetics along with model settings are included in Supporting Information S1 (Tables S1-S4).The main difference between the previous modeling work and the current work is that the sediment was conceptualized as a homogeneous reactor in one dimension previously.This assumption cannot explain the observed vertical heterogeneity of porewater CH 4 in sediments.Hence, we developed the 2D heterogeneous model to further investigate how heterogeneity can shape the sediment redox condition.

Representation of Heterogeneity in Permeability and Organic Carbon
Heterogeneity in sediment permeability is represented by the binary composition of low-permeable silt and highpermeable sand, of which the silt content increases from 10% to 90% in increments of 10%.Ten stochastic realizations were generated for each silt fraction using TPROGS-a Markov Chain based geostatistical model (Carle, 1999).For different realizations, centimeter-scale silt or sand lenses were randomly distributed, mimicking the highly heterogeneous fluvial sedimentation environments (Figure S1 in Supporting Information S1).The permeability of silt and sand were set to 10 13 and 10 11 m 2 , respectively, which are typical values for low-and high-permeable riverbed sediments, respectively (Hatch et al., 2010).The porosity is set to be identical for both silt and sand using the calibrated value 43% (Song et al., 2018).
Two types of heterogeneity in organic carbon are considered.One is the infiltration-induced decrease of POC concentration with depth (Brunke (1999); see further discussion below), and the other is a depth-dependent decrease combined with the texture-induced POC concentrations (typical of more heterogeneous deposits, e.g., Krause et al. (2009)).For infiltration-induced POC heterogeneity, the concentration of POC is assumed to decrease exponentially with depth, where C POC | z=0 is the POC concentration at the SWI, z is the depth below the SWI in unit of centimeter and n is an adjustable parameter.The exponential decrease in filtered particle concentration is predicted from colloid filtration theory (Yao et al., 1971;Harvey & Garabedian, 1991), and is consistent with recent field studies of POM infiltration into Hanford 300 Area sediments (Roden et al., 2023).In addition, this relationship has been used successfully in wetland studies where extensive field observations were used for validation, and has also been employed in other modeling studies of riverbed sediment metabolism (Chen, Chen, et al., 2023;Walter & Heimann, 2000).Five different cases of infiltration-induced decrease of POC are considered with the empirical parameter n in Equation 1 increasing from 5 to 100.Larger n indicates POC decreases more slowly with depth.
The POC concentration at the SWI was assumed to be constant as 0.6% for all realizations; thus, variations in the depth distribution of POC are assumed to be the result of differences in the efficiency of POC-enriched particle retention.For texture-induced POC heterogeneity, the POC concentration of silt was fixed at 0.5% by weight, such that heterogeneity in silt content translated directly into heterogeneity in POC content.This assumption is consistent with the observed correlation between sediment POC and % silt + clay content of Hanford 300 Area sediments (Chen, Chen, et al., 2023;Hou et al., 2017), where POC concentrations ranged from 0.2% to 1.2% for silt + clay contents ranging from 20% to 50%, with an average value of 0.54 ± 0.32%.

Types of Bottom Boundary Conditions
The above-mentioned variations in sediment permeability and POC content yielded a total of 900 simulations (9 (silt fractions) × 10 (stochastic permeability realizations) × 5 (infiltration-induced POC heterogeneity indicated by n) × 2 (infiltration-induced only + infiltration-induced combined with texture-induced POC heterogeneity)).
The bottom boundary condition was set to either oxic (0.2 P air O 2 ) or anoxic (0 μM O 2 ) for each of these simulations, resulting in an overall total of 1,800 simulations.Simulated annual mean O 2 and CH 4 concentrations were used as an indicator of the impact of the oxic versus anoxic lower boundary conditions on sediment redox conditions.The high DO concentration at the bottom boundary is based on the fact that the local Hanford 300 Area shallow groundwater, which is in direct hydrological exchange with HZ fluids, has been inferred to be welloxygenated (Ahmed et al., 2012).Recent measurements of deep HZ (200 cm) DO using a novel fluid pumping and in-line optical sensor system at the Hanford 300 Area study site have confirmed that this is the case (Kaufman et al., 2022).Furthermore, direct in situ measurements of sediment DO concentration using optical sensors embedded directly in the sediment indicate that rapid upward flux of oxic deep HZ fluids leads an increase in shallow (20 cm) alluvial riverbed sediment DO concentration after periods of O 2 depletion during times of relatively low fluid flow rate (Roden et al., 2023).All of these observations are consistent with the presence of oxic groundwater at the base of the 2-m thick alluvial sediment layer depicted in our model, and with the assumption of an oxic lower boundary condition in previous reactive transport simulations of Hanford 300 Area alluvial sediment DOC metabolism (Song et al., 2020).This observation contrasts with the lake or ocean sediments where DO is typically depleted at depth.Penetration of DO throughout the alluvial sediment fluid flow can be explained by both downward flow of aerobic river water and upward flow of aerobic groundwater through high hydraulic conductivity zones within the sediment column.

Influence of Textural Heterogeneity on Physical Transport
Vertical hydraulic gradients dictate flow direction and the relative magnitudes of fluid fluxes for a stochastically generated permeability field (Figure 1a).In the Hanford Reach of the Columbia River, flow conditions are highly dynamic due to seasonal patterns of runoff and to dam control of riverflow in the upstream, which create a unique fluvial environment for investigating complex patterns of redox change in sediments (Chen et al., 2022;Chen, Yang, et al., 2023;Zachara et al., 2020).The flooding season is from April to June during which time the river stage is high and the hydrologic exchange flow is dominated by downwelling.In other seasons, flow direction oscillates between upwelling and downwelling as a result of rainfall patterns as well as daily hydroelectric power generation, indicating a frequent mixing between the river water and groundwater.The permeability fields with low silt content (10%) represent the condition of silt lenses in sandy riverbeds (Figure 1b).Similarly, high silt content (90%) indicates the existence of sand lenses in the silty riverbeds (Figure 1f).Such conceptualization for sediment heterogeneity is in line with the field observations that riverbed sediments have distinct hydraulic conductivities that may vary in space and time (Hatch et al., 2010;Hou et al., 2019;Krause et al., 2009).Although the Hanford site itself is more representative of the former situation, the goal here is to gain general insight into how heterogeneous permeability fields are likely to influence sediment redox conditions.It is found that annual mean fluid flux decreases with increasing silt content, though different realizations for a certain silt content may introduce some uncertainty for the calculated mean flux (Figure 1i).The conservative (no consumption) river tracer and groundwater tracer can penetrate through the sediments when silt fractions are low, which means the river water and groundwater are indeed well mixed under dynamic flow condition at Hanford site (Figures S2 and S3 in Supporting Information S1).The mixing ratio for surface water may vary depending on heterogeneity and flow direction.

Influence of Heterogeneity in POC Content on DO Dynamics
Infiltration-induced POC heterogeneity is assumed to decrease exponentially with depth (Figure 1g).For n = 5 and 25, the POC content at the depth of ∼0.2 and ∼1 m decreases to less than 1% of that at the SWI, respectively.The combined manifestation of both texture-induced and infiltration-induced heterogeneity in POC is illustrated in Figure 1h, where the randomly distributed low-permeable silts with high organic content is expected to create a complex spatial pattern of POC enriched zones (Figure S4 in Supporting Information S1).Consistent with this expectation, the abundance of DO is strongly associated with the silt content of the sediments during flooding events (Figures 2a-2f).When the upwelling groundwater is assumed to be saturated with DO, sediments can be  resupplied with oxygen during both downwelling and upwelling (Figure 2a).Nevertheless, when the silt content is greater than 30%, most of the sediment column is rendered anoxic (Figures 2d-2f).
For the 10% silt case (silt lenses in sandy sediments), an anoxic zone is observed at the location where relatively large silt lenses exist even though the bulk sediment is saturated with DO and no textured-induced high POC content is considered (Figure 2b).This is because the transport of DO through the low-permeable silt is dominated by diffusion and DO supply to the lenses is not able to compensate for DO consumption by aerobic respiration in the event time scale.This mechanism can be illustrated with the spatial distribution of Damkohler (Da) number which is the ratio of transport timescale to DO consumption timescale (Figure S5 in Supporting Information S1).
It is noteworthy that the anoxic zone may disappear if the river stage is maintained at the peak level because the DO consumption timescale is still higher than the transport timescale (Da < 1).However, the flooding inferred from the hydrological conditions at the Hanford 300 Area typically receded soon after the peak, and hence a constant anoxic zone was maintained in low-permeable silts.
These results demonstrate the potential for anoxic conditions to arise at the event timescale, especially in small rivers and streams where the summer flooding is mild.The penetration depth of the DO plume strongly depends on the high-permeability (sand) preferential flow path.The existence of silt lens across the sediment may limit DO in the region above it (Figures 2c-2f).Within the anoxic zone a series of anaerobic reactions takes place depending on silt/POC content and the availability of TEAs (Figures 2g-2k).Denitrification is the dominant anaerobic pathway, followed by the progressive onset of iron reduction, sulfate reduction and methanogenesis.
The simulated presence of anoxic zones in otherwise bulk aerobic Hanford 300 Area sediments is consistent with previous observations of both geochemical and microbiological indicators of anerobic microbial metabolism in materials recovered by freeze coring (Moser et al., 2003).
In the absence of texture-induced heterogeneity, the abundance of POC has a more continuous depth-dependent influence on the DO distribution, where a more gradual decline in POC abundance with depth (increasing value of n in Equation 1) leads to a shallower depth of DO penetration (Figures 2i-2p).The consideration of textureinduced heterogeneity in POC does not markedly alter DO patterns at the event timescale, except that organicrich silts becomes more reduced as metabolism transitions from denitrification to methanogenesis (Figure S6 in Supporting Information S1).Babey et al. (2022) recently pointed out that low-permeable and organic-rich silts can facilitate the formation of a reduction zone at the downgradient side by releasing large amounts of DOC in alluvial aquifer sediments.This effect is not observed in our models due to the frequent changes of flow direction in the HZ.
Temporal change of DO in sediments is directly controlled by flow conditions (Figures S7 and S8 in Supporting Information S1).Downwelling flow is the universal way to bring DO into the sediments considering that the oxygen content in upwelling groundwater is site-specific and could vary to a large degree.Although DO in river water can vary, depending on river temperature and light (Zhi et al., 2023), in general river waters are welloxygenated as a result of being in direct contact with atmospheric O 2 .The oxygen front represents the interface between the aerobic respiration zone and the anaerobic respiration zone (Figure S9 in Supporting Information S1).The simulated annual dynamics of DO implies that using a single or a few snapshots of vertical porewater solute concentration to infer the biogeochemical processes, as commonly done in numerous studies, may lead to biased conclusions in fluvial environments.The simulated alternation between oxic and anoxic condition is also confirmed by in situ measurements of sediment dissolved O 2 using novel optic sensors deployed directly in Hanford 300 Area riverbed sediments (Roden et al., 2023).Porewater DOC concentrations measured at a certain depth may not necessarily reflect the POC abundance at that depth in situations where mobile DOC can be redistributed during advective/dispersive transport (Figures S10a in Supporting Information S1).As the silt content increases, DOC concentration at a given depth could serve as a better indicator of the in situ POC abundance for that depth (Figures S10b-S10d in Supporting Information S1) due to restricted solute transport.

Influence of Heterogeneity in Permeability and POC Content on Overall Sediment Redox Conditions
If DO and dissolved methane (DM) are used as end-member indicators of redox conditions in the sediments, it can be observed that the POC and silt content have comparable influences on redox conditions at the seasonal time scale (Figure 3).Less annual mean DO is observed with increasing silt content without considering the texturedinduced heterogeneity in POC (Figure 3a), which agrees with the numerous field observations that fine sediments with low permeability are the hot spots of anaerobic biogeochemical reactions and potential sources of methane production (Crawford et al., 2017;Zhang et al., 2020).Because of this effect, the variation in annual mean DO relative to the median value is larger when the silt content is 50%, implying that the concentration of DO is less predictable with increased heterogeneity in permeability.The combined influence of silt and POC content has a complex influence on sediment DO dynamics.Changes in POC abundance have negligible impact on DO when silt content is above 50%, due to the consistent exhaustion of DO.With silt contents less than 50%, changes in POC abundance lead to greater variability in DO because of the interplay between DO consumption and renewal via advective fluid flux.With lower POC, DO depletion is less extensive and overall variability in DO concentration is smaller.With higher POC, DO consumption is more rapid, such that periods of major DO depletion arise when input through fluid flux is not sufficient to keep up with demand-a dynamic that leads to large temporal variations in DO concentration.In contrast to DO, concentrations of DM increase monotonously with increasing silt content and POC abundance, as other electron acceptors become depleted and methanogenesis becomes a more significant pathway for organic carbon metabolism (Figure 3b).When texture-induced heterogeneity in POC is taken into consideration, the role of infiltration-induced (i.e., depth-dependent) POC heterogeneity becomes less important (Figures 3c and 3d).Collectively these results suggest that, at the seasonal timescale, heterogeneity in POC is likely to have comparable influence on sediment redox condition in comparison with heterogeneity in permeability.
The simulated annual mean DO content for the entire sediment domain was, as expected, higher with an oxic (upwelling groundwater saturated with DO) versus an anoxic bottom boundary condition (Figure 4a).However, for the upper half of the spatial domain where POC is most abundant, the influence of bottom boundary DO content was significant only for permeable sandy sediments (silt fractions =10%, Figure 4b).The role of bottom boundary DO in governing redox conditions decreased gradually and was minor as the silt fraction increased to >30% (Figure 4b).For some realizations, simulated DO in the upper half of the spatial domain was actually higher with an anoxic bottom boundary than with an oxic bottom boundary.This situation arose when silt lenses were located at shallow and deep depths, which impeded vertical DO transfer during both downwelling and upwelling fluid flow.Similar patterns to those shown Figure 4 were observed as POC content increases (n > 25, Figures S11-S13 in Supporting Information S1), even without considering textured-induced heterogeneity in POC (Figure S14 in Supporting Information S1).

Implications for Riverbed Biogeochemical Cycling
The important role of stream/riverbed sediments with respect to ecosystem metabolism is widely recognized due to the high metabolism of sediments compared to water columns (Minshall et al., 1983).This study and previous work (Chen, Chen, et al., 2023;Kaufman et al., 2022;Roden et al., 2023) in the Hanford Reach of the Columbia River demonstrates that redox conditions in the riverbed can exhibit complex spatiotemporal dynamics that are closely related to fluctuations in hydrologic exchange flow.Increase of the river stage during seasonal flooding causes downwelling flow, thereby creating oxic conditions in permeable riverbed sediments.On an event timescale, daily or weekly changes in river water level linked to variations in precipitation or (in the case of the Hanford Reach of the Columbia River) dam control may result in the frequent fluctuation of redox condition in sediments.On a time scale of decades, it has been demonstrated that anthropogenic activities such as groundwater pumping and climate change have resulted in the flow conditions of rivers and streams to convert from gaining to loosing (de Graaf et al., 2019).The results of this study suggest that sediments are more likely to experience oxic conditions for loosing rivers where the time scale of oxygen-consumption is larger than fluid residence time as DO-saturated riverwater intrudes into the riverbed.
Organic matter in both particulate and dissolved forms is the essential material for sustaining microbial metabolism in riverbed sediments.Sediment organic matter is primarily renewed via two mechanisms in fluvial environments.One is deposition of POC delivered to the sediment surface during periods of high river flow and seasonal or other episodic detrital organic matter input (Krause et al., 2022).Inputs of higher plant detritus, typically referred to "coarse particulate organic matter" (CPOM), has been studied extensively in streams and rivers as a basis for lotic ecosystem food chains (Wetzel, 2001).Some fraction of CPOM is transformed (by physical and biological activity) over time into more fine-grained (generally < 1 mm) particles, referred to as "fine particulate organic matter" (FPOM).Such particles are subject to both downstream export (e.g., Cushing et al. (1993)) as well as vertical transport into porous (sandy) stream/riverbed sediments (e.g., Brunke (1999)).Input of POC in the form of FPOM (which may include biomass of microalgae and attached microbiota) to sediments during downwelling fluid flow and gravitational settling is referred to as "depth filtration" in the stream ecology literature (Brunke, 1999).A recent field experiment deployed POC-free sediment traps at the Hanford 300 Area to constrain potential rates of suspended POC infiltration (i.e., depth filtration) into the riverbed (Roden et al., 2023).These experiments showed maximum POC accumulation at within the upper few cm of sediment, which declined exponentially with depth as expected for filtration-driven particle retention (Harvey & Garabedian, 1991).However, there was significant variability in the POC content within all depth intervals.This variability may have been tied to variations in the FPOM filtration efficiency of the material in the traps, for example, related to fine-scale variations in % silt content that arose during filling of the traps with combusted riverbed sediment.Previous studies with both natural photosynthetic biomass (Huettel & Rusch, 2000) and Cryptosporidium parvum oocysts (Atwill Edward et al., 2002;Harter et al., 2000) have shown that the efficiency of filtration of silt-sized (ca. 10 μm diameter) suspended particles is strongly influenced by the grain size of the porous medium, with finer-grained materials being much more efficient in particle filtration.Hence, variations in sediment silt content may lead to lateral and depth-dependent heterogeneity in sediment POC content superimposed on the overall decline in POC content with depth.The considerations lend support for the combined depiction of POC heterogeneity based on silt content and filtration effects adopted in this study.
It is noteworthy that the organic-rich riparian aquifer could also deliver dissolved organic matter into the riverbeds by lateral groundwater discharge flow (Wang et al., 2023).Disentangling the contributions of different sources, such as surface water and lateral groundwater exchange, to the autochthonous organic matter in sediments remains a major challenge in HZ biogeochemistry.High-resolution techniques (e.g., FT-ICR) for identifying the composition of organic matter in sediments may be a potential solution for the problem (Stegen et al., 2016).
The influence of low-permeable sediments or heterogeneity in permeability on flow and transport has been well studied in the groundwater community (Jakobsen, 2007;Perzan et al., 2021;Yan et al., 2016).However, the conceptual model for the HZ remains incomplete, due in large part to lack of information on the complex interplay between fluid flow rate/direction and biogeochemical dynamics.The view of HZ as a homogeneous "reactor" that processes dissolved organic matter and nutrients introduced from the surface water (e.g., Boulton et al. (1998)) may need to be revised, because in many situations metabolism of organic materials endogenous to the HZ itself are likely to be quantitatively more important than processing of solutes entering from the river itself.The widely adopted travel-time-based modeling approaches simulates nutrient transformation in the HZ relative to the flowing channel with different conceptual models being developed to reproduce the observed tailing behavior in stream tests for both nonreactive and reactive tracers (Fang et al., 2020;Painter, 2018Painter, , 2021)).Travel time is a surrogate for the degree of reaction in HZ such as denitrification and has the advantage of easy upscaling to the watershed-scale.By contrast, reactive transport modeling can explicitly simulate the biogeochemical processes in the context of physiochemical heterogeneity, which is more flexible in terms of representing different aerobic and anaerobic reactions but more computationally expensive.It is noteworthy that an upscaling approach has been developed for quantifying riverbed redox conditions at the reach-to basin-scale with computational costs linearly increasing with the number of simulated reaches (Chen, Yang, et al., 2023).

Conclusion
Riverbed redox conditions exhibit complex spatiotemporal patterns, which are influenced dramatically by hydrological fluctuations, as well as heterogeneity in sediment permeability and POC content.This study presents a Geophysical Research Letters 10.1029/2023GL107761 novel analysis of how such heterogeneity may dictate biogeochemical responses to hydrodynamic changes on both event and seasonal timescales.These findings highlight the importance of accurate characterization of heterogeneity in both permeability and POC content for predicting redox dynamics in riverbed sediments.

Figure 1 .
Figure 1.Hydrologic and biogeochemical conditions of the reactive transport model.(a) Vertical hydraulic gradients (J) between 10/1/2017 and 10/1/2018 at the Hanford reach of the Columbia River (Chen, Chen, et al., 2023).J was calculated from the measured river stage and the hydraulic head in sediments.The dashed line indicates the peak of summer flooding.(b-f) Selected realizations of permeability fields with the silt fraction increasing from 10% (b), 30% (c), 50% (d), 70% (e) to 90% (f).The silt and sand lenses are represented with black and light orange, respectively.The green box in b indicates relatively large silt lenses.(g) Infiltration-induced vertical distributions of POC with n in Equation 1 varying from 5 to 100.(h) Spatial distribution of POC considering infiltration-induced and texture-induced heterogeneity (20% silt).(i) The annual mean fluid flux with the silt fraction increasing from 10% to 90%.Ten stochastic realizations of permeability fields are generated for each case of silt fraction.

Figure 2 .
Figure 2. Spatial profiles of biogeochemical processes considering heterogeneity in permeability and infiltration-induced POC distribution during the flooding event.(a) O 2 flux across the top and bottom boundaries (positive: flow in, negative: flow out) with 10% silt.(b-f) Spatial profiles of dissolved oxygen at the time of peak flooding (dashed line in Figure 1a) with the silt fraction increasing from 10% (b), 30% (c), 50% (d), 70% (e) to 90% (f).The green box in (b) corresponds to the silt lenses shown in Figure 1b.(g-k) Reaction rates of aerobic respiration (g), denitrification (h), iron reduction (i), sulfate reduction (j), and methanogenesis (k) at the peak flooding.(l-p) Reaction rates of aerobic respiration with n = 5 (l), 25 (m), 50 (n), 75 (o), and 100 (p) at the peak flooding.The upwelling groundwater is assumed to be saturated with O 2 .

Figure 3 .
Figure 3. Redox conditions of riverbed sediments considering heterogeneity in permeability and infiltration-induced/textureinduced POC distribution.(a) The annual mean concentration of O 2 with the silt fraction increasing from 10% to 90% and n in Equation 1 increasing from 5 to 100 for POC abundance.For each case of n and silt fraction, 10 realizations are simulated, corresponding to the 10 stochastic permeability fields.For heterogeneity in POC, only infiltration-induced heterogeneity (smooth decrease with depth) is considered.(b) The annual mean concentration of CH 4 with the silt fraction increasing from 10% to 90% and n increasing from 5 to 100.The upwelling groundwater is assumed to be saturated with O 2 .Only infiltrationinduced heterogeneity in POC is considered.(c-d) Annual mean O 2 and CH 4 correspond to (a) and (b), respectively, considering texture-induced and infiltration-induced heterogeneity in POC.

Figure 4 .
Figure 4. Influence of bottom boundary conditions (oxic vs. anoxic) on the dissolved oxygen.(a) The annual mean concentration of O 2 for the entire sediment.Blue and orange represent the conditions that upwelling groundwater is under anoxic and oxic conditions, respectively.n is 25, in which the concentration of POC below the depth of 1 m decreases to less than 1% of that at the SWI.Both texture-induced and infiltration-induced heterogeneity in POC are considered.(b) The annual mean concentration of O 2 for the upper 1 m of the sediment.