The Olewigerbach is a third-order tributary of the River Mosel, located south of the city of Trier in southwest Germany. The catchment drains a 25 km² watershed with a total stream length of 14 km (Krein and Schorer, 2000). The area above the experimental site drains 8.01 km². The elevation difference between the headwaters and the outlet is approximately 300 m. The stream has a pluvial regime, receiving an average annual precipitation of 745 mm and exhibiting a mean discharge of 106 L s⁻¹ recorded between 2010 and 2023 at the study site. The catchment's geology is dominated by Devonian schists, primarily argillaceous slates. The shallow soil layer (∼ 1.3 m) contains impermeable clay pockets underlain by colluvial or bedrock argillaceous slates (Krein and Symader, 2000). The study area is situated in the headwaters of the catchment at an elevation of approximately 290 m above sea level, characterized by steep hillslopes.

Drought conditions were defined as periods when streamflow fell below 5 % (5.3 L s⁻¹) of the mean discharge, following thresholds set by Yevjevich (1967) and Zelenhasić and Salvai (1987).

Field measurements were conducted from 2020 to 2023. The dataset comprises 28 differential discharge gauging campaigns, utilizing NaCl slug injections, to estimate HT during runoff events, over the stream section equipped with sampling wells. We monitored 68 runoff events with precipitation totals ranging up to 60 mm. Corresponding streamflow responses showed peak discharges between 20 and 700 L s⁻¹. Antecedent soil moisture conditions varied from dry at summer drought events to near saturation at winter events (Table 1 and Fig. 3). While these event descriptions provide hydrological context, the classification of events in this study is based on hysteresis behaviour and HT rather than on precipitation or discharge metrics. Hysteresis loops were recorded based on stream water levels and groundwater levels measured at two groundwater wells (GW1 and GW2) located close to the stream (Fig. 1).

Groundwater levels were continuously logged in 10 min intervals using Orpheus Mini Level Loggers (OTT GmbH). At the upstream gauging station three piezometers were installed in a transect: one in the stream (289.47 m a.s.l.) and two at the groundwater wells, GW1 (288.63 m a.s.l.) and GW2 (288.65 m a.s.l.). GW1 is located 1.5 m from the streambank, and GW2 is 3.7 m away. Groundwater levels were monitored at a depth of 1.3 m. Based on the recorded hydrograph data; 68 events were selected for hysteresis analysis using the h-Index developed by Zuecco et al. (2016). Corresponding HT measurements (n=28) were available for 20 of these events (Table 1). We identified three severe drought periods: 27 March to 1 April 2020; 24 April to 28 September 2020; and 6 September to 20 October 2021 (Fig. 3). Precipitation and modelled soil moisture data (AMBAV, DWD) for the study area were obtained from the Deutscher Wetterdienst (DWD) weather station located near the Olewigerbach catchment (Station: Trier-Petrisberg; ID: 05100; https://opendata.dwd.de/climate_environment/, last access: 17 April 2026).

Streamflow data were monitored at the sampling site (Fig. 1). We collected 144 water samples – 48 each from the stream, GW1, and GW2 – across three identical sampling spots, including two additional downstream locations without piezometers. Sampling was conducted under various baseflow conditions (Table1; BFI). The experimental infrastructure is drawn out in detail in Bäthke and Schuetz (2024).

The majority of the events analysed in summer showed groundwater levels of the near stream well (GW1) below the stream water level (Fig. 3). However, in winter groundwater is often observed to be higher than the stream, especially at events during high soil moisture conditions. Analysing the hysteresis between the stream and the respective groundwater well, we observed differences between hysteretic behaviour during all seasons. During summer events the hydraulic head becomes negative, but also the hydraulic head difference between the two observed wells (Fig. 3). Precipitation events correspond with the on-site logged hydrograph dynamics as well as the soil moisture data (Fig. 4c).

Higher catchment soil moisture contents corresponded to higher stream flow and hydrograph reaction velocity to rain events. During drought conditions at stream flow below 5 % of mean discharge, we observed the highest HT [% m⁻¹] values (Fig. 5), while absolute net exchange presents itself as very low (Fig. 5).

We observed extended HT during drought, and distinct seasonality of the h-Index (Fig. 5) between winter and summer (including the drought period). The shift between seasons of the h-Index corresponds with the crossing of its moving average of the zero line (Fig. 5b and c), with h-values during the summer mostly negative index values (counterclockwise hysteresis) and during the winter positive index values (clockwise hysteresis) (Fig. 3). However, there is a shift in seasonality between the two groundwater wells, with the well closer to the stream (GW1) showing the seasonal shift earlier and transitions later in the year compared to the second well (GW2). The distance between wells is 2.2 m, which appears to correspond to one month in delay. Summer rainfall events show a marginal impact on low flows, just initializing sharp spikes in stream flow for a very short period (Fig. 3). During winter, frequent rain events sustain higher baseflow (Fig. 4b). Antecedent moisture conditions provide the boundary for discharge generation (Table 1). The Olewigerbach is during both seasons a highly reactive stream with increased discharge amplitudes (Fig. 4b). Seasonality, reflected in precipitation and soil moisture data, aligns with h-Index values and hydraulic gradients at the Olewigerbach (Fig. 5).

Comparing the h-Indices, both wells at the experimental site show significant differences (Wilcoxon rank-sum test) in h-Indices between summer and winter (Fig. 6).

The first well tends towards more negative h-Index values (counterclockwise) compared to the second well. Seasonally, both wells show significant differences between h-Index values. During the summer period, the h-Index values present decreasing values, shifting from clockwise to counterclockwise behaviour (Fig. 6). Additionally, compared with the h-Index values tend further towards negative values during the drought periods. However, the drought dataset is small in observed events (n=9). h-Index values are substantially larger during drought conditions compared to winter, indicating an increase in hysteresis loop size.

Thus, showing the increased extended duration of the total counterclockwise hysteresis between the stream water level and the groundwater level. Comparing summer HT to drought event HT, the significance level of the correlation decreases from summer to drought (Fig. 6). Comparing the h-Index values of both wells to HT, as the part of discharge exchanged per distance (% m⁻¹), both wells behave similar. The h-Index shows a significant correlation with overall hydrological turnover (Fig. 7a, b), as well as with its components, gross loss (Fig. 7c, d) and gross gain (Fig. 7e, f). A comparable pattern is evident across all presented correlations: drought measurements cluster at the lowest levels of the h-Index range, showing the highest observed HT values (Fig. 7). These drought measurements were taken during periods of lowest discharge, whereas winter measurements correspond to periods of highest discharge. Additionally, we observe a positive correlation between discharge and h-Index values at both wells (Fig. 7g and h).

The mean hydraulic gradient between groundwater wells and the stream during single events, defined as negative toward the stream, shows a significant correlation with h-Index values (Fig. 8a and b; Eq. 12).

Steeper negative gradients and more negative h-Index values were primarily observed during summer and drought events. In contrast, winter events exhibited both clockwise and counterclockwise hysteresis under less steep gradients. Positive gradients generally indicate gaining conditions, while negative gradients reflect losing conditions. Between the wells (Fig. 8c and d; Eq. 13), most summer and drought events displayed negative gradients and counterclockwise hysteresis. Based on hydraulic gradient and h-Index we divided the observations into four behavioural clusters (I–IV):

Cluster I: Mostly Winter events and heavy summer rain events; characterized by flat or positive gradients and counterclockwise hysteresis (groundwater rises before surface water).

Hysteretic behaviour and the associated HT induced mixing through the hyporheic zone and the stream bank, allow for exchange of solutes between GW and SW. Hence, we compared the sampling results by distinguishing seasons (Summer and Winter), Hysteresis direction (h<0, h>0) and hydraulic conditions based on the gradient data (Losing or Gaining) (Fig. 9).

We observed differences for distribution of potassium samples between seasons. The winter events, positive h-Index show consistently the highest values in the stream with similar concentrations in both wells. However, for summer events, negative h-Indices and losing conditions exhibit high potassium values at the stream gradually declining with distance towards the wells. In winter potassium is in a similar range at the wells (Fig. 9a). Generally, silica is rising from groundwater to stream in concentration while potassium shows the opposite (Fig. 9a). We detected redox-sensitive species (Fe, Mn, NO3-) in all samples, with the highest concentrations of iron and manganese consistently observed in near-stream groundwater wells. This suggests heterogeneous redox conditions in the riparian groundwater compared to the stream. For DOC concentrations tended to decline from stream to GW during the summer months while the opposite occurred during winter. Nitrate concentrations were generally variable with, on average, higher stream concentrations compared to GW (Fig. 9b). Nitrate is constantly present during our observations, regardless of redox potential indicated by the concentrations of iron and manganese ions. Manganese and iron are simultaneously present at the groundwater wells, often in similarly high concentrations, and show clear temporal patterns. During summer, concentrations of both elements were higher at GW1 than at GW2, suggesting intensified redox activity closer to the stream. These conditions coincided with hydraulic gradients indicative of losing conditions and counterclockwise hysteresis, indicating that stream-derived organic carbon inputs may stimulate microbial activity in the near-stream sediments.

High-resolution hydraulic gradient data allowed comparison of streamflow events with sufficient resolution to minimize noise. This allowed us to find seasonal variations in hysteretic patterns during the observation period (Fig. 6), showing that hysteresis between GW and SW levels are connected to seasonal hydrological states, comparable to those found in the literature (e.g., Zuecco et al., 2016; Lloyd et al., 2016a; Gelmini et al., 2022). The diagnostic use of hysteresis utilizes lagged hydraulic responses as indicators of system state rather than causal mechanisms.

Although the h-Index showed only slight differences in hysteresis between the wells (Figs. 3–5), correlation analysis reveals significant spatial and temporal variability in groundwater-stream interactions (Figs. 6 and 7). These variations reflect the dynamic nature of the hyporheic zone, which changes seasonally, particularly between winter, summer, and drought conditions (e.g., Wondzell, 2011; Cardenas, 2015; Harvey and Bencala, 1993; Malzone et al., 2016; Wroblicky et al., 1998). As observed by Gelmini et al. (2022), our study identifies clear seasonality in hysteretic behaviour. Specifically, HT varies with seasonal changes in hydraulic gradients and discharge magnitudes (Figs. 7 and 8). The h-Index serves as an event-based parameter to describe GW-SW interactions by incorporating hydraulic preconditions and event characteristics (Table 1). Distinct h-Index value ranges correspond to increased portions of stream water engaged in turnover processes (Fig. 7), aligning with studies emphasizing pressure-head variations and their influence on seasonal groundwater-stream connectivity (Brunner et al., 2009, 2011). The seasonal transition is marked by declining h-Index values (including directional reversals) from winter to summer, coinciding with decreasing stream discharge and increasing hydrological turnover. These findings indicate that both the direction and magnitude of hysteresis control the intensity and direction of groundwater–surface-water exchange, as evidenced by the HT measurements. This is consistent with findings from Payn et al. (2009), Covino et al. (2011), and others. Beyond direction, the h-Index also reflects temporal dynamics between groundwater and surface water levels during events (Zuecco et al., 2016; Lloyd et al., 2016b), offering a physical framework for understanding HT processes. Thus, emphasizing that HT reflects exchange-driven renewal processes rather than storage duration, distinguishing it conceptually from residence time metrics. Hydraulic gradients between the stream and shallow groundwater serve as preconditions for runoff events, modulating GW-SW interactions (e.g., Zimmer and McGlynn, 2017; Voltz et al., 2013) a distinct seasonal pattern emerges, steeper hydraulic gradients during summer events correspond to counterclockwise hysteresis, while gentler gradients in autumn and winter are associated with clockwise hysteresis (Fig. 8). In addition, the riparian zone at the study site is covert by vegetation, which further influence GW–SW interactions. Root water uptake and evapotranspiration can modify near-stream hydraulic gradients, particularly during low-flow and drought conditions (Tabacchi et al., 2000), thereby potentially reinforcing strong gradients and affecting exchange fluxes between stream and the riparian zone (Wondzell et al., 2010; Boano et al., 2014). While these processes were not explicitly quantified in this study, they likely interact with the observed hysteresis patterns and hydrological turnover and should be considered when interpreting GW–SW exchange dynamics.

This relationship is reflected in our results, where counterclockwise hysteresis (negative h-Index) predominantly occurs under steep negative hydraulic gradients, indicating losing conditions, whereas positive or near-zero gradients are commonly associated with clockwise hysteresis and gaining conditions (Fig. 8 and corresponding h-Index values in Sect. 3.3). Soil moisture controls the unsaturated hydraulic conductivity, governing occurrence of infiltration and overland flow (Horton, 1933). Thus, controlling threshold behaviour and the hydraulic response to rainfall events (e.g. Zehe and Sivapalan, 2009). Based on this hypothesis we identified three system states during the observations, winter, summer and drought conditions. These patterns reflect groundwater storage states and hyporheic zone variability. Hydraulic gradient changes not only influence hysteretic loop direction but also reflect groundwater table dynamics, hyporheic zone extent, and riparian bank storage (e.g., Brunner et al., 2009; Gu et al., 2012; Malzone et al., 2016). Elevated stream stages promote counterclockwise hysteresis and hyporheic zone expansion through increased bankstorage (Gu et al., 2012). Conversely, groundwater discharge initially reduces the hyporheic zone's extent (Cardenas and Wilson, 2007).

Our results demonstrate that dynamic local gradients strongly influence hyporheic exchange fluxes in the form of observed HT magnitude (Fig. 8). Thus, the Olewigerbach exhibits distinct seasonal system states that shape the water volume at the surface-groundwater boundary layer and thus shaping exchange in the form of HT. HT significantly correlates with h-Index values, connecting system state driven response with GW-SW exchange. We propose addressing HT as a proportion of total discharge, considering the extent of hyporheic zone involvement relative to streamflow conditions. Under low surface discharge, hyporheic zone expansion potentially increases the ratio of hyporheic area to streamflow, engaging a greater fraction of stream water in HT.

This is demonstrated by the observed clustering of hysteresis index and hydraulic gradients reveal distinct hydrological states driven by seasonal conditions and event timing. We show that plotting events with hydraulic gradient against h-Index value four behavioural clusters can be identified (Fig. 8). Cluster I is dominated by winter events with counterclockwise hysteresis and flat or positive gradients, suggests storage influence where groundwater responds ahead of streamflow. This may be indicative of a hyporheic response occurring during early infiltration phases. Cluster II, with winter-dominated events and clockwise hysteresis under gaining conditions, likely indicates surface runoff or upstream input overwhelming the local GW response. These conditions reduce the influence of subsurface exchange, leading to stream-dominated hysteretic behaviour. Cluster III shows the strongest expression of hyporheic connectivity under summer and drought conditions. This description of present-day summer and drought system states may support the interpretation of possible HT process shifts if hydrological regime changes occur in the future. Steep negative gradients and counterclockwise loops here point to deeper, prolonged infiltration into the stream bank enabling intense mixing, reflected by the high HT values during those conditions (Fig. 7). Cluster IV shows the temporal variability and spatial heterogeneity of riparian processes. Particularly, the flatter gradients between the two wells are more susceptible to transient fluctuations during rainfall events, resulting in mixed hysteresis responses.

Additionally, our results show that gaining and losing conditions are not exclusively associated with one hysteresis direction. Similar observations were reported by Gelmini et al. (2022), who found that hysteresis behaviour, captured by the h-Index can vary depending on antecedent conditions and event timing, rather than being strictly determined by the direction of exchange. However, this cluster fitted the least number of events during our observed period (Figs. 8 and 10). These findings show that combining h-Index and hydraulic gradients allows detailed classification of hydrological states at event scale. Our results support Hypothesis (1): the direction and magnitude of the h-Index exhibit a clear and consistent seasonal pattern, with clockwise hysteresis dominating under winter gaining conditions and counterclockwise hysteresis prevailing during summer and drought losing conditions. We also accept Hypothesis (2), since hydrological turnover (HT), as an independent measure of exchange, strongly co-varied with h-Index values. Periods of intensified HT aligned with negative h-Index values under steep losing gradients, whereas low HT occurred under positive, or near-zero gradients associated with gaining conditions.

Counterclockwise hysteresis and high HT are associated with enhanced riparian bank storage. Vertical expansion of the hyporheic zone at Olewigerbach is likely limited by shallow sediments and small stream size (Malzone et al., 2016). Instead, horizontal bankstorage expansion dominates, forming storage capacities (Fig. 10) along the streambank (Gu et al., 2012). The occurrence of bank storage is associated with counterclockwise hysteresis between stream water and groundwater levels, which occurs primarily during summer (Cluster III), while winter is dominated by clockwise hysteresis patterns (Fig. 10). This is further promoted by the losing conditions observed at the site, as similarly described in the Selke catchment (Trauth et al., 2015). Short-time alternating water level gradients during and after stream flow events create oxic-anoxic transition zones in the riparian zone could promote microbial processes (Knorr et al., 2009). Thus, these transition zones created by redox fluctuation within the riparian zone are shaping redox hot spots. This is supported by elevated concentrations of DOC, Fe²⁺, and Mn²⁺ at GW1 during summer losing conditions, fitting with the concept of the hydrological framework of redox-active compounds in aquatic systems explaining the simultaneous occurrence of ions along the redox ladder. Thus, our observed enhanced biogeochemical activity is likely to be driven by redox fluctuations (Fig. 9). Such conditions influence chemical turnover, pollutant degradation and solute transport. In redox-sensitive environments, such as the riparian zone microbial reduction processes are understood to occur in a sequence with electron acceptors becoming progressively depleted (Zehnder and Stumm, 1988). This sequence is influenced by factors such as redox potential, pH, electron acceptor availability, and the presence of bioavailable organic matter. While our study lacks direct measurements of dissolved oxygen (DO) and pH, which would have strengthened the interpretation of redox processes, we infer redox conditions based on observed concentrations of DOC, Fe²⁺, and Mn²⁺. However, point-scale measurements of DO or pH in riparian groundwater would still represent spatially mixed signals, because small-scale heterogeneity in flow paths and reaction zones are difficult to resolve with standard sampling alone (Frei et al., 2012; Knorr et al., 2009; Peiffer et al., 2021; Alewell et al., 2008). We interpret that, during summer and under counterclockwise hysteresis conditions, microbial activity is increased, although the significance tests indicate that DOC concentrations are comparable across the sampling sites and that Fe²⁺ and Mn²⁺ do not differ significantly between the groundwater wells. Thus, during summer and under counterclockwise hysteresis conditions the increased microbial activity and oxygen depletion is based on the co-occurrence of high DOC concentrations, elevated Fe²⁺ and Mn²⁺ in GW1 during losing conditions and counterclockwise hysteresis, likely driven by enhanced inflow of labile organic carbon from the stream into near-stream sediments (Smith and Arah, 1984). This is evidenced by elevated DOC concentrations and redox-sensitive solutes at GW1 (Fig. 9). However, our observations revealed the co-occurrence of nitrate, manganese, and iron in groundwater and stream samples. This simultaneous presence of multiple electron acceptors suggests that the redox sequence is not strictly adhered to under field conditions, a finding consistent with Alewell et al. (2008) and others. Such deviations from idealized zonation likely result from heterogeneous flow paths, dynamic hydraulic gradients (Peiffer et al., 2021; Frei et al., 2012; Knorr et al., 2009), and continuous mixing at the GW-SW interface (Gu et al., 2012). Such conditions foster overlapping redox states, enabling parallel rather than sequential redox processes. This highlights the riparian zone as a hydrologically and biogeochemically active layer, supporting diverse redox environments (e.g. Hill and Cardaci, 2004; Carlyle and Hill, 2001). Our findings emphasize the coexistence of multiple redox processes. The resulting shifts between oxic and anoxic conditions are critical for degrading pollutants and supporting biogeochemical activity in the riparian zone. These dynamics underscore the importance of water-stage fluctuations in enhancing chemical exchanges and microbial activity at the GW-SW boundary (Gu et al., 2012), supporting Hypothesis (3) Counterclockwise hysteresis under summer losing conditions coincided with elevated DOC, Fe²⁺ and Mn²⁺ at the near-stream well, indicating that local hysteresis-driven exchange promotes redox fluctuations and the formation of riparian redox hot spots. Overall, the observed SW–GW level hysteresis reflects the system's exchange-driven redox response and should therefore be interpreted as a diagnostic indicator, not as a mechanistic explanation.