https://orcid.org/0009-0001-5327-6591
ABSTRACT
In recent years, the significant change in the runoff–sediment distribution in the upper Yangtze River has led to an increased sediment contribution from the Minjiang River Basin (MRB) to the Three Gorges Reservoir. However, previous studies on sediment load changes in the MRB have focused mainly on annual-scale characteristics, thereby neglecting features driven by floods. Therefore, this study focused on examining the changes in flood-event sediment loads in the MRB based on mathematical statistics and comprehensive measurement data. The results indicated that human activities, climate change, and seismic events have caused an increasing trend in the flood-scale sediment load in the upper MRB and a decreasing trend in the lower MRB. The changes in the flood-scale sediment modulus at low runoff erosion power levels were greater than those at high runoff erosion power levels at different abrupt-change stages. During extreme flood events, the actual sediment concentration in the MRB remained below the sediment-carrying capacity. This study provides novel insights into water resource management during the flood season in the MRB and similar basins.
HIGHLIGHTS
Changes in the sediment load at the flood-event scale show opposite trends upstream and downstream of the Minjiang River Basin (MRB).
After an abrupt change, the variation in sediment transport is smaller under high-flow conditions than under low-flow conditions.
Under extreme flood conditions, the actual sediment concentration in the MRB does not reach the sediment-carrying capacity.
INTRODUCTION
River systems play a crucial role in the global hydrological cycle, and the sediment load of rivers is vital for geochemical cycles and material transport processes (Syvitski et al. 2022; Kong et al. 2023). However, climate change and human activities, such as deforestation, urbanization, and mining, have led to significant changes in the characteristics of sediment transport in global river networks (Dethier et al. 2022; Ostad-Ali-Askari 2022; Yunus et al. 2022). Changes in river sediment transport characteristics can affect the normal operation of water conservancy projects, increase flood control risks, and impact river biodiversity, water quality, and carbon cycling processes (Li et al. 2021; Talebmorad & Ostad-Ali-Askari 2022; Yin et al. 2023). Therefore, exploring the sediment transport characteristics of river systems is essential for watershed conservation and the comprehensive utilization of water resources.
Many studies have focused on investigating the characteristics of sediment transport in various rivers from different perspectives, and particular attention has been given to changes in river sediment transport properties and their influencing factors (Huang et al. 2020; Li et al. 2021; Goldberg et al. 2021; Biggs et al. 2022; Shin et al. 2023; Yin et al. 2023; Ayeni & Royall 2024). Yin et al. (2023) analyzed the sediment transport characteristics of six rivers in eastern China from the 1950s to 2020 and indicated a decreasing trend in the sediment load across all rivers, and this decrease was caused primarily by human activities. In contrast, Li et al. (2021) discovered an increasing trend in the sediment load of rivers in Asian high mountain areas, which was influenced mainly by climate change. Research has provided a wealth of information on the changing trends of long-term sediment transport in rivers, but most studies are based on annual-scale data. In rainstorm-driven rivers, sediment transport primarily occurs during the flood season and can even be concentrated during a few flood events (Lyu et al. 2020; Serra et al. 2022; Tuset et al. 2022). Analyzing annual-scale sediment transport data can reduce high-intensity sediment transport phenomena during flood periods and mask sediment transport characteristics during flood events. As research has progressed, studies have focused on the characteristics of flood-scale sediment transport in basins (Diaf et al. 2020; Tian et al. 2022; Tuset et al. 2022). Diaf et al. (2020) analyzed sediment transport characteristics during 11 floods in the Mekerra region of Algeria between 1980 and 2012 and found that the flood-scale sediment transport characteristics in semiarid areas were greatly influenced by the spatial distribution of precipitation, with similar probabilities for both clockwise and counterclockwise patterns. Tuset et al. (2022) analyzed the sediment transport characteristics of floods in Mediterranean mountainous areas from 2005 to 2008 and reported that the counterclockwise pattern was the most common hysteresis pattern. Tian et al. (2022) analyzed 330 flood events in the Gushanchuan Basin of the middle reaches of the Yellow River from 1955 to 2018 and indicated a highly positive correlation between the sediment load during flood events and runoff with respect to the proportions of complex, counterclockwise, and figure-eight patterns exceeding 90%.
This study focused on the Minjiang River Basin (MRB) in China, which is a major sediment-producing area in the upper reaches of the Yangtze River (YZR). The MR is a typical rainstorm-driven river. The MRB is the primary location for constructing cascade reservoirs in the upper reaches of the YZR. Additionally, owing to its unique geological structure, the MRB frequently experiences events such as landslides and avalanches. Moreover, the upper reaches of the MR are located in the Longmenshan fault zone, which experienced the Wenchuan earthquake in 2008. Therefore, the sediment transport characteristics of the MRB are complex. Currently, most related studies of the MR are based on the annual scale. For example, Wu et al. (2020) found that the annual sediment load in the MRB decreased from 1954 to 2015 and that reservoir construction was the main influencing factor. Wang et al. (2023) analyzed sediment transport data from 1970 to 2019 and discovered a significant decreasing trend in the annual sediment load in the MRB, with human activities contributing over 93% to the total sediment load. Zhang et al. (2024) found a significant reduction in sediment transport in the MRB from 1956 to 2021, which was attributed mainly to human activities, especially the construction of large reservoirs within the basin. With respect to individual flood events, Liu et al. (2023) preliminarily analyzed sediment transport characteristics during flood events at the Gaochang (GC) hydrological station at the outlet of the MR. However, the MRB encompasses various complex climates and underlying surface conditions, with significant differences in runoff and sediment production characteristics among various regions. Analyzing only data from the basin outlet does not provide a comprehensive understanding of the sediment transport characteristics of the entire basin. Since the construction of cascade reservoirs in the lower reaches of the Jinsha River in 2013, the contribution of the MRB to the Three Gorges Reservoir (TGR) has increased. Therefore, to optimize TGR operation, the flood-event sediment transport characteristics of the MR must be examined urgently.
This paper aimed to provide a comprehensive analysis of sediment transport characteristics at the flood-event scale in the MR. This analysis was based on measured data from representative hydrological stations, i.e., the Zhenjiangguan (ZJG) and GC stations. First, the trends and abrupt changes in the flood-event sediment load were identified. Then, flood hysteresis loops and runoff erosion power were employed to study the changes in the sediment load at different abrupt-change stages. Finally, the impacts of climate change, human activities, and seismic events on the changes in the sediment transport characteristics of the basin were analyzed.
STUDY AREA
The YZR originates from the Tibetan Plateau, flows into the East China Sea, spans a distance of approximately 6,300 km, and exhibits a basin covering approximately 1.8 × 106 km2. It is one of the world's largest rivers in terms of runoff and sediment load, with an average yearly runoff of 878.2 × 109 m3 and an average annual sediment load of 134 × 106 t (Zhou et al. 2020). Upstream of the YZR, the TGR is the world's largest reservoir in terms of scale and installed capacity.
(a) Location of the MRB and distribution of the major rivers and lakes in the YRB. (b) Distribution of the basic meteorological stations, hydrological stations, and large reservoirs in the MRB. (c) Temporal variations in the proportion of the sediment yield in the MRB (yellow bar) to the sediment inflow to the TGR.
(a) Location of the MRB and distribution of the major rivers and lakes in the YRB. (b) Distribution of the basic meteorological stations, hydrological stations, and large reservoirs in the MRB. (c) Temporal variations in the proportion of the sediment yield in the MRB (yellow bar) to the sediment inflow to the TGR.
The MRB is located in the monsoon zone and is influenced by the southwest and southeast monsoons. Precipitation events are concentrated during the flood season (May–October), with the seasonal rainfall accounting for 80–85% of the annual total amount (Zhai et al. 2022). In recent years, the MR has played an increasingly significant role in determining the sediment load inflow to the TGR. Since 1955, the contribution of the annual sediment yield from the MRB to the TGR has increased from 10 to 40% by 2020, becoming the primary source of the sediment load of the TGR (Figure 1(c)).
This study focused primarily on the main stem of the MR, especially at two hydrological stations (ZJG and GC). The sediment transport dynamics at ZJG reflect the sediment yield conditions in the upper reaches, whereas the sediment transport characteristics at GC reflect the sediment yield dynamics in the lower reaches. Additionally, ZJG is located within the area affected by the Wenchuan earthquake, and this area is crucial for analyzing the impact of earthquakes on the sediment yield. Analyzing these two stations provides a better understanding of the sediment transport characteristics of the MRB at the flood-event scale.
MATERIALS AND METHODS
Data sources
In this study, daily discharge and suspended sediment concentration (SSC) data at ZJG and GC from 1957 to 2020 were provided by the Yangtze River Water Commission and the Hydrological Bureau of Sichuan Province. Daily precipitation data for the 1957–2020 period from National Meteorological Stations were retrieved from the China Meteorological Data Network (http://data.cma.cn/). Additionally, the analysis incorporated normalized difference vegetation index (NDVI) data with a spatial resolution of 8 × 8 km and a temporal resolution of 15 days for 1982–2020, as published by Li et al. (2023). Figure 1(b) shows the geographical distribution of the major hydrological and meteorological stations and the large reservoirs within the basin.
Methods
Flowchart of this study.
Mathematical statistics method
In this study, the mathematical statistics method was used to analyze the trend and mutation of the annual flood-scale runoff and sediment load at ZJG and GC. The Mann–Kendall method (M–K test) was employed to examine the trend, and the cumulative anomaly method was adopted to assess abrupt-change points.
(1) M–K test
For a given significance level α, if 
, the sequence X exhibits a significant trend at the confidence level α. Conversely, the trend is considered statistically insignificant. Z > 0 denotes an upward trend in the sequence, whereas Z < 0 indicates a downward trend.
(2) Cumulative anomaly method
is the mean value of the time series X.Cumulative anomaly curves were generated from the cumulative anomaly values across various time intervals (Oliveira & da Silva Quaresma 2017; Zhou et al. 2020). The identification of the turning points of these curves allowed for the preliminary determination of abrupt changes within the data. Further validation of the identified abrupt points requires further significance tests of the obtained results.
Runoff–sediment relationship analysis
In this study, the runoff–sediment relationship analysis method was employed to analyze the relationship between runoff and sediment yield, as well as the asynchronous characteristics of flood and sediment peaks during a given flood event.
(1) Runoff–sediment relationship curves
is the flood peak modulus of the flood event, m3/(s·km2); and H is the runoff depth of the flood event, mm.
Previous studies have indicated that the runoff erosion power provides a more accurate depiction of the impacts of natural rainfall and surface conditions on flood erosion and the sediment yield than metrics such as the rainfall erosion power and runoff depth (Liu et al. 2023; Jiang et al. 2024; Yang et al. 2024). In this study, the relationship between the runoff erosion power and the sediment transport modulus was investigated to determine the runoff–sediment relationship in the MRB.
(2) Hysteresis analysis
The relationship between the flow discharge (Q) and the SSC during a given flood event is usually shown as a loop (also referred to as the SSC–Q hysteresis loop; Williams 1989). Hysteresis loops showing the relationship between the SSC and Q during a flood event provide insights into the asynchronous characteristics of flood and sediment peaks (Khosravi et al. 2019; Guo et al. 2024; Ranjan & Roshni 2024). Different hysteresis loops indicate disparities in sediment origins and precipitation areas during flood events. Williams (1989) categorized hysteresis curves into distinct modes based on variations in the SSC–Q relationship. Typical modes include clockwise, counterclockwise, figure-eight, and complex loops, with their criteria provided in Table 1.
Criteria of various hysteresis modesa (Williams 1989)
| Hysteresis modes | Criteria |
|---|---|
| Clockwise loops | (SSC/Q)r > (SSC/Q)f, for all values of Q |
| Counterclockwise loops | (SSC/Q)r < (SSC/Q)f, for all values of Q |
| Figure-eight loops | (SSC/Q)r > (SSC/Q)f, for one range of Q values |
| (SSC/Q)r < (SSC/Q)f, for another range of Q values | |
| Complex loops | Contains at least two or more modes |
| Hysteresis modes | Criteria |
|---|---|
| Clockwise loops | (SSC/Q)r > (SSC/Q)f, for all values of Q |
| Counterclockwise loops | (SSC/Q)r < (SSC/Q)f, for all values of Q |
| Figure-eight loops | (SSC/Q)r > (SSC/Q)f, for one range of Q values |
| (SSC/Q)r < (SSC/Q)f, for another range of Q values | |
| Complex loops | Contains at least two or more modes |
a(SSC/Q)r denotes the rising limb of a flood event. (SSC/Q)f denotes the falling limb of a flood event.
RESULTS
Trends in the annual flood-event runoff and sediment load
The annual flood-event runoff and sediment load trends were initially investigated to characterize sediment transport at the flood-event scale in the MRB. A total of 441 flood events at ZJG and 422 flood events at GC were identified on the basis of daily runoff data from 1957 to 2020, following the methods proposed by Liu et al. (2023) and Zhang et al. (2023).
Temporal variations in the annual flood runoff and sediment transport in the MRB: (a) ZJG and (b) GC.
Temporal variations in the annual flood runoff and sediment transport in the MRB: (a) ZJG and (b) GC.
Table 2 provides the results of the M–K test for the annual flood runoff and sediment load. At GC, a significant decreasing trend was observed in the annual sediment load at the flood-event scale, with a confidence level of α = 95%. In contrast, the increasing trend in the annual sediment load at the flood-event scale at ZJG was not statistically significant.
Results from the trends and abrupt changes in the annual flood-event runoff and sediment load
| Station name | Z-value of the M–K test | Abrupt-change years of runoff | Abrupt-change years of the sediment load | ||
|---|---|---|---|---|---|
| Time series | Runoff | Sediment load | |||
| ZJG | 1957–2020 | 0.75 | 1.43 | 1995, 2009 | 1973, 2012a |
| GC | 1957–2020 | −2.05 | −3.59 | 1969, 1994 | 1995, 2006a |
| Station name | Z-value of the M–K test | Abrupt-change years of runoff | Abrupt-change years of the sediment load | ||
|---|---|---|---|---|---|
| Time series | Runoff | Sediment load | |||
| ZJG | 1957–2020 | 0.75 | 1.43 | 1995, 2009 | 1973, 2012a |
| GC | 1957–2020 | −2.05 | −3.59 | 1969, 1994 | 1995, 2006a |
aThis change point passed the significance test.
Abrupt changes in the annual flood-event runoff and sediment load
The cumulative anomaly curve method was employed to analyze the abrupt changes in the annual runoff and sediment load at the flood-event scale. Table 2 provides a summary of the results regarding the abrupt-change points of the annual flood runoff and sediment load. The abrupt-change year for the annual sediment load was identified as 2012 at ZJG and 2006 at GC.
The annual sediment load at ZJG increased from 0.50 × 106 to 0.92 × 106 t after the observed change, an increase of 81.89%. In contrast, the annual sediment load at GC decreased from 47.75 × 106 to 23.21 × 106 t after the abrupt change, a decrease of 51.39%.
Changes in the W–E relationship at the flood-event scale
The runoff–sediment relationship was analyzed, which is defined as the relationship between the runoff erosion power (E) and the sediment transport modulus (W) of each flood event. The flood sequence at each station was divided into two stages on the basis of the mutation year, and the change in the runoff–sediment relationship at the flood-event scale was examined.
Relationships between the runoff erosion power (E) and the sediment load modulus (W) at the flood-event scale: (a) ZJG and (b) GC.
Relationships between the runoff erosion power (E) and the sediment load modulus (W) at the flood-event scale: (a) ZJG and (b) GC.
A comparison of the runoff–sediment relationship curves before and after the abrupt change at the two stations revealed that coefficient a at ZJG increased and that exponent b decreased. In contrast, those indices at GC exhibited opposite trends. These results indicate that the current sediment supply above ZJG increased, whereas the overall sediment supply in the MRB decreased. The change in exponent b indicated that the sediment-carrying capacity of the river above ZJG was reduced, whereas it increased in the MRB.
As shown in Figure 4, when E was low, a difference could be observed in the sediment load at each station before and after the abrupt change. However, when E was high, the sediment load converged. This occurred because at low E levels, the sediment load is influenced primarily by the sediment supply, whereas at high E levels, it is limited by the sediment-carrying capacity.
Changes in the hysteresis patterns at the flood-event scale
SSC–Q hysteresis patterns for typical flood-event processes. (a) Counterclockwise pattern; (b) clockwise pattern; (c) figure-eight pattern; and (d) complex loop pattern.
SSC–Q hysteresis patterns for typical flood-event processes. (a) Counterclockwise pattern; (b) clockwise pattern; (c) figure-eight pattern; and (d) complex loop pattern.
Figure 5(c) shows a figure-eight pattern, in which the SSC increases rapidly with Q at the high stage of the flood but decreases slowly during the falling limb. These results suggest that significant sediment is stored in the river channel before the flood, and new sediment sources, such as bank failure, emerge during the flood. Figure 5(d) illustrates a complex hysteresis loop, which is a combination of various flood types over extended durations. Multiple flood and sediment peaks may occur during a single flood event, and the SSC is influenced by the superposition of these events, resulting in a complex hysteresis curve.
The SSC–Q relationships of flood events at ZJG and GC were analyzed on the basis of their hysteresis patterns. These patterns were categorized into four types: clockwise, counterclockwise, figure-eight, and complex loops. Table 3 provides the distributions of the flood events with different hysteresis patterns at the two stations before and after the mutation year. The dominant hysteresis pattern upstream of ZJG was consistently the clockwise pattern, and the proportion increased from 66.84 to 75.41%. Conversely, the dominant mode at GC shifted from the clockwise pattern, which accounted for 40.06% of the overall pre-change mode, to the counterclockwise pattern, representing 44.76% of the overall post-change mode.
Number of flood events divided into different hysteresis modes at the different stations
| PeriodsPatterns | GC | ZJG | ||
|---|---|---|---|---|
| 1957–2006 | 2007–2020 | 1957–2012 | 2013–2020 | |
| Clockwise | 127 | 27 | 254 | 46 |
| Counterclockwise | 76 | 47 | 7 | 2 |
| Figure-eight | 63 | 19 | 26 | 2 |
| Complex loop | 51 | 12 | 93 | 11 |
| PeriodsPatterns | GC | ZJG | ||
|---|---|---|---|---|
| 1957–2006 | 2007–2020 | 1957–2012 | 2013–2020 | |
| Clockwise | 127 | 27 | 254 | 46 |
| Counterclockwise | 76 | 47 | 7 | 2 |
| Figure-eight | 63 | 19 | 26 | 2 |
| Complex loop | 51 | 12 | 93 | 11 |
The use of hysteresis modes to determine sediment sources is a common practice, as shown by Fortesa et al. (2020) and Yu et al. (2023). The prevalence of the clockwise pattern at ZJG increased further after 2012, which is consistent with the above findings regarding heightened sediment transport during flood events at ZJG. Thus, the sediment at ZJG primarily originated from ravines or rivers nearby. Conversely, the predominant flood hysteresis pattern at GC shifted from the clockwise pattern to the counterclockwise pattern after 2006, indicating a change in the sediment source from downstream of the MR to a more distant upstream area.
DISCUSSION
Reasons for the changes in the sediment transport characteristics of the MRB
The upper and lower MRB experienced abrupt changes in runoff–sediment relationships and sediment sources at the flood-event scale in 2012 and 2006, respectively. The hysteresis patterns and sediment load per runoff erosion power also differed before and after the changes. On the basis of research on this basin and other similar basins (Li et al. 2020; Wu et al. 2020; Wang & Sun 2021), the changes in the sediment transport characteristics of the MRB were analyzed from three aspects: climate change, human activities, and seismic events.
Impact of climate change
(a) Spatial distribution of multiyear average precipitation in the MRB. (b) Temporal variations in the annual precipitation in the MRB.
(a) Spatial distribution of multiyear average precipitation in the MRB. (b) Temporal variations in the annual precipitation in the MRB.
Impact of human activities
(a) Spatial distribution of the NDVI in the MRB, corresponding to 1982 and 2020. (b) Relationships between the annual sediment load and the NDVI in the MRB.
(a) Spatial distribution of the NDVI in the MRB, corresponding to 1982 and 2020. (b) Relationships between the annual sediment load and the NDVI in the MRB.
As shown in Figure 7(b), a negative correlation was observed between the annual sediment yield and the annual NDVI. This finding highlights the significant role of increased vegetation coverage due to soil and water conservation efforts in reducing the sediment yield of the basin, which is consistent with the findings of Zhou et al. (2020) and Wang et al. (2023).
Cumulative reservoir capacity and annual sediment load in the MRB from 1957 to 2020.
Cumulative reservoir capacity and annual sediment load in the MRB from 1957 to 2020.
Impact of seismic events
(a) Distribution of the major faults and geologic hazards in the MRB. (b) Distribution of the intensities of three major earthquakes of magnitude VI and above.
(a) Distribution of the major faults and geologic hazards in the MRB. (b) Distribution of the intensities of three major earthquakes of magnitude VI and above.
Relationships between sediment concentration (SC) and sediment-carrying capacity (SCC) during typical flood events in 2020.
Relationships between sediment concentration (SC) and sediment-carrying capacity (SCC) during typical flood events in 2020.
As shown in Figure 10, under extreme flood conditions, the sediment concentration only reached 85% of the sediment-carrying capacity, indicating that the MR has yet to reach a saturated sediment transport state. Thus, the sediment concentration in the MRB could further increase. On the one hand, sedimentation is a critical issue affecting the service life of reservoirs, and currently, the construction of cascade reservoirs is underway in the MRB. On the other hand, navigation channel regulation projects have also been constructed in the MRB (Deng & Ge 2020), and the unsaturated sediment transport state may induce phenomena such as bank failure that can affect navigation. Therefore, in the future, sediment monitoring in the MRB should be strengthened.
CONCLUSION
In this study, runoff and sediment data at ZJG in the upstream MRB and GC in the downstream MRB were analyzed to investigate the sediment transport characteristics at the flood-event scale. The specific conclusions are as follows:
(1) Significant differences were observed in the sediment transport characteristics during flood events in the MRB from 1957 to 2020, with an increasing trend upstream and a decreasing trend downstream. The annual flood sediment loads exhibited abrupt changes in 2012 and 2006 in these two areas.
(2) Differences in the sediment load corresponding to the runoff erosion power were observed during different periods, and the variation in flood-scale sediment load under high runoff erosion power levels was smaller than that under low runoff erosion power levels.
(3) Influenced by the sediment supply sources, the proportion of clockwise hysteresis patterns at ZJG increased from 66.84 to 75.41% after 2012, whereas the hysteresis pattern at GC shifted from a predominantly clockwise pattern (40.06%) to a predominantly counterclockwise pattern (44.76%) after 2006.
(4) Affected by the Wenchuan earthquake, although the sediment load during flood events increased in the upper MRB, the sediment concentration has not yet reached the sediment-carrying capacity. A further increase in the sediment concentration upstream could occur, and enhanced sediment monitoring is needed.
(5) In this study, the equation of Zhang Ruijin was used to calculate the sediment-carrying capacity, and the recommended parameter values for the upper reaches of the YRB were directly employed. In the future, methods that are more suitable for calculating the sediment-carrying capacity in the MRB should be explored further.
AUTHOR CONTRIBUTION
Z.W. conceptualized (lead) the work, performed data curation (lead), investigated (equal) the study, carried out methodology (lead); wrote the original draft (lead), and reviewed and edited (lead) the article. D.L. conducted funding acquisition (lead), supervised (lead) the work, and wrote, reviewed, and edited (equal) the work. S.L. conceptualized (supported) the work; performed data curation (equal), investigated (equal) the work; found resources (equal), and validated (equal), wrote, reviewed, and edited (equal) the work. J.H. supported methodology (supporting), found resources (equal), and wrote, reviewed, and edited (equal) the article. R.X. performed data curation (equal) and wrote, reviewed, and edited the article (equal).
ACKNOWLEDGEMENTS
The work was financially supported by the National Natural Science Foundation of China (U2243241).
DATA AVAILABILITY STATEMENT
Data cannot be made publicly available; readers should contact the corresponding author for details.
CONFLICT OF INTEREST
The authors declare there is no conflict.
