The problem begins with a deceptively simple assumption. For decades, riverbank filtration and managed aquifer recharge systems have been evaluated as if the hydraulic gradient between a river and its aquifer were roughly constant. Water infiltrates, sediments and microbes remove some pollutants, and cleaner water emerges at the well. But field conditions are rarely steady. Storms raise river stage within hours, seasonal recharge resets regional groundwater levels, coastal tides alternate infiltration and discharge over hourly cycles, and human pumping and dam operations bend the flow field at will. Each of these forcings changes the magnitude and direction of water exchange, expands or contracts the active exchange zone, and redistributes the travel and residence times of every water parcel crossing the boundary.

The consequences for organic micropollutants, or OMPs, are profound. When a flood raises the river head above the adjacent groundwater, the prevailing gradient can weaken or reverse entirely. River water loaded with carbamazepine, sulfamethoxazole, benzotriazole, and other persistent compounds is pushed into bank sediments and riparian aquifers that routine conditions never reach. Field monitoring at Danube riverbank filtration transects between 2014 and 2016 captured this directly: two flood events increased riverine loads of these compounds, and groundwater concentrations during flooding could exceed simultaneous river concentrations and extend farther into the aquifer than routine monitoring would suggest. The review is careful to note that this result cannot be read as event-specific degradation or formation from synchronous concentration pairs alone, because the sampled groundwater may reflect antecedent source concentrations, mixing, and a distribution of travel times.

Here the review introduces one of its central integrative concepts: hydrologically realized mobility. Molecular properties such as polarity, charge, and hydrophobicity constrain sorption and persistence, but they do not uniquely predict whether a compound will actually cross the surface water–groundwater interface during a given event. Hydrodynamics determine whether intrinsic mobility is expressed as cross-interface transport. The testable proposition is that the same compound and source pulse will yield different arrival times and exported mass when forcing redistributes pathways, even if chemistry is unchanged. Paired conservative-tracer and OMP breakthrough across contrasting events would support this proposition; unchanged OMP response after demonstrable pathway and river-fraction changes would weaken it. Mobility, in this framing, is an event-specific outcome to be quantified, not a compound property to be assumed.

Once contaminants enter the interface, the event sequence unfolds in stages. During infiltration, the instantaneous contaminant input rate is the product of the surface-water concentration and the normal Darcy flux integrated over the infiltrating boundary. Advection controls the mean direction of transport, while dispersion and media heterogeneity distribute contaminant mass among pathways of different lengths and residence times. A single input pulse can therefore produce distributed arrival times and breakthrough magnitudes. When flow reverses and the river stage falls, infiltrated water drains gradually back to the channel as bank-storage return flow, and some infiltrated river water remains stored in the aquifer long after the storm. Hydrological modeling shows that bank-storage water may return to the river after recession or remain in the aquifer, with event timing changing water ages and return-flow fractions. For OMPs, the corresponding contaminant-mass routes are hypotheses until parent concentrations, fluxes, and stored pools are measured through the event.

Retardation and temporary storage add further complexity. Solid–water partitioning can slow OMP movement relative to the carrier water, and in heterogeneous media reversible sorption and slow mass exchange contribute to retardation and tailing. A field injection experiment resolved depth-dependent pesticide retardation against a conservative tracer, showing that sorption behavior can vary substantially within a single aquifer. But a delayed aqueous tail alone does not demonstrate that OMP mass was stored in sediment or a slow domain. Verifying contaminant storage requires a parent-compound mass balance, a measured sorbed pool, or a release perturbation. The review distinguishes three hydrologic-response classes: mobile–persistent compounds that traverse the interface rapidly or over long distances with little loss, mobile–reactive compounds whose fate depends on reactive exposure and pathway timing, and reversibly retarded compounds that show delayed arrival, tailing, and possible clean-water release. These classes are hypotheses about dominant responses, not fixed compound labels, and a compound may move between classes across sites or hydrologic phases.

Remobilization is where the story turns from storage to secondary source. When a flood recedes, stage declines, or slow pathways reconnect, untransformed OMP mass in slow-exchange domains or on reversible sorption sites may return to the mobile phase. Laboratory evidence is direct: in triplicate riverbed-sediment columns fed with contaminant-free tap water, previously retained compounds entered the effluent, demonstrating remobilization under controlled conditions. The field-scale significance of such release during flood recovery remains untested. Temporary storage can switch the functional role of the interface across event phases: enhanced infiltration creates a temporary sink that suppresses immediate output, whereas untransformed stored mass can become a potential secondary source during recovery. The interface can be described as a potential secondary source only when mass balance, parent compound–transformation product relations, or conservative–reactive tracer comparisons exclude transformation and dilution. Whether delayed release creates an environmentally relevant secondary pulse remains unresolved.

Reaction conditions add a dynamic dimension. Infiltrating surface water supplies dissolved oxygen and electron acceptors, while respiration consumes them along flow paths, creating redox zonation that shifts with temperature, organic carbon supply, and exchange. At the wastewater-impacted River Erpe, suspect screening along preset 35-centimeter hyporheic flow paths retained 130 aqueous trace organic contaminants after quality filters; 58 showed decreases, 63 showed no attenuation, and seven increased. Among the 58 attenuated compounds, 51 had higher estimated rates in the oxic than the suboxic segment. Recent simulated riverbank filtration experiments further show that sequential oxidizing–reducing conditions can improve removal of some recalcitrant OMPs, including irbesartan, benzotriazole, and gabapentin, relative to exclusively oxidizing or reducing conditions, accompanied by distinct redox-driven microbial specialization. The sequence of reaction environments encountered by a contaminant may be more informative than a single redox measurement. The review frames this as a dynamic reaction area: a compound-specific, time-dependent volume where transported OMP mass encounters conditions favorable for transformation. The testable proposition is that OMP transformation rises when event-driven pathway redistribution increases overlap between the OMP pulse and favorable reaction conditions, after conservative transport and mixing are controlled.

Antecedent state, or the memory of the interface, may be the most underappreciated control. Earlier low-flow, high-flow, drought, or stage-fluctuation periods change pore-water composition, redox state, microbial communities, and stored solute mass. Process-based simulations of successive peak flows show that event separation relative to system memory controls whether residence-time effects carry into a later peak. Intermittent-flow OMP columns provide controlled evidence that drying changes microbial composition and transformation routes for selected compounds. A thermo-hydro-biogeochemical model found that low mean stage during drought amplified the effects of high-frequency stage fluctuations on hyporheic temperature and biogeochemical reaction patterns. These sources support a hypothesis of history-dependent OMP fate, but direct event-resolved field tests remain scarce. The testable prediction is that matched current forcing and source inputs yield different OMP breakthrough or recovery when measured pre-event storage or reaction states differ.

The methodological implications are stark. Low-frequency and asynchronous sampling can miss the hydrologic phases that control transport entirely. At one Rhine riverbank filtration site, modeling linked high floods to river-to-well travel times shorter than eight days compared with a routine 25-to-40-day range, while low-flow conditions lengthened modeled residence to about 60 days. A fixed calendar lag may miss the contributing source period. Synchronous surface-water and groundwater concentration ratios are only apparent contrasts, not parcel-based removal fractions or proof of transformation. The review prescribes a hierarchy: constrain hydrodynamics and travel-time distributions first with groundwater levels, electrical conductivity, stable isotopes, and other conservative tracers, and only then estimate compound-specific retardation and reaction parameters. Validation should be partitioned by hydrologic state or independent event, not randomly. A model demonstrates cross-state predictive skill only if it reproduces the hydraulic response, travel-time distribution, arrival time, and OMP breakthrough in independent events without recalibrating key parameters. If dispersion, sorption, or degradation parameters must be adjusted for every flood, the result is event-specific fitting rather than an early-warning model.

Exposure assessment must also change. The review proposes event-resolved metrics: first arrival, peak concentration, exposure duration, cumulative breakthrough mass, and recovery time, each with explicit uncertainty and decision thresholds. At a full-scale riverbank filtration system in the Shepperton gravels, monitored flood inundation raised dissolved organic carbon to approximately 140 percent of baseline and mean well dissolved oxygen to approximately 440 percent of baseline, while micro-organic contaminant detections also increased. Most flood water-quality effects occurred during the first two to three weeks, with a return toward baseline conditions within about six weeks. The end of inundation is not necessarily the end of elevated exposure. Recovery time deserves particular emphasis, and compound- and site-specific decision thresholds and sensitive receptors would be required to extend these transport metrics into a formal health or ecological risk assessment.

The central knowledge gap remains direct event-scale field mass balances linking OMP storage to delayed release. Within the literature corpus assessed, transient water exchange and changing water ages are supported mainly by hydrological studies; direct event-scale field mass balances were not identified. Direct remobilization is demonstrated in a controlled clean-water column test, whereas a field event-scale balance from OMP input through storage and delayed release was not found. The next priority is event-triggered monitoring from rising stage through recovery, paired with time-dependent source–response matching and pathway-resolved reaction measurements. The central question shifts from whether one flood increases an OMP concentration to the following: under which combinations of antecedent state, inter-event interval, and hydrologic disturbance does OMP breakthrough or delayed release exceed a site-defined exposure threshold, and how long does that condition persist after the event? Until independently validated OMP event data are available, this remains a research agenda. But the message is already clear: the riverbank is not a filter. It is a dynamic, history-laden archive that opens and closes with every storm, and we have only just begun to read it.

Subject of Research: Effects of transient surface water–groundwater exchange on the transport of organic micropollutants: temporary storage, retardation, and remobilization

Article Title: Effects of transient surface water–groundwater exchange on the transport of organic micropollutants: temporary storage, retardation, and remobilization

Article References: Effects of transient surface water–groundwater exchange on the transport of organic micropollutants: temporary storage, retardation, and remobilization. (n.d.). https://doi.org/10.3389/frwa.2026.1980968

Image Credits: AI Generated

DOI: 10.3389/frwa.2026.1980968

Keywords: Effects, transient, surface, water, groundwater, exchange, transport, organic, micropollutants, temporary, storage, retardation