Abstract
Confluences act as crucial intersection in a fluvial system as they regulate the flow, sediment, and geomorphological stability. Two of the natural occurrences that have an effect on hydro-morphological processes that characterize the channel are sediment movement and stream discharge. However, assessments of suspended sediment distribution patterns across channel confluences and functions of tributaries on sediment transfer in downstream streams are still inadequately explored. To fulfil this, a year-long field investigation was conducted around river confluence at various time scales during the hydrological season. The bathymetry of the confluence was performed using digital echo-sounding equipment, electro-magnetic current meter, and differential global positioning equipment. This study uses in situ measurement and empirical techniques to quantify the spatiotemporal distribution of suspended sediment particles. Furthermore, Yang's formula was used to analyze the sediment concentration, and several statistical standards were employed to establish a differentiation between predicted and observed data. The results illustrated that a good agreement has been achieved among the calculated and measured statistics based on the correlation coefficient value. According to the findings, 8% fewer errors in the quantification of suspended sediment were induced in the tributary than in the main river because of the very low flow of the tributary. The present ramifications can help increase our database and understanding of the dynamics of major river confluences.
HIGHLIGHTS
A year-long field investigation was carried out to assess short-term suspended sediment rates and stream flow rates.
Statistical analysis was concluded between observed and predicted data.
The grain size of bed material varied significantly among individual channels in a variety of fluvial circumstances.
The tributary channel reflects the fluctuations along the longitudinal gradient of sediment rate.
INTRODUCTION
River junctions are realized as important fluvial characteristics since both thorough and widespread hydro-physical and geo-ecological processes occur at this junction (Rice et al. 2008; Edward & Latrubesse 2015; Yuan et al. 2019; Gualtieri et al. 2020). Hydro-morphodynamics in rivers of any size (large- or small-scale rivers) are unpredictable and complex, mainly in river networks where flows converge or diverge (Parsons et al. 2007; Leite Ribeiro et al. 2012; Constantinescu et al. 2016). Significant modifications in flow hydrodynamics, bed morphology, and environmental and ecological features happen at confluence due to the possible mixing of the sediment and pollutants from separate flows, adjusting to the post-confluence planform geometry (Nazari-Giglou et al. 2016; Tang et al. 2018; Gualtieri et al. 2019; Rhoads 2020; Yuan et al. 2022). Numerous academic fields, inclusive of sedimentology (e.g. Constantinescu et al. 2012; Zhang et al. 2014; Rhoads & Johnson 2018), hydraulics (Constantinescu et al. 2012, 2016; Sukhodolov & Sukhodolova 2019); and ecology (Benda et al. 2004; Blettler et al. 2014) have attempted to comprehend the function of fluvial networks. Confluences are naturally occurring elements in river network systems that control the movement of flow, sediment, and geomorphological stability. It is crucial to predict the large amount of sediment loads slightly more precisely corresponding to the carrying capacity of the river, especially for river networks. Various planform geometries, velocity ratios, bed discordance, and junction angles were used to assess the movement of sediments in river systems (Riley et al. 2014; Guillen Ludena et al. 2017; Sukhodolov et al. 2017; Lyubimova et al. 2020; Rhoads 2020). The fluctuation of the mixed flows at the surface level in the junction relies upon the streamflow condition and the annual hydrological regime. The structure and extent of river flow are changed owing to varying rainfall patterns, and this has an impact on sediment yield and transport patterns of sediment from tropical river basins. The dispersal of suspended sediment in the channel junction is influenced by the sediment deliveries over the main and lateral flows. The flow discharge and sediment discharge undergo drastic variations at the junctions (Szupiany et al. 2012; Khosronejad et al. 2016), which necessitate adjusting the interdependent channel form factors. The abrupt adaptation in the hydro-sedimentological environment of the receiver stream brought on by the tributary inflow of water and sediment at river junctions results in changes in physical conditions (Biron & Lane 2008). Although the tributaries have usual basin characteristics and their linking with the major stream changes in the sedimentation that causes the hydro-morphological alteration of the receiving downstream river. Therefore, the inputs and contributions of the tributaries have a significant impact on the concentrations of suspended sediment characteristics within the downstream channel. Quantifying the sediment load in the streams is one of the major fluvial issues that scientists are currently dealing with because it directly influences several hydro- morphological problems in channel beds. According to numerous empirical equations that are available or from actual quantification of sediment taken around the timespan, the suspended sediment load (SSL) of the stream is typically approximated by Khanchoul et al. (2009). However, it continues to be a major issue for researchers to estimate SSL in small rivers with fast discharge changes.
Numerous studies have been undertaken in the past by Ferguson & Hoey (2008), Espinoza et al. (2012), Guillen Ludena et al. (2017), Gualtieri et al. (2018) to recognize the procedures intricate in motion of sediment through the estuary and the river network system. To examine the relationship among flow dynamics and bed morphology with downstream stream (Ragno et al. 2021), field investigations (Rhoads et al. 2009; Riley et al. 2014; Sukhodolov et al. 2017; Ianniruberto et al. 2018; Lane et al. 2008; Lewis et al. 2020; Yuan et al. 2021, 2022), laboratory tests (Escauriaza et al. 2012; Guillen Ludena et al. 2017; Cheng & Constantinescu 2018; Yuan et al. 2018; Wang et al. 2019; Yu et al. 2020; Zhang & Lin 2021; Balouchi et al. 2022) and modelling (Baranya & Jozsa 2013; Sukhodolov et al. 2017; Jiang et al. 2022) were carried out. At the junction of the Parana and Paraguay Rivers, measurements from a field survey by Parsons et al. (2007) showed how the cross-sectional velocity dispersion affects the flow and suspended bed sediment concentration, with the effect becoming more pronounced as the aspect ratio (W/H) of the channels rises. Confluences are highlighted as a site of discontinuance in the hydro-sediment continuance in a river system (Rice et al. 2008; Duncan et al. 2009). The studies of hydro-sediment dynamics and associated problems in the junction hydrodynamic region through Leite Ribeiro et al. (2012), Martín-Vide et al. (2015), and Nazari-Giglou et al. (2016) are significant in this context. According to Szupiany et al. (2012), the characteristics of rivers, like the development of islands, the growth of sizable lateral bars, and the presence of turbulent zones, might affect the transport of sediment. Artificial neural networks (ANNs) and machine learning (ML) were used by Tachi et al. (2016) to estimate suspended sediment discharge, but these methods did not demonstrate accuracy in forecasting suspended sediment concentration (SSC) based on the flow rate. The intricate interplay between water output and sediment concentration made it impossible to forecast SSC using artificial methods.
Literature surveys have developed our perception of the trend of riverine suspended sediment loads; however, our understanding of suspended sediment motion in these rivers is very limited. Accurately measuring sediment loads in alluvial rivers is a challenging and unrealistic task, which significantly impedes our knowledge. In most alluvial rivers, it is difficult and rare to acquire direct measurements of sediment transport, because sand and other particles of various sizes are typically combined with sediments. Following the review of the literature, it was found that no thorough research has been completed on the dynamics of suspended sediment transfer, or the movement of sediment in the water column, from any perspective, particularly with regard to the impact of unsteady rainfall and river discharge on sediment concentrations. So far, the quantification of the post-confluence channel's suspended sediment distribution patterns is still not realized. Therefore, it is necessary to assess the suspended sediment discharge under various hydraulic conditions at the junctions. This study examines the spatiotemporal changeability of suspended sediment distribution and estimates SSL for certain time periods following a junction over several hydrological seasons by employing both field measurement and empirical methods. In the present study, field data from the Gumti River and Pitragang tributary are used to calculate the total suspended sediment (TSS) transport rate using method of Yang (1979). Thus, it is hoped that a thorough investigation of this confluence would contribute to our understanding of river confluences and facilitate the better direction of this crucial river network system.
MATERIALS AND METHODS
Study area


Field procedures
Dimension and average hydraulic parameters of the junction at the time of bathymetric survey
Channel . | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() |
---|---|---|---|---|---|---|---|---|---|
Main (U/S) | 75.46 | 1.518 | 87.93 | 0.768 | 0.58 | 49.71 | 67,497 | 0.198 | 2.31 |
Main (D/S) | 84.92 | 1.451 | 109.06 | 0.885 | 0.63 | 58.53 | 96,528 | 0.235 | 2.55 |
Tributary | 35.18 | 0.914 | 20.98 | 0.652 | 0.61 | 38.49 | 13,689 | 0.218 | 1.18 |
Channel . | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() | ![]() |
---|---|---|---|---|---|---|---|---|---|
Main (U/S) | 75.46 | 1.518 | 87.93 | 0.768 | 0.58 | 49.71 | 67,497 | 0.198 | 2.31 |
Main (D/S) | 84.92 | 1.451 | 109.06 | 0.885 | 0.63 | 58.53 | 96,528 | 0.235 | 2.55 |
Tributary | 35.18 | 0.914 | 20.98 | 0.652 | 0.61 | 38.49 | 13,689 | 0.218 | 1.18 |
Schematic sketches of measurement cross-sections showing the sampling locations.








Here, W is the width of rivers (m), H is the water height (m), is the discharge in
,
is the mean flow velocity
,
is the water density
,
is the viscosity of water
, S is the slope of confluence channel,
is the hydraulic depth (m), M is the momentum flux
,
is Froude number of flow,
is Reynolds number of flow, and
is the aspect ratio.
Water levels were observed during the field study period (February 2021 and February 2022).
Water levels were observed during the field study period (February 2021 and February 2022).
Mean water quality parameters at the confluence downstream
Sl. no. . | Temperature (oC) . | Conductivity (μS/cm) . | Turbidity (NTU) . | pH . | SSC (mg/l) . |
---|---|---|---|---|---|
15–20 Feb 2021 | 28.4 | 134 | 102 | 7.2 | 121 |
24–30 May 2021 | 29.3 | 207 | 136 | 7.8 | 185 |
18–23 Aug 2021 | 28.1 | 232 | 158 | 8.2 | 194 |
14–19 Nov 2021 | 27.8 | 189 | 113 | 8.0 | 170 |
17–22 Feb 2022 | 28.1 | 116 | 97 | 7.5 | 104 |
Sl. no. . | Temperature (oC) . | Conductivity (μS/cm) . | Turbidity (NTU) . | pH . | SSC (mg/l) . |
---|---|---|---|---|---|
15–20 Feb 2021 | 28.4 | 134 | 102 | 7.2 | 121 |
24–30 May 2021 | 29.3 | 207 | 136 | 7.8 | 185 |
18–23 Aug 2021 | 28.1 | 232 | 158 | 8.2 | 194 |
14–19 Nov 2021 | 27.8 | 189 | 113 | 8.0 | 170 |
17–22 Feb 2022 | 28.1 | 116 | 97 | 7.5 | 104 |
Grain size particle
Percentages of clay, silt, and sand fractions for each branch channel
Branch channels . | % soil . | ||
---|---|---|---|
Clay . | Silt (mm) . | Sand . | |
![]() | 0.002–0.0475 . | 0.0475–2 . | |
Tributary | 1.04 | 4.68 | 94.28 |
Main channel (U/S) | 1.52 | 5.44 | 93.04 |
Main channel (D/S) | 1.76 | 7.31 | 90.93 |
Branch channels . | % soil . | ||
---|---|---|---|
Clay . | Silt (mm) . | Sand . | |
![]() | 0.002–0.0475 . | 0.0475–2 . | |
Tributary | 1.04 | 4.68 | 94.28 |
Main channel (U/S) | 1.52 | 5.44 | 93.04 |
Main channel (D/S) | 1.76 | 7.31 | 90.93 |
Measurement of SSL
Direct method



Empirical method








RESULTS AND DISCUSSION
Comparison between predicted and observed sediment concentration
At first glance, it seems that a minor difference in expected suspended sediment concentration between the two different methodologies (i.e. field data and computer data). Therefore, a statistical analysis is required to identify the specific distinctions between them. The accuracy of the predicted method was statistically analyzed by comparing it with observed data. Calculations for the mean absolute percentage error (MAPE), root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), and linear correlation coefficient were made as part of the statistical study (Table 4). The following equations are used to generate these statistical measures (Krause et al. 2005; Najafzadeh & Tafarojnoruz 2016). At this point, it is emphasized that Yang's calculations were used to get the overall sediment concentration.
Measured and derived statistical criteria for the confluence river
Channel . | Months/Statistical criteria . | 15–20 Feb 2021 . | 24–30 May 2021 . | 18–23 Aug 2021 . | 14–19 Nov 2021 . | 17–22 Feb 2022 . | Mean value . |
---|---|---|---|---|---|---|---|
MC | MAPE | 0.285 | 0.744 | 0.898 | 0.995 | 0.337 | 0.652 |
TC | 0.431 | 0.672 | 0.414 | 0.372 | 0.524 | 0.483 | |
MC | RMSE | 10.58 | 23.31 | 37.12 | 29.61 | 10.43 | 23.21 |
TC | 8.14 | 12.53 | 12.88 | 17.35 | 8.02 | 11.78 | |
MC | NSE | 0.481 | –2.092 | −4.596 | −2.884 | −0.342 | −1.887 |
TC | −1.148 | −2.975 | −0.309 | −1.453 | −2.697 | −1.716 | |
MC | r | 0.923 | 0.883 | 0.879 | 0.952 | 0.771 | 0.882 |
TC | 0.937 | 0.944 | 0.896 | 0.857 | 0.858 | 0.898 | |
MC | R2 | 0.851 | 0.739 | 0.773 | 0.929 | 0.894 | 0.837 |
TC | 0.878 | 0.891 | 0.826 | 0.803 | 0.806 | 0.841 |
Channel . | Months/Statistical criteria . | 15–20 Feb 2021 . | 24–30 May 2021 . | 18–23 Aug 2021 . | 14–19 Nov 2021 . | 17–22 Feb 2022 . | Mean value . |
---|---|---|---|---|---|---|---|
MC | MAPE | 0.285 | 0.744 | 0.898 | 0.995 | 0.337 | 0.652 |
TC | 0.431 | 0.672 | 0.414 | 0.372 | 0.524 | 0.483 | |
MC | RMSE | 10.58 | 23.31 | 37.12 | 29.61 | 10.43 | 23.21 |
TC | 8.14 | 12.53 | 12.88 | 17.35 | 8.02 | 11.78 | |
MC | NSE | 0.481 | –2.092 | −4.596 | −2.884 | −0.342 | −1.887 |
TC | −1.148 | −2.975 | −0.309 | −1.453 | −2.697 | −1.716 | |
MC | r | 0.923 | 0.883 | 0.879 | 0.952 | 0.771 | 0.882 |
TC | 0.937 | 0.944 | 0.896 | 0.857 | 0.858 | 0.898 | |
MC | R2 | 0.851 | 0.739 | 0.773 | 0.929 | 0.894 | 0.837 |
TC | 0.878 | 0.891 | 0.826 | 0.803 | 0.806 | 0.841 |
MC, main channel; TC, tributary channel.
The MAPE is an indicator of the quality of the approximation between predicted and measured values, given the magnitude of the actual quantity value. It is considered that the prediction is reasonably accurate if the MAPE value is less than 5%. When the MAPE is more than 10% but less than 25%, the accuracy is poor but acceptable, and when the MAPE is more than 25%, the accuracy is extremely low.
The RMSE computes the average discrepancy between the values predicted by a model and the actual values. It provides an estimate of the model's accuracy or how effectively it can anticipate the desired outcome. The RMSE ranges between 0 and +∞. The correlation between measured and calculated values improves with a lower RMSE.
NSE can be quantitatively defined by the correctness of model outputs. NSE ranges from −∞ to 1, with 1 being the optimal value.
Correlation coefficient r expresses the degree of mutual linear dependence between the variables and
, and ranges between values of −1 and +1. Here
is the measured sediment concentration;
is the predicted sediment concentration;
is the average value of
;
is the mean value of
and N is the number of data.
The magnitudes of the RMSE and NSE, supported the Yang's formula, can be considered to be rather satisfactory in the comparison between the observed and predicted approaches for the channel confluence described above. Also acceptable is the degree of linear dependency amid the predicted and the observed sediment concentration. The outcomes indicate that the values of NSE are clearly non-negative and there is a better identical between observed and predicted SSC with a mean determination coefficient of R2 = 0.84. The MAPE value for both predicted and measured values is less than 1% suggesting a high level of performance and accuracy in the anticipated results. According to the statistical requirements of this study, which are displayed in Table 4, Yang's approach adequately describes the suspended sediment concentrations in the Gumti and Pitragang rivers, and the findings are adequate.
Another important finding from this study is that, with average water discharge of the amount of suspended sediment that reaches the entrance of the confluence is rather small roughly 85 ton/year. During the bathymetric survey, the downstream tributary and Gumti River had mean SSC of 46 and 94 mg/l, respectively. The highest amount of suspended sediment was observed in August in the main river and in November in the tributary (Table 5). The performance of the results is based on a comparison of the quantification of suspended sediment yields applying two separate methods. The results emerging from empirical formulas usually differ from the measured data. In the current work, the empirical formula uses a mathematical formula to estimate the sediment load of the dataset based on sample data. The direct approach uses a dataset to calculate the SSL without making any assumptions. Although, the direct method is expensive, time-consuming and laborious. However, compared to the empirical method, the direct method is significantly more exact and accurate. The primary cause of error for the direct method was due to the heterogeneity of samples collected from the river, and a large enough spatial and temporal variation of the concentration and rate of sediment transport. This error source was random and could not be eliminated. In the present study, the relative error of the empirical method was 6.23%, while, the relative error of the direct method was less than 3.15%. Moreover, the difference in suspended sediment rate is dependent on the extent to which its tributary has an impact. There were eight percentages (8%) fewer errors in the quantification of suspended sediment induced in the tributary than in the main river because of the very low flow of the tributary. Fewer errors were found in the analysis of suspended sediment generated during the cloudy period than in the dry period.
Shows the observed and predicted suspended sediment transport rate (g/s)
Channel/Months . | Results . | 15–20 Feb 2021 . | 24–30 May 2021 . | 18–23 Aug 2021 . | 14–19 Nov 2021 . | 17–22 Feb 2022 . | Mean (g/s) . |
---|---|---|---|---|---|---|---|
Main | Calculated | 379 | 764 | 1,070 | 736 | 260 | 642 |
Observed | 304 | 458 | 885 | 371 | 192 | 442 | |
Tributary | Calculated | 42 | 54 | 116 | 140 | 61 | 83 |
Observed | 28 | 30 | 84 | 103 | 43 | 58 |
Channel/Months . | Results . | 15–20 Feb 2021 . | 24–30 May 2021 . | 18–23 Aug 2021 . | 14–19 Nov 2021 . | 17–22 Feb 2022 . | Mean (g/s) . |
---|---|---|---|---|---|---|---|
Main | Calculated | 379 | 764 | 1,070 | 736 | 260 | 642 |
Observed | 304 | 458 | 885 | 371 | 192 | 442 | |
Tributary | Calculated | 42 | 54 | 116 | 140 | 61 | 83 |
Observed | 28 | 30 | 84 | 103 | 43 | 58 |
Suspended sediment transport and sediment discharge
Based on the present study, two factors of interest that are believed to have fundamental effects on the confluence hydrodynamic zone (CHZ) are hypothesized to be tributary sediment yield and runoff volume. The study affirms that the dimension of the tributary compared to the mainstream affects sediment movement inside the CHZ of the main channel. Furthermore, the suspended load estimation suggests that notable spatiotemporal changeability can happen in sediment supplies from upstream to downstream reach of confluence. The yearly hydrological regime had an impact on the water mixture in August 2021; water mixing rates were highest followed by those in June and October. It is probable that the tributary was carrying clay (of which 1% were smaller than 0.002 mm), suspended silt (of which 5% were less than 0.05 mm) and sand (94%). On the other hand, the Gumti River was carrying some clay (of which 2% were smaller than 0.002 mm), silt (of which 7% of the particles were less than 0.05 mm), and fine sand (91%) (Figure 4). The average grain size of the suspended particles was identical in the Gumti River and Pitragang tributary. However, the grain size of bed material varied significantly among individual channels. The mean grain size of the tributary was only 0.27 mm and its bed material was much finer than that of the main river. The tributary river experienced severe erosion due to the increased suspended load during the high velocity survey periods.
Display the scatter plot of measured and predicted suspended sediment concentration rate: (a) tributary channel and (b) main channel (d/s).
Display the scatter plot of measured and predicted suspended sediment concentration rate: (a) tributary channel and (b) main channel (d/s).
Uncertainty and constraints in suspended load estimation
The calculation of sediment loads (Q = suspended load, Qs + bed load, Qb) in this work is to quantify SSL Qs repeatedly along transects in river confluences over a 1-year period. Most previous studies that quantified sediment loads along the large river made use of 2D or 3D models that included bedform contour height data from multi-beam echo-sounders. Because we did not have such measures; we merely employed Yang's equation and performed our tests in a laboratory setting. Additionally, we measured the SSC at each individual section to confirm our estimations of the confluence's suspended load transport rates. In order to quantify daily Qs from Equation (4), it is necessary to calculate annual riverbed height as well as additional daily variables like flow velocity, river width, and cross-sectional area at the three branch channels. The estimations of all these limits were susceptible to potential errors. This is a significant source of uncertainty in the estimations of suspended load in this study. The uncertainty in the average flow rate is governed by errors in the cross-sectional area. The estimation of river width using the earth explorer and arc map tools could lead to inaccuracy in the estimate of cross-sectional area. Additionally, because the stream was assumed to be a rectangle and the cross-sectional areas were calculated as a function of flow depth and channel breadth, the cross-sectional areas at all positions were possibly exaggerated. This study was unable to discover a way to calculate the cross-sectional area such that it perfectly matched the cross-section of the semi-oval riverbed. Although there are certain ambiguities in the study, the estimations of suspended load rates are reasonable.
CONCLUSIONS
An instant joining of two channels with separate flow and sediment discharge at a river confluence results in unusual hydro-sediment and morphological conditions. The hydrological conditions of the year and the mixed homogenized area determine the surface water flow and sediment mixing of two distinct upstream channels along the downstream channel. In the current study, two approaches (conventional and empirical) were used to relate sediment concentration and water discharge for better sediment yield estimation. Field surveys were conducted every 2 months intervals through February 2021 and 2022 to investigate the spatiotemporal variations of suspended sediments (suspended sediment transport rate) of a river confluence between the Gumti River and the outflow channel of Pitragang tributary. Bed material and SSC samples were gathered from respective chosen points and analyzed to characterize the samples. A statistical study was applied to assess the accuracy of the approach.
The following highlights could be concluded from this study:
Attempts have been made to investigate the SSL dynamicity of confluence according to field data acquired from the confluence site in order to foresee the impact of a tributary to generate interruption within the longitudinal pattern of suspended load as a function of spatial tributary watershed attribute. The study confirms that the tributary junction reflects locations where the longitudinal gradient of sediment rate changes at specified locations. The tributary has had a significant influence on changing the local sediment discharge of the main channel depending on the size of their catchment corresponding to the main channel. The amount of suspended loads appeared to rise as river discharge and stage increased upstream. At upstream locations, suspended loads appeared to increase with the increasing local suspended particles, river discharge and river stage. The most important issue to control sediment transfer in the CHZ is found to be the sediment output and runoff volume from tributary catchments. The outcome demonstrated that, in many instances correlation coefficient value R2 > 0.84 , and the estimated and actual suspended sediment discharge matched well. The measured transport rates of suspended sediment ranged from 642 to 442 g/s, which is in contrast to what Yang's formula computed. This work suggests that a large alluvial river can convey sediment with significant spatiotemporal variability, and that forthcoming sediment management in the stream should discover ways to release coarse and fine sediments held in the examined river reach. The results suggest that Yang's formula can be successfully used for the determination of the total sediment concentration in the Gumti River and Pitraganga tributary. Therefore, it is hoped that the work presented here would provide a valuable knowledge basis and inspire creative or innovative ideas for fruitful and cogent future research, particularly in the area of river confluences. From there, future efforts could be focused on managing and restoring the confluences.
ACKNOWLEDGEMENTS
This research was financially supported by the Science and Engineering Research Board (SERB), Ministry of Science and Technology, Government of India (Grant no. EMR/2016/005371).
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.