Density-Driven Mixing and Stratified Flow Dynamics in Paldang Reservoir Under Variable Hydraulic Conditions

Abstract

This study investigated density-driven mixing and stratified flow dynamics in Paldang Reservoir, a river-type reservoir formed at the confluence of the South Han River, North Han River, and Gyeongan Stream in South Korea. High-resolution field observations were conducted under varying hydrologic and hydraulic conditions using an Acoustic Doppler Current Profiler (ADCP) and multi-parameter water quality sensors (EXO2). Spatial distributions of flow velocity, water temperature, and electrical conductivity (EC) were analyzed to evaluate tributary interaction and mixing behavior within the reservoir. Distinct spatial mixing structures associated with tributary inflow heterogeneity and hydraulic operation conditions were identified. During flood-season conditions, highly turbid and high-conductivity inflow from the South Han River propagated beneath the North Han River inflow, generating density-driven lower-layer intrusion near the confluence region. Under intermittent discharge conditions at the Cheongpyeong Dam, unstable upper- and lower-layer separation structures and localized reverse-flow behavior developed. In contrast, continuous discharge conditions promoted stable tributary propagation and persistent stratified mixing structures. Case-based Richardson number (Ri) estimates further indicated localized shear-driven mixing at low-Ri inflow sections and relatively stable stratification at high-Ri sections, providing quantitative support for the observed spatial heterogeneity in density-driven mixing. Overall, spatial mixing in Paldang Reservoir was governed by tributary density contrasts and further shaped by hydraulic operation conditions. These findings improve understanding of density-driven mixing processes in river-type reservoirs under varying hydraulic conditions.

1. Introduction

Hydrodynamic mixing behavior in river-type reservoirs is strongly influenced by tributary interactions, density differences, and hydraulic operation conditions. Mixing processes occurring at tributary confluences play a critical role in controlling circulation patterns, vertical stratification, constituent transport, and water-quality variability within reservoirs [1,2]. In particular, density differences generated by variations in temperature and water-quality characteristics among tributaries can produce vertically stratified flow structures, lower-layer intrusion, and incomplete mixing near confluence zones [3].

River confluences and regulated reservoirs have received increasing attention because density-driven mixing processes influence ecological connectivity, contaminant transport, and drinking-water quality. Understanding these processes is therefore important not only for local reservoir management but also for improving hydrodynamic interpretation in regulated river-type reservoirs worldwide. Previous studies have demonstrated that confluence mixing behavior is governed by flow velocity, discharge ratio, density contrast, and channel geometry [4,5,6,7]. Field observations using Acoustic Doppler Current Profilers (ADCPs) and multi-parameter water-quality sensors have further shown that density differences between tributaries can substantially modify flow structures and mixing patterns.

Recent field and numerical investigations have highlighted that three-dimensional mixing processes at river confluences are controlled by the combined effects of momentum ratio, density contrast, bed discordance, junction geometry, and hydraulic residence time [8,9,10]. More recently, studies employing the Mixing Proximity Index (MPI) have successfully quantified horizontal mixing processes at confluences. However, most previous investigations have focused on relatively short river reaches, idealized channel configurations, or horizontal mixing characteristics. Consequently, field-based studies simultaneously resolving three-dimensional hydrodynamics, density-driven stratification, tributary intrusion, and hydraulic-operation effects in regulated river-type reservoirs remain limited.

In regulated reservoirs, tributary interactions are continuously modified by upstream dam and weir operations. Variations in discharge conditions can alter residence time, longitudinal momentum, buoyancy-driven circulation, and mixing interfaces, thereby enhancing or suppressing stratification depending on hydraulic forcing conditions. Despite the recognized importance of these processes, quantitative field evidence describing how density-related water-mass differences and hydraulic regulation jointly control vertical separation, lower-layer intrusion, and incomplete mixing remains insufficient.

Paldang Reservoir is a river-type reservoir located at the confluence of the South Han River, North Han River, and Gyeongan Stream in South Korea. The reservoir is characterized by relatively short residence time and strong hydrodynamic variability resulting from large tributary inflows and upstream hydraulic structure operations. Distinct temperature and electrical conductivity (EC) characteristics are observed among tributaries because of differences in watershed conditions and hydraulic regulation patterns [1,3]. Water temperature directly influences water density, while EC behaves as a relatively conservative tracer over the short transport times considered in this study. Consequently, the combined use of temperature and EC provides an effective means of identifying density-driven mixing processes, tributary propagation pathways, and stratification interfaces.

The major hydraulic structures influencing tributary inflow conditions include Cheongpyeong Dam, Ipo Weir, and Paldang Dam (Figure 1). Discharge operations from these structures directly modify tributary inflow characteristics and spatial mixing behavior within the reservoir. As a result, spatial mixing structures, circulation patterns, and vertical stratification continuously evolve under varying hydrologic and hydraulic conditions.

Therefore, the primary research gap addressed in this study is the lack of integrated field observations capable of simultaneously evaluating three-dimensional hydrodynamics, density-driven tributary intrusion, and hydraulic-operation effects within a regulated river-type reservoir. To address this gap, the objective of this study was to investigate spatial mixing behavior and stratification dynamics in Paldang Reservoir under contrasting hydraulic operation conditions using integrated ADCP measurements and high-resolution water-quality observations. Particular attention was given to density-driven tributary interactions, lower-layer intrusion processes, and the influence of dam discharge operations on three-dimensional mixing structures and vertical flow separation. By combining hydrodynamic observations with water-quality distributions, this study provides new insights into density-controlled mixing mechanisms in regulated river-type reservoirs and offers information that may contribute to reservoir operation and water-quality management in similar systems worldwide.

2. Materials and Methods

2.1. Study Area and Monitoring Strategy

Paldang Reservoir, a river-type reservoir located in South Korea, was selected as the study area for this investigation. The reservoir is situated at the confluence of the South Han River, North Han River, and Gyeongan Stream and exhibits complex hydrodynamic characteristics resulting from tributary interactions and hydraulic structure operations. Owing to the convergence of multiple tributaries with distinct hydraulic and water-quality characteristics, the reservoir provides an ideal environment for investigating density-driven mixing processes, stratification dynamics, and spatial flow interactions [11,12].

The survey area encompassed the South Han River, North Han River, Gyeongan Stream, Paldang Dam region, and the areas surrounding the Sonae and Jokja islands (Figure 2). Survey transects were designed to cover the principal tributary inflow corridors, confluence regions, hydraulic transition zones, water-intake areas, and downstream reservoir reaches influenced by dam operation. These locations represent the primary pathways through which tributary water masses enter, interact, and propagate within the reservoir. In addition, the transect layout was configured to intersect expected lateral and vertical mixing interfaces, thereby enabling the observation of both longitudinal transport processes and cross-sectional stratification structures.

High-resolution field monitoring was conducted using a survey boat equipped with an Acoustic Doppler Current Profiler (ADCP-M9, SonTek, San Diego, CA, USA) and a multi-parameter water-quality sonde (YSI EXO2, Yellow Springs, OH, USA). The monitoring platform followed predefined zigzag survey transects while continuously recording hydraulic and water-quality data throughout the reservoir. Simultaneously, vertical profiling measurements were conducted at representative locations to characterize water-column structure and stratification conditions. This integrated monitoring strategy enabled detailed evaluation of three-dimensional hydrodynamic behavior and density-related mixing processes.

Before each field survey, the EXO2 sonde was inspected and calibrated following the manufacturer’s recommended procedures for temperature, conductivity, DO, pH, turbidity, and Chl-a sensors. ADCP measurements were synchronized with GPS positioning data, and boat speed was maintained as low and stable as possible to reduce navigation-induced velocity noise. Measurement uncertainty was controlled through pre-survey calibration, synchronized positioning, post-survey screening, and comparison of spatial patterns among adjacent observations.

The survey configuration and observation spacing were designed to resolve spatial heterogeneity associated with tributary inflows, mixing interfaces, and density-driven circulation. Monitoring intervals were determined considering channel geometry, bathymetry, and the characteristic scales of mixing structures identified in previous investigations. The resulting survey design provided continuous spatial coverage while maintaining sufficient resolution to detect flow separation, lower-layer intrusion, and stratification interfaces.

The horizontal survey spacing was designed to intersect the main tributary corridors, confluence zone, island-induced hydraulic transition zones, and downstream reservoir reach. The spacing was selected to be sufficiently fine to capture lateral gradients in EC, temperature, and flow velocity while maintaining continuous boat-based coverage over the reservoir-scale survey domain. Vertically, ADCP bin settings and EXO2 profiling intervals were selected to resolve near-surface, mid-depth, and near-bottom variations while avoiding unreliable measurements close to the bed and water surface. This configuration was appropriate for detecting the expected stratification interfaces, lower-layer intrusion, and vertical flow separation in the relatively shallow confluence sections.

Because the width and water depth varied among the survey transects, a single fixed spatial averaging grid was not appropriate for all cross-sections. Therefore, ADCP velocity data were spatially averaged according to the scale and hydraulic characteristics of each transect. The horizontal averaging interval ranged from 15 to 40 m, and the vertical ADCP cell interval ranged from 0.3 to 1.0 m (

Table 1. Spatial averaging configuration applied to the ADCP measurements. Horizontal averaging intervals were selected according to the transect width, whereas vertical intervals correspond to the ADCP depth-cell size used for valid velocity averaging.

Section	BJ1	NJ1	BN	PJ1	PJ2	PS	PH3	PH2	PH1	KS1	KS2	KS3
Horizontal averaging interval (m)	20	20	40	30	20	30	25	20	25	20	15	25
Vertical ADCP cell interval (m)	0.6	0.7	0.7	0.8	0.8	0.8	1.0	1.0	1.0	0.3	0.3	0.4
Transect distance (m)	500	400	1400	1100	650	1000	750	500	700	500	400	750
Mean depth (m)	10	13	13	17	16	15	21	21	22	4	4	6

). Depth-averaged velocities were calculated using only valid ADCP velocity cells after quality screening. These averaging intervals were selected to reduce local measurement noise while preserving the major flow structures associated with tributary inflow, confluence mixing, island-induced hydraulic transitions, lower-layer intrusion, and vertical stratification [13].

Hydraulic measurements were acquired using the ADCP under low-speed navigation conditions. Depth-averaged velocity fields and cross-sectional flow structures were derived from the ADCP observations [7,14] and subsequently used to evaluate tributary interactions and reservoir-scale mixing behavior. To ensure data reliability, raw ADCP measurements were subjected to quality-control procedures including the removal of invalid ensembles, bottom-tracking errors, side-lobe interference, and isolated velocity spikes. Velocity data were further spatially averaged where appropriate to reduce navigation-related noise while preserving major hydrodynamic features. Additional quality-assurance procedures included instrument calibration, GPS synchronization, and post-processing verification of signal correlation and velocity consistency.

Invalid ADCP ensembles were excluded when navigation records were missing, water depth was unreliable, or velocity values were physically inconsistent with neighboring ensembles. Bottom-tracking errors were screened by comparing bottom-track velocity, GPS trajectory, and depth continuity along each transect. Measurements affected by side-lobe interference near the bed were removed because acoustic reflection from the bed can contaminate velocity estimates in the lowest measurable cells. Isolated velocity spikes were identified from abrupt deviations from adjacent ensembles and surrounding vertical bins and were removed before spatial averaging. These procedures reduced measurement noise while preserving coherent flow structures such as tributary propagation, reverse flow, and lower-layer intrusion.

For ADCP quality control, beam correlation was evaluated using the internal quality-control algorithm of the RiverSurveyor software rather than an externally specified threshold. Accordingly, no additional user-defined correlation-coefficient threshold was applied. Correlation scores recorded by RiverSurveyor for SmartPulseHD samples were inspected together with velocity consistency, bottom-track continuity, and depth continuity during post-processing. The minimum beam signal-to-noise ratio (SNR) threshold for water-velocity calculation was maintained at the default RiverSurveyor value of 1.0 dB. This manufacturer-recommended setting was retained to avoid introducing an arbitrary site-specific filtering criterion [15].

Water-quality observations were conducted simultaneously along the survey transects. Water temperature, pH, dissolved oxygen (DO), electrical conductivity (EC), chlorophyll-a (Chl-a), turbidity, total organic carbon (TOC), total nitrogen (TN), and total phosphorus (TP) were continuously measured [16]. Horizontal spatial distributions were obtained during boat-based surveys, whereas vertical profiles were collected through water-column observations at selected stations. Geographic positioning system (GPS) data were recorded concurrently to define the precise spatial locations of all measurements.

The manufacturer-stated accuracy or specification values of the principal instruments used in the field survey are summarized in

Table 2. Manufacturer-stated accuracy or specification values of the instruments used in this study [17,18].

Instrument	Variable	Accuracy
ADCP	Velocity	±0.25% of measured velocity ±2 mm/s
	Depth	±1% of measured depth
EXO2	Temperature	±0.2 °C
	Conductivity	±1% of reading or 2 µS/cm
	Turbidity	±2% of reading or 0.3 FNU%
	pH	±0.1 pH unit
	Dissolved oxygen	±0.1 mg/L or ±1% of reading
	Chlorophyll-a	relative fluorescence measurement; used as auxiliary water-quality indicator

. Because repeated stationary ADCP observations were not conducted at every transect, total field measurement uncertainty could not be independently estimated from repeated observations. Therefore, measurement uncertainty was controlled and reported based on manufacturer-stated instrument specifications, pre-survey EXO2 calibration, GPS synchronization, low-speed navigation, removal of invalid ADCP ensembles, screening of bottom-tracking errors and side-lobe interference, and post-processing verification of spatial consistency among adjacent observations [17,18,19].

The monitored water-quality variables may originate from different watershed and in-reservoir sources. Turbidity is primarily associated with storm-driven soil erosion, suspended sediment transport, and possible bed-sediment resuspension. EC reflects dissolved ions derived from watershed geology, urban runoff, wastewater inputs, and dam-regulated tributary water masses. TN and TP are commonly supplied by agricultural runoff, urban non-point sources, wastewater effluent, and internal release from sediments. TOC may originate from terrestrial organic matter, wastewater inputs, and in-reservoir biological production, whereas Chl-a reflects algal biomass controlled by nutrient availability, light, temperature, and residence time. DO is affected by atmospheric reaeration, photosynthesis, respiration, and organic matter decomposition. These sources are important for tributary-reservoir mixing because different tributaries can transport different constituent loads, and incomplete mixing or density-driven intrusion can determine whether these materials remain in the surface layer, move laterally along a tributary corridor, or are transported downstream through the lower layer.

Although multiple water-quality parameters were monitored, water temperature and EC were selected as the primary indicators for evaluating density-driven mixing processes because they directly influence water density and behave as relatively conservative tracers during short-term hydrodynamic events. Turbidity was additionally used to identify suspended-sediment-rich inflows during flood conditions. Other variables, including DO, Chl-a, TOC, TN, and TP, were primarily used to characterize overall water-quality conditions and support interpretation of tributary-specific characteristics.

Although the absolute change in freshwater density caused by temperature and dissolved ions during flood events is generally small, even small relative density differences can be hydrodynamically important in weakly mixed reservoir environments. In Paldang Reservoir, tributary inflows with contrasting temperature, EC, and suspended sediment concentrations can retain distinct water-mass properties over short event timescales. Therefore, temperature and EC were used not to imply large bulk-density changes, but as practical indicators of relative density contrast and conservative water-mass tracing during tributary-reservoir mixing.

Three survey cases were investigated under contrasting hydrologic and hydraulic conditions. CASE1 represented a post-flood condition characterized by elevated tributary discharge resulting from antecedent precipitation and delayed watershed runoff. Although precipitation during the survey period was negligible, hydraulic conditions remained strongly influenced by preceding storm events. CASE2 and CASE3 represented normal-flow conditions under intermittent and continuous dam-discharge operations, respectively. These contrasting cases were selected to evaluate how tributary inflow characteristics and hydraulic regulation influence density-driven mixing, vertical stratification, and reservoir circulation patterns within Paldang Reservoir.

2.2. Spatial Interpolation and Data Processing

Spatial interpolation was applied to reconstruct continuous hydraulic and water-quality distributions from discrete field observations obtained during the monitoring surveys. Because complete spatial coverage of the reservoir could not be achieved under field conditions, interpolation techniques were required to estimate spatial flow structures, mixing interfaces, and water-quality gradients from strategically distributed observation data [20,21,22,23].

The interpolation procedure assumed that neighboring observation points exhibit similar hydraulic and water-quality characteristics owing to spatial autocorrelation within the reservoir system [24]. Using this principle, continuous spatial distributions of flow velocity, water temperature, and electrical conductivity (EC) were reconstructed to evaluate tributary interactions, density-driven mixing processes, and stratification structures within Paldang Reservoir.

Several interpolation approaches have been widely applied in hydrologic and environmental studies, including inverse distance weighting (IDW), Natural Neighbor, Spline, and Kriging methods [25,26]. Among these approaches, Empirical Bayesian Kriging (EBK) was selected in the present study because it effectively represents spatial variability within heterogeneous environmental datasets while explicitly accounting for uncertainty associated with semivariogram estimation [27,28].

Unlike IDW, which estimates values primarily according to distance weighting [29], EBK incorporates spatial covariance relationships among neighboring observations through a geostatistical framework. Furthermore, conventional Kriging methods typically rely on a single fitted semivariogram model, whereas EBK repeatedly simulates semivariogram parameters using a Bayesian approach and incorporates their uncertainty into the interpolation process. This procedure reduces interpolation bias caused by spatial heterogeneity and improves prediction reliability for irregularly distributed field observations.

In the EBK workflow, multiple semivariogram realizations were generated iteratively to characterize uncertainty associated with spatial covariance estimation. The probability distributions of these simulated semivariograms were subsequently integrated to estimate interpolation surfaces and prediction uncertainty simultaneously. Because semivariogram parameters are automatically updated during the iterative simulation process, subjective selection of a single semivariogram model is minimized, resulting in more stable interpolation performance for complex environmental datasets.

This characteristic is particularly advantageous for river-type reservoirs where abrupt hydraulic transitions, density-driven flow separation, and incomplete mixing frequently occur. In Paldang Reservoir, multiple tributary inflows with distinct hydraulic and water-quality characteristics generate complex mixing interfaces that are difficult to represent using deterministic interpolation methods. Therefore, EBK was considered suitable for reconstructing spatial patterns associated with tributary propagation, density-driven intrusion, and stratification processes.

Spatial interpolation and raster generation were performed using ArcGIS Pro 3.0 (Esri Inc., Redlands, CA, USA). For each variable, the interpolation workflow consisted of exploratory data screening, semivariogram simulation, spatial prediction, and cross-validation. Leave-one-out cross-validation was conducted to evaluate interpolation performance using statistical indicators including mean error (ME), root-mean-square error (RMSE), and standardized prediction error. These diagnostics were used to assess whether the interpolated surfaces adequately reproduced the observed spatial gradients while minimizing artificial boundaries and interpolation artifacts.

The reconstructed hydraulic and water-quality distributions were subsequently integrated with field observations to provide continuous spatial representations of flow velocity, water temperature, and EC. These interpolated datasets formed the basis for analyses of tributary interaction, density-driven mixing behavior, lower-layer intrusion, and vertical stratification within the reservoir.

2.3. Data Analysis and Quantitative Mixing Metrics

Integrated hydraulic and water-quality datasets were analyzed to evaluate three-dimensional mixing behavior and tributary interactions within Paldang Reservoir. Spatial analyses were performed using the interpolated hydraulic and water-quality distributions described in Section 2.2, together with the original field observations. The integrated dataset enabled interpretation of circulation structures, mixing interfaces, stratification patterns, and tributary propagation pathways under different hydrologic and hydraulic conditions [30,31,32,33].

To support quantitative comparisons among survey cases, tributary discharge ratios and fractional inflow contributions were calculated using measured inflow conditions. The South Han-to-North Han discharge ratio decreased from 5.49 in CASE1 to 2.68 in CASE2 and 1.16 in CASE3, indicating a progressive transition from strongly South Han-dominated hydraulic forcing toward more balanced tributary contributions. Similarly, the South Han River contribution to the combined South Han–North Han inflow decreased from 84.6% in CASE1 to 72.8% in CASE2 and 53.8% in CASE3. These indicators were used to characterize differences in tributary forcing conditions among survey periods and to support interpretation of observed mixing behavior.

Density-driven mixing processes were evaluated through integrated analyses of flow velocity, water temperature, and electrical conductivity (EC). Water temperature and EC were selected as the principal indicators because both variables directly influence water density and behave as relatively conservative tracers during short-term hydrodynamic events. Consequently, spatial gradients of temperature and EC provided effective indicators for identifying tributary intrusion pathways, stratification interfaces, and incomplete mixing processes within the reservoir.

To provide a quantitative indicator of stratification strength, Richardson number (Ri) values were calculated at the major tributary inflow cross-sections (BJ1, NJ1, and PH1) for each survey case. The cross-section codes represent the North Han River inflow section (BJ1), the South Han River inflow section (NJ1), and the section near Paldang Dam (PH1). The Gyeongan Stream inflow section (KS1) was excluded from the Ri analysis because its shallow water depth can make vertical density and velocity gradients highly sensitive to small measurement errors, resulting in potentially overestimated or underestimated Ri values. Ri was used as a dimensionless stability index linking vertical density gradients with velocity shear, expressed conceptually as Ri = gHΔρ/(ρU²), where g is gravitational acceleration, rho is water density, d rho/dz is the vertical density gradient, and du/dz is vertical velocity shear. The vertical density gradient was estimated from field-measured vertical water-temperature profiles collected using the EXO2 sonde at each representative cross-section. Representative flow velocities were obtained from ADCP-derived depth-averaged velocities measured at the corresponding cross-sections. The calculated Ri values were used for relative comparison of stratification stability among the three survey cases rather than for absolute stability assessment.

In this study, Ri values lower than 0.25 were interpreted as dynamically unstable or mixing-prone conditions, whereas values greater than 0.25 were interpreted as relatively stable stratification conditions. The Ri analysis was used to complement the qualitative interpretation of EC, temperature, turbidity, and velocity distributions.

Density-related mixing behavior was interpreted by combining hydraulic observations with spatial distributions of temperature and EC. Areas exhibiting persistent thermal and conductivity gradients together with vertically separated flow structures were interpreted as density-controlled mixing interfaces. Particular attention was given to identifying lower-layer intrusion, upper-layer separation, and incomplete mixing associated with tributaries possessing contrasting hydraulic and water-quality characteristics.

Longitudinal, transverse, and vertical analyses were conducted to evaluate spatial variations in flow structure and mixing behavior throughout the reservoir. ADCP-derived velocity fields were integrated with interpolated water-quality distributions to investigate circulation patterns, tributary propagation pathways, and the development of mixing interfaces. Cross-sectional analyses further focused on the formation of stratification structures and their modification under contrasting hydraulic operation conditions.

Comparisons among the three survey cases enabled evaluation of the influence of tributary inflow characteristics, hydrologic forcing, and dam discharge operation on reservoir mixing dynamics. Through integrated spatial visualization and cross-sectional analyses, multidimensional datasets describing density-driven mixing behavior and stratification processes were generated. These analyses provided the basis for evaluating the role of hydraulic regulation in controlling spatial mixing structures within Paldang Reservoir.

The analytical framework developed in this study provides a practical approach for investigating density-driven mixing processes in regulated river-type reservoirs and may support hydrodynamic interpretation and water-quality management in other multi-tributary reservoir systems.

3. Results

3.1. Hydrologic Conditions and Flow Regime Characteristics

Hydrologic and hydraulic conditions corresponding to the three survey cases are summarized in Figure 3 and observed in Paldang Reservoir, thereby enabling evaluation of their influence on density-driven mixing and stratification processes (

Table 3. Hydrologic and hydraulic conditions for each survey case in Paldang Reservoir.

Case	Hydrologic Condition	Survey Day Precipitation (mm)	Air Temperature (°C)	North Han River Discharge (m3/s)	South Han River Discharge (m3/s)	Gyeongan Stream Discharge (m3/s)
CASE1 (23.09.22)	Flood-season condition	0	18.40	175.3	962.2	16.61
CASE2 (23.10.27)	Normal-flow condition	0	13.85	57.13	153.2	2.33
CASE3 (24.04.17)	Normal-flow condition	0	14.56	263.63	306.6	2.77

). The survey campaigns were intentionally designed to represent contrasting tributary inflow conditions and dam-operation scenarios commonly.

CASE1 represented a post-flood hydraulic condition characterized by elevated tributary inflows and a pronounced discharge imbalance between the South Han River and North Han River systems. Although no precipitation occurred during the survey itself, the investigation was conducted three days after a major precipitation event. Consequently, substantial watershed runoff generated during the antecedent precipitation period continued to influence tributary discharge conditions during the field survey. The discharges of the North Han River, South Han River, and Gyeongan Stream were 175.3, 962.2, and 16.61 m³/s, respectively, resulting in a South Han-to-North Han discharge ratio of 5.49. This indicates that hydraulic forcing within the reservoir was strongly dominated by inflow from the South Han River.

CASE1 was not associated with precipitation during the survey day itself; rather, the elevated inflow reflected delayed watershed runoff following antecedent precipitation and increased upstream discharge through the Ipo Weir. Because runoff from the watershed can continue to enter the river network after precipitation has ceased, tributary discharge remained high during the field campaign. The high South Han River discharge and South Han-to-North Han discharge ratio therefore indicate that CASE1 represented a post-flood inflow condition rather than a direct precipitation-period observation.

CASE2 represented a normal-flow condition associated with intermittent discharge operation at the Cheongpyeong Dam. Tributary inflows were substantially lower than those observed during CASE1, with corresponding discharges of 57.13, 153.2, and 8.16 m³/s for the North Han River, South Han River, and Gyeongan Stream, respectively. The South Han-to-North Han discharge ratio decreased to 2.68, indicating weaker but still dominant hydraulic influence from the South Han River. Under these intermittent discharge conditions, downstream momentum supplied by the North Han River was periodically reduced, creating favorable conditions for unstable mixing interfaces and density-driven tributary intrusion within the confluence region.

CASE3 represented a normal-flow condition characterized by continuous discharge operation at the Cheongpyeong Dam. The discharges of the North Han River, South Han River, and Gyeongan Stream were 263.63, 306.6, and 2.77 m³/s, respectively. The corresponding South Han-to-North Han discharge ratio decreased to 1.16, indicating the most balanced tributary inflow condition among the three survey cases. Continuous discharge from the Cheongpyeong Dam maintained relatively stable downstream momentum in the North Han River system, thereby suppressing reverse intrusion from the South Han River and promoting persistent longitudinal flow propagation.

Temporal variations in tributary discharge and hydraulic operation conditions are presented in Figure 3. CASE1 was characterized by elevated inflow conditions associated with antecedent precipitation and increased discharge from the Ipo Weir (Figure 3a). CASE2 exhibited alternating discharge and non-discharge periods resulting from hydropower operation at the Cheongpyeong Dam (Figure 3b), whereas CASE3 showed relatively continuous and stable discharge conditions throughout the survey period (Figure 3c).

Long-term hydrologic characteristics were further evaluated using flow-duration curves (FDCs) constructed from discharge records for 2014–2025 (Figure 4). For each hydraulic structure, discharge values were arranged in descending order from the largest to the smallest value, and the rank-based exceedance percentage was calculated for each discharge value to construct the FDC. Because an FDC represents the percentage of time that discharge equals or exceeds a given value, it provides a useful basis for distinguishing post-flood, high-flow, and normal-flow hydraulic conditions in river and reservoir analyses. The FDC of the Cheongpyeong Dam exhibited a relatively gentle slope, indicating that discharge conditions were strongly regulated by hydropower operation and maintained within a comparatively narrow range. In contrast, the FDC of the Ipo Weir displayed substantially greater variability, reflecting stronger dependence on natural hydrologic conditions and watershed runoff processes (Figure 4).

The locations of the three survey cases within the long-term flow-duration distributions are indicated by colored markers in Figure 4. CASE1 corresponded to relatively high-flow conditions, whereas CASE2 and CASE3 represented normal-flow conditions. However, despite similar overall flow regimes, CASE2 and CASE3 differed substantially in hydraulic operation patterns because of intermittent and continuous dam discharge conditions, respectively. In the FDC-based survey sequence, CASE1, CASE2, and CASE3 correspond to the second, third, and sixth field campaigns, respectively.

These contrasting hydrologic and hydraulic conditions provide a useful framework for evaluating how tributary inflow characteristics and dam operation jointly influence tributary interaction, density-driven mixing, stratification development, and vertical flow structures within Paldang Reservoir. The results presented in the following sections are interpreted within this hydrologic context.

3.2. Spatial Mixing Behavior Based on Water Temperature and EC Distributions

Spatial distributions of electrical conductivity (EC) and water temperature for the three survey cases are presented in Figure 5. Distinct spatial mixing structures associated with tributary interactions were observed under different hydrologic and hydraulic conditions. In all cases, clear spatial heterogeneity persisted between the South Han River and North Han River inflows, indicating incomplete mixing and the preservation of tributary-specific water-mass characteristics within the reservoir.

During CASE1, corresponding to post-flood conditions with a South Han-to-North Han discharge ratio of 5.49, the South Han River exerted dominant hydraulic control on the reservoir system. Relatively high EC water originating from the South Han River propagated toward the downstream reservoir region, whereas lower EC water associated with the North Han River remained concentrated within the upstream tributary corridor (Figure 5a). Water-temperature distributions exhibited a similar pattern, with relatively warm water extending from the South Han River toward the central confluence zone (Figure 5b). Despite the elevated inflow conditions and enhanced downstream transport, distinct thermal and conductivity gradients remained identifiable after tributary convergence. This observation indicates that complete turbulent mixing did not occur during downstream propagation and that tributary water masses largely retained their original density characteristics.

CASE2 exhibited substantially different spatial patterns under intermittent discharge conditions from the Cheongpyeong Dam. The South Han-to-North Han discharge ratio decreased to 2.68, reducing the dominance of the South Han River and allowing density-related mixing processes to become more pronounced. Distinct EC contrasts remained evident between the tributaries, and the mixing boundary became more clearly defined within the central confluence region (Figure 5c). Water-temperature distributions showed similarly strong spatial separation, with warm water occupying the North Han River corridor and cooler water associated with the South Han River inflow (Figure 5d). Compared with CASE1, the persistence of thermal and conductivity gradients over a broader spatial extent suggests that intermittent discharge operation reduced longitudinal momentum and limited lateral mixing, thereby promoting the maintenance of stratified mixing interfaces.

CASE3 represented continuous discharge conditions with the most balanced tributary inflows among the three cases (South Han-to-North Han discharge ratio = 1.16). Under these conditions, EC distributions indicated more continuous downstream propagation of tributary water masses, particularly within the North Han River corridor (Figure 5e). Water-temperature patterns also exhibited smoother longitudinal transitions than those observed during CASE2 (Figure 5f). Nevertheless, distinct thermal and conductivity gradients remained identifiable near the confluence region, demonstrating that complete mixing was not achieved even under sustained hydraulic forcing. Continuous discharge maintained stable downstream momentum and organized longitudinal transport pathways; however, density-related differences between tributaries continued to exert a strong influence on reservoir-scale mixing behavior.

Comparison of the three survey cases reveals that both tributary forcing and hydraulic regulation played important roles in controlling spatial mixing structures within Paldang Reservoir. Under post-flood conditions (CASE1), strong South Han River inflow promoted rapid downstream transport while maintaining distinct tributary signatures. Under intermittent discharge conditions (CASE2), reduced momentum enhanced the persistence of mixing interfaces and stratified flow structures. Under continuous discharge conditions (CASE3), tributary transport became more organized, but density-driven mixing boundaries remained evident throughout the confluence region.

Overall, the EC and temperature fields indicate that tributary inflows with contrasting density characteristics formed persistent mixing interfaces and spatially heterogeneous water masses within Paldang Reservoir. The maintenance of distinct thermal and conductivity gradients after tributary convergence indicates that density-driven tributary interactions play a fundamental role in governing reservoir-scale mixing behavior. Furthermore, the results show that hydraulic regulation modifies the extent and persistence of these mixing interfaces by altering tributary momentum, residence time, and downstream transport pathways. These findings provide an important basis for interpreting the three-dimensional flow structures and stratification patterns discussed in the following sections.

3.3. Spatial Flow-Velocity Structure and Hydraulic Interaction

Spatial distributions of depth-averaged flow velocity for the three survey cases are presented in Figure 6. The observed velocity fields revealed substantial differences in hydrodynamic structure according to tributary inflow conditions and hydraulic operation patterns. These variations directly influenced tributary propagation pathways, advective transport processes, and the persistence of density-driven mixing interfaces within Paldang Reservoir.

During CASE1, corresponding to post-flood conditions, strong downstream-oriented flow propagation developed throughout the reservoir owing to elevated tributary inflows (Figure 6a). The South Han River exhibited substantially higher flow velocities than the North Han River, with representative velocities of approximately 0.30 m/s and 0.07 m/s, respectively. Following tributary convergence near sections BN and PJ1, flow velocities remained relatively high throughout the downstream reservoir reach, indicating efficient longitudinal transport toward Paldang Dam. Localized flow acceleration was observed near Sonae Island and the dam approach channel, where channel constriction enhanced downstream momentum. These results demonstrate that hydrodynamic conditions during CASE1 were primarily controlled by strong South Han River forcing, consistent with the high discharge ratio observed during this period.

The coexistence of strong downstream transport and persistent thermal and conductivity gradients suggests that advective transport alone was insufficient to achieve complete mixing. Instead, tributary water masses retained distinct physicochemical characteristics during downstream propagation, indicating that density-driven processes continued to influence mixing behavior even under relatively energetic flow conditions.

CASE2 exhibited markedly different hydrodynamic characteristics under intermittent discharge conditions from the Cheongpyeong Dam (Figure 6b). Flow velocities throughout the reservoir were substantially lower than those observed during CASE1, with most regions exhibiting velocities below 0.05 m/s. Localized flow acceleration occurred around Jokja Island, where maximum velocities approached 0.10 m/s; however, large portions of the reservoir remained characterized by weak advective transport. Particularly stagnant conditions developed within the Gyeongan Stream region, where flow velocities decreased to approximately 0.02 m/s.

The reduction in downstream momentum under intermittent discharge conditions likely increased water residence time and reduced lateral exchange between tributary water masses. These hydraulic conditions are consistent with the pronounced thermal and conductivity gradients observed in Figure 5 and provide favorable conditions for the maintenance of incomplete mixing and vertically stratified flow structures.

CASE3 represented continuous discharge conditions and exhibited more spatially organized flow propagation than CASE2 (Figure 6c). Average flow velocities within the South Han River and North Han River corridors were approximately 0.08 m/s, and continuous downstream transport was maintained throughout the reservoir. Following tributary convergence, flow velocities remained relatively stable along the main channel, while locally intensified flow conditions developed near the PH2 section, where velocities increased to approximately 0.12 m/s.

A notable feature of CASE3 was the occurrence of localized reverse-flow behavior near KS3 on the left side of Sonae Island. This reverse-flow pattern indicates the formation of a recirculation zone associated with tributary interaction and stable hydraulic forcing conditions. Such circulation structures may substantially increase local residence time, delay lateral mixing, and promote the persistence of density-driven stratification within the confluence region.

Comparison of the three survey cases demonstrates that hydraulic regulation strongly influenced both the magnitude and spatial organization of reservoir flow structures. CASE1 was characterized by strong advective transport associated with flood-season inflows, whereas CASE2 exhibited reduced momentum and weak circulation under intermittent discharge conditions. CASE3 maintained more stable downstream transport but also generated localized recirculation structures capable of preserving density gradients.

When interpreted together with the EC and temperature distributions presented in Figure 5, these results indicate that reservoir mixing behavior is governed by the interaction between advective transport and density-driven processes. Strong advection promotes tributary propagation and downstream transport, whereas reduced momentum and localized circulation favor the persistence of density gradients and incomplete mixing. Consequently, hydraulic regulation influences not only flow velocity but also the development, maintenance, and spatial extent of density-controlled mixing interfaces within Paldang Reservoir. These findings provide the hydrodynamic basis for understanding the three-dimensional stratification structures discussed in the following section.

3.4. Cross-Sectional Stratification and Mixing Structure

Cross-sectional distributions of turbidity, electrical conductivity (EC), and water temperature for the three survey cases are presented in Figure 7. The observed vertical structures provide direct evidence of density-driven mixing processes occurring within Paldang Reservoir. In particular, distinct lower-layer intrusion, upper-layer separation, and vertically stratified flow structures were consistently observed near the tributary confluence region upstream of Jokja Island, indicating that tributary water masses remained only partially mixed following confluence.

During CASE1, corresponding to post-flood conditions, strong contrasts in turbidity, EC, and temperature developed between the South Han River and North Han River inflows (Figure 7a). Surface turbidity concentrations were approximately 42.7 mg/L in the South Han River and 9.1 mg/L in the North Han River, resulting in a turbidity difference of 33.6 mg/L. The highly turbid South Han River water propagated beneath the North Han River water mass after confluence, producing a distinct lower-layer intrusion structure. EC distributions exhibited a similar pattern, with relatively high-conductivity South Han River water concentrated within the lower portion of the cross section and lower-conductivity North Han River water occupying the upper layer. These observations indicate that density contrasts associated with suspended solids, temperature, and dissolved constituents collectively contributed to vertical flow separation during the post-flood period.

The case-based Ri results further supported the interpretation of heterogeneous mixing and stratification among the three survey cases (North Han River inflow, whereas higher Ri values at NJ1 (4.58) and PH1 (0.53) indicated relatively stable stratification at those inflow sections. CASE2 was interpreted as mostly stratified because NJ1 and PH1 showed high Ri values (1.40 and 2.41), although the lower BJ1 Ri value (0.16) indicated localized mixing. In CASE3, PH1 remained relatively stable (Ri = 0.35), while low Ri values at BJ1 and NJ1 (0.05 and 0.09) suggested mixing-prone conditions near the inflow sections. These results indicate that the stability of density-driven mixing interfaces varied by both survey case and inflow cross-section (

Table 4. Richardson number (Ri) values calculated at the major inflow cross-sections (BJ1, NJ1, and PH1) for each case (Ri < 0.25 indicates dynamically unstable or mixing-prone conditions, whereas Ri > 0.25 indicates relatively stable stratification; KS1 was excluded because of shallow-water limitations).

Case	BJ1 Ri	NJ1 Ri	PH1 Ri	Dominant Condition	Interpretation
CASE1	0.02	4.58	0.53	Mixing-prone	Low BJ1 Ri indicates localized shear-driven mixing, whereas higher NJ1 and PH1 Ri values indicate relatively stable stratification.
CASE2	0.16	1.40	2.41	Mostly stratified	High NJ1 and PH1 Ri values indicate stable stratification; the low BJ1 value indicates localized mixing.
CASE3	0.05	0.09	0.35	Stratified	The PH1 value indicates relatively stable stratification, whereas low BJ1 and NJ1 values indicate localized mixing near the inflow sections.

). In CASE1, the low Ri value at BJ1 (0.02) indicated localized shear-driven mixing near the North Han River inflow, whereas the higher Ri values at NJ1 (4.58) and PH1 (0.53) indicated relatively stable stratification at those sections.

CASE2 exhibited the most pronounced density-driven stratification among the three survey cases (Figure 7b). Surface EC values were approximately 280.4 μS/cm in the South Han River and 100.8 μS/cm in the North Han River, corresponding to an EC contrast of 179.6 μS/cm. This strong conductivity difference generated a distinct two-layer structure in which high-density South Han River water intruded beneath the North Han River water mass. Intermittent non-discharge conditions at the Cheongpyeong Dam reduced downstream momentum from the North Han River, allowing lower-layer intrusion to extend farther into the confluence region. Opposing propagation directions between upper and lower layers were locally observed, indicating unstable stratification interfaces and partial reverse-flow behavior. These results suggest that reduced advective transport enhanced the persistence of density gradients and promoted incomplete vertical mixing.

CASE3 displayed substantially different cross-sectional characteristics despite maintaining a large conductivity contrast between tributaries (approximately 164.2 μS/cm; Figure 7c). Continuous discharge from the Cheongpyeong Dam sustained downstream momentum within the North Han River corridor and suppressed large-scale reverse intrusion from the South Han River. Consequently, lower-layer intrusion structures became less distinct than those observed during CASE2, and more stable longitudinal transport developed throughout the confluence region. Nevertheless, persistent EC and temperature gradients remained evident within the water column, indicating that density-driven stratification continued to influence vertical flow organization even under sustained hydraulic forcing.

Comparison of the three survey cases reveals that the intensity and spatial extent of lower-layer intrusion were strongly controlled by both tributary density contrasts and hydraulic regulation. Although substantial density differences existed during all survey periods, the resulting stratification structures varied according to hydrodynamic forcing conditions. Under post-flood conditions (CASE1), strong downstream transport coexisted with density-driven lower-layer intrusion. Under intermittent discharge conditions (CASE2), reduced momentum allowed density-driven processes to dominate, resulting in the strongest stratification and most extensive lower-layer intrusion. Under continuous discharge conditions (CASE3), stable downstream momentum partially suppressed reverse intrusion while maintaining persistent density interfaces.

Overall, the cross-sectional analyses demonstrate that tributary density differences alone do not determine mixing behavior within Paldang Reservoir. Instead, the interaction between density contrasts and hydraulic regulation governs the formation, persistence, and spatial extent of lower-layer intrusion and vertically stratified mixing structures. These observations provide direct field evidence that dam operation can substantially modify density-driven mixing processes in regulated river-type reservoirs. Furthermore, the identified stratification patterns establish the hydrodynamic foundation for interpreting the circulation structures and density-controlled transport mechanisms discussed in the following section.

4. Discussion

The present study demonstrates that density-driven mixing in Paldang Reservoir is governed by the combined effects of tributary density contrasts and hydraulic regulation. Although discharge magnitude influenced overall flow intensity and tributary propagation, the observed mixing structures could not be explained solely by hydrologic forcing. Rather, dam-controlled flow conditions interacted with density gradients to determine the formation, persistence, and spatial extent of stratified mixing interfaces.

Previous studies have shown that density differences at river confluences can generate vertically separated flow structures, lower-layer intrusion, and incomplete mixing even after tributary convergence [3,6,34]. The results obtained in Paldang Reservoir are consistent with these observations, demonstrating that water masses possessing different thermal and conductivity characteristics may remain partially separated over considerable downstream distances.

However, unlike many previous studies that primarily focused on natural river confluences, laboratory experiments, or idealized numerical simulations, the present study demonstrates how hydraulic regulation imposed by upstream dam operation modifies density-driven mixing under actual reservoir conditions. Field observations revealed that changes in dam discharge altered tributary momentum, residence time, circulation patterns, and the persistence of density-driven mixing interfaces. These findings extend previous confluence studies by demonstrating that hydraulic operation is an additional controlling factor governing three-dimensional mixing behavior in regulated river-type reservoirs.

The case-based Ri values provide quantitative support for the coexistence of mixing-prone and stratified conditions within Paldang Reservoir. Low Ri values at selected inflow sections indicate conditions favorable for vertical mixing, whereas higher Ri values indicate stronger resistance to vertical mixing and relatively stable density stratification. These results are consistent with the observed lower-layer intrusion, localized reverse-flow behavior, and persistent conductivity gradients. Thus, density-driven mixing in the reservoir cannot be characterized by a single basin-scale condition; rather, it varies according to survey case, tributary density contrasts, and local hydraulic forcing.

One of the most important findings of this study is that the strongest lower-layer intrusion was observed during CASE2 rather than CASE1, despite the larger discharge magnitude observed during the flood-season condition. During CASE2, intermittent discharge operation at the Cheongpyeong Dam reduced downstream momentum within the North Han River system and weakened advective transport across the confluence region. As a result, density differences between the South Han River and North Han River became the dominant controlling mechanism, allowing high-conductivity South Han River water to intrude beneath the North Han River water mass and generate pronounced two-layer flow structures. This observation indicates that reduced hydraulic forcing may enhance density-driven stratification even when overall discharge conditions are lower than those observed during flood periods.

In contrast, CASE3 demonstrated that continuous discharge operation can partially suppress density-driven intrusion despite the persistence of substantial conductivity and temperature differences between tributaries. Continuous downstream momentum maintained stable longitudinal transport pathways and limited reverse intrusion from the South Han River into the North Han River corridor. Nevertheless, distinct thermal and conductivity interfaces remained evident, indicating that hydraulic regulation modified, but did not eliminate, density-driven mixing processes. These findings suggest that dam operation primarily controls the spatial expression of density-driven mixing rather than the existence of density gradients themselves.

The velocity observations further revealed the importance of localized circulation structures in controlling reservoir-scale mixing behavior. In particular, the recirculation zone observed near KS3 during CASE3 suggests that topographic controls and tributary interaction can create localized regions of prolonged residence time. Such circulation zones may delay lateral mixing, preserve density gradients, and promote heterogeneous water-quality distributions even when large-scale flow conditions appear relatively stable. Similar circulation-induced mixing limitations have been reported in other confluence and reservoir systems, emphasizing the importance of considering both density effects and local hydrodynamic structures when interpreting mixing behavior.

The integrated use of hydraulic observations together with water temperature and electrical conductivity (EC) proved effective for identifying tributary propagation pathways, density-driven intrusion, and incomplete mixing within Paldang Reservoir. Water temperature reflected density differences between tributary inflows, whereas EC served as a relatively conservative tracer during the short transport times associated with the field surveys, allowing clear identification of tributary water masses and mixing interfaces. In contrast, variables such as TP, TN, Chl-a, TOC, and DO were more strongly influenced by biogeochemical processes and local source inputs and therefore served primarily as supporting indicators of water-quality conditions rather than direct tracers of hydraulic mixing.

Specifically, DO can be increased by photosynthesis and atmospheric reaeration but decreased by respiration and microbial decomposition of organic matter [35]. Chl-a can increase through algal growth under favorable nutrient, light, and residence-time conditions and decrease through settling, grazing, or flushing. TOC can be modified by external organic matter inputs, algal production, microbial degradation, and sediment-water exchange. TN can be transformed through mineralization, nitrification, denitrification, biological uptake, and settling of particulate nitrogen. TP can be modified by algal uptake, particulate settling, sediment resuspension, and release from bottom sediments under reducing conditions. For this reason, these variables are less conservative than temperature and EC over short-term hydrodynamic events and were interpreted as supporting indicators rather than primary tracers of physical mixing.

From a reservoir-management perspective, the results indicate that hydraulic regulation can substantially influence the location, persistence, and intensity of density-driven mixing interfaces. Because these interfaces affect constituent transport pathways, water-quality distribution, and ecological connectivity, operational decisions at upstream hydraulic structures may indirectly influence reservoir water-quality conditions. Consequently, consideration of density-driven mixing processes may improve the interpretation of reservoir monitoring data and support more effective management of regulated river-type reservoirs.

Several limitations should be acknowledged. First, the analyses were based on representative field surveys rather than continuous long-term monitoring, and therefore short-term temporal variability could not be fully resolved. Second, the reconstructed spatial distributions depend on observation density and interpolation performance, although cross-validation procedures were applied to evaluate interpolation reliability. Third, the Richardson-number analysis was limited to representative inflow cross-sections for the three survey cases and was intended to provide relative comparisons of stratification stability rather than continuous three-dimensional stability fields across the entire reservoir. Future studies incorporating high-frequency monitoring, direct density profiling, expanded Richardson-number analysis, and three-dimensional hydrodynamic modeling would provide additional insight into stratification strength and mixing efficiency. Despite these limitations, the present study provides rare field-based evidence of density-driven mixing processes in a regulated river-type reservoir and establishes a practical framework for interpreting tributary interactions under variable hydraulic operation conditions.

Another limitation is that phosphorus- and algae-related variables were not analyzed as primary indicators of mixing dynamics. Although TP and Chl-a are important indicators of reservoir eutrophication and ecological water quality, their concentrations are strongly influenced by biological uptake, algal growth, settling, sediment release, and other transformation processes. Therefore, the present study focused on water temperature and EC as the primary indicators of density-driven mixing, while TP and Chl-a were used to support the interpretation of overall water-quality conditions. Future studies integrating hydrodynamic observations with nutrient cycling and algal dynamics would provide a more comprehensive understanding of the ecological consequences of density-driven mixing.

5. Conclusions

This study investigated spatial mixing behavior and stratification dynamics in Paldang Reservoir using integrated field observations of flow velocity, water temperature, turbidity, and electrical conductivity (EC) under contrasting tributary inflow and hydraulic operation conditions. Based on the results obtained from three representative survey cases, the following conclusions can be drawn:

(1) Spatial mixing behavior within Paldang Reservoir was strongly controlled by density differences between tributary inflows. Distinct thermal and conductivity gradients persisted after tributary convergence, indicating incomplete mixing and the preservation of tributary-specific water-mass characteristics throughout the reservoir.

(2) Density-driven lower-layer intrusion developed consistently near the confluence region upstream of Jokja Island. During the post-flood condition (CASE1), highly turbid and high-conductivity South Han River water propagated beneath the North Han River inflow, producing clear vertically stratified flow structures. These observations provide direct field evidence of density-controlled tributary interaction within a regulated river-type reservoir.

(3) Hydraulic operation substantially modified the spatial extent and intensity of density-driven mixing structures. Intermittent discharge conditions (CASE2) weakened downstream momentum and promoted unstable two-layer flow structures, reverse-flow behavior, and pronounced lower-layer intrusion. In contrast, continuous discharge conditions (CASE3) maintained stable downstream transport pathways and partially suppressed reverse intrusion while preserving persistent density interfaces.

(4) Integrated analyses of flow velocity, water temperature, turbidity, EC, and case-based Richardson number proved effective for identifying tributary propagation pathways, density-driven mixing interfaces, and stratification structures. In particular, EC functioned as a relatively conservative tracer for tracking tributary water masses, while Richardson number provided a quantitative indicator for comparing relative stratification stability among the survey cases. Overall, these findings demonstrate that density-driven mixing in Paldang Reservoir results from the interaction between tributary density contrasts and hydraulic regulation rather than from hydrologic forcing alone.

The principal contribution of this study is the field-based demonstration that density-driven mixing in a regulated river-type reservoir is governed not only by tributary inflow characteristics but also by hydraulic regulation imposed by upstream dam operation. By integrating high-resolution ADCP measurements with spatial water-quality observations, this study provides direct evidence that hydraulic regulation modifies tributary momentum, lower-layer intrusion, and three-dimensional stratified mixing processes under natural reservoir conditions. These findings suggest that adaptive dam-operation strategies considering tributary density differences may improve intake-water management and reduce the downstream transport of sediment-rich lower-layer water during high-flow events.

From a management perspective, the findings indicate that density-driven mixing and stratification should be considered when evaluating water-quality transport, intake-water conditions, and reservoir operation strategies. Depending on the transported constituents, persistent stratification may either limit rapid contaminant dispersion or facilitate the downstream transport of sediment-rich and pollutant-rich lower-layer water masses. Therefore, coordinated dam-operation strategies, event-based monitoring during high-inflow periods, and watershed-scale management of turbidity and nutrient sources are recommended to reduce adverse water-quality impacts.

Future studies integrating high-frequency monitoring, expanded Richardson-number analysis, direct density profiling, and three-dimensional hydrodynamic modeling would further improve understanding of density-controlled mixing processes and support adaptive management of regulated river-type reservoirs.